Journal 5

krgzbtl

AppliedBehaviorAnalysisCooper.pdf

Home >Psychology homework help >Journal 5

9 781292 023212

ISBN 978-1-29202-321-2

Applied Behavior Analysis Cooper Heron Heward

Second Edition

A pplied B

ehavior A nalysis Cooper H

eron H ew

ard Second E dition

Applied Behavior Analysis Cooper Heron Heward

Second Edition

Pearson Education Limited Edinburgh Gate Harlow Essex CM20 2JE England and Associated Companies throughout the world

Visit us on the World Wide Web at: www.pearsoned.co.uk

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without either the prior written permission of the publisher or a licence permitting restricted copying in the United Kingdom issued by the Copyright Licensing Agency Ltd, Saffron House, 6–10 Kirby Street, London EC1N 8TS.

All trademarks used herein are the property of their respective owners. The use of any trademark in this text does not vest in the author or publisher any trademark ownership rights in such trademarks, nor does the use of such trademarks imply any affi liation with or endorsement of this book by such owners.

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library

Printed in the United States of America

ISBN 10: 1-292-02321-X ISBN 13: 978-1-292-02321-2

Table of Contents

P E A R S O N C U S T O M L I B R A R Y

Glossary

1John O. Cooper, Timothy E. Heron, William L. Heward

1. Definition and Characteristics of Applied Behavior Analysis

22John O. Cooper, Timothy E. Heron, William L. Heward

2. Basic Concepts

44John O. Cooper, Timothy E. Heron, William L. Heward

3. Selecting and Defining Target Behaviors

68John O. Cooper, Timothy E. Heron, William L. Heward

4. Measuring Behavior

92John O. Cooper, Timothy E. Heron, William L. Heward

5. Improving and Assessing the Quality of Behavioral Measurement

122

122John O. Cooper, Timothy E. Heron, William L. Heward

6. Constructing and Interpreting Graphic Displays of Behavioral Data

146

146John O. Cooper, Timothy E. Heron, William L. Heward

7. Analyzing Behavior Change: Basic Assumptions and Strategies

178

178John O. Cooper, Timothy E. Heron, William L. Heward

8. Reversal and Alternating Treatments Designs

196

196John O. Cooper, Timothy E. Heron, William L. Heward

9. Multiple Baseline and Changing Criterion Designs

220

220John O. Cooper, Timothy E. Heron, William L. Heward

10. Planning and Evaluating Applied Behavior Analysis Research

245

245John O. Cooper, Timothy E. Heron, William L. Heward

11. Positive Reinforcement

276

276John O. Cooper, Timothy E. Heron, William L. Heward

12. Negative Reinforcement

311

311John O. Cooper, Timothy E. Heron, William L. Heward

13. Schedules of Reinforcement

324

324John O. Cooper, Timothy E. Heron, William L. Heward

14. Punishment by Stimulus Presentation

344

344John O. Cooper, Timothy E. Heron, William L. Heward

15. Punishment by Removal of a Stimulus

374

374John O. Cooper, Timothy E. Heron, William L. Heward

16. Motivating Operations

390

390John O. Cooper, Timothy E. Heron, William L. Heward

17. Stimulus Control

408

408John O. Cooper, Timothy E. Heron, William L. Heward

18. Imitation

426

426John O. Cooper, Timothy E. Heron, William L. Heward

19. Chaining

434

434John O. Cooper, Timothy E. Heron, William L. Heward

20. Shaping

454

454John O. Cooper, Timothy E. Heron, William L. Heward

21. Extinction

468

468John O. Cooper, Timothy E. Heron, William L. Heward

22. Differential Reinforcement

481

481John O. Cooper, Timothy E. Heron, William L. Heward

23. Antecedent Interventions

498

498John O. Cooper, Timothy E. Heron, William L. Heward

24. Functional Behavior Assessment

510

510John O. Cooper, Timothy E. Heron, William L. Heward

25. Verbal Behavior

536

536John O. Cooper, Timothy E. Heron, William L. Heward

26. Contingency Contracting, Token Economy, and Group Contingencies

558

558John O. Cooper, Timothy E. Heron, William L. Heward

27. Self-Management

583

583John O. Cooper, Timothy E. Heron, William L. Heward

28. Generalization and Maintenance of Behavior Change

622

622John O. Cooper, Timothy E. Heron, William L. Heward

29. Ethical Considerations for Applied Behavior Analysts

664

664John O. Cooper, Timothy E. Heron, William L. Heward

Bibliography

685

685John O. Cooper, Timothy E. Heron, William L. Heward

III

729

729Index

A-B design A two-phase experimental design consisting of a pre-treatment baseline condition (A) followed by a treatment condition (B).

A-B-A design A three-phase experimental design consisting of an initial baseline phase (A) until steady state respond- ing (or countertherapeutic trend) is obtained, an interven- tion phase in which the treatment condition (B) is implemented until the behavior has changed and steady state responding is obtained, and a return to baseline con- ditions (A) by withdrawing the independent variable to see whether responding “reverses” to levels observed in the initial baseline phase. (See A-B-A-B design, reversal design, withdrawal design.)

A-B-A-B design An experimental design consisting of (1) an initial baseline phase (A) until steady state responding (or countertherapeutic trend) is obtained, (2) an initial inter- vention phase in which the treatment variable (B) is im- plemented until the behavior has changed and steady state responding is obtained, (3) a return to baseline conditions (A) by withdrawing the independent variable to see whether responding “reverses” to levels observed in the initial baseline phase, and (4) a second intervention phase (B) to see whether initial treatment effects are replicated (also called reversal design, withdrawal design).

abative effect (of a motivating operation) A decrease in the current frequency of behavior that has been reinforced by the stimulus that is increased in reinforcing effectiveness by the same motivating operation. For example, food in- gestion abates (decreases the current frequency of) be- havior that has been reinforced by food.

ABC recording See anecdotal observation. abolishing operation (AO) A motivating operation that de-

creases the reinforcing effectiveness of a stimulus, object, or event. For example, the reinforcing effectiveness of food is abolished as a result of food ingestion.

accuracy (of measurement) The extent to which observed values, the data produced by measuring an event, match the true state, or true values, of the event as it exists in nature. (See observed value and true value.)

adjunctive behavior Behavior that occurs as a collateral ef- fect of a schedule of periodic reinforcement for other be-

havior; time-filling or interim activities (e.g., doodling, idle talking, smoking, drinking) that are induced by sched- ules of reinforcement during times when reinforcement is unlikely to be delivered. Also called schedule-induced behavior.

affirmation of the consequent A three-step form of reason- ing that begins with a true antecedent–consequent (if-A- then-B) statement and proceeds as follows: (1) If A is true, then B is true; (2) B is found to be true; (3) therefore, A is true. Although other factors could be responsible for the truthfulness of A, a sound experiment affirms several if-A- then-B possibilities, each one reducing the likelihood of factors other than the independent variable being respon- sible for the observed changes in behavior.

alternating treatments design An experimental design in which two or more conditions (one of which may be a no- treatment control condition) are presented in rapidly alter- nating succession (e.g., on alternating sessions or days) independent of the level of responding; differences in re- sponding between or among conditions are attributed to the effects of the conditions (also called concurrent schedule design, multielement design, multiple schedule design).

alternative schedule Provides reinforcement whenever the requirement of either a ratio schedule or an interval sched- ule—the basic schedules that makeup the alternative schedule—is met, regardless of which of the component schedule’s requirements is met first.

anecdotal observation A form of direct, continuous obser- vation in which the observer records a descriptive, tem- porally sequenced account of all behavior(s) of interest and the antecedent conditions and consequences for those behaviors as those events occur in the client’s natural en- vironment (also called ABC recording).

antecedent An environmental condition or stimulus change existing or occurring prior to a behavior of interest.

antecedent intervention A behavior change strategy that manipulates contingency-independent antecedent stimuli (motivating operations). (See noncontingent reinforce- ment, high-probability request sequence, and functional communication training. Contrast with antecedent con- trol, a behavior change intervention that manipulates

Glossary

contingency-dependent consequence events to affect stim- ulus control.)

antecedent stimulus class A set of stimuli that share a com- mon relationship. All stimuli in an antecedent stimulus class evoke the same operant behavior, or elicit the same respondent behavior. (See arbitrary stimulus class, fea- ture stimulus class.)

applied behavior analysis (ABA) The science in which tac- tics derived from the principles of behavior are applied to improve socially significant behavior and experimentaton is used to identify the variables responsible for the im- provement in behavior.

arbitrary stimulus class Antecedent stimuli that evoke the same response but do not resemble each other in physical form or share a relational aspect such as bigger or under (e.g., peanuts, cheese, coconut milk, and chicken breasts are members of an arbitrary stimulus class if they evoke the response “sources of protein”). (Compare to feature stim- ulus class.)

artifact An outcome or result that appears to exist because of the way it is measured but in fact does not correspond to what actually occurred.

ascending baseline A data path that shows an increasing trend in the response measure over time. (Compare with descending baseline.)

audience Anyone who functions as a discriminative stimulus evoking verbal behavior. Different audiences may control different verbal behavior about the same topic because of a differential reinforcement history. Teens may describe the same event in different ways when talking to peers ver- sus parents.

autoclitic A secondary verbal operant in which some aspect of a speaker’s own verbal behavior functions as an SD or an MO for additional speaker verbal behavior. The auto- clitic relation can be thought of as verbal behavior about verbal behavior.

automatic punishment Punishment that occurs independent of the social mediation by others (i.e., a response product serves as a punisher independent of the social environ- ment).

automatic reinforcement Reinforcement that occurs inde- pendent of the social mediation of others (e.g., scratching an insect bite relieves the itch).

automaticity (of reinforcement) Refers to the fact that be- havior is modified by its consequences irrespective of the person’s awareness; a person does not have to recognize or verbalize the relation between her behavior and a rein- forcing consequence, or even know that a consequence has occurred, for reinforcement to “work.” (Contrast with automatic reinforcement.)

aversive stimulus In general, an unpleasant or noxious stim- ulus; more technically, a stimulus change or condition that functions (a) to evoke a behavior that has terminated it in the past; (b) as a punisher when presented following be- havior, and/or (c) as a reinforcer when withdrawn follow- ing behavior.

avoidance contingency A contingency in which a response prevents or postpones the presentation of a stimulus. (Compare with escape contingency.)

B-A-B design A three-phase experimental design that begins with the treatment condition. After steady state respond- ing has been obtained during the initial treatment phase (B), the treatment variable is withdrawn (A) to see whether responding changes in the absence of the independent vari- able. The treatment variable is then reintroduced (B) in an attempt to recapture the level of responding obtained dur- ing the first treatment phase.

backup reinforcers Tangible objects, activities, or privileges that serve as reinforcers and that can be purchased with tokens.

backward chaining A teaching procedure in which a trainer completes all but the last behavior in a chain, which is performed by the learner, who then receives re- inforcement for completing the chain. When the learner shows competence in performing the final step in the chain, the trainer performs all but the last two behaviors in the chain, the learner emits the final two steps to com- plete the chain, and reinforcement is delivered. This se- quence is continued until the learner completes the entire chain independently.

backward chaining with leaps ahead A backward chain- ing procedure in which some steps in the task analysis are skipped; used to increase the efficiency of teaching long behavior chains when there is evidence that the skipped steps are in the learner’s repertoire.

bar graph A simple and versatile graphic format for sum- marizing behavioral data; shares most of the line graph’s features except that it does not have distinct data points representing successive response measures through time. Also called a histogram.

baseline A condition of an experiment in which the inde- pendent variable is not present; data obtained during base- line are the basis for determining the effects of the independent variable; a control condition that does not necessarily mean the absence of instruction or treatment, only the absence of a specific independent variable of ex- perimental interest.

baseline logic A term sometimes used to refer to the exper- imental reasoning inherent in single-subject experimental designs; entails three elements: prediction, verification, and replication. (See steady state strategy.)

behavior The activity of living organisms; human behavior includes everything that people do. A technical definition: “that portion of an organism’s interaction with its envi- ronment that is characterized by detectable displacement in space through time of some part of the organism and that results in a measurable change in at least one aspect of the environment” (Johnston & Pennypacker, 1993a, p. 23). (See operant behavior, respondent behavior, re- sponse, response class.)

behavior-altering effect (of a motivating operation) An alteration in the current frequency of behavior that has

Glossary

been reinforced by the stimulus that is altered in effec- tiveness by the same motivating operation. For example, the frequency of behavior that has been reinforced with food is increased or decreased by food deprivation or food ingestion.

behavior chain A sequence of responses in which each re- sponse produces a stimulus change that functions as con- ditioned reinforcement for that response and as a discriminative stimulus for the next response in the chain; reinforcement for the last response in a chain maintains the reinforcing effectiveness of the stimulus changes pro- duced by all previous responses in the chain.

behavior chain interruption strategy An intervention that relies on the participant’s skill in performing the critical el- ements of a chain independently; the chain is interrupted occasionally so that another behavior can be emitted.

behavior chain with a limited hold A contingency that spec- ifies a time interval by which a behavior chain must be completed for reinforcement to be delivered.

behavior change tactic A technologically consistent method for changing behavior derived from one or more princi- ples of behavior (e.g., differential reinforcement of other behavior, response cost); possesses sufficient generality across subjects, settings, and/or behaviors to warrant its codification and dissemination.

behavior checklist A checklist that provides descriptions of specific skills (usually in hierarchical order) and the con- ditions under which each skill should be observed. Some checklists are designed to assess one particular behavior or skill area. Others address multiple behaviors or skill areas. Most use a Likert scale to rate responses.

behavior trap An interrelated community of contingen- cies of reinforcement that can be especially powerful, producing substantial and long-lasting behavior changes. Effective behavior traps share four essential features: (a) They are “baited” with virtually irresistible rein- forcers that “lure” the student to the trap; (b) only a low- effort response already in the student’s repertoire is necessary to enter the trap; (c) once inside the trap, in- terrelated contingencies of reinforcement motivate the student to acquire, extend, and maintain targeted aca- demic and/or social skills; and (d) they can remain ef- fective for a long time because students shows few, if any, satiation effects.

behavioral assessment A form of assessment that involves a full range of inquiry methods (observation, interview, testing, and the systematic manipulation of antecedent or consequence variables) to identify probable antecedent and consequent controlling variables. Behavioral assess- ment is designed to discover resources, assets, significant others, competing contingencies, maintenance and gener- ality factors, and possible reinforcer and/or punishers that surround the potential target behavior.

behavioral contract See contingency contract. behavioral contrast The phenomenon in which a change in

one component of a multiple schedule that increases or

decreases the rate of responding on that component is ac- companied by a change in the response rate in the oppo- site direction on the other, unaltered component of the schedule.

behavioral cusp A behavior that has sudden and dramatic consequences that extend well beyond the idiosyncratic change itself because it exposes the person to new envi- ronments, reinforcers, contingencies, responses, and stim- ulus controls. (See pivotal behavior.)

behavioral momentum A metaphor to describe a rate of re- sponding and its resistance to change following an alter- ation in reinforcement conditions. The momentum metaphor has also been used to describe the effects produced by the high-probability (high-p) request sequence.

behaviorism The philosophy of a science of behavior; there are various forms of behaviorism. (See methodological behaviorism, radical behaviorism.)

believability The extent to which the researcher convinces herself and others that the data are trustworthy and de- serve interpretation. Measures of interobserver agreement (IOA) are the most often used index of believability in ap- plied behavior analysis. (See interobserver agreement (IOA).)

bonus response cost A procedure for implementing response cost in which the person is provided a reservoir of rein- forcers that are removed in predetermined amounts con- tingent on the occurrence of the target behavior.

calibration Any procedure used to evaluate the accuracy of a measurement system and, when sources of error are found, to use that information to correct or improve the measurement system.

celeration The change (acceleration or deceleration) in rate of responding over time; based on count per unit of time (rate); expressed as a factor by which responding is ac- celerating or decelerating (multiplying or dividing); dis- played with a trend line on a Standard Celeration Chart. Celeration is a generic term without specific reference to accelerating or decelerating rates of response. (See Standard Celeration Chart.)

celeration time period A unit of time (e.g., per week, per month) in which celeration is plotted on a Standard Cel- eration Chart. (See celeration and celeration trend line.)

celeration trend line The celeration trend line is measured as a factor by which rate multiplies or divides across the celeration time periods (e.g., rate per week, rate per month, rate per year, and rate per decade). (See celeration.)

chained schedule A schedule of reinforcement in which the response requirements of two or more basic schedules must be met in a specific sequence before reinforcement is delivered; a discriminative stimulus is correlated with each component of the schedule.

chaining Various procedures for teaching behavior chains. (See backward chaining, backward chaining with leaps ahead, behavior chain, forward chaining.)

changing criterion design An experimental design in which an initial baseline phase is followed by a series of treatment

Glossary

phases consisting of successive and gradually changing criteria for reinforcement or punishment. Experimental control is evidenced by the extent the level of responding changes to conform to each new criterion.

clicker training A term popularized by Pryor (1999) for shaping behavior using conditioned reinforcement in the form of an auditory stimulus. A handheld device produces a click sound when pressed. The trainer pairs other forms of reinforcement (e.g., edible treats) with the click sound so that the sound becomes a conditioned reinforcer.

component analysis Any experiment designed to identify the active elements of a treatment condition, the relative contributions of different variables in a treatment pack- age, and/or the necessary and sufficient components of an intervention. Component analyses take many forms, but the basic strategy is to compare levels of responding across successive phases in which the intervention is implemented with one or more components left out.

compound schedule A schedule of reinforcement consist- ing of two or more elements of continuous reinforcement (CRF), the four intermittent schedules of reinforcement (FR, VR, FI, VI), differential reinforcement of various rates of responding (DRH, DRL), and extinction. The el- ements from these basic schedules can occur successively or simultaneously and with or without discriminative stim- uli; reinforcement may be contingent on meeting the re- quirements of each element of the schedule independently or in combination with all elements.

concept formation A complex example of stimulus control that requires stimulus generalization within a class of stim- uli and discrimination between classes of stimuli.

concurrent schedule (conc) A schedule of reinforcement in which two or more contingencies of reinforcement (ele- ments) operate independently and simultaneously for two or more behaviors.

conditional probability The likelihood that a target behav- ior will occur in a given circumstance; computed by cal- culating (a) the proportion of occurrences of behavior that were preceded by a specific antecedent variable and (b) the proportion of occurrences of problem behavior that were followed by a specific consequence. Conditional probabilities range from 0.0 to 1.0; the closer the condi- tional probability is to 1.0, the stronger the relationship is between the target behavior and the antecedent/conse- quence variable.

conditioned motivating operation (CMO) A motivating op- eration whose value-altering effect depends on a learning history. For example, because of the relation between locked doors and keys, having to open a locked door is a CMO that makes keys more effective as reinforcers, and evokes behavior that has obtained such keys.

conditioned negative reinforcer A previously neutral stim- ulus change that functions as a negative reinforcer because of prior pairing with one or more negative reinforcers. (See negative reinforcer; compare with unconditioned neg- ative reinforcer).

conditioned punisher A previously neutral stimulus change that functions as a punisher because of prior pair- ing with one or more other punishers; sometimes called secondary or learned punisher. (Compare with uncon- ditioned punisher.)

conditioned reflex A learned stimulus–response functional relation consisting of an antecedent stimulus (e.g., sound of refrigerator door opening) and the response it elicits (e.g., salivation); each person’s repertoire of conditioned reflexes is the product of his or her history of interactions with the environment (ontogeny). (See respondent con- ditioning, unconditioned reflex.)

conditioned reinforcer A stimulus change that functions as a reinforcer because of prior pairing with one or more other reinforcers; sometimes called secondary or learned reinforcer.

conditioned stimulus (CS) The stimulus component of a con- ditioned reflex; a formerly neutral stimulus change that elicits respondent behavior only after it has been paired with an unconditioned stimulus (US) or another CS.

confidentiality Describes a situation of trust insofar as any in- formation regarding a person receiving or having received services may not be discussed with or otherwise made available to another person or group, unless that person has provided explicit authorization for release of such information.

conflict of interest A situation in which a person in a position of responsibility or trust has competing professional or personal interests that make it difficult to fulfill his or her duties impartially.

confounding variable An uncontrolled factor known or sus- pected to exert influence on the dependent variable.

consequence A stimulus change that follows a behavior of in- terest. Some consequences, especially those that are im- mediate and relevant to current motivational states, have significant influence on future behavior; others have little effect. (See punisher, reinforcer.)

contingency Refers to dependent and/or temporal relations between operant behavior and its controlling variables. (See contingent, three-term contingency.)

contingency contract A mutually agreed upon document be- tween parties (e.g., parent and child) that specifies a con- tingent relationship between the completion of specified behavior(s) and access to specified reinforcer(s).

contingency reversal Exchanging the reinforcement contin- gencies for two topographically different responses. For example, if Behavior A results in reinforcement on an FR 1 schedule of reinforcement and Behavior B results in rein- forcement being withheld (extinction), a contingency re- versal consists of changing the contingencies such that Behavior A now results in extinction and Behavior B re- sults in reinforcement on an FR 1 schedule.

contingent Describes reinforcement (or punishment) that is delivered only after the target behavior has occurred.

contingent observation A procedure for implementing time- out in which the person is repositioned within an existing

Glossary

setting such that observation of ongoing activities remains, but access to reinforcement is lost.

continuous measurement Measurement conducted in a man- ner such that all instances of the response class(es) of in- terest are detected during the observation period.

continuous reinforcement (CRF) A schedule of reinforce- ment that provides reinforcement for each occurrence of the target behavior.

contrived contingency Any contingency of reinforcement (or punishment) designed and implemented by a behavior analyst or practitioner to achieve the acquisition, mainte- nance, and/or generalization of a targeted behavior change. (Contrast with naturally existing contingency.)

contrived mediating stimulus Any stimulus made functional for the target behavior in the instructional setting that later prompts or aids the learner in performing the target be- havior in a generalization setting.

copying a text An elementary verbal operant that is evoked by a nonvocal verbal discriminative stimulus that has point-to-point correspondence and formal similarity with the controlling response.

count A simple tally of the number of occurrences of a be- havior. The observation period, or counting time, should always be noted when reporting count measures.

counting time The period of time in which a count of the number of responses emitted was recorded.

cumulative record A type of graph on which the cumula- tive number of responses emitted is represented on the ver- tical axis; the steeper the slope of the data path, the greater the response rate.

cumulative recorder A device that automatically draws cu- mulative records (graphs) that show the rate of response in real time; each time a response is emitted, a pen moves upward across paper that continuously moves at a con- stant speed.

data The results of measurement, usually in quantifiable form; in applied behavior analysis, it refers to measures of some quantifiable dimension of a behavior.

data path The level and trend of behavior between succes- sive data points; created by drawing a straight line from the center of each data point in a given data set to the center of the next data point in the same set.

delayed multiple baseline design A variation of the multi- ple baseline design in which an initial baseline, and per- haps intervention, are begun for one behavior (or setting, or subject), and subsequent baselines for additional be- haviors are begun in a staggered or delayed fashion.

dependent group contingency A contingency in which re- inforcement for all members of a group is dependent on the behavior of one member of the group or the behavior of a select group of members within the larger group.

dependent variable The variable in an experiment measured to determine if it changes as a result of manipulations of the independent variable; in applied behavior analysis, it repre- sents some measure of a socially significant behavior. (See target behavior; compare with independent variable.)

deprivation The state of an organism with respect to how much time has elapsed since it has consumed or contacted a particular type of reinforcer; also refers to a procedure for increasing the effectiveness of a reinforcer (e.g., with- holding a person’s access to a reinforcer for a specified period of time prior to a session). (See motivating oper- ation; contrast with satiation.)

descending baseline A data path that shows a decreasing trend in the response measure over time. (Compare with ascending baseline.)

descriptive functional behavior assessment Direct obser- vation of problem behavior and the antecedent and con- sequent events under naturally occurring conditions.

determinism The assumption that the universe is a lawful and orderly place in which phenomena occur in relation to other events and not in a willy-nilly, accidental fashion.

differential reinforcement Reinforcing only those responses within a response class that meet a specific criterion along some dimension(s) (i.e., frequency, topography, duration, latency, or magnitude) and placing all other responses in the class on extinction. (See differential reinforcement of alternative behavior, differential reinforcement of in- compatible behavior, differential reinforcement of other behavior, discrimination training, shaping.)

differential reinforcement of alternative behavior (DRA) A procedure for decreasing problem behavior in which re- inforcement is delivered for a behavior that serves as a de- sirable alternative to the behavior targeted for reduction and withheld following instances of the problem behavior (e.g., reinforcing completion of academic worksheet items when the behavior targeted for reduction is talk-outs).

differential reinforcement of diminishing rates (DRD) A schedule of reinforcement in which reinforcement is pro- vided at the end of a predetermined interval contingent on the number of responses emitted during the interval being fewer than a gradually decreasing criterion based on the in- dividual’s performance in previous intervals (e.g., fewer than five responses per 5 minutes, fewer than four responses per 5 minutes, fewer than three responses per 5 minutes).

differential reinforcement of high rates (DRH) A sched- ule of reinforcement in which reinforcement is provided at the end of a predetermined interval contingent on the number of responses emitted during the interval being greater than a gradually increasing criterion based on the individual’s performance in previous intervals (e.g., more than three responses per 5 minutes, more than five re- sponses per 5 minutes, more than eight responses per 5 minutes).

differential reinforcement of incompatible behavior (DRI) A procedure for decreasing problem behavior in which re- inforcement is delivered for a behavior that is topograph- ically incompatible with the behavior targeted for reduction and withheld following instances of the prob- lem behavior (e.g., sitting in seat is incompatible with walking around the room).

Glossary

differential reinforcement of low rates (DRL) A schedule of reinforcement in which reinforcement (a) follows each occurrence of the target behavior that is separated from the previous response by a minimum interresponse time (IRT), or (b) is contingent on the number of responses within a period of time not exceeding a predetermined cri- terion. Practitioners use DRL schedules to decrease the rate of behaviors that occur too frequently but should be maintained in the learner’s repertoire. (See full-session DRL, interval DRL, and spaced-responding DRL.)

differential reinforcement of other behavior (DRO) A pro- cedure for decreasing problem behavior in which rein- forcement is contingent on the absence of the problem behavior during or at specific times (i.e., momentary DRO); sometimes called differential reinforcement of zero rates of responding or omission training). (See fixed-interval DRO, fixed-momentary DRO, variable-interval DRO, and variable-momentary DRO.)

direct measurement Occurs when the behavior that is mea- sured is the same as the behavior that is the focus of the in- vestigation. (Contrast with indirect measurement.)

direct replication An experiment in which the researcher attempts to duplicate exactly the conditions of an earlier experiment.

discontinuous measurement Measurement conducted in a manner such that some instances of the response class(es) of interest may not be detected.

discrete trial Any operant whose response rate is controlled by a given opportunity to emit the response. Each discrete re- sponse occurs when an opportunity to respond exists. Discrete trial, restricted operant, and controlled operant are synonymous technical terms. (Contrast with free operant.)

discriminated avoidance A contingency in which respond- ing in the presence of a signal prevents the onset of a stim- ulus from which escape is a reinforcer (See also discriminative stimulus, discriminated operant, free- operant avoidance, and stimulus control.)

discriminated operant An operant that occurs more fre- quently under some antecedent conditions than under oth- ers. (See discriminative stimulus [SD], stimulus control.)

discriminative stimulus (SD) A stimulus in the presence of which responses of some type have been reinforced and in the absence of which the same type of responses have oc- curred and not been reinforced; this history of differential reinforcement is the reason an SD increases the momentary frequency of the behavior. (See differential reinforce- ment, stimulus control, stimulus discrimination train- ing, and stimulus delta [S�].)

double-blind control A procedure that prevents the subject and the observer(s) from detecting the presence or absence of the treatment variable; used to eliminate confounding of results by subject expectations, parent and teacher expec- tations, differential treatment by others, and observer bias. (See placebo control.)

DRI/DRA reversal technique An experimental technique that demonstrates the effects of reinforcement; it uses dif-

ferential reinforcement of an incompatible or alternative behavior (DRI/DRA) as a control condition instead of a no-reinforcement (baseline) condition. During the DRI/ DRA condition, the stimulus change used as reinforce- ment in the reinforcement condition is presented contin- gent on occurrences of a specified behavior that is either incompatible with the target behavior or an alternative to the target behavior. A higher level of responding during the reinforcement condition than during the DRI/DRA condition demonstrates that the changes in behavior are the result of contingent reinforcement, not simply the pre- sentation of or contact with the stimulus event. (Compare with DRO reversal technique and noncontingent rein- forcement (NCR) reversal technique.)

DRO reversal technique An experimental technique for demonstrating the effects of reinforcement by using dif- ferential reinforcement of other behavior (DRO) as a con- trol condition instead of a no-reinforcement (baseline) condition. During the DRO condition, the stimulus change used as reinforcement in the reinforcement con- dition is presented contingent on the absence of the tar- get behavior for a specified time period. A higher level of responding during the reinforcement condition than during the DRO condition demonstrates that the changes in behavior are the result of contingent reinforcement, not simply the presentation of or contact with the stim- ulus event. (Compare with DRI/DRA reversal tech- nique and noncontingent reinforcement (NCR) reversal technique.)

duration A measure of the total extent of time in which a behavior occurs.

echoic An elementary verbal operant involving a response that is evoked by a verbal discriminative stimulus that has point-to-point correspondence and formal similarity with the response.

ecological assessment An assessment protocol that ac- knowledges complex interrelationships between envi- ronment and behavior. An ecological assessment is a method for obtaining data across multiple settings and persons.

empiricism The objective observation of the phenomena of interest; objective observations are “independent of the in- dividual prejudices, tastes, and private opinions of the sci- entist. . . . Results of empirical methods are objective in that they are open to anyone’s observation and do not de- pend on the subjective belief of the individual scientist” (Zuriff, 1985, p. 9).

environment The conglomerate of real circumstances in which the organism or referenced part of the organism ex- ists; behavior cannot occur in the absence of environment.

escape contingency A contingency in which a response ter- minates (produces escape from) an ongoing stimulus. (Compare with avoidance contingency.)

escape extinction Behaviors maintained with negative rein- forcement are placed on escape extinction when those be- haviors are not followed by termination of the aversive

Glossary

stimulus; emitting the target behavior does not enable the person to escape the aversive situation.

establishing operation (EO) A motivating operation that establishes (increases) the effectiveness of some stimu- lus, object, or event as a reinforcer. For example, food de- privation establishes food as an effective reinforcer.

ethical codes of behavior Statements that provide guide- lines for members of professional associations when de- ciding a course of action or conducting professional duties; standards by which graduated sanctions (e.g., rep- rimand, censure, expulsion) can be imposed for deviating from the code.

ethics Behaviors, practices, and decisions that address such basic and fundamental questions as: What is the right thing to do? What’s worth doing? What does it mean to be a good behavior analytic practitioner?

event recording Measurement procedure for obtaining a tally or count of the number of times a behavior occurs.

evocative effect (of a motivating operation) An increase in the current frequency of behavior that has been reinforced by the stimulus that is increased in reinforcing effective- ness by the same motivating operation. For example, food deprivation evokes (increases the current frequency of) be- havior that has been reinforced by food.

exact count-per-interval IOA The percentage of total in- tervals in which two observers recorded the same count; the most stringent description of IOA for most data sets ob- tained by event recording.

exclusion time-out A procedure for implementing time-out in which, contingent on the occurrence of a target behav- ior, the person is removed physically from the current en- vironment for a specified period.

experiment A carefully controlled comparison of some mea- sure of the phenomenon of interest (the dependent vari- able) under two or more different conditions in which only one factor at a time (the independent variable) differs from one condition to another.

experimental analysis of behavior (EAB) A natural science approach to the study of behavior as a subject matter in its own right founded by B. F. Skinner; methodological features include rate of response as a basic dependent vari- able, repeated or continuous measurement of clearly de- fined response classes, within-subject experimental comparisons instead of group design, visual analysis of graphed data instead of statistical inference, and an em- phasis on describing functional relations between behav- ior and controlling variables in the environment over formal theory testing.

experimental control Two meanings: (a) the outcome of an experiment that demonstrates convincingly a functional relation, meaning that experimental control is achieved when a predictable change in behavior (the dependent variable) can be reliably produced by manipulating a spe- cific aspect of the environment (the independent variable); and (b) the extent to which a researcher maintains precise control of the independent variable by presenting it, with-

drawing it, and/or varying its value, and also by eliminat- ing or holding constant all confounding and extraneous variables. (See confounding variable, extraneous vari- able, and independent variable.)

experimental design The particular type and sequence of conditions in a study so that meaningful comparisons of the effects of the presence and absence (or different values) of the independent variable can be made.

experimental question A statement of what the researcher seeks to learn by conducting the experiment; may be pre- sented in question form and is most often found in a pub- lished account as a statement of the experiment’s purpose. All aspects of an experiment’s design should follow from the experimental question (also called the research question).

explanatory fiction A fictitious or hypothetical variable that often takes the form of another name for the observed phe- nomenon it claims to explain and contributes nothing to a functional account or understanding of the phenomenon, such as “intelligence” or “cognitive awareness” as expla- nations for why an organism pushes the lever when the light is on and food is available but does not push the lever when the light is off and no food is available.

external validity The degree to which a study’s findings have generality to other subjects, settings, and/or behaviors. (Compare to internal validity.)

extinction (operant) The discontinuing of a reinforcement of a previously reinforced behavior (i.e., responses no longer produce reinforcement); the primary effect is a decrease in the frequency of the behavior until it reaches a prereinforced level or ultimately ceases to occur. (See extinction burst, spontaneous recovery; compare respondent extinction)

extinction burst An increase in the frequency of responding when an extinction procedure is initially implemented.

extraneous variable Any aspect of the experimental setting (e.g., lighting, temperature) that must be held constant to prevent unplanned environmental variation.

fading A procedure for transferring stimulus control in which features of an antecedent stimulus (e.g., shape, size, position, color) controlling a behavior are gradu- ally changed to a new stimulus while maintaining the current behavior; stimulus features can be faded in (en- hanced) or faded out (reduced).

feature stimulus class Stimuli that share common physical forms or structures (e.g., made from wood, four legs, round, blue) or common relative relationships (e.g., bigger than, hotter than, higher than, next to). (Compare to arbitrary stimulus class.)

fixed interval (FI) A schedule of reinforcement in which re- inforcement is delivered for the first response emitted fol- lowing the passage of a fixed duration of time since the last response was reinforced (e.g., on an FI 3-minute schedule, the first response following the passage of 3 min- utes is reinforced).

fixed-interval DRO (FI-DRO) A DRO procedure in which reinforcement is available at the end of intervals of fixed duration and delivered contingent on the absence of the

Glossary

problem behavior during each interval. (See differential reinforcement of other behavior (DRO).)

fixed-momentary DRO (FM-DRO) A DRO procedure in which reinforcement is available at specific moments of time, which are separated by a fixed amount of time, and delivered contingent on the problem not occurring at those moments. (See differential reinforcement of other be- havior (DRO).)

fixed ratio (FR) A schedule of reinforcement requiring a fixed number of responses for reinforcement (e.g., an FR 4 schedule reinforcement follows every fourth response).

fixed-time schedule (FT) A schedule for the delivery of non- contingent stimuli in which a time interval remains the same from one delivery to the next.

formal similarity A situation that occurs when the control- ling antecedent stimulus and the response or response product (a) share the same sense mode (e.g., both stimu- lus and response are visual, auditory, or tactile) and (b) physically resemble each other. The verbal relations with formal similarity are echoic, coping a text, and imitation as it relates to sign language.

forward chaining A method for teaching behavior chains that begins with the learner being prompted and taught to perform the first behavior in the task analysis; the trainer completes the remaining steps in the chain. When the learner shows competence in performing the first step in the chain, he is then taught to perform the first two be- haviors in the chain, with the training completing the chain. This process is continued until the learner completes the entire chain independently.

free operant Any operant behavior that results in minimal displacement of the participant in time and space. A free operant can be emitted at nearly any time; it is discrete, it requires minimal time for completion, and it can produce a wide range of response rates. Examples in ABA include (a) the number of words read during a 1-minute counting period, (b) the number of hand slaps per 6 seconds, and (c) the number of letter strokes written in 3 minutes. (Con- trast with discrete trial.)

free-operant avoidance A contingency in which responses at any time during an interval prior to the scheduled onset of an aversive stimulus delays the presentation of the aver- sive stimulus. (Contrast with discriminated avoidance.)

frequency A ratio of count per observation time; often ex- pressed as count per standard unit of time (e.g., per minute, per hour, per day) and calculated by dividing the number of responses recorded by the number of standard units of time in which observations were conducted; used inter- changeably with rate.

full-session DRL A procedure for implementing DRL in which reinforcement is delivered at the end of the session if the total number of responses emitted during the ses- sion does not exceed a criterion limit. (See differential reinforcement of low rates (DRL).)

function-altering effect (relevant to operant relations) A relatively permanent change in an organism’s repertoire

of MO, stimulus, and response relations, caused by re- inforcement, punishment, an extinction procedure, or a recovery from punishment procedure. Respondent function-altering effects result from the pairing and un- pairing of antecedent stimuli.

function-based definition Designates responses as members of the targeted response class solely in terms of their com- mon effect on the environment.

functional analysis (as part of functional behavior assess- ment) An analysis of the purposes (functions) of prob- lem behavior, wherein antecedents and consequences representing those in the person’s natural routines are arranged within an experimental design so that their sep- arate effects on problem behavior can be observed and measured; typically consists of four conditions: three test conditions—contingent attention, contingent escape, and alone—and a control condition in which problem behav- ior is expected to be low because reinforcement is freely available and no demands are placed on the person.

functional behavior assessment (FBA) A systematic method of assessment for obtaining information about the pur- poses (functions) a problem behavior serves for a person; results are used to guide the design of an intervention for decreasing the problem behavior and increasing appro- priate behavior.

functional communication training (FCT) An antecedent in- tervention in which an appropriate communicative behavior is taught as a replacement behavior for problem behavior usually evoked by an establishing operation (EO); involves differential reinforcement of alternative behavior (DRA).

functional relation A verbal statement summarizing the re- sults of an experiment (or group of related experiments) that describes the occurrence of the phenomena under study as a function of the operation of one or more spec- ified and controlled variables in the experiment in which a specific change in one event (the dependent variable) can be produced by manipulating another event (the inde- pendent variable), and that the change in the dependent variable was unlikely the result of other factors (con- founding variables); in behavior analysis expressed as b = f (x1), (x2), . . . , where b is the behavior and x1, x2, etc., are environmental variables of which the behavior is a function.

functionally equivalent Serving the same function or pur- pose; different topographies of behavior are functionally equivalent if they produce the same consequences.

general case analysis A systematic process for identifying and selecting teaching examples that represent the full range of stimulus variations and response requirements in the generalization setting(s). (See also multiple exemplar training and teaching sufficient examples.)

generalization A generic term for a variety of behavioral processes and behavior change outcomes. (See general- ization gradient, generalized behavior change, response generalization, response maintenance, setting/situation generalization, and stimulus generalization.)

Glossary

generalization across subjects Changes in the behavior of people not directly treated by an intervention as a func- tion of treatment contingencies applied to other people.

generalization probe Any measurement of a learner’s per- formance of a target behavior in a setting and/or stimulus situation in which direct training has not been provided.

generalization setting Any place or stimulus situation that differs in some meaningful way from the instructional set- ting and in which performance of the target behavior is desired. (Contrast with instructional setting.)

generalized behavior change A behavior change that has not been taught directly. Generalized outcomes take one, or a combination of, three primary forms: response maintenance, stimulus/setting generalization, and response generaliza- tion. Sometimes called generalized outcome.

generalized conditioned punisher A stimulus change that, as a result of having been paired with many other punish- ers, functions as punishment under most conditions be- cause it is free from the control of motivating conditions for specific types of punishment.

generalized conditioned reinforcer A conditioned reinforcer that as a result of having been paired with many other re- inforcers does not depend on an establishing operation for any particular form of reinforcement for its effectiveness.

generic (tact) extension A tact evoked by a novel stimulus that shares all of the relevant or defining features associ- ated with the original stimulus.

graph A visual format for displaying data; reveals relations among and between a series of measurements and rele- vant variables.

group contingency A contingency in which reinforcement for all members of a group is dependent on the behavior of (a) a person within the group, (b) a select group of mem- bers within the larger group, or (c) each member of the group meeting a performance criterion. (See dependent group contingency, independent group contingency, interdependent group contingency.)

habilitation Habilitation (adjustment) occurs when a per- son’s repertoire has been changed such that short- and long-term reinforcers are maximized and short- and long- term punishers are minimized.

habit reversal A multiple-component treatment package for reducing unwanted habits such as fingernail biting and muscle tics; treatment typically includes self-awareness training involving response detection and procedures for identifying events that precede and trigger the response; competing response training; and motivation techniques including self-administered consequences, social support systems, and procedures for promoting the generalization and maintenance of treatment gains.

habituation A decrease in responsiveness to repeated pre- sentations of a stimulus; most often used to describe a re- duction of respondent behavior as a function of repeated presentation of the eliciting stimulus over a short span of time; some researchers suggest that the concept also ap- plies to within-session changes in operant behavior.

hallway time-out A procedure for implementing time-out in which, contingent on the occurrence of an inappropriate behavior, the student is removed from the classroom to a hallway location near the room for a specified period of time.

hero procedure Another term for a dependent group contin- gency (i.e., a person earns a reward for the group).

high-probability (high-p) request sequence An antecedent intervention in which two to five easy tasks with a known history of learner compliance (the high-p requests) are presented in quick succession immediately before re- questing the target task, the low-p request. Also called int- erspersed requests, pretask requests, or behavioral momentum.

higher order conditioning Development of a conditioned reflex by pairing of a neutral stimulus (NS) with a condi- tioned stimulus (CS). Also called secondary conditioning.

history of reinforcement An inclusive term referring in gen- eral to all of a person’s learning experiences and more specifically to past conditioning with respect to particular response classes or aspects of a person’s repertoire. (See ontogeny.)

hypothetical construct A presumed but unobserved process or entity (e.g., Freud’s id, ego, and superego).

imitation A behavior controlled by any physical movement that serves as a novel model excluding vocal-verbal be- havior, has formal similarity with the model, and imme- diately follows the occurrence of the model (e.g., within seconds of the model presentation). An imitative behavior is a new behavior emitted following a novel antecedent event (i.e., the model). (See formal similarity; contrast with echoic.)

impure tact A verbal operant involving a response that is evoked by both an MO and a nonverbal stimulus; thus, the response is part mand and part tact. (See mand and tact.)

independent group contingency A contingency in which reinforcement for each member of a group is dependent on that person’s meeting a performance criterion that is in ef- fect for all members of the group.

independent variable The variable that is systematically ma- nipulated by the researcher in an experiment to see whether changes in the independent variable produce reliable changes in the dependent variable. In applied behavior analysis, it is usually an environmental event or condition antecedent or consequent to the dependent variable. Sometimes called the intervention or treatment variable. (Compare with dependent variable.)

indirect functional assessment Structured interviews, check- lists, rating scales, or questionnaires used to obtain infor- mation from people who are familiar with the person exhibiting the problem behavior (e.g., teachers, parents, caregivers, and/or the individual him- or herself); used to identify conditions or events in the natural environment that correlate with the problem behavior.

indirect measurement Occurs when the behavior that is measured is in some way different from the behavior of

Glossary

interest; considered less valid than direct measurement be- cause inferences about the relation between the data ob- tained and the actual behavior of interest are required. (Contrast with direct measurement.)

indiscriminable contingency A contingency that makes it difficult for the learner to discriminate whether the next response will produce reinforcement. Practitioners use in- discriminable contingencies in the form of intermittent schedules of reinforcement and delayed rewards to pro- mote generalized behavior change.

informed consent When the potential recipient of services or participant in a research study gives his explicit permission before any assessment or treatment is provided. Full dis- closure of effects and side effects must be provided. To give consent, the person must (a) demonstrate the capac- ity to decide, (b) do so voluntarily, and (c) have adequate knowledge of all salient aspects of the treatment.

instructional setting The environment where instruction oc- curs; includes all aspects of the environment, planned and unplanned, that may influence the learner’s acquisition and generalization of the target behavior. (Contrast with generalization setting.)

interdependent group contingency A contingency in which reinforcement for all members of a group is dependent on each member of the group meeting a performance criterion that is in effect for all members of the group.

intermittent schedule of reinforcement (INT) A contin- gency of reinforcement in which some, but not all, occur- rences of the behavior produce reinforcement.

internal validity The extent to which an experiment shows convincingly that changes in behavior are a function of the independent variable and not the result of uncontrolled or unknown variables. (Compare to external validity.)

interobserver agreement (IOA) The degree to which two or more independent observers report the same observed values after measuring the same events.

interresponse time (IRT) A measure of temporal locus; defined as the elapsed time between two successive responses.

interval-by-interval IOA An index of the agreement be- tween observers for data obtained by interval recording or time sampling measurement; calculated for a given ses- sion or measurement period by comparing the two ob- servers’ recordings of the occurrence or nonoccurrence of the behavior in each observation interval and dividing the number of intervals of agreement by the total number of intervals and multiplying by 100. Also called the point- by-point or total interval IOA. (Compare to scored-interval IOA and unscored-interval IOA.)

interval DRL A procedure for implementing DRL in which the total session is divided into equal intervals and rein- forcement is provided at the end of each interval in which the number of responses during the interval is equal to or below a criterion limit. (See differential reinforcement of low rates (DRL).)

intraverbal An elementary verbal operant that is evoked by a verbal discriminative stimulus and that does not have point-to-point correspondence with that verbal stimulus.

irreversibility A situation that occurs when the level of re- sponding observed in a previous phase cannot be repro- duced even though the experimental conditions are the same as they were during the earlier phase.

lag reinforcement schedule A schedule of reinforcement in which reinforcement is contingent on a response being different in some specified way (e.g., different topogra- phy) from the previous response (e.g., Lag 1) or a speci- fied number of previous responses (e.g., Lag 2 or more).

latency See response latency. level The value on the vertical axis around which a series of

behavioral measures converge. level system A component of some token economy systems

in which participants advance up (or down) through a suc- cession of levels contingent on their behavior at the cur- rent level. The performance criteria and sophistication or difficulty of the behaviors required at each level are higher than those of preceding levels; as participants advance to higher levels, they gain access to more desirable rein- forcers, increased privileges, and greater independence.

limited hold A situation in which reinforcement is available only during a finite time following the elapse of an FI or VI interval; if the target response does not occur within the time limit, reinforcement is withheld and a new inter- val begins (e.g., on an FI 5-minute schedule with a limited hold of 30 seconds, the first correct response following the elapse of 5 minutes is reinforced only if that response occurs within 30 seconds after the end of the 5-minute interval).

line graph Based on a Cartesian plane, a two-dimensional area formed by the intersection of two perpendicular lines. Any point within the plane represents a specific relation between the two dimensions described by the intersecting lines. It is the most common graphic format for displaying data in applied behavior analysis.

listener Someone who provides reinforcement for verbal be- havior. A listener may also serve as an audience evoking verbal behavior. (Contrast with speaker.)

local response rate The average rate of response during a smaller period of time within a larger period for which an overall response rate has been given. (See overall re- sponse rate.)

magnitude The force or intensity with which a response is emitted; provides important quantitative parameters used in defining and verifying the occurrence of some response classes. Responses meeting those criteria are measured and reported by one or more fundamental or derivative measures such as frequency, duration, or latency. Some- times called amplitude.

maintenance Two different meanings in applied behavior analysis: (a) the extent to which the learner continues to perform the target behavior after a portion or all of the

Glossary

intervention has been terminated (i.e., response mainte- nance), a dependent variable or characteristic of behavior; and (b) a condition in which treatment has been discon- tinued or partially withdrawn, an independent variable or experimental condition.

mand An elementary verbal operant that is evoked by an MO and followed by specific reinforcement.

massed practice A self-directed behavior change technique in which the person forces himself to perform an unde- sired behavior (e.g., a compulsive ritual) repeatedly, which sometimes decreases the future frequency of the behavior.

matching law The allocation of responses to choices avail- able on concurrent schedules of reinforcement; rates of responding across choices are distributed in proportions that match the rates of reinforcement received from each choice alternative.

matching-to-sample A procedure for investigating condi- tional relations and stimulus equivalence. A matching-to- sample trial begins with the participant making a response that presents or reveals the sample stimulus; next, the sam- ple stimulus may or may not be removed, and two or more comparison stimuli are presented. The participant then se- lects one of the comparison stimuli. Responses that select a comparison stimulus that matches the sample stimulus are reinforced, and no reinforcement is provided for re- sponses selecting the nonmatching comparison stimuli.

mean count-per-interval IOA The average percentage of agreement between the counts reported by two observers in a measurement period comprised of a series of smaller counting times; a more conservative measure of IOA than total count IOA.

mean duration-per-occurrence IOA An IOA index for du- ration per occurrence data; also a more conservative and usually more meaningful assessment of IOA for total du- ration data calculated for a given session or measurement period by computing the average percentage of agreement of the durations reported by two observers for each oc- currence of the target behavior.

measurement bias Nonrandom measurement error; a form of inaccurate measurement in which the data consistently overestimate or underestimate the true value of an event.

measurement by permanent product A method of mea- suring behavior after it has occurred by recording the ef- fects that the behavior produced on the environment.

mentalism An approach to explaining behavior that assumes that a mental, or “inner,” dimension exists that differs from a behavioral dimension and that phenomena in this di- mension either directly cause or at least mediate some forms of behavior, if not all.

metaphorical (tact) extension A tact evoked by a novel stim- ulus that shares some, but not all, of the relevant features of the original stimulus.

methodological behaviorism A philosophical position that views behavioral events that cannot be publicly observed as outside the realm of science.

metonymical (tact) extension A tact evoked by a novel stim- ulus that shares none of the relevant features of the origi- nal stimulus configuration, but some irrelevant yet related feature has acquired stimulus control.

mixed schedule (mix) A compound schedule of reinforce- ment consisting of two or more basic schedules of rein- forcement (elements) that occur in an alternating, usually random, sequence; no discriminative stimuli are correlated with the presence or absence of each element of the sched- ule, and reinforcement is delivered for meeting the re- sponse requirements of the element in effect at any time.

momentary time sampling A measurement method in which the presence or absence of behaviors are recorded at pre- cisely specified time intervals. (Contrast with interval recording.)

motivating operation (MO) An environmental variable that (a) alters (increases or decreases) the reinforcing or pun- ishing effectiveness of some stimulus, object, or event; and (b) alters (increases or decreases) the current frequency of all behavior that has been reinforced or punished by that stimulus, object, or event. (See abative effect, abolishing operation (AO), behavior-altering effect, evocative ef- fect, establishing operation (EO), value-altering effect.)

multielement design See alternating treatments design. multiple baseline across behaviors design A multiple base-

line design in which the treatment variable is applied to two or more different behaviors of the same subject in the same setting.

multiple baseline across settings design A multiple base- line design in which the treatment variable is applied to the same behavior of the same subject across two or more different settings, situations, or time periods.

multiple baseline across subjects design A multiple base- line design in which the treatment variable is applied to the same behavior of two or more subjects (or groups) in the same setting.

multiple baseline design An experimental design that begins with the concurrent measurement of two or more behaviors in a baseline condition, followed by the application of the treatment variable to one of the behaviors while baseline conditions remain in effect for the other behavior(s). After maximum change has been noted in the first behavior, the treatment variable is applied in sequential fashion to each of the other behaviors in the design. Experimental control is demonstrated if each behavior shows similar changes when, and only when, the treatment variable is introduced.

multiple control (of verbal behavior) There are two types of multiple control: (a) convergent multiple control occurs when a single verbal response is a function of more than one variable and (b) what is said has more than one an- tecedent source of control. Divergent multiple control oc- curs when a single antecedent variable affects the strength of more than one responses.

multiple exemplar training Instruction that provides the learner with practice with a variety of stimulus conditions,

Glossary

response variations, and response topographies to ensure the acquisition of desired stimulus controls response forms; used to promote both setting/situation generaliza- tion and response generalization. (See teaching sufficient examples.)

multiple probe design A variation of the multiple baseline design that features intermittent measures, or probes, dur- ing baseline. It is used to evaluate the effects of instruction on skill sequences in which it is unlikely that the subject can improve performance on later steps in the sequence before learning prior steps.

multiple schedule (mult) A compound schedule of rein- forcement consisting of two or more basic schedules of re- inforcement (elements) that occur in an alternating, usually random, sequence; a discriminative stimulus is correlated with the presence or absence of each element of the schedule, and reinforcement is delivered for meet- ing the response requirements of the element in effect at any time.

multiple treatment interference The effects of one treatment on a subject’s behavior being confounding by the influence of another treatment administered in the same study.

multiple treatment reversal design Any experimental de- sign that uses the experimental methods and logic of the reversal tactic to compare the effects of two or more ex- perimental conditions to baseline and/or to one another (e.g., A-B-A-B-C-B-C, A-B-A-C-A-D-A-C-A-D, A-B-A- B-B+C-B-B+C).

naive observer An observer who is unaware of the study’s pur- pose and/or the experimental conditions in effect during a given phase or observation period. Data obtained by a naive observer are less likely to be influenced by observers’ expectations.

naturally existing contingency Any contingency of rein- forcement (or punishment) that operates independent of the behavior analyst’s or practitioner’s efforts; includes socially mediated contingencies contrived by other peo- ple and already in effect in the relevant setting. (Contrast with contrived contingency.)

negative punishment A response behavior is followed im- mediately by the removal of a stimulus (or a decrease in the intensity of the stimulus), that decreases the future fre- quency of similar responses under similar conditions; sometimes called Type II punishment. (Contrast with positive punishment.)

negative reinforcer A stimulus whose termination (or re- duction in intensity) functions as reinforcement. (Contrast with positive reinforcer.)

neutral stimulus (NS) A stimulus change that does not elicit respondent behavior. (Compare to conditioned stimulus (CS), unconditioned stimulus (US).)

noncontingent reinforcement (NCR) A procedure in which stimuli with known reinforcing properties are presented on fixed-time (FT) or variable-time (VT) schedules com- pletely independent of behavior; often used as an an-

tecedent intervention to reduce problem behavior. (See fixed-time schedule (FT), variable-time schedule (VT).)

noncontingent reinforcement (NCR) reversal technique An experimental control technique that demonstrates the effects of reinforcement by using noncontingent rein- forcement (NCR) as a control condition instead of a no- reinforcement (baseline) condition. During the NCR condition, the stimulus change used as reinforcement in the reinforcement condition is presented on a fixed or vari- able time schedule independent of the subject’s behavior. A higher level of responding during the reinforcement con- dition than during the NCR condition demonstrates that the changes in behavior are the result of contingent rein- forcement, not simply the presentation of or contact with the stimulus event. (Compare with DRI/DRA reversal technique, DRO reversal technique.)

nonexclusion time-out A procedure for implementing time- out in which, contingent on the occurrence of the target behavior, the person remains within the setting, but does not have access to reinforcement, for a specified period.

normalization As a philosophy and principle, the belief that people with disabilities should, to the maximum extent pos- sible, be physically and socially integrated into the main- stream of society regardless of the degree or type of disability. As an approach to intervention, the use of pro- gressively more typical settings and procedures “to estab- lish and/or maintain personal behaviors which are as culturally normal as possible” (Wolfensberger, 1972, p. 28).

observed value A measure produced by an observation and measurement system. Observed values serve as the data that the researcher and others will interpret to form con- clusions about an investigation. (Compare with true value.)

observer drift Any unintended change in the way an ob- server uses a measurement system over the course of an in- vestigation that results in measurement error; often entails a shift in the observer’s interpretation of the original def- initions of the target behavior subsequent to being trained. (See measurement bias, observer reactivity.)

observer reactivity Influence on the data reported by an observer that results from the observer’s awareness that others are evaluating the data he reports. (See also meas- urement bias and observer drift.)

ontogeny The history of the development of an individual organism during its lifetime. (See history of reinforce- ment; compare to phylogeny.)

operant behavior Behavior that is selected, maintained, and brought under stimulus control as a function of its conse- quences; each person’s repertoire of operant behavior is a product of his history of interactions with the environment (ontogeny).

operant conditioning The basic process by which operant learning occurs; consequences (stimulus changes imme- diately following responses) result in an increased (rein- forcement) or decreased (punishment) frequency of the same type of behavior under similar motivational and

Glossary

environmental conditions in the future. (See motivating operation, punishment, reinforcement, response class, stimulus control.)

overall response rate The rate of response over a given time period. (See local response rate.)

overcorrection A behavior change tactic based on positive punishment in which, contingent on the problem behavior, the learner is required to engage in effortful behavior di- rectly or logically related to fixing the damage caused by the behavior. Forms of overcorrection are restitutional overcorrection and positive practice overcorrection. (See positive practice overcorrection, restitutional over- correction.)

parametric analysis An experiment designed to discover the differential effects of a range of values of an independent variable.

parsimony The practice of ruling out simple, logical expla- nations, experimentally or conceptually, before consider- ing more complex or abstract explanations.

partial-interval recording A time sampling method for mea- suring behavior in which the observation period is divided into a series of brief time intervals (typically from 5 to 10 seconds). The observer records whether the target behavior occurred at any time during the interval. Partial-interval recording is not concerned with how many times the be- havior occurred during the interval or how long the behav- ior was present, just that it occurred at some point during the interval; tends to overestimate the proportion of the obser- vation period that the behavior actually occurred.

partition time-out An exclusion procedure for implement- ing time-out in which, contingent on the occurrence of the target behavior, the person remains within the time-in set- ting, but stays behind a wall, shield, or barrier that restricts the view.

percentage A ratio (i.e., a proportion) formed by combining the same dimensional quantities, such as count (number ÷ number) or time (duration ÷ duration; latency ÷ latency); expressed as a number of parts per 100; typically expressed as a ratio of the number of responses of a certain type per total number of responses (or opportunities or intervals in which such a response could have occurred). A percentage presents a proportional quantity per 100.

philosophic doubt An attitude that the truthfulness and va- lidity of all scientific theory and knowledge should be con- tinually questioned.

phylogeny The history of the natural evolution of a species. (Compare to ontogeny.)

pivotal behavior A behavior that, when learned, produces corresponding modifications or covariation in other un- trained behaviors. (Compare to behavioral cusp.)

placebo control A procedure that prevents a subject from detecting the presence or absence of the treatment vari- able. To the subject the placebo condition appears the same as the treatment condition (e.g., a placebo pill con- tains an inert substance but looks, feels, and tastes exactly

like a pill that contains the treatment drug). (See double- blind control.)

planned activity check (PLACHECK) A variation of mo- mentary time sampling in which the observer records whether each person in a group is engaged in the target behavior at specific points in time; provides a measure of “group behavior.”

planned ignoring A procedure for implementing time-out in which social reinforcers—usually attention, physical contact, and verbal interaction—are withheld for a brief period contingent on the occurrence of the target behavior.

point-to-point correspondence A relation between the stim- ulus and response or response product that occurs when the beginning, middle, and end of the verbal stimulus matches the beginning, middle, and end of the verbal response. The verbal relations with point-to-point correspondence are echoic, copying a text, imitation as it relates to sign lan- guage, textual, and transcription.

positive practice overcorrection A form of overcorrection in which, contingent on an occurrence of the target behavior, the learner is required to repeated a correct form of the be- havior, or a behavior incompatible with the problem be- havior, a specified number of times; entails an educative component. (See overcorrection, restitutional overcor- rection.)

positive punishment A behavior is followed immediately by the presentation of a stimulus that decreases the future fre- quency of the behavior; sometimes called Type I punish- ment. (Contrast with negative punishment.)

positive reinforcement Occurs when a behavior is followed immediately by the presentation of a stimulus that in- creases the future frequency of the behavior in similar con- ditions (Contrast to negative reinforcement.)

positive reinforcer A stimulus whose presentation or onset functions as reinforcement. (Contrast with negative rein- forcer.)

postreinforcement pause The absence of responding for a period of time following reinforcement; an effect com- monly produced by fixed interval (FI) and fixed ratio (FR) schedules of reinforcement.

practice effects Improvements in performance resulting from opportunities to perform a behavior repeatedly so that baseline measures can be obtained.

prediction A statement of the anticipated outcome of a presently unknown or future measurement; one of three components of the experimental reasoning, or baseline logic, used in single-subject research designs. (See repli- cation, verification.)

Premack principle A principle that states that making the opportunity to engage in a high-probability behavior contingent on the occurrence of a low-frequency behavior will function as reinforcement for the low-frequency be- havior. (See also response-deprivation hypothesis.)

principle of behavior A statement describing a functional re- lation between behavior and one or more of its controlling

Glossary

variables with generality across organisms, species, set- tings, behaviors, and time (e.g., extinction, positive rein- forcement); an empirical generalization inferred from many experiments demonstrating the same functional relation

procedural fidelity See treatment integrity. programming common stimuli A tactic for promoting set-

ting/situation generalization by making the instructional setting similar to the generalization setting; the two-step process involves (1) identifying salient stimuli that char- acterize the generalization setting and (2) incorporating those stimuli into the instructional setting.

progressive schedule of reinforcement A schedule that sys- tematically thins each successive reinforcement opportu- nity independent of the individual’s behavior; progressive ratio (PR) and progressive interval (PI) schedules are thinned using arithmetic or geometric progressions.

punisher A stimulus change that decreases the future fre- quency of behavior that immediately precedes it. (See aversive stimulus, conditioned punisher, unconditioned punisher.)

punishment Occurs when stimulus change immediately fol- lows a response and decreases the future frequency of that type of behavior in similar conditions. (See negative pun- ishment, positive punishment.)

radical behaviorism A thoroughgoing form of behaviorism that attempts to understand all human behavior, including private events such as thoughts and feelings, in terms of controlling variables in the history of the person (on- togeny) and the species (phylogeny).

rate A ratio of count per observation time; often expressed as count per standard unit of time (e.g., per minute, per hour, per day) and calculated by dividing the number of re- sponses recorded by the number of standard units of time in which observations were conducted; used interchange- ably with frequency. The ratio is formed by combining the different dimensional quantities of count and time (i.e., count time). Ratios formed from different dimensional quantities retain their dimensional quantities. Rate and frequency in behavioral measurement are synonymous terms. (Contrast with percentage.)

ratio strain A behavioral effect associated with abrupt in- creases in ratio requirements when moving from denser to thinner reinforcement schedules; common effects in- clude avoidance, aggression, and unpredictable pauses or cessation in responding.

reactivity Effects of an observation and measurement pro- cedure on the behavior being measured. Reactivity is most likely when measurement procedures are obtrusive, espe- cially if the person being observed is aware of the ob- server’s presence and purpose.

recovery from punishment procedure The occurrence of a previously punished type of response without its punish- ing consequence. This procedure is analogous to the ex- tinction of previously reinforced behavior and has the effect of undoing the effect of the punishment.

reflex A stimulus–response relation consisting of an antecedent stimulus and the respondent behavior it elicits (e.g., bright light–pupil contraction). Unconditioned and conditioned re- flexes protect against harmful stimuli, help regulate the in- ternal balance and economy of the organism, and promote reproduction. (See conditioned reflex, respondent be- havior, respondent conditioning, unconditioned reflex.)

reflexive conditioned motivating operation (CMO-R) A stimulus that acquires MO effectiveness by preceding some form of worsening or improvement. It is exemplified by the warning stimulus in a typical escape–avoidance proce- dure, which establishes its own offset as reinforcement and evokes all behavior that has accomplished that offset.

reflexivity A type of stimulus-to-stimulus relation in which the learner, without any prior training or reinforcement for doing so, selects a comparison stimulus that is the same as the sample stimulus (e.g., A = A). Reflexivity would be demonstrated in the following matching-to-sample proce- dure: The sample stimulus is a picture of a tree, and the three comparison stimuli are a picture of a mouse, a pic- ture of a cookie, and a duplicate of the tree picture used as the sample stimulus. The learner selects the picture of the tree without specific reinforcement in the past for making the tree-picture-to-tree-picture match. (It is also called generalized identity matching.) (See stimulus equiva- lence; compare to transitivity, symmetry.)

reinforcement Occurs when a stimulus change immediately follows a response and increases the future frequency of that type of behavior in similar conditions. (See negative reinforcement, positive reinforcement.)

reinforcer A stimulus change that increases the future fre- quency of behavior that immediately precedes it. (See conditioned reinforcer, unconditioned reinforcer.)

reinforcer-abolishing effect (of a motivating operation) A decrease in the reinforcing effectiveness of a stimulus, ob- ject, or event caused by a motivating operation. For ex- ample, food ingestion abolishes (decreases) the reinforcing effectiveness of food.

reinforcer assessment Refers to a variety of direct, empiri- cal methods for presenting one or more stimuli contingent on a target response and measuring their effectiveness as reinforcers.

reinforcer-establishing effect (of a motivating operation) An increase in the reinforcing effectiveness of a stimulus, object, or event caused by a motivating operation. For ex- ample, food deprivation establishes (increases) the rein- forcing effectiveness of food.

relevance of behavior rule Holds that only behaviors likely to produce reinforcement in the person’s natural environ- ment should be targeted for change.

reliability (of measurement) Refers to the consistency of measurement, specifically, the extent to which repeated measurement of the same event yields the same values.

repeatability Refers to the fact that a behavior can occur re- peatedly through time (i.e., behavior can be counted); one of the three dimensional quantities of behavior from which

Glossary

all behavioral measurements are derived. (See count, fre- quency, rate, celeration, temporal extent, and temporal locus.)

repertoire All of the behaviors a person can do; or a set of behaviors relevant to a particular setting or task (e.g., gar- dening, mathematical problem solving).

replication (a) Repeating conditions within an experiment to determine the reliability of effects and increase internal validity. (See baseline logic, prediction, verification.) (b) Repeating whole experiments to determine the gener- ality of findings of previous experiments to other subjects, settings, and/or behaviors. (See direct replication, ex- ternal validity, systematic replication.)

resistance to extinction The relative frequency with which operant behavior is emitted during extinction.

respondent behavior The response component of a reflex; behavior that is elicited, or induced, by antecedent stimuli. (See reflex, respondent conditioning.)

respondent conditioning A stimulus–stimulus pairing pro- cedure in which a neutral stimulus (NS) is presented with an unconditioned stimulus (US) until the neutral stimulus becomes a conditioned stimulus that elicits the conditioned response (also called classical or Pavlovian conditioning). (See conditioned reflex, higher order conditioning.)

respondent extinction The repeated presentation of a con- ditioned stimulus (CS) in the absence of the unconditioned stimulus (US); the CS gradually loses its ability to elicit the conditioned response until the conditioned reflex no longer appears in the individual’s repertoire.

response A single instance or occurrence of a specific class or type of behavior. Technical definition: an “action of an organism’s effector. An effector is an organ at the end of an efferent nerve fiber that is specialized for altering its envi- ronment mechanically, chemically, or in terms of other en- ergy changes” (Michael, 2004, p. 8). (See response class.)

response blocking A procedure in which the therapist phys- ically intervenes as soon as the learner begins to emit a problem behavior to prevent completion of the targeted behavior.

response class A group of responses of varying topography, all of which produce the same effect on the environment.

response cost The contingent loss of reinforcers (e.g., a fine), producing a decrease of the frequency of behavior; a form of negative punishment.

response-deprivation hypothesis A model for predicting whether contingent access to one behavior will function as reinforcement for engaging in another behavior based on whether access to the contingent behavior represents a re- striction of the activity compared to the baseline level of engagement. (See Premack principle.)

response differentiation A behavior change produced by differential reinforcement: Reinforced members of the current response class occur with greater frequency, and unreinforced members occur less frequently (undergo ex- tinction); the overall result is the emergence of a new re- sponse class.

response generalization The extent to which a learner emits untrained responses that are functionally equivalent to the trained target behavior. (Compare to response mainte- nance and setting/situation generalization.)

response latency A measure of temporal locus; the elapsed time from the onset of a stimulus (e.g., task direction, cue) to the initiation of a response.

response maintenance The extent to which a learner con- tinues to perform the target behavior after a portion or all of the intervention responsible for the behavior’s initial appearance in the learner’s repertoire has been terminated. Often called maintenance, durability, behavioral persis- tence, and (incorrectly) resistance to extinction. (Com- pare to response generalization and setting/situation generalization.)

restitutional overcorrection A form of overcorrection in which, contingent on the problem behavior, the learner is required to repair the damage or return the environment to its original state and then to engage in additional behavior to bring the environment to a condition vastly better than it was in prior to the misbehavior. (See overcorrection and positive practice overcorrection.)

reversal design Any experimental design in which the re- searcher attempts to verify the effect of the independent variable by “reversing” responding to a level obtained in a previous condition; encompasses experimental designs in which the independent variable is withdrawn (A-B- A-B) or reversed in its focus (e.g., DRI/DRA). (See A-B-A design, A-B-A-B design, B-A-B, DRI/DRA reversal technique, DRO reversal technique, noncontingent re- inforcement (NCR) reversal technique.)

rule-governed behavior Behavior controlled by a rule (i.e., a verbal statement of an antecedent-behavior-consequence contingency); enables human behavior (e.g., fastening a seatbelt) to come under the indirect control of temporally remote or improbable but potentially significant conse- quences (e.g., avoiding injury in an auto accident). Often used in contrast to contingency-shaped behavior, a term used to indicate behavior selected and maintained by con- trolled, temporally close consequences.

satiation A decrease in the frequency of operant behavior presumed to be the result of continued contact with or con- sumption of a reinforcer that has followed the behavior; also refers to a procedure for reducing the effectiveness of a reinforcer (e.g., presenting a person with copious amounts of a reinforcing stimulus prior to a session). (See motivating operation; contrast with deprivation.)

scatterplot A two-dimensional graph that shows the relative distribution of individual measures in a data set with re- spect to the variables depicted by the x and y axes. Data points on a scatterplot are not connected.

schedule of reinforcement A rule specifying the environ- mental arrangements and response requirements for rein- forcement; a description of a contingency of reinforcement.

schedule thinning Changing a contingency of reinforcement by gradually increasing the response ratio or the extent of

Glossary

the time interval; it results in a lower rate of reinforcement per responses, time, or both.

science A systematic approach to the understanding of nat- ural phenomena (as evidenced by description, prediction, and control) that relies on determinism as its fundamental assumption, empiricism as its primary rule, experimenta- tion as its basic strategy, replication as a requirement for believability, parsimony as a value, and philosophic doubt as its guiding conscience.

scored-interval IOA An interobserver agreement index based only on the intervals in which either observer recorded the occurrence of the behavior; calculated by di- viding the number of intervals in which the two observers agreed that the behavior occurred by the number of inter- vals in which either or both observers recorded the occur- rence of the behavior and multiplying by 100. Scored-interval IOA is recommended as a measure of agreement for behaviors that occur at low rates because it ignores the intervals in which agreement by chance is highly likely. (Compare to interval-by-interval IOA and unscored-interval IOA.)

selection by consequences The fundamental principle un- derlying operant conditioning; the basic tenet is that all forms of (operant) behavior, from simple to complex, are selected, shaped, and maintained by their consequences during an individual’s lifetime; Skinner’s concept of se- lection by consequences is parallel to Darwin’s concept of natural selection of genetic structures in the evolution of species.

self-contract Contingency contract that a person makes with himself, incorporating a self-selected task and reward as well as personal monitoring of task completions and self- delivery of the reward.

self-control Two meanings: (a) A person’s ability to “delay gratification” by emitting a response that will produce a larger (or higher quality) delayed reward over a response that produces a smaller but immediate reward (sometimes considered impulse control); (b) A person’s behaving in a certain way so as to change a subsequent behavior (i.e., to self-manage her own behavior). Skinner (1953) concep- tualized self-control as a two-response phenomenon: The controlling response affects variables in such a way as to change the probability of the controlled response. (See self-management.)

self-evaluation A procedure in which a person compares his performance of a target behavior with a predetermined goal or standard; often a component of self-management. Sometimes called self-assessment.

self-instruction Self-generated verbal responses, covert or overt, that function as rules or response prompts for a desired behavior; as a self-management tactic, self-instruction can guide a person through a behavior chain or sequence of tasks.

self-management The personal application of behavior change tactics that produces a desired change in behavior.

self-monitoring A procedure whereby a person systemati- cally observes his behavior and records the occurrence or

nonoccurrence of a target behavior. (Also called self- recording or self-observation.)

semilogarithmic chart A two-dimensional graph with a log- arithmic scaled y axis so that equal distances on the verti- cal axis represent changes in behavior that are of equal proportion. (See Standard Celeration Chart.)

sensory extinction The process by which behaviors main- tained by automatic reinforcement are placed on extinction by masking or removing the sensory consequence.

sequence effects The effects on a subject’s behavior in a given condition that are the result of the subject’s experi- ence with a prior condition.

setting/situation generalization The extent to which a learner emits the target behavior in a setting or stimulus sit- uation that is different from the instructional setting.

shaping Using differential reinforcement to produce a series of gradually changing response classes; each response class is a successive approximation toward a terminal be- havior. Members of an existing response class are selected for differential reinforcement because they more closely resemble the terminal behavior. (See differential rein- forcement, response class, response differentiation, successive approximations.)

single-subject designs A wide variety of research designs that use a form of experimental reasoning called baseline logic to demonstrate the effects of the independent variable on the behavior of individual subjects. (Also called single- case, within-subject, and intra-subject designs) (See also alternating treatments design, baseline logic, changing criterion design, multiple baseline design, reversal de- sign, steady state strategy.)

social validity Refers to the extent to which target behaviors are appropriate, intervention procedures are acceptable, and important and significant changes in target and collateral be- haviors are produced.

solistic (tact) extension A verbal response evoked by a stim- ulus property that is only indirectly related to the proper tact relation (e.g., Yogi Berra’s classic malapropism: “Baseball is ninety percent mental; the other half is physical.”

spaced-responding DRL A procedure for implementing DRL in which reinforcement follows each occurrence of the target behavior that is separated from the previous re- sponse by a minimum interresponse time (IRT). (See diff- erential reinforcement of low rates (DRL).)

speaker Someone who engages in verbal behavior by emit- ting mands, tacts, intraverbals, autoclitics, and so on. A speaker is also someone who uses sign language, gestures, signals, written words, codes, pictures, or any form of ver- bal behavior. (Contrast with listener.)

split-middle line of progress A line drawn through a series of graphed data points that shows the overall trend in the data; drawn through the intersections of the vertical and horizontal middles of each half of the charted data and then adjusted up or down so that half of all the data points fall on or above and half fall on or below the line.

Glossary

spontaneous recovery A behavioral effect associated with extinction in which the behavior suddenly begins to occur after its frequency has decreased to its prereinforcement level or stopped entirely.

stable baseline Data that show no evidence of an upward or downward trend; all of the measures fall within a relatively small range of values. (See steady state responding.)

Standard Celeration Chart A multiply–divide chart with

can accommodate response rates as low as 1 per 24 hours (0.000695 per minute) to as high as 1,000 per minute. It enables the standardized charting of celeration, a factor by which rate of behavior multiplies or divides per unit of time. (See semilogarithmic chart.)

steady state responding A pattern of responding that ex- hibits relatively little variation in its measured dimensional quantities over a period of time.

steady state strategy Repeatedly exposing a subject to a given condition while trying to eliminate or control extra- neous influences on the behavior and obtaining a stable pattern of responding before introducing the next condi- tion.

stimulus “An energy change that affects an organism through its receptor cells” (Michael, 2004, p. 7).

stimulus class A group of stimuli that share specified com- mon elements along formal (e.g., size, color), temporal (e.g., antecedent or consequent), and/or functional (e.g., discriminative stimulus) dimensions.

stimulus control A situation in which the frequency, latency, duration, or amplitude of a behavior is altered by the pres- ence or absence of an antecedent stimulus. (See discrimi- nation, discriminative stimulus.)

stimulus delta (S�) A stimulus in the presence of which a given behavior has not produced reinforcement in the past. (Contrast with discriminative stimulus (SD).)

stimulus discrimination training The conventional proce- dure requires one behavior and two antecedent stimulus conditions. Responses are reinforced in the presence of one stimulus condition, the SD, but not in the presence of the other stimulus, the S�.

stimulus equivalence The emergence of accurate responding to untrained and nonreinforced stimulus–stimulus rela- tions following the reinforcement of responses to some stimulus–stimulus relations. A positive demonstration of reflexivity, symmetry, and transitivity is necessary to meet the definition of equivalence.

stimulus generalization When an antecedent stimulus has a history of evoking a response that has been reinforced in its presence, the same type of behavior tends to be evoked by stimuli that share similar physical properties with the controlling antecedent stimulus.

stimulus generalization gradient A graphic depiction of the extent to which behavior that has been reinforced in the presence of a specific stimulus condition is emitted in the presence of other stimuli. The gradient shows relative degree of stimulus generalization and stimulus control (or

discrimination). A flat slope across test stimuli shows a high degree of stimulus generalization and relatively little discrimination between the trained stimulus and other stimuli; a slope that drops sharply from its highest point corresponding to the trained stimulus indicates a high de- gree of stimulus control (discrimination) and relatively lit- tle stimulus generalization.

stimulus preference assessment A variety of procedures used to determine the stimuli that a person prefers, the rel- ative preference values (high versus low) of those stimuli, the conditions under which those preference values remain in effect, and their presumed value as reinforcers.

stimulus–stimulus pairing A procedure in which two stim- uli are presented at the same time, usually repeatedly for a number of trials, which often results in one stimulus ac- quiring the function of the other stimulus.

successive approximations The sequence of new response classes that emerge during the shaping process as the re- sult of differential reinforcement; each successive response class is closer in form to the terminal behavior than the response class it replaces.

surrogate conditioned motivating operation (CMO-S) A stimulus that acquires its MO effectiveness by being paired with another MO and has the same value-altering and behavior-altering effects as the MO with which it was paired.

symmetry A type of stimulus-to-stimulus relationship in which the learner, without prior training or reinforcement for doing so, demonstrates the reversibility of matched sample and comparison stimuli (e.g., if A = B, then B = A). Symmetry would be demonstrated in the following matching-to-sample procedure: The learner is taught, when presented with the spoken word car (sample stimulus A), to select a compari- son picture of a car (comparison B). When presented with the picture of a car (sample stimulus B), without additional train- ing or reinforcement, the learner selects the comparison spo- ken word car (comparison A). (See stimulus equivalence; compare to reflexivity, transitivity.)

systematic desensitization A behavior therapy treatment for anxieties, fears, and phobias that involves substituting one response, generally muscle relaxation, for the unwanted behavior—the fear and anxiety. The client practices re- laxing while imagining anxiety-producing situations in a sequence from the least fearful to the most fearful.

systematic replication An experiment in which the re- searcher purposefully varies one or more aspects of an ear- lier experiment. A systematic replication that reproduces the results of previous research not only demonstrates the reliability of the earlier findings but also adds to the ex- ternal validity of the earlier findings by showing that the same effect can be obtained under different conditions.

tact An elementary verbal operant evoked by a nonverbal discriminative stimulus and followed by generalized con- ditioned reinforcement.

tandem schedule A schedule of reinforcement identical to the chained schedule except, like the mix schedule, the

six base-10 (or 10, ÷ 10) cycles on the vertical axis that×

Glossary

tandem schedule does not use discriminative stimuli with the elements in the chain. (See chained schedule, mixed schedule.)

target behavior The response class selected for intervention; can be defined either functionally or topographically.

task analysis The process of breaking a complex skill or se- ries of behaviors into smaller, teachable units; also refers to the results of this process.

teaching loosely Randomly varying functionally irrelevant stimuli within and across teaching sessions; promotes set- ting/situation generalization by reducing the likelihood that (a) a single or small group of noncritical stimuli will acquire exclusive control over the target behavior and (2) the learner’s performance of the target behavior will be impeded or “thrown off” should he encounter any of the “loose” stimuli in the generalization setting.

teaching sufficient examples A strategy for promoting gen- eralized behavior change that consists of teaching the learner to respond to a subset of all of the relevant stimu- lus and response examples and then assessing the learner’s performance on untrained examples. (See multiple ex- emplar training.)

temporal extent Refers to the fact that every instance of be- havior occurs during some amount of time; one of the three dimensional quantities of behavior from which all behav- ioral measurements are derived. (See repeatability and temporal locus.)

temporal locus Refers to the fact that every instance of be- havior occurs at a certain point in time with respect to other events (i.e., when in time behavior occurs can be mea- sured); often measured in terms of response latency and interresponse time (IRT); one of the three dimensional quantities of behavior from which all behavioral measure- ments are derived. (See repeatability, temporal extent.)

terminal behavior The end product of shaping. textual An elementary verbal operant involving a response

that is evoked by a verbal discriminative stimulus that has point-to-point correspondence, but not formal similarity, between the stimulus and the response product.

three-term contingency The basic unit of analysis in the analysis of operant behavior; encompasses the temporal and possibly dependent relations among an antecedent stimulus, behavior, and consequence.

time-out from positive reinforcement The contingent with- drawal of the opportunity to earn positive reinforcement or the loss of access to positive reinforcers for a specified time; a form of negative punishment (also called time-out).

time-out ribbon A procedure for implementing nonexclu- sion time-out in which a child wears a ribbon or wristband that becomes discriminative for receiving reinforcement. Contingent on misbehavior, the ribbon is removed and ac- cess to social and other reinforcers are unavailable for a specific period. When time-out ends, the ribbon or band is returned to the child and time-in begins.

time sampling A measurement of the presence or absence of behavior within specific time intervals. It is most use-

ful with continuous and high-rate behaviors. (See momentary time sampling, partial-interval recording, and whole-interval recording.)

token An object that is awarded contingent on appropriate behavior and that serves as the medium of exchange for backup reinforcers.

token economy A system whereby participants earn gener- alized conditioned reinforcers (e.g., tokens, chips, points) as an immediate consequence for specific behaviors; par- ticipants accumulate tokens and exchange them for items and activities from a menu of backup reinforcers. (See generalized conditioned reinforcer.)

topography The physical form or shape of a behavior. topography-based definition Defines instances of the tar-

geted response class by the shape or form of the behavior. total count IOA The simplest indicator of IOA for event

recording data; based on comparing the total count recorded by each observer per measurement period; cal- culated by dividing the smaller of the two observers’ counts by the larger count and multiplying by 100.

total duration IOA A relevant index of IOA for total dura- tion measurement; computed by dividing the shorter of the two durations reported by the observers by the longer duration and multiplying by 100.

total-task chaining A variation of forward chaining in which the learner receives training on each behavior in the chain during each session.

transcription An elementary verbal operant involving a spoken verbal stimulus that evokes a written, typed, or fin- ger-spelled response. Like the textual, there is point-to- point correspondence between the stimulus and the response product, but no formal similarity.

transitive conditioned motivating operation (CMO-T) An environmental variable that, as a result of a learning history, establishes (or abolishes) the reinforcing effec- tiveness of another stimulus and evokes (or abates) the be- havior that has been reinforced by that other stimulus.

transitivity A derived (i.e., untrained) stimulus-stimulus re- lation (e.g., A = C, C = A) that emerges as a product of training two other stimulus-stimulus relations (e.g., A = B and B = C). For example, transitivity would be demon- strated if, after training the two stimulus-stimulus rela- tions shown in 1 and 2 below, the relation shown in 3 emerges without additional instruction or reinforcement: (1) If A (e.g., spoken word bicycle) = B (e.g., the picture

of a bicycle) (see Figure 3), and (2) B (the picture of a bicycle) = C (e.g., the written word

bicycle) (see Figure 4), then (3) C (the written word bicycle) = A (the spoken name,

bicycle) (see Figure 5). (See stimulus equivalence; compare to reflexivity, symmetry.)

treatment drift An undesirable situation in which the inde- pendent variable of an experiment is applied differently during later stages than it was at the outset of the study.

treatment integrity The extent to which the independent variable is applied exactly as planned and described and no

Glossary

other unplanned variables are administered inadvertently along with the planned treatment. Also called procedural fidelity.

trend The overall direction taken by a data path. It is de- scribed in terms of direction (increasing, decreasing, or zero trend), degree (gradual or steep), and the extent of variability of data points around the trend. Trend is used in predicting future measures of the behavior under un- changing conditions.

trial-by-trial IOA An IOA index for discrete trial data based on comparing the observers’ counts (0 or 1) on a trial-by- trial, or item-by-item, basis; yields a more conservative and meaningful index of IOA for discrete trial data than does total count IOA.

trials-to-criterion A special form of event recording; a mea- sure of the number of responses or practice opportunities needed for a person to achieve a preestablished level of accuracy or proficiency.

true value A measure accepted as a quantitative description of the true state of some dimensional quantity of an event as it exists in nature. Obtaining true values requires “spe- cial or extraordinary precautions to ensure that all possi- ble sources of error have been avoided or removed” (John- ston & Pennypacker, 1993a, p. 136). (Compare with observed value.)

Type I error An error that occurs when a researcher con- cludes that the independent variable had an effect on the dependent variable, when no such relation exists; a false positive. (Contrast with Type II error.)

Type II error An error that occurs when a researcher con- cludes that the independent variable had no effect on the dependent variable, when in truth it did; a false negative. (Contrast with Type I error.)

unconditioned motivating operation (UMO) A motivating operation whose value-altering effect does not depend on a learning history. For example, food deprivation increases the reinforcing effectiveness of food without the necessity of any learning history.

unconditioned negative reinforcer A stimulus that func- tions as a negative reinforcer as a result of the evolution- ary development of the species (phylogeny); no prior learning is involved (e.g., shock, loud noise, intense light, extreme temperatures, strong pressure against the body). (See negative reinforcer; compare with conditioned neg- ative reinforcer.)

unconditioned punisher A stimulus change that decreases the frequency of any behavior that immediately precedes it irrespective of the organism’s learning history with the stimulus. Unconditioned punishers are products of the evo- lutionary development of the species (phylogeny), mean- ing that all members of a species are more or less susceptible to punishment by the presentation of uncon- ditioned punishers (also called primary or unlearned pun- ishers). (Compare with conditioned punisher.)

unconditioned reflex An unlearned stimulus–response func- tional relation consisting of an antecedent stimulus (e.g.,

food in mouth) that elicits the response (e.g., salivation); a product of the phylogenic evolution of a given species; all biologically intact members of a species are born with similar repertoires of unconditioned reflexes. (See con- ditioned reflex.)

unconditioned reinforcer A stimulus change that increases the frequency of any behavior that immediately precedes it irrespective of the organism’s learning history with the stimulus. Unconditioned reinforcers are the product of the evolutionary development of the species (phylogeny). Also called primary or unlearned reinforcer. (Compare with conditioned reinforcer.)

unconditioned stimulus (US) The stimulus component of an unconditioned reflex; a stimulus change that elicits re- spondent behavior without any prior learning.

unpairing Two kinds: (a) The occurrence alone of a stimu- lus that acquired its function by being paired with an al- ready effective stimulus, or (b) the occurrence of the stimulus in the absence as well as in the presence of the ef- fective stimulus. Both kinds of unpairing undo the result of the pairing: the occurrence alone of the stimulus that became a conditioned reinforcer; and the occurrence of the unconditioned reinforcer in the absence as well as in the presence of the conditioned reinforcer.

unscored-interval IOA An interobserver agreement index based only on the intervals in which either observer recorded the nonoccurrence of the behavior; calculated by dividing the number of intervals in which the two ob- servers agreed that the behavior did not occur by the num- ber of intervals in which either or both observers recorded the nonoccurrence of the behavior and multiplying by 100. Unscored-interval IOA is recommended as a measure of agreement for behaviors that occur at high rates because it ignores the intervals in which agreement by chance is highly likely. (Compare to interval-by-interval IOA, scored-interval IOA.)

validity (of measurement) The extent to which data obtained from measurement are directly relevant to the target be- havior of interest and to the reason(s) for measuring it.

value-altering effect (of a motivating operation) An alter- ation in the reinforcing effectiveness of a stimulus, object, or event as a result of a motivating operation. For exam- ple, the reinforcing effectiveness of food is altered as a re- sult of food deprivation and food ingestion.

variability The frequency and extent to which multiple mea- sures of behavior yield different outcomes.

variable baseline Data points that do not consistently fall within a narrow range of values and do not suggest any clear trend.

variable interval (VI) A schedule of reinforcement that pro- vides reinforcement for the first correct response follow- ing the elapse of variable durations of time occurring in a random or unpredictable order. The mean duration of the intervals is used to describe the schedule (e.g., on a VI 10- minute schedule, reinforcement is delivered for the first response following an average of 10 minutes since the last

Glossary

reinforced response, but the time that elapses following the last reinforced response might range from 30 seconds or less to 25 minutes or more).

variable-interval DRO (VI-DRO) A DRO procedure in which reinforcement is available at the end of intervals of variable duration and delivered contingent on the absence of the problem behavior during the interval. (See differ- ential reinforcement of other behavior (DRO).)

variable-momentary DRO (VM-DRO) A DRO procedure in which reinforcement is available at specific moments of time, which are separated by variable amounts of time in random sequence, and delivered if the problem is not oc- curring at those times. (See differential reinforcement of other behavior (DRO).)

variable ratio (VR) A schedule of reinforcement requiring a varying number of responses for reinforcement. The number of responses required varies around a random number; the mean number of responses required for rein- forcement is used to describe the schedule (e.g., on a VR 10 schedule an average of 10 responses must be emitted for reinforcement, but the number of responses required fol- lowing the last reinforced response might range from 1 to 30 or more).

variable-time schedule (VT) A schedule for the delivery of noncontingent stimuli in which the interval of time from one delivery to the next randomly varies around a given time. For example, on a VT 1-minute schedule, the delivery- to-delivery interval might range from 5 seconds to 2 min- utes, but the average interval would be 1 minute.

verbal behavior Behavior whose reinforcement is mediated by a listener; includes both vocal-verbal behavior (e.g.,

saying “Water, please” to get water) and nonvocal-verbal behavior (pointing to a glass of water to get water). En- compasses the subject matter usually treated as language and topics such as thinking, grammar, composition, and understanding.

verification One of three components of the experimental reasoning, or baseline logic, used in single-subject research designs; accomplished by demonstrating that the prior level of baseline responding would have remained un- changed had the independent variable not been introduced. Verifying the accuracy of the original prediction reduces the probability that some uncontrolled (confounding) vari- able was responsible for the observed change in behavior. (See prediction, replication.)

visual analysis A systematic approach for interpreting the results of behavioral research and treatment programs that entails visual inspection of graphed data for vari- ability, level, and trend within and between experimental conditions.

whole-interval recording A time sampling method for mea- suring behavior in which the observation period is divided into a series of brief time intervals (typically from 5 to 15 seconds). At the end of each interval, the observer records whether the target behavior occurred throughout the entire interval; tends to underestimate the proportion of the ob- servation period that many behaviors actually occurred.

withdrawal design A term used by some authors as a syn- onym for A-B-A-B design; also used to describe experi- ments in which an effective treatment is sequentially or partially withdrawn to promote the maintenance of behav- ior changes. (See A-B-A-B design, reversal design.)

Definition and Characteristics of Applied Behavior Analysis

Key Terms

applied behavior analysis (ABA) behaviorism determinism empiricism experiment experimental analysis of behavior

(EAB)

explanatory fiction functional relation hypothetical construct mentalism methodological behaviorism parsimony

philosophic doubt radical behaviorism replication science

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 2: Definition and Characteristics

2-1 Explain behavior in accordance with the philosophical assumptions of behavior analysis, such as the lawfulness of behavior, empiricism, experimental analysis, and parsimony.

2-2 Explain determinism as it relates to behavior analysis.

2-3 Distinguish between mentalistic and environmental explanations of behavior.

2-4 Distinguish among the experimental analysis of behavior, applied behavior analysis, and behavioral technologies.

2-5 Describe and explain behavior, including private events, in behavior analytic (nonmentalistic) terms.

2-6 Use the dimensions of applied behavior analysis (Baer, Wolf, & Risley 1968) for evaluating interventions to determine if they are behavior analytic.

Content Area 3: Principles, Processes, and Concepts

3-10 Define and provide examples of functional relations.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®). All rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and questions about this document must be submitted directly to the BACB.

Applied behavior analysis is a science devoted to the understanding and improvement of human behavior. But other fields have similar

intent. What sets applied behavior analysis apart? The answer lies in its focus, goals, and methods. Applied be- havior analysts focus on objectively defined behaviors of social significance; they intervene to improve the behav- iors under study while demonstrating a reliable relation- ship between their interventions and the behavioral improvements; and they use the methods of scientific in- quiry—objective description, quantification, and con- trolled experimentation. In short, applied behavior analysis, or ABA, is a scientific approach for discovering environmental variables that reliably influence socially significant behavior and for developing a technology of behavior change that takes practical advantage of those discoveries.

This chapter provides a brief outline of the history and development of behavior analysis, discusses the phi- losophy that underlies the science, and identifies the defining dimensions and characteristics of applied be- havior analysis. Because applied behavior analysis is an applied science, we begin with an overview of some pre- cepts fundamental to all scientific disciplines.

Some Basic Characteristics and a Definition of Science Used properly, the word science refers to a systematic approach for seeking and organizing knowledge about the natural world. Before offering a definition of science, we discuss the purpose of science and the basic assump- tions and attitudes that guide the work of all scientists, ir- respective of their fields of study.

Purpose of Science

The overall goal of science is to achieve a thorough un- derstanding of the phenomena under study—socially im- portant behaviors, in the case of applied behavior analysis. Science differs from other sources of knowl- edge or ways we obtain knowledge about the world around us (e.g., contemplation, common sense, logic, au- thority figures, religious or spiritual beliefs, political cam- paigns, advertisements, testimonials). Science seeks to discover nature’s truths. Although it is frequently mis- used, science is not a tool for validating the cherished or preferred versions of “the truth” held by any group, cor- poration, government, or institution. Therefore, scientific knowledge must be separated from any personal, politi- cal, economic, or other reasons for which it was sought.

Different types of scientific investigations yield knowledge that enables one or more of three levels of un-

derstanding: description, prediction, and control. Each level of understanding contributes to the scientific knowl- edge base of a given field.

Description

Systematic observation enhances the understanding of a given phenomenon by enabling scientists to describe it accurately. Descriptive knowledge consists of a collec- tion of facts about the observed events that can be quan- tified, classified, and examined for possible relations with other known facts—a necessary and important activity for any scientific discipline. The knowledge obtained from descriptive studies often suggests possible hy- potheses or questions for additional research.

For example, White (1975) reported the results of observing the “natural rates” of approval (verbal praise or encouragement) and disapproval (criticisms, reproach) by 104 classroom teachers in grades 1 to 12. Two major findings were that (a) the rates of teacher praise dropped with each grade level, and (b) in every grade after second, the rate at which teachers delivered statements of disap- proval to students exceeded the rate of teacher approval. The results of this descriptive study helped to stimulate dozens of subsequent studies aimed at discovering the factors responsible for the disappointing findings or at increasing teachers’ rates of praise (e.g., Alber, Heward, & Hippler, 1999; Martens, Hiralall, & Bradley, 1997; Sutherland, Wehby, & Yoder, 2002; Van Acker, Grant, & Henry, 1996).

Prediction

A second level of scientific understanding occurs when repeated observations reveal that two events consistently covary with each other. That is, in the presence of one event (e.g., approaching winter) another event occurs (or fails to occur) with some specified probability (e.g., cer- tain birds fly south). When systematic covariation be- tween two events is found, this relationship—termed a correlation—can be used to predict the relative proba- bility that one event will occur, based on the presence of the other event.

Because no variables are manipulated or controlled by the researcher, correlational studies cannot demon- strate whether any of the observed variables are respon- sible for the changes in the other variable(s), and no such relations should be inferred (Johnston & Pennypacker, 1993a). For example, although a strong correlation ex- ists between hot weather and an increased incidence of drowning deaths, we should not assume that a hot and humid day causes anyone to drown. Hot weather also correlates with other factors, such as an increased num- ber of people (both swimmers and nonswimmers) seeking

Definition and Characteristics of Applied Behavior Analysis

relief in the water, and many instances of drowning have been found to be a function of factors such as the use of alcohol or drugs, the relative swimming skills of the vic- tims, strong riptides, and the absence of supervision by lifeguards.

Results of correlational studies can, however, sug- gest the possibility of causal relations, which can then be explored in later studies. The most common type of cor- relational study reported in the applied behavior analysis literature compares the relative rates or conditional prob- abilities of two or more observed (but not manipulated) variables (e.g., Atwater & Morris, 1988; Symons, Hoch, Dahl, & McComas, 2003; Thompson & Iwata, 2001). For example, McKerchar and Thompson (2004) found correlations between problem behavior exhibited by 14 preschool children and the following consequent events: teacher attention (100% of the children), presentation of some material or item to the child (79% of the children), and escape from instructional tasks (33% of the children). The results of this study not only provide empirical val- idation for the social consequences typically used in clin- ical settings to analyze the variables maintaining children’s problem behavior, but also increase confidence in the prediction that interventions based on the findings from such assessments will be relevant to the conditions that occur naturally in preschool classrooms (Iwata et al., 1994). In addition, by revealing the high probabilities with which teachers responded to problem behavior in ways that are likely to maintain and strengthen it, Mc- Kerchar and Thompson’s findings also point to the need to train teachers in more effective ways to respond to problem behavior.

Control

The ability to predict with a certain degree of confidence is a valuable and useful result of science; prediction en- ables preparation. However, the greatest potential bene- fits from science can be derived from the third, and highest, level of scientific understanding—control. Evi- dence of the kinds of control that can been derived from scientific findings in the physical and biological sciences surrounds us in the everyday technologies we take for granted: pasteurized milk and the refrigerators we store it in; flu shots and the automobiles we drive to go get them; aspirin and the televisions that bombard us with news and advertisements about the drug.

Functional relations, the primary products of basic and applied behavior analytic research, provide the kind of scientific understanding that is most valuable and use- ful to the development of a technology for changing behavior. A functional relation exists when a well- controlled experiment reveals that a specific change in one event (the dependent variable) can reliably be pro-

1Skinner (1953) noted, that although telescopes and cyclotrons give us a “dramatic picture of science in action” (p. 12), and science could not have advanced very far without them, such devices and apparatus are not science themselves. “Nor is science to be identified with precise measurement. We can measure and be mathematical without being scientific at all, just as we may be scientific with these aids” (p. 12). Scientific instruments bring sci- entists into greater contact with their subject matter and, with measure- ment and mathematics, enable a more precise description and control of key variables.

duced by specific manipulations of another event (the independent variable), and that the change in the depen- dent variable was unlikely to be the result of other extra- neous factors (confounding variables).

Johnston and Pennypacker (1980) described func- tional relations as “the ultimate product of a natural sci- entific investigation of the relation between behavior and its determining variables” (p. 16).

Such a “co-relation” is expressed as y = f (x), where x is the independent variable or argument of the function, and y is the dependent variable. In order to determine if an observed relation is truly functional, it is necessary to demonstrate the operation of the values of x in isolation and show that they are sufficient for the production of y. . . . [H]owever, a more powerful relation exists if ne- cessity can be shown (that y occurs only if x occurs). The most complete and elegant form of empirical inquiry in- volves applying the experimental method to identifying functional relations. (Johnston & Pennypacker, 1993a, p. 239)

The understanding gained by the scientific discovery of functional relations is the basis of applied technologies in all fields. Almost all of the research studies cited in this text are experimental analyses that have demonstrated or discovered a functional relation between a target behav- ior and one or more environmental variables. However, it is important to understand that functional relations are also correlations (Cooper, 2005), and that:

In fact, all we ever really know is that two events are related or “co-related” in some way. To say that one “causes” another is to say that one is solely the result of the other. To know this, it is necessary to know that no other factors are playing a contributing role. This is vir- tually impossible to know because it requires identifying all possible such factors and then showing that they are not relevant. (Johnston & Pennypacker, 1993a, p. 240)

Attitudes of Science Science is first of all a set of attitudes.

—B. F. Skinner (1953, p. 12)

The definition of science lies not in test tubes, spec- trometers, or electron accelerators, but in the behavior of scientists. To begin to understand any science, we need to look past the apparatus and instrumentation that are most readily apparent and examine what scientists do.1 The

Definition and Characteristics of Applied Behavior Analysis

pursuit of knowledge can properly be called science when it is carried out according to general methodological pre- cepts and expectations that define science. Although there is no “scientific method” in the sense of a prescribed set of steps or rules that must be followed, all scientists share a fundamental assumption about the nature of events that are amenable to investigation by science, general notions about basic strategy, and perspectives on how to view their findings. These attitudes of science—determinism, empiricism, experimentation, replication, parsimony, and philosophic doubt—constitute a set of overriding as- sumptions and values that guide the work of all scientists (Whaley & Surratt, 1968).

Determinism

Science is predicated on the assumption of determinism. Scientists presume that the universe, or at least that part of it they intend to probe with the methods of science, is a lawful and orderly place in which all phenomena occur as the result of other events. In other words, events do not just happen willy-nilly; they are related in system- atic ways to other factors, which are themselves physical phenomena amenable to scientific investigation.

If the universe were governed by accidentalism, a philosophical position antithetical to determinism that holds that events occur by accident or without cause, or by fatalism, the belief that events are predetermined, the scientific discovery and technological use of functional relations to improve things would be impossible.

If we are to use the methods of science in the field of human affairs, we must assume behavior is lawful and determined. We must expect to discover what a man does is the result of specifiable conditions and that once these conditions have been discovered, we can anticipate and to some extent determine his actions. (Skinner, 1953, p. 6)

Determinism plays a pivotal dual role in the conduct of scientific practice: It is at once a philosophical stance that does not lend itself to proof and the confirmation that is sought by each experiment. In other words, the scien- tist first assumes lawfulness and then proceeds to look for lawful relations (Delprato & Midgley, 1992).

Empiricism

Scientific knowledge is built on, above all, empiricism— the practice of objective observation of the phenomena of interest. Objectivity in this sense means “independent of the individual prejudices, tastes, and private opinions of the scientist. . . . Results of empirical methods are ob- jective in that they are open to anyone’s observation and do not depend on the subjective belief of the individual scientist” (Zuriff, 1985, p. 9).

In the prescientific era, as well as in nonscientific and pseudoscientific activities today, knowledge was (and is) the product of contemplation, speculation, personal opinion, authority, and the “obvious” logic of common sense. The scientist’s empirical attitude, however, de- mands objective observation based on thorough descrip- tion, systematic and repeated measurement, and precise quantification of the phenomena of interest.

As in every scientific field, empiricism is the fore- most rule in the science of behavior. Every effort to un- derstand, predict, and improve behavior hinges on the behavior analyst’s ability to completely define, system- atically observe, and accurately and reliably measure oc- currences and nonoccurrences of the behavior of interest.

Experimentation

Experimentation is the basic strategy of most sciences. Whaley and Surratt (1968) used the following anecdote to introduce the need for conducting experiments.

A man who lived in a suburban dwelling area was sur- prised one evening to see his neighbor bow to the four winds, chant a strange melody, and dance around his front lawn beating a small drum. After witnessing the same ritual for over a month, the man became over- whelmed with curiosity and decided to look into the matter.

“Why do you go through this same ritual each evening?” the man asked his neighbor.

“It keeps my house safe from tigers,” the neighbor replied.

“Good grief!” the man said. “Don’t you know there isn’t a tiger within a thousand miles of here?”

“Yeah,” the neighbor smiled. “Sure works, doesn’t it!” (pp. 23–2 to 23–3)

When events are observed to covary or occur in close temporal sequence, a functional relation may exist, but other factors may be responsible for the observed values of the dependent variable. To investigate the possible ex- istence of a functional relation, an experiment (or better, a series of experiments) must be performed in which the factor(s) suspected of having causal status are systemat- ically controlled and manipulated while the effects on the event under study are carefully observed. In dis- cussing the meaning of the term experimental, Dinsmoor (2003) noted that

two measures of behavior may be found to covary at a statistically significant level, but it is not thereby made clear which factor is the cause and which is the effect, or indeed whether the relations between the two is not the product of a third, confounded factor, with which both of them happened to covary. Suppose, for example, it is found that students with good grades have more dates than those with lower grades. Does this imply that high

Definition and Characteristics of Applied Behavior Analysis

grades make people socially attractive? That dating is the royal road to academic success? That it pays to be smart? Or that financial security and free time contribute both to academic and to social success? (p. 152)

Reliably predicting and controlling any phenomena, including the presence of tigers in one’s backyard, re- quires the identification and manipulation of the factors that cause those phenomena to act as they do. One way that the individual described previously could use the ex- perimental method to evaluate the effectiveness of his rit- ual would be to first move to a neighborhood in which tigers are regularly observed and then systematically ma- nipulate the use of his antitiger ritual (e.g., one week on, one week off, one week on) while observing and record- ing the presence of tigers under the ritual and no-ritual conditions.

The experimental method is a method for isolating the relevant variables within a pattern of events. . . . [W]hen the experimental method is employed, it is possible to change one factor at a time (independent variable) while leaving all other aspects of the situation the same, and then to observe what effect this change has on the target behavior (dependent variable). Ideally, a functional rela- tion may be obtained. Formal techniques of experimen- tal control are designed to make sure that the conditions being compared are otherwise the same. Use of the ex- perimental method serves as a necessary condition (sine quo non) to distinguish the experimental analysis of be- havior from other methods of investigation. (Dinsmoor, 2003, p. 152)

Thus, an experiment is a carefully conducted com- parison of some measure of the phenomenon of interest (the dependent variable) under two or more different con- ditions in which only one factor at a time (the indepen- dent variable) differs from one condition to another.

Replication

The results of a single experiment, no matter how well the study was designed and conducted and no matter how clear and impressive the findings, are never sufficient to earn an accepted place among the scientific knowledge base of any field. Although the data from a single exper- iment have value in their own right and cannot be dis- counted, only after an experiment has been replicated a number of times with the same basic pattern of results do scientists gradually become convinced of the findings.

Replication—the repeating of experiments (as well as repeating independent variable conditions within ex- periments)—“pervades every nook and cranny of the ex- perimental method” (Johnston & Pennypacker, 1993a,

p. 244). Replication is the primary method with which scientists determine the reliability and usefulness of their findings and discover their mistakes (Johnston & Penny- packer, 1980; 1993a; Sidman, 1960). Replication—not the infallibility or inherent honesty of scientists—is the primary reason science is a self-correcting enterprise that eventually gets it right (Skinner, 1953).

How many times must an experiment be repeated with the same results before the scientific community ac- cepts the findings? There is no set number of replications required, but the greater the importance of the findings for either theory or practice, the greater the number of repli- cations to be conducted.

Parsimony

One dictionary definition of parsimony is great frugal- ity, and in a special way this connotation accurately de- scribes the behavior of scientists. As an attitude of science, parsimony requires that all simple, logical ex- planations for the phenomenon under investigation be ruled out, experimentally or conceptually, before more complex or abstract explanations are considered. Parsi- monious interpretations help scientists fit their findings within the field’s existing knowledge base. A fully par- simonious interpretation consists only of those elements that are necessary and sufficient to explain the phenom- enon at hand. The attitude of parsimony is so critical to scientific explanations that it is sometimes referred to as the Law of Parsimony (Whaley & Surratt, 1968), a “law” derived from Occam’s Razor, credited to William of Occam (1285–1349), who stated: “One should not in- crease, beyond what is necessary, the number of entities required to explain anything” (Mole, 2003). In other words, given a choice between two competing and com- pelling explanations for the same phenomenon, one should shave off extraneous variables and choose the sim- plest explanation, the one that requires the fewest assumptions.

Philosophic Doubt

The attitude of philosophic doubt requires the scientist to continually question the truthfulness of what is re- garded as fact. Scientific knowledge must always be viewed as tentative. Scientists must constantly be will- ing to set aside their most cherished beliefs and findings and replace them with the knowledge derived from new discoveries.

Good scientists maintain a healthy level of skepti- cism. Although being skeptical of others’ research may be

Definition and Characteristics of Applied Behavior Analysis

easy, a more difficult but critical characteristic of scien- tists is that they remain open to the possibility—as well as look for evidence—that their own findings or inter- pretations are wrong. As Oliver Cromwell (1650) stated in another context: “I beseech you . . . think it possible you may be mistaken.” For the true scientist, “new find- ings are not problems; they are opportunities for further investigation and expanded understanding” (Todd & Mor- ris, 1993, p. 1159).

Practitioners should be as skeptical as researchers. The skeptical practitioner not only requires scientific ev- idence before implementing a new practice, but also eval- uates continually its effectiveness once the practice has been implemented. Practitioners must be particularly skeptical of extraordinary claims made for the effective- ness of new theories, therapies, or treatments (Jacobson, Foxx, & Mulick, 2005; Maurice, 2006).

Claims that sound too good to be true are usually just that. . . . Extraordinary claims require extraordinary evi- dence. (Sagan, 1996; Shermer, 1997)

What constitutes extraordinary evidence? In the strictest sense, and the sense that should be employed when evaluating claims of educational effectiveness, ev- idence is the outcome of the application of the scientific method to test the effectiveness of a claim, a theory, or a practice. The more rigorously the test is conducted, the more often the test is replicated, the more extensively the test is corroborated, the more extraordinary the evi- dence. Evidence becomes extraordinary when it is extra- ordinarily well tested. (Heward & Silvestri, 2005, p. 209)

We end our discussion of philosophic doubt with two pieces of advice, one from Carl Sagan, the other from B. F. Skinner: “The question is not whether we like the conclusion that emerges out of a train of reasoning, but whether the conclusion follows from the premise or start- ing point and whether that premise is true” (Sagan, 1996, p. 210). “Regard no practice as immutable. Change and be ready to change again. Accept no eternal verity. Ex- periment” (Skinner, 1979, p. 346).

Other Important Attitudes and Values

The six attitudes of science that we have examined are necessary features of science and provide an important context for understanding applied behavior analysis. However, the behavior of most productive and success- ful scientists is also characterized by qualities such as thoroughness, curiosity, perseverance, diligence, ethics, and honesty. Good scientists acquire these traits because behaving in such ways has proven beneficial to the progress of science.

2Informative and interesting descriptions of the history of behavior analy- sis can be found in Hackenberg (1995), Kazdin (1978), Michael (2004), Pierce and Epling (1999), Risley (1997, 2005), Sidman (2002), Skinner (1956, 1979), Stokes (2003), and in a special section of articles in the fall 2003 issue of The Behavior Analyst.

A Definition of Science

There is no universally accepted, standard definition of science. We offer the following definition as one that en- compasses the previously discussed purposes and atti- tudes of science, irrespective of the subject matter. Science is a systematic approach to the understanding of natural phenomena—as evidenced by description, pre- diction, and control—that relies on determinism as its fundamental assumption, empiricism as its prime direc- tive, experimentation as its basic strategy, replication as its necessary requirement for believability, parsimony as its conservative value, and philosophic doubt as its guid- ing conscience.

A Brief History of the Development of Behavior Analysis Behavior analysis consists of three major branches. Be- haviorism is the philosophy of the science of behavior, basic research is the province of the experimental analy- sis of behavior (EAB), and developing a technology for improving behavior is the concern of applied behavior analysis (ABA). Applied behavior analysis can be fully understood only in the context of the philosophy and basic research traditions and findings from which it evolved and remains connected today. This section pro- vides an elementary description of the basic tenets of be- haviorism and outlines some of the major events that have marked the development of behavior analysis.2 Table 1 lists major books, journals, and professional organiza- tions that have contributed to the advancement of behav- ior analysis since the 1930s.

Stimulus–Response Behaviorism of Watson

Psychology in the early 1900s was dominated by the study of states of consciousness, images, and other men- tal processes. Introspection, the act of carefully observ- ing one’s own conscious thoughts and feelings, was a primary method of investigation. Although the authors of several texts in the first decade of the 20th century de- fined psychology as the science of behavior (see Kazdin, 1978), John B. Watson is widely recognized as the spokesman for a new direction in the field of psychology.

Definition and Characteristics of Applied Behavior Analysis

Table 1 Representative Selection of Books, Journals, and Organizations That Have Played a Major Role in the Development and Dissemination of Behavior Analysis

Decade Books Journals Organizations

1930s The Behavior of Organisms—Skinner (1938)

1940s Walden Two—Skinner (1948)

1950s Principles of Psychology—Keller & Schoenfeld Journal of the Experimental Society for the Experimental Analysis (1950) Analysis of Behavior (1958) of Behavior (SEAB) (1957)

Science and Human Behavior—Skinner (1953)

Schedules of Reinforcement—Ferster & Skinner (1957)

Verbal Behavior—Skinner (1957)

1960s Tactics for Scientific Research—Sidman (1960) Journal of Applied Behavior American Psychological Association -

Child Development, Vols. I & II—Bijou & Baer Analysis (1968) Division 25 Experimental Analysis of

(1961, 1965) Behavior (1964)

The Analysis of Behavior—Holland Experimental Analysis of Behaviour

& Skinner (1961) Group (UK) (1965)

Research in Behavior Modification—Krasner & Ullmann (1965)

Operant Behavior: Areas of Research and Application—Honig (1966)

The Analysis of Human Operant Behavior— Reese (1966)

Contingencies of Reinforcement: A Theoretical Analysis—Skinner (1969)

1970s Beyond Freedom and Dignity—Skinner (1971) Behaviorism (1972) Norwegian Association for Behavior

Elementary Principles of Behavior—Whaley (became Behavior and Analysis (1973)

& Malott (1971) Philosophy in 1990) Midwestern Association for Behavior

Contingency Management in Education and Revista Mexicana de Analysis (MABA) (1974)

Other Equally Exciting Places—Malott (1974) Analisis de la Conducta Mexican Society of Behavior Analysis (1975) (1975)About Behaviorism—Skinner (1974) Behavior Modification (1977) Association for Behavior Analysis Applying Behavior-Analysis Procedures with Journal of Organizational (formerly MABA) (1978)Children and Youth—Sulzer-Azaroff & Mayer Behavior Management(1977) (1977)

Learning—Catania (1979) Education & Treatment of Children (1977)

The Behavior Analyst (1978)

1980s Strategies and Tactics for Human Behavioral Journal of Precision Teaching Society for the Advancement of Research—Johnston & Pennypacker (1980) and Celeration (originally, Behavior Analysis (1980)

Behaviorism: A Conceptual Reconstruction— Journal of Precision Teaching) Cambridge Center for Behavioral Zuriff (1985) (1980) Studies (1981)

Recent Issues in the Analysis of Behavior— Analysis of Verbal Behavior Japanese Association for Behavior Skinner (1989) (1982) Analysis (1983)

Behavioral Interventions (1986)

Japanese Journal of Behavior Analysis (1986)

Behavior Analysis Digest (1989)

Definition and Characteristics of Applied Behavior Analysis

In his influential article, “Psychology as the Behaviorist Views It,” Watson (1913) wrote:

Psychology as the behaviorist views it is a purely objective experimental branch of natural science. Its theoretical goal is the prediction and control of behavior. Introspection forms no essential part of its methods, nor is the scientific value of its data dependent upon the readiness with which they lend themselves to interpreta- tion in terms of consciousness. (p. 158)

Watson argued that the proper subject matter for psy- chology was not states of mind or mental processes but observable behavior. Further, the objective study of be- havior as a natural science should consist of direct ob- servation of the relationships between environmental stimuli (S) and the responses (R) they evoke. Watsonian behaviorism thus became known as stimulus–response (S–R) psychology. Although there was insufficient sci- entific evidence to support S–R psychology as a workable explanation for most behavior, Watson was confident that his new behaviorism would indeed lead to the prediction and control of human behavior and that it would allow practitioners to improve performance in areas such as ed-

ucation, business, and law. Watson (1924) made bold claims concerning human behavior, as illustrated in this famous quotation:

Give me a dozen healthy infants, well-formed, and my own specified world to bring them up in and I’ll guaran- tee to take any one at random and train him to become any type of specialist I might select—doctor, lawyer, artist, merchant-chief and, yes, even beggar-man and thief, regardless of his talents, penchants, tendencies, abilities, vocations, and race of his ancestors. I am going beyond my facts and I admit it, but so have the advo- cates of the contrary and they have been doing it for many thousands of years. (p. 104)

It is unfortunate that such extraordinary claims were made, exaggerating the ability to predict and control human behavior beyond the scientific knowledge avail- able. The quotation just cited has been used to discredit Watson and continues to be used to discredit behaviorism in general, even though the behaviorism that underlies contemporary behavior analysis is fundamentally differ- ent from the S–R paradigm. Nevertheless, Watson’s con- tributions were of great significance: He made a strong

Table 1 (continued )

Decade Books Journals Organizations

1990s Concepts and Principles of Behavior Behavior and Social Issues Accreditation of Training Programs in Analysis—Michael (1993) (1991) Behavior Analysis (Association for

Behavior Analysis) (1993)

Radical Behaviorism: The Philosophy and the Journal of Behavioral Behavior Analyst Certification Board Science—Chiesa (1994) Education (1991) (BACB) (1998)

Equivalence Relations and Behavior—Sidman Council of Directors of Graduate (1994) Programs in Behavior Analysis

Functional Analysis of Problem Behavior— Journal of Positive Behavior (Association for Behavior Analysis)

Repp & Horner (1999) Interventions (1999) (1999)

The Behavior Analyst Today (1999)

2000s European Journal of First Board Certified Behavior Behavior Analysis (2000) Analysts (BCBA) and Board Certified

Behavioral Technology Associate Behavior Analysts (BCABA)

Today (2001) credentialed by the BACB (2000)

Behavioral Development European Association for Behaviour

Bulletin (2002) Analysis (2002)

Journal of Early and Intensive Behavior Intervention (2004)

Brazilian Journal of Behavior Analysis (2005)

International Journal of Behavioral Consultation and Therapy (2005)

Definition and Characteristics of Applied Behavior Analysis

B. F. Skinner (left) in his Indiana University lab circa 1945 and (right) circa 1967. B. F. Skinner Foundation (left); Julie S. Vargas/B. F. Skinner Foundation (right).

3For an interesting biography and scholarly examination of J. B. Watson’s contributions, see Todd and Morris (1994).

case for the study of behavior as a natural science on a par with the physical and biological sciences.3

Experimental Analysis of Behavior

The experimental branch of behavior analysis formally began in 1938 with the publication of B. F. Skinner’s The Behavior of Organisms (1938/1966). The book summa- rized Skinner’s laboratory research conducted from 1930 to 1937 and brought into perspective two kinds of be- havior: respondent and operant.

Respondent behavior is reflexive behavior as in the tradition of Ivan Pavlov (1927/1960). Respondents are elicited, or “brought out,” by stimuli that immediately precede them. The antecedent stimulus (e.g., bright light) and the response it elicits (e.g., pupil constriction) form a functional unit called a reflex. Respondent behaviors are essentially involuntary and occur whenever the elic- iting stimulus is presented.

Skinner was “interested in giving a scientific account of all behavior, including that which Descartes had set aside as ‘willed’and outside the reach of science” (Glenn, Ellis, & Greenspoon, 1992, p. 1330). But, like other psy- chologists of the time, Skinner found that the S–R para- digm could not explain a great deal of behavior, particularly behaviors for which there were no apparent antecedent causes in the environment. Compared to reflexive behav- ior with its clear eliciting events, much of the behavior of organisms appeared spontaneous or “voluntary.” In an at- tempt to explain the mechanisms responsible for “volun- tary” behavior, other psychologists postulated mediating variables inside the organism in the form of hypothetical constructs such as cognitive processes, drives, and free 4In The Behavior of Organisms, Skinner called the conditioning of re-

spondent behavior Type S conditioning and the conditioning of operant behavior Type R conditioning, but these terms were soon dropped. Re- spondent and operant conditioning and the three-term contingency are fur- ther defined and discussed in the next chapter entitled “Basic Concepts.”

will. Skinner took a different tack. Instead of creating hyp- othetical constructs, presumed but unobserved entities that could not be manipulated in an experiment, Skinner continued to look in the environment for the determinants of behavior that did not have apparent antecedent causes (Kimball, 2002; Palmer, 1998).

He did not deny that physiological variables played a role in determining behavior. He merely felt that this was the domain of other disciplines, and for his part, re- mained committed to assessing the causal role of the en- vironment. This decision meant looking elsewhere in time. Through painstaking research, Skinner accumu- lated significant, if counterintuitive, evidence that be- havior is changed less by the stimuli that precede it (though context is important) and more by the conse- quences that immediately follow it (i.e., consequences that are contingent upon it). The essential formulation for this notion is S–R–S, otherwise known as the three–term contingency. It did not replace the S–R model—we still salivate, for instance, if we smell food cooking when we are hungry. It did, however, account for how the environment “selects” the great part of learned behavior.

With the three-term contingency Skinner gave us a new paradigm. He achieved something no less profound for the study of behavior and learning than Bohr’s model of the atom or Mendel’s model of the gene. (Kimball, 2002, p. 71)

Skinner called the second type of behavior operant behavior.4 Operant behaviors are not elicited by preced- ing stimuli but instead are influenced by stimulus changes that have followed the behavior in the past. Skinner’s

Definition and Characteristics of Applied Behavior Analysis

R es

p o

n se

Time in Minutes

60 120

Original Conditioning All responses to the lever were reinforced. The first three reinforcements were apparently ineffective. The fourth is followed by a rapid increase in rate.

Figure 1 The first data set presented in B. F. Skinner’s The Behavior of Organisms: An Experimental Analysis (1938). From The Behavior of Organisms: An Experimental Analysis by B. F. Skinner, p. 67. Original copyright 1938 by Appleton- Century. Copyright 1991 by B. F. Skinner Foundation, Cambridge, MA. Used by permission.

5Most of the methodological features of the experimental approach pio- neered by Skinner (e.g., repeated measurement of rate or frequency of re- sponse as the primary dependent variable, within-subject experimental comparisons, visual analysis of graphic data displays) continue to charac- terize both basic and applied research in behavior analysis.

most powerful and fundamental contribution to our un- derstanding of behavior was his discovery and experi- mental analyses of the effects of consequences on behavior. The operant three-term contingency as the pri- mary unit of analysis was a revolutionary conceptual breakthrough (Glenn, Ellis, & Greenspoon, 1992).

Skinner (1938/1966) argued that the analysis of op- erant behavior “with its unique relation to the environ- ment presents a separate important field of investigation” (p. 438). He named this new science the experimental analysis of behavior and outlined the methodology for its practice. Simply put, Skinner recorded the rate at which a single subject (he initially used rats and later, pi- geons) emitted a given behavior in a controlled and stan- dardized experimental chamber.

The first set of data Skinner presented in The Be- havior of Organisms was a graph that “gives a record of the resulting change in behavior” (p. 67) when a food pellet was delivered immediately after a rat pressed a lever (see Figure 1). Skinner noted that the first three times that food followed a response “had no observable effect” but that “the fourth response was followed by an appreciable increase in rate showing a swift acceleration to a maximum” (pp. 67–68).

Skinner’s investigative procedures evolved into an elegant experimental approach that enabled clear and powerful demonstrations of orderly and reliable func- tional relations between behavior and various types of environmental events.5 By systematically manipulating the arrangement and scheduling of stimuli that preceded and followed behavior in literally thousands of labora- tory experiments from the 1930s through the 1950s, Skin- ner and his colleagues and students discovered and verified the basic principles of operant behavior that con- tinue to provide the empirical foundation for behavior

6Skinner, who many consider the most eminent psychologist of the 20th century (Haagbloom et al., 2002), authored or coauthored 291 primary- source works (see Morris & Smith, 2003, for a complete bibliography). In addition to Skinner’s three-volume autobiography (Particulars of My Life, 1976; The Shaping of a Behaviorist, 1979; A Matter of Consequences, 1983), numerous biographical books and articles have been written about Skinner, both before and after his death. Students interested in learning about Skinner should read B.F. Skinner: A Life by Daniel Bjork (1997), B.F. Skinner, Organism by Charles Catania (1992), Burrhus Frederic Skin- ner (1904–1990): A Thank You by Fred Keller (1990), B. F. Skinner—The Last Few Days by his daughter Julie Vargas (1990), and Skinner as Self- Manager by Robert Epstein. Skinner’s contributions to ABA are described by Morris, Smith, and Altus (2005).

analysis today. Description of these principles of behav- ior—general statements of functional relations between behavior and environmental events—and tactics for changing behavior derived from those principles consti- tute a major portion of this text.

Skinner’s Radical Behaviorism

In addition to being the founder of the experimental analysis of behavior, B. F. Skinner wrote extensively on the philosophy of that science.6 Without question, Skin- ner’s writings have been the most influential both in guid- ing the practice of the science of behavior and in proposing the application of the principles of behavior to new areas. In 1948 Skinner published Walden Two, a fictional account of how the philosophy and principles of behavior might be used in a utopian community. This was followed by his classic text, Science and Human Be- havior (1953), in which he speculated on how the prin- ciples of behavior might be applied to complex human behavior in areas such as education, religion, govern- ment, law, and psychotherapy.

Much of Skinner’s writing was devoted to the de- velopment and explanation of his philosophy of behav- iorism. Skinner began his book About Behaviorism (1974) with these words:

Behaviorism is not the science of human behavior; it is the philosophy of that science. Some of the questions it asks are these: Is such a science really possible? Can it account for every aspect of human behavior? What

Definition and Characteristics of Applied Behavior Analysis

methods can it use? Are its laws as valid as those of physics and biology? Will it lead to a technology, and if so, what role will it play in human affairs? (p. 1)

The behaviorism that Skinner pioneered differed sig- nificantly (indeed, radically) from other psychological theories, including other forms of behaviorism. Although there were, and remain today, many psychological mod- els and approaches to the study of behavior, mentalism is the common denominator among most.

In general terms, mentalism may be defined as an ap- proach to the study of behavior which assumes that a mental or “inner” dimension exists that differs from a behavioral dimension. This dimension is ordinarily re- ferred to in terms of its neural, psychic, spiritual, subjec- tive, conceptual, or hypothetical properties. Mentalism further assumes that phenomena in this dimension either directly cause or at least mediate some forms of behav- ior, if not all. These phenomena are typically designated as some sort of act, state, mechanism, process, or entity that is causal in the sense of initiating or originating. Mentalism regards concerns about the origin of these phenomena as incidental at best. Finally, mentalism holds that an adequate causal explanation of behavior must appeal directly to the efficacy of these mental phe- nomena. (Moore, 2003, pp. 181–182)

Hypothetical constructs and explanatory fictions are the stock and trade of mentalism, which has dominated Western intellectual thought and most psychological the- ories (Descartes, Freud, Piaget), and it continues to do so into the 21st century. Freud, for example, created a complex mental world of hypothetical constructs—the id, ego, and superego—that he contended were key to understanding a person’s actions.

Hypothetical constructs—“theoretical terms that refer to a possibly existing, but at the moment unobserved process or entity” (Moore, 1995, p. 36)—can neither be observed nor experimentally manipulated (MacCorquo- dale & Meehl, 1948; Zuriff, 1985). Free will, readiness, innate releasers, language acquisition devices, storage and retrieval mechanisms for memory, and information processing are all examples of hypothetical constructs that are inferred from behavior. Although Skinner (1953, 1974) clearly indicated that it is a mistake to rule out events that influence our behavior because they are not accessible to others, he believed that using presumed but unobserved mentalistic fictions (i.e., hypothetical con- structs) to explain the causes of behavior contributed nothing to a functional account.

Consider a typical laboratory situation. A food- deprived rat pushes a lever each time a light comes on and receives food, but the rat seldom pushes the lever when the light is off (and if it does, no food is delivered).

When asked to explain why the rat pushes the lever only when the light is on, most will say that the rat has “made the association” between the light being on and food being delivered when the lever is pressed. As a result of making that association, the animal now “knows” to press the lever only when the light is on. Attributing the rat’s be- havior to a hypothetical cognitive process such as asso- ciating or to something called “knowledge” adds nothing to a functional account of the situation. First, the envi- ronment (in this case, the experimenter) paired the light and food availability for lever presses, not the rat. Second, the knowledge or other cognitive process that is said to explain the observed behavior is itself unexplained, which begs for still more conjecture.

The “knowledge” that is said to account for the rat’s performance is an example of an explanatory fiction, a fictitious variable that often is simply another name for the observed behavior that contributes nothing to an un- derstanding of the variables responsible for developing or maintaining the behavior. Explanatory fictions are the key ingredient in “a circular way of viewing the cause and ef- fect of a situation” (Heron, Tincani, Peterson, & Miller, 2005, p. 274) that give a false sense of understanding.

Turning from observed behavior to a fanciful inner world continues unabated. Sometimes it is little more than a linguistic practice. We tend to make nouns of ad- jectives and verbs and must then find a place for the things the nouns are said to represent. We say that a rope is strong and before long we are speaking of its strength. We call a particular kind of strength tensile, and then ex- plain that the rope is strong because it possesses tensile strength. The mistake is less obvious but more trouble- some when matters are more complex. . . .

Consider now a behavioral parallel. When a person has been subject to mildly punishing consequences in walking on a slippery surface, he may walk in a manner we describe as cautious. It is then easy to say that he walks with caution or that he shows caution. There is no harm in this until we begin to say that he walks carefully because of his caution. (Skinner, 1974, pp. 165–166, emphasis added)

Some believe that behaviorism rejects all events that cannot be operationally defined by objective assessment. Accordingly, Skinner is thought to have rejected all data from his system that could not be independently verified by other people (Moore, 1984). Moore (1985) called this operational view “a commitment to truth by agreement” (p. 59). This common view of the philosophy of behav- iorism is limited; in reality, there are many kinds of be- haviorism—structuralism, methodological behaviorism, and forms of behaviorism that use cognitions as causal factors (e.g., cognitive behavior modification and social learning theory), in addition to the radical behaviorism of Skinner.

Definition and Characteristics of Applied Behavior Analysis

Structuralism and methodological behaviorism do reject all events that are not operationally defined by ob- jective assessment (Skinner, 1974). Structuralists avoid mentalism by restricting their activities to descriptions of behavior. They make no scientific manipulations; accordingly, they do not address questions of causal fac- tors. Methodological behaviorists differ from the struc- turalists by using scientific manipulations to search for functional relations between events. Uncomfortable with basing their science on unobservable phenomena, some early behaviorists either denied the existence of “inner variables” or considered them outside the realm of a sci- entific account. Such an orientation is often referred to as methodological behaviorism.

Methodological behaviorists also usually acknowl- edge the existence of mental events but do not consider them in the analysis of behavior (Skinner, 1974). Method- ological behaviorists’ reliance on public events, exclud- ing private events, restricts the knowledge base of human behavior and discourages innovation in the science of be- havior. Methodological behaviorism is restrictive because it ignores areas of major importance for an understand- ing of behavior.

Contrary to popular opinion, Skinner did not object to cognitive psychology’s concern with private events (i.e., events taking place “inside the skin”) (Moore, 2000). Skinner was the first behaviorist to view thoughts and feelings (he called them “private events”) as behavior to be analyzed with the same conceptual and experimental tools used to analyze publicly observable behavior, not as phenomena or variables that exist within and operate via principles of a separate mental world.

Essentially, Skinner’s behaviorism makes three major assumptions regarding the nature of private events: (a) Private events such as thoughts and feelings are behavior; (b) behavior that takes place within the skin is distin- guished from other (“public”) behavior only by its inac- cessibility; and (c) private behavior is influenced by (i.e., is a function of ) the same kinds of variables as publicly accessible behavior.

We need not suppose that events which take place within an organism’s skin have special properties for that rea- son. A private event may be distinguished by its limited accessibility but not, so far as we know, by any special structure of nature. (Skinner, 1953, p. 257)

By incorporating private events into an overall con- ceptual system of behavior, Skinner created a radical be- haviorism that includes and seeks to understand all human behavior. “What is inside the skin, and how do we know about it? The answer is, I believe, the heart of radical behaviorism” (Skinner, 1974, p. 218). The proper connotations of the word radical in radical behaviorism

are far-reaching and thoroughgoing, connoting the phi- losophy’s inclusion of all behavior, public and private. Radical is also an appropriate modifier for Skinner’s form of behaviorism because it represents a dramatic depar- ture from other conceptual systems in calling for

probably the most drastic change ever proposed in our way of thinking about man. It is almost literally a matter of turning the explanation of behavior inside out. (Skin- ner, 1974, p. 256)

Skinner and the philosophy of radical behaviorism acknowledge the events on which fictions such as cog- nitive processes are based. Radical behaviorism does not restrict the science of behavior to phenomena that can be detected by more than one person. In the context of rad- ical behaviorism, the term observe implies “coming into contact with” (Moore, 1984). Radical behaviorists con- sider private events such as thinking or sensing the stim- uli produced by a damaged tooth to be no different from public events such as oral reading or sensing the sounds produced by a musical instrument. According to Skinner (1974), “What is felt or introspectively observed is not some nonphysical world of consciousness, mind, or men- tal life but the observer’s own body” (pp. 18–19).

The acknowledgment of private events is a major aspect of radical behaviorism. Moore (1980) stated it concisely:

For radical behaviorism, private events are those events wherein individuals respond with respect to certain stim- uli accessible to themselves alone. . . . The responses that are made to those stimuli may themselves be public, i.e., observable by others, or they may be private, i.e., accessible only to the individual involved. Nonetheless, to paraphrase Skinner (1953), it need not be supposed that events taking place within the skin have any special properties for that reason alone. . . . For radical behav- iorism, then, one’s responses with respect to private stimuli are equally lawful and alike in kind to one’s re- sponses with respect to public stimuli. (p. 460)

Scientists and practitioners are affected by their own social context, and institutions and schools are dominated by mentalism (Heward & Cooper, 1992; Kimball, 2002). A firm grasp of the philosophy of radical behaviorism, in addition to knowledge of principles of behavior, can help the scientist and practitioner resist the mentalistic ap- proach of dropping the search for controlling variables in the environment and drifting toward explanatory fic- tions in the effort to understand behavior. The principles of behavior and the procedures presented in this text apply equally to public and private events.

Definition and Characteristics of Applied Behavior Analysis

A thorough discussion of radical behaviorism is far beyond the scope of this text. Still, the serious student of applied behavior analysis should devote considerable study to the original works of Skinner and to other authors who have critiqued, analyzed, and extended the philo- sophical foundations of the science of behavior.7 (See Box 1 for Don Baer’s perspectives on the meaning and importance of radical behaviorism.)

Applied Behavior Analysis

One of the first studies to report the human application of principles of operant behavior was conducted by Fuller (1949). The subject was an 18-year-old boy with pro- found developmental disabilities who was described in the language of the time as a “vegetative idiot.” He lay on his back, unable to roll over. Fuller filled a syringe with a warm sugar-milk solution and injected a small amount of the fluid into the young man’s mouth every time he moved his right arm (that arm was chosen because he moved it infrequently). Within four sessions the boy was moving his arm to a vertical position at a rate of three times per minute.

The attending physicians . . . thought it was impossible for him to learn anything—according to them, he had not learned anything in the 18 years of his life—yet in four experimental sessions, by using the operant condi- tioning technique, an addition was made to his behavior which, at this level, could be termed appreciable. Those who participated in or observed the experiment are of the opinion that if time permitted, other responses could be conditioned and discriminations learned. (Fuller, 1949, p. 590)

During the 1950s and into the early 1960s re- searchers used the methods of the experimental analysis of behavior to determine whether the principles of be- havior demonstrated in the laboratory with nonhuman subjects could be replicated with humans. Much of the early research with human subjects was conducted in clinic or laboratory settings. Although the participants typically benefited from these studies by learning new behaviors, the researchers’ major purpose was to deter- mine whether the basic principles of behavior discovered in the laboratory operated with humans. For example, Bijou (1955, 1957, 1958) researched several principles of behavior with typically developing subjects and peo- ple with mental retardation; Baer (1960, 1961, 1962) ex- amined the effects of punishment, escape, and avoidance

7Excellent discussions of radical behaviorism can be found in Baum (1994); Catania and Harnad (1988); Catania and Hineline (1996); Chiesa (1994); Lattal (1992); Lee (1988); and Moore (1980, 1984, 1995, 2000, 2003).

contingencies on preschool children; Ferster and DeMyer (1961, 1962; DeMyer & Ferster, 1962) conducted a sys- tematic study of the principles of behavior using children with autism as subjects; and Lindsley (1956, 1960) as- sessed the effects of operant conditioning on the behav- ior of adults with schizophrenia. These early researchers clearly established that the principles of behavior are ap- plicable to human behavior, and they set the stage for the later development of applied behavior analysis.

The branch of behavior analysis that would later be called applied behavior analysis (ABA) can be traced to the 1959 publication of Ayllon and Michael’s paper ti- tled “The Psychiatric Nurse as a Behavioral Engineer.” The authors described how direct care personnel in a state hospital used a variety of techniques based on the prin- ciples of behavior to improve the functioning of residents with psychotic disorders or mental retardation. During the 1960s many researchers began to apply principles of behavior in an effort to improve socially important be- havior, but these early pioneers faced many problems. Laboratory techniques for measuring behavior and for controlling and manipulating variables were sometimes unavailable, or their use was inappropriate in applied set- tings. As a result, the early practitioners of applied behavior analysis had to develop new experimental procedures as they went along. There was little funding for the new discipline, and researchers had no ready out- let for publishing their studies, making it difficult to com- municate among themselves about their findings and solutions to methodological problems. Most journal ed- itors were reluctant to publish studies using an experi- mental method unfamiliar to mainstream social science, which relied on large numbers of subjects and tests of statistical inference.

Despite these problems it was an exciting time, and major new discoveries were being made regularly. For example, many pioneering applications of behavior prin- ciples to education occurred during this period (see, e.g., O’Leary & O’Leary, 1972; Ulrich, Stachnik, & Mabry 1974), from which were derived teaching procedures such as contingent teacher praise and attention (Hall, Lund, & Jackson, 1968), token reinforcement systems (Birnbrauer, Wolf, Kidder, & Tague, 1965), curriculum design (Becker, Englemann, & Thomas, 1975), and programmed instruction (Bijou, Birnbrauer, Kidder, & Tague, 1966; Markle, 1962). The basic methods for reliably improving student performance developed by those early applied behavior analysts provided the foundation for behavioral approaches to curriculum design, instructional methods, classroom management, and the generalization and main- tenance of learning that continue to be used decades later (cf., Heward et al., 2005).

University programs in applied behavior analysis were begun in the 1960s and early 1970s at Arizona State

Definition and Characteristics of Applied Behavior Analysis

Box 1 What Is Behaviorism?

Don Baer loved the science of behavior. He loved to write about it, and he loved to talk about it. Don was famous for his unparalleled ability to speak extemporaneously about complex philosophical, experimental, and professional issues in a way that always made thorough conceptual, practical, and human sense. He did so with the vocabulary and syntax of a great author and the accomplished delivery of a master storyteller. The only thing Don knew better than his audience was his science.

On three occasions, in three different decades, graduate students and faculty in the special education program at Ohio State University were fortunate to have Professor Baer serve as Distinguished Guest Faculty for a doctoral seminar, Contemporary Issues in Special Education and Applied Behavior Analysis. The questions and responses that follow were selected from transcripts of two of Professor Baer’s three OSU teleconference seminars.

If a person on the street approached you and asked, “What’s behaviorism?” how would you reply?

The key point of behaviorism is that what people do can be understood. Traditionally, both the layperson and the psychologist have tried to understand behavior by see- ing it as the outcome of what we think, what we feel, what we want, what we calculate, and etcetera. But we don’t have to think about behavior that way. We could look upon it as a process that occurs in its own right and has its own causes. And those causes are, very often, found in the external environment.

Behavior analysis is a science of studying how we can arrange our environments so they make very likely the behaviors we want to be probable enough, and they make unlikely the behaviors we want to be improbable. Behaviorism is understanding how the environment works so that we can make ourselves smarter, more or- ganized, more responsible; so we can encounter fewer punishments and fewer disappointments. A central point of behaviorism is this: We can remake our environment to accomplish some of that much more easily than we can remake our inner selves.

An interviewer once asked Edward Teller, the physi- cist who helped develop the first atomic bomb, “Can you explain to a nonscientist what you find so fasci-

nating about science, particularly physics?” Teller replied, “No.” I sense that Teller was suggesting that a nonscientist would not be able to comprehend, un- derstand, or appreciate physics and his fascination with it. If a nonscientist asked you, “What do you find so fascinating about science, particularly the science of human behavior?” what would you say?

Ed Morris organized a symposium on just this topic a couple of years ago at the Association for Behavior Analysis annual convention, and in that symposium, Jack Michael commented on the fact that although one of our discipline’s big problems and challenges is communi- cating with our society about who we are, what we do, and what we can do, he didn’t find it reasonable to try to summarize what behavior analysis is to an ordinary per- son in just a few words. He gave us this example: Imag- ine a quantum physicist is approached at a cocktail party by someone who asks, “What is quantum physics?” Jack said that the physicist might very well answer, and prob- ably should answer, “I can’t tell you in a few words. You should register for my course.”

I’m very sympathetic with Jack’s argument. But I also know, as someone who’s confronted with the poli- tics of relating our discipline to society, that although it may be a true answer, it’s not a good answer. It’s not an answer that people will hear with any pleasure, or in- deed, even accept. I think such an answer creates only resentment. Therefore, I think we have to engage in a bit of honest show business. So, if I had to somehow state some connotations of what holds me in the field, I guess I would say that since I was a child I always found my biggest reinforcer was something called understanding. I liked to know how things worked. And of all of the things in the world there are to understand, it became clear to me that the most fascinating was what people do. I started with the usual physical science stuff, and it was intriguing to me to understand how radios work, and how electricity works, and how clocks work, etcetera. But when it became clear to me that we could also learn how people work—not just biologically, but behav- iorally—I thought that’s the best of all. Surely, everyone must agree that that’s the most fascinating subject mat- ter. That there could be a science of behavior, of what we do, of who we are? How could you resist that?

Adapted from “Thursday Afternoons with Don: Selections from Three Teleconference Seminars on Applied Behavior Analysis” by W. L. Heward & C. L. Wood (2003). In K. S. Budd & T. Stokes (Eds.), A Small Matter of Proof: The Legacy of Donald M. Baer (pp. 293–310). Reno, NV: Context Press. Used by permission.

Definition and Characteristics of Applied Behavior Analysis

University, Florida State University, the University of Illi- nois, Indiana University, the University of Kansas, the University of Oregon, the University of Southern Illinois, the University of Washington, West Virginia University, and Western Michigan University, among others. Through their teaching and research, faculty at each of these pro- grams made major contributions to the rapid growth of the field.8

Two significant events in 1968 mark that year as the formal beginning of contemporary applied behavior analysis. First, the Journal of Applied Behavior Analysis (JABA) began publication. JABA was the first journal in the United States to deal with applied problems that gave researchers using methodology from the experimental analysis of behavior an outlet for publishing their find- ings. JABA was and continues to be the flagship journal of applied behavior analysis. Many of the early articles in JABA became model demonstrations of how to conduct and interpret applied behavior analysis, which in turn led to improved applications and experimental methodology.

The second major event of 1968 was the publication of the paper, “Some Current Dimensions of Applied Be- havior Analysis” by Donald M. Baer, Montrose M. Wolf, and Todd R. Risley. These authors, the founding fathers of the new discipline, defined the criteria for judging the adequacy of research and practice in applied behavior analysis and outlined the scope of work for those in the science. Their 1968 paper has been the most widely cited publication in applied behavior analysis, and it remains the standard description of the discipline.

Defining Characteristics of Applied Behavior Analysis Baer, Wolf, and Risley (1968) recommended that applied behavior analysis should be applied, behavioral, ana- lytic, technological, conceptually systematic, effective, and capable of appropriately generalized outcomes. In 1987 Baer and colleagues reported that the “seven self- conscious guides to behavior analytic conduct” (p. 319) they had offered 20 years earlier “remain functional; they still connote the current dimensions of the work usually called applied behavior analysis” (p. 314). As we write this book, nearly 40 years have passed since Baer, Wolf, and Risley’s seminal paper was published, and we find that the seven dimensions they posed continue to serve as the primary criteria for defining and judging the value of applied behavior analysis.

8Articles describing the histories of the applied behavior analysis programs at five of these universities can be found in the winter 1994 issue of JABA.

Applied

The applied in applied behavior analysis signals ABA’s commitment to affecting improvements in behaviors that enhance and improve people’s lives. To meet this crite- rion, the researcher or practitioner must select behaviors to change that are socially significant for participants: social, language, academic, daily living, self-care, voca- tional, and/or recreation and leisure behaviors that im- prove the day-to-day life experience of the participants and/or affect their significant others (parents, teachers, peers, employers) in such a way that they behave more positively with and toward the participant.

Behavioral

At first it may seem superfluous to include such an ob- vious criterion—of course applied behavior analysis must be behavioral. However, Baer and colleagues (1968) made three important points relative to the behavioral cri- terion. First, not just any behavior will do; the behavior chosen for study must be the behavior in need of im- provement, not a similar behavior that serves as a proxy for the behavior of interest or the subject’s verbal de- scription of the behavior. Behavior analysts conduct stud- ies of behavior, not studies about behavior. For example, in a study evaluating the effects of a program to teach school children to get along with one another, an applied behavior analyst would directly observe and measure clearly defined classes of interactions between and among the children instead of using indirect measures such as the children’s answers on a sociogram or responses to a questionnaire about how they believe they get along with one another.

Second, the behavior must be measurable; the precise and reliable measurement of behavior is just as critical in applied research as it is in basic research. Applied re- searchers must meet the challenge of measuring socially significant behaviors in their natural settings, and they must do so without resorting to the measurement of non- behavioral substitutes.

Third, when changes in behavior are observed dur- ing an investigation, it is necessary to ask whose behav- ior has changed. Perhaps only the behavior of the observers has changed. “Explicit measurement of the re- liability of human observers thus becomes not merely good technique, but a prime criterion of whether the study was appropriately behavioral” (Baer et al., 1968, p. 93). Or perhaps the experimenter’s behavior has changed in an unplanned way, making it inappropriate to attribute any observed change in the subject’s behavior to the inde- pendent variables that were manipulated. The applied be- havior analyst should attempt to monitor the behavior of all persons involved in a study.

Definition and Characteristics of Applied Behavior Analysis

Analytic

A study in applied behavior analysis is analytic when the experimenter has demonstrated a functional relation be- tween the manipulated events and a reliable change in some measurable dimension of the targeted behavior. In other words, the experimenter must be able to control the occurrence and nonoccurrence of the behavior. Some- times, however, society does not allow the repeated manipulation of important behaviors to satisfy the re- quirements of experimental method. Therefore, applied behavior analysts must demonstrate control to the great- est extent possible, given the restraints of the setting and behavior; and then they must present the results for judg- ment by the consumers of the research. The ultimate issue is believability: Has the researcher achieved experimen- tal control to demonstrate a reliable functional relation?

The analytic dimension enables ABA to not only demonstrate effectiveness, but also to provide the “acid test proof” of functional and replicable relations between the interventions it recommends and socially significant outcomes (D. M. Baer, October 21, 1982, personal communication).

Because we are a data- and design-based discipline, we are in the remarkable position of being able to prove that behavior can work in the way that our technology pre- scribes. We are not theorizing about how behavior can work; we are describing systematically how it has worked many times in real-world applications, in de- signs too competent and with measurement systems too reliable and valid to doubt. Our ability to prove that be- havior can work that way does not, of course, establish that behavior cannot work any other way: we are not in a discipline that can deny any other approaches, only in one that can affirm itself as knowing many of its sufficient conditions at the level of experimental proof . . . our subject matter is behavior change, and we can specify some actionable sufficient conditions for it. (D. M. Baer, personal communication, October 21, 1982, emphasis in original).

Technological

A study in applied behavior analysis is technological when all of its operative procedures are identified and described with sufficient detail and clarity “such that a reader has a fair chance of replicating the application with the same results” (Baer et al., 1987, p. 320).

It is not enough to say what is to be done when the sub- ject makes response R1; it is essential also whenever possible to say what is to be done if the subject makes the alternative responses, R2, R3, etc. For example, one may read that temper tantrums in children are often ex- tinguished by closing the child in his room for the dura- tion of the tantrums plus ten minutes. Unless that

procedure description also states what should be done if the child tries to leave the room early, or kicks out the window, or smears feces on the walls, or begins to make strangling sounds, etc., it is not precise technological de- scription. (Baer et al., 1968, pp. 95–96)

No matter how powerful its effects in any given study, a behavior change method will be of little value if practitioners are unable to replicate it. The development of a replicable technology of behavior change has been a defining characteristic and continuing goal of ABA from its inception. Behavioral tactics are replicable and teachable to others. Interventions that cannot be repli- cated with sufficient fidelity to achieve comparable out- comes are not considered part of the technology.

A good check of the technological adequacy of a pro- cedural description is to have a person trained in applied behavior analysis carefully read the description and then act out the procedure in detail. If the person makes any mistakes, adds any operations, omits any steps, or has to ask any questions to clarify the written description, then the description is not sufficiently technological and re- quires improvement.

Conceptually Systematic

Although Baer and colleagues (1968) did not state so ex- plicitly, a defining characteristic of applied behavior analysis concerns the types of interventions used to im- prove behavior. Although there are an infinite number of tactics and specific procedures that can be used to alter behavior, almost all are derivatives and/or combinations of a relatively few basic principles of behavior. Thus, Baer and colleagues recommended that research reports of applied behavior analysis be conceptually systematic, meaning that the procedures for changing behavior and any interpretations of how or why those procedures were effective should be described in terms of the relevant principle(s) from which they were derived.

Baer and colleagues (1968) provided a strong ra- tionale for the use of conceptual systems in applied be- havior analysis. First, relating specific procedures to basic principles might enable the research consumer to derive other similar procedures from the same principle(s). Sec- ond, conceptual systems are needed if a technology is to become an integrated discipline instead of a “collection of tricks.” Loosely related collections of tricks do not lend themselves to systematic expansion, and they are difficult to learn and to teach in great number.

Effective

An effective application of behavioral techniques must improve the behavior under investigation to a practical degree. “In application, the theoretical importance of a

Definition and Characteristics of Applied Behavior Analysis

variable is usually not at issue. Its practical importance, specifically its power in altering behavior enough to be socially important, is the essential criterion” (Baer et al., 1968, p. 96). Whereas some investigations produce re- sults of theoretical importance or statistical significance, to be judged effective an applied behavior analysis study must produce behavior changes that reach clinical or so- cial significance.

How much a given behavior of a given subject needs to change for the improvement to be considered socially important is a practical question. Baer and colleagues stated that the answer is most likely to come from the people who must deal with the behavior; they should be asked how much the behavior needs to change. The ne- cessity of producing behavioral changes that are mean- ingful to the participant and/or those in the participant’s environment has pushed behavior analysts to search for “robust” variables, interventions that produce large and consistent effects on behavior (Baer, 1977a).

When they revisited the dimension of effectiveness 20 years later, Baer, Wolf, and Risley (1987) recom- mended that the effectiveness of ABA also be judged by a second kind of outcome: the extent to which changes in the target behaviors result in noticeable changes in the reasons those behaviors were selected for change origi- nally. If such changes in the subjects’ lives do not occur, ABA may achieve one level of effectiveness yet fail to achieve a critical form of social validity (Wolf, 1978).

We may have taught many social skills without examin- ing whether they actually furthered the subject’s social life; many courtesy skills without examining whether anyone actually noticed or cared; many safety skills without examining whether the subject was actually safer thereafter; many language skills without measuring whether the subject actually used them to interact differ- ently than before; many on-task skills without measur- ing the actual value of those tasks; and, in general, many survival skills without examining the subject’s actual subsequent survival. (Baer et al., 1987, p. 322)

Generality

A behavior change has generality if it lasts over time, ap- pears in environments other than the one in which the in- tervention that initially produced it was implemented, and/or spreads to other behaviors not directly treated by the intervention. A behavior change that continues after the original treatment procedures are withdrawn has gen- erality. And generality is evident when changes in tar- geted behavior occur in nontreatment settings or situations as a function of treatment procedures. Gener- ality also exists when behaviors change that were not the focus of the intervention. Although not all instances of generality are adaptive (e.g., a beginning reader who has

just learned to make the sound for the letter p in words such as pet and ripe, might make the same sound when seeing the letter p in the word phone), desirable general- ized behavior changes are important outcomes of an ap- plied behavior analysis program because they represent additional dividends in terms of behavioral improvement.

Some Additional Characteristics of ABA

Applied behavior analysis offers society an approach to- ward solving problems that is accountable, public, doable, empowering, and optimistic (Heward, 2005). Although not among ABA’s defining dimensions, these character- istics should help increase the extent to which decision makers and consumers in many areas look to behavior analysis as a valuable and important source of knowl- edge for achieving improvements.

Accountable

The commitment of applied behavior analysts to effec- tiveness, their focus on accessible environmental vari- ables that reliably influence behavior, and their reliance on direct and frequent measurement to detect changes in behavior yield an inescapable and socially valuable form of accountability. Direct and frequent measurement—the foundation and most important component of ABA prac- tices—enables behavior analysts to detect their successes and, equally important, their failures so they can make changes in an effort to change failure to success (Bushell & Bear, 1994; Greenwood & Maheady, 1997).

Failure is always informative in the logic of behavior analysis, just as it is in engineering. The constant reac- tion to lack of progress [is] a definitive hallmark of ABA. (Baer, 2005, p. 8)

Gambrill (2003) described the sense of accountabil- ity and self-correcting nature of applied behavior analy- sis very well.

Applied behavior analysis is a scientific approach to un- derstanding behavior in which we guess and critically test ideas, rather than guess and guess again. It is a process for solving problems in which we learn from our mistakes. Here, false knowledge and inert knowl- edge are not valued. (p. 67)

Public

“Everything about ABA is visible and public, explicit and straightforward. . . . ABA entails no ephemeral, mysti- cal, or metaphysical explanations; there are no hidden

Definition and Characteristics of Applied Behavior Analysis

treatments; there is no magic” (Heward, 2005, p. 322). The transparent, public nature of ABA should raise its value in fields such as education, parenting and child care, employee productivity, geriatrics, health and safety, and social work—to name only a few—whose goals, meth- ods, and outcomes are of vital interest to many con- stituencies.

Doable

Classroom teachers, parents, coaches, workplace super- visors, and sometimes the participants themselves im- plemented the interventions found effective in many ABA studies. This demonstrates the pragmatic element of ABA. “Although ‘doing ABA’ requires far more than learning to administer a few simple procedures, it is not prohibitively complicated or arduous. As many teachers have noted, implementing behavioral strategies in the classroom . . . might best be described as good old- fashioned hard work” (Heward, 2005, p. 322).

Empowering

ABA gives practitioners real tools that work. Knowing how to do something and having the tools to accomplish it instills confidence in practitioners. Seeing the data showing behavioral improvements in one’s clients, stu- dents, or teammates, or in oneself, not only feels good, but also raises one’s confidence level in assuming even more difficult challenges in the future.

Optimistic

Practitioners knowledgeable and skilled in behavior analysis have genuine cause to be optimistic for four rea- sons. First, as Strain and Joseph (2004) noted:

The environmental view promoted by behaviorism is es- sentially optimistic; it suggests that (except for gross ge- netic factors) all individuals possess roughly equal potential. Rather than assuming that individuals have some essential internal characteristic, behaviorists as- sume that poor outcomes originate in the way the envi- ronment and experience shaped the individual’s current behavior. Once these environmental and experiential factors are identified, we can design prevention and in- tervention programs to improve the outcomes. . . . Thus, the emphasis on external control in the behavioral ap- proach . . . offers a conceptual model that celebrates the possibilities for each individual. (Strain et al., 1992, p. 58)

Second, direct and continuous measurement enables practitioners to detect small improvements in perfor- mance that might otherwise be overlooked. Third, the

more often a practitioner uses behavioral tactics with pos- itive outcomes (the most common result of behaviorally based interventions), the more optimistic she becomes about the prospects for future success.

A sense of optimism, expressed by the question “Why not?” has been a central part of ABA and has had an enormous impact on its development from its earliest days. Why can’t we teach a person who does not yet talk to talk? Why shouldn’t we go ahead and try to change the environments of young children so that they will dis- play more creativity? Why would we assume that this person with a developmental disability could not learn to do the same things that many of us do? Why not try to do it? (Heward, 2005, p. 323)

Fourth, ABA’s peer-reviewed literature provides many examples of success in teaching students who had been considered unteachable. ABA’s continuous record of achievements evokes a legitimate feeling of optimism that future developments will yield solutions to behav- ioral challenges that are currently beyond the existing technology. For example, in response to the perspective that some people have disabilities so severe and profound that they should be viewed as ineducable, Don Baer of- fered this perspective:

Some of us have ignored both the thesis that all persons are educable and the thesis that some persons are inedu- cable, and instead have experimented with ways to teach some previously unteachable people. Those experiments have steadily reduced the size of the apparently inedu- cable group relative to the obviously educable group. Clearly, we have not finished that adventure. Why pre- dict its outcome, when we could simply pursue it, and just as well without a prediction? Why not pursue it to see if there comes a day when there is such a small class of apparently ineducable persons left that it consists of one elderly person who is put forward as ineducable. If that day comes, it will be a very nice day. And the next day will be even better. (D. M. Baer, February 15, 2002, personal communication, as cited in Heward, 2006, p. 473)

Definition of Applied Behavior Analysis We began this chapter by stating that applied behavior analysis is concerned with the improvement and under- standing of human behavior. We then described some of the attitudes and methods that are fundamental to scien- tific inquiry, briefly reviewed the development of the sci- ence and philosophy of behavior analysis, and examined the characteristics of ABA. All of that provided neces- sary context for the following definition of applied be- havior analysis:

Definition and Characteristics of Applied Behavior Analysis

Applied behavior analysis is the science in which tactics derived from the principles of behavior are applied systematically to improve socially signifi- cant behavior and experimentation is used to iden- tify the variables responsible for behavior change.

This definition includes six key components. First, the practice of applied behavior analysis is guided by the attitudes and methods of scientific inquiry. Second, all behavior change procedures are described and imple- mented in a systematic, technological manner. Third, not any means of changing behavior qualifies as applied behavior analysis: Only those procedures conceptually derived from the basic principles of behavior are cir- cumscribed by the field. Fourth, the focus of applied be- havior analysis is socially significant behavior. The fifth and sixth parts of the definition specify the twin goals of applied behavior analysis: improvement and under- standing. Applied behavior analysis seeks to make mean- ingful improvement in important behavior and to produce an analysis of the factors responsible for that improve- ment.

Four Interrelated Domains of Behavior Analytic Science and Professional Practices Guided by That Science

The science of behavior analysis and its application to human problems consists of four domains: the three branches of behavior analysis—behaviorism, EAB, and ABA—and professional practice in various fields that is informed and guided by that science. Figure 2 identi- fies some of the defining features and characteristics of these four interrelated domains. Although most behavior analysts work primarily in one or two of the domains shown in Figure 2, it is common for a behavior analyst to function in multiple domains at one time or another (Hawkins & Anderson, 2002; Moore & Cooper, 2003).

A behavior analyst who pursues theoretical and con- ceptual issues is engaged in behaviorism, the philosoph- ical domain of behavior analysis. A product of such work is Delprato’s (2002) discussion of the importance of countercontrol (behavior by people experiencing aver- sive control by others that helps them escape and avoid the control while not reinforcing and sometimes punish- ing the controller’s responses) toward an understanding of effective interventions for interpersonal relations and cultural design.

The experimental analysis of behavior is the basic research branch of the science. Basic research consists of experiments in laboratory settings with both human and nonhuman subjects with a goal of discovering and

9Schedules of reinforcement are discussed in chapter entitled “Schedules of Reinforcement.” 10Examples and discussions of bridge or translational research can be found in the winter 1994 and winter 2003 issues of JABA; Fisher and Mazur (1997); Lattal and Neef (1996); and Vollmer and Hackenberg (2001).

clarifying fundamental principles of behavior. An exam- ple is Hackenberg and Axtell’s (1993) experiments in- vestigating how choices made by humans are affected by the dynamic interaction of schedules of reinforcement that entail short- and long-term consequences.9

Applied behavior analysts conduct experiments aimed at discovering and clarifying functional relations between socially significant behavior and its controlling variables, with which they can contribute to the further development of a humane and effective technology of behavior change. An example is research by Tarbox, Wal- lace, and Williams (2003) on the assessment and treat- ment of elopement (running or walking away from a caregiver without permission), a behavior that poses great danger for young children and people with disabilities.

The delivery of behavior analytic professional ser- vices occurs in the fourth domain. Behavior analysis prac- titioners design, implement, and evaluate behavior change programs that consist of behavior change tactics derived from fundamental principles of behavior discovered by basic researchers, and that have been experimentally val- idated for their effects on socially significant behavior by applied researchers. An example is when a therapist providing home-based treatment for a child with autism embeds frequent opportunities for the child to use his emerging social and language skills in the context of nat- uralistic, daily routines and ensures that the child’s re- sponses are followed with reinforcing events. Another example is a classroom teacher trained in behavior analy- sis who uses positive reinforcement and stimulus fading to teach students to identify and classify fish into their respective species by the shape, size, and location of their fins.

Although each of the four domains of ABA can be defined and practiced in its own right, none of the do- mains are, or should be, completely independent of and uninformed by developments in the others. Both the sci- ence and the application of its findings benefit when the four domains are interrelated and influence one another (cf., Critchfield & Kollins, 2001; Lattal & Neef; 1996; Stromer, McComas, & Rehfeldt, 2000). Evidence of the symbiotic relations between the basic and applied do- mains is evident in research that “bridges” basic and ap- plied areas and in applied research that translates the knowledge derived from basic research “into state-of- the-art clinical practices for use in the community” (Ler- man, 2003, p. 415).10

Definition and Characteristics of Applied Behavior Analysis

Figure 2 Some comparisons and relationships among the four domains of behavior analysis science and practice.

Behaviorism Experimental Analysis Applied Behavior Practice Guided of Behavior (EAB) Analysis (ABA) by Behavior Analysis

The Science of Behavior Analysis

The Application of Behavior Analysis

Province Theory and Basic research Applied research Helping people behave more philosophy successfully

Primary activity Conceptual and Design, conduct, Design, conduct, Design, implement, and philosophical interpret, and report interpret, and report evaluate behavior change analysis basic experiments applied experiments programs

Primary goal Theoretical Discover and clarify A technology for Improvements in the lives and product account of all basic principles of improving socially of participants/clients as a

behavior consistent behavior; functional significant behavior; result of changes in their with existing data relations between functional relations behavior

behavior and control- between socially ling variables significant behavior

and controlling variables

Secondary goals Identify areas in Identify questions for Identify questions for Increased efficiency in which empirical EAB and/or ABA to EAB and/or ABA to achieving primary goal; may data are absent investigate further; investigate further; identify questions for ABA and/or conflict and raise theoretical raise theoretical and EAB suggest resolutions issues issues

Agreement with As much as Complete—Although Complete—Although As much as possible, but existing database possible, but theory differences among differences among practitioners must often deal

must go beyond data sets exist, EAB data sets exist, ABA with situations not covered database by provides the basic provides the applied by existing data design research database research database

Testability Partially—All Mostly—Technical Mostly—Same Partially—All behavior and behavior and limitations preclude limitations as EAB variables of interest are not variables of interest measurement and plus those posed by accessible (e.g., a student’s are not accessible experimental applied settings (e.g., home life) (e.g., phylogenic manipulation of some ethical concerns, contingencies) variables uncontrolled events)

Scope Most Least Wide scope As much scope as As much scope as Narrow scope because because theory the EAB database the ABA database practitioner’s primary focus is attempts to enables enables helping the specific situation account for all behavior

Precision Least Most Minimal precision As much precision as As much precision Maximum precision is sought is possible because EAB’s current tech- as ABA’s current to change behavior most ef- experimental data nology for experi- technology for fectively in specific instance do not exist for all mental control and experimental control behavior encom- the researcher’s skills and the researcher’s passed by theory enable skills enable

▲ ▲

Definition and Characteristics of Applied Behavior Analysis

The Promise and Potential of ABA

In a paper titled, “A Futuristic Perspective for Applied Behavior Analysis,” Jon Bailey (2000) stated that

It seems to me that applied behavior analysis is more relevant than ever before and that it offers our citizens, parents, teachers, and corporate and government leaders advantages that cannot be matched by any other psycho- logical approach. . . . I know of no other approach in psychology that can boast state-of-the-art solutions to the most troubling social ills of the day. (p. 477)

We, too, believe that ABA’s pragmatic, natural sci- ence approach to discovering environmental variables that reliably influence socially significant behavior and to developing a technology to take practical advantage of those discoveries offers humankind its best hope for solv- ing many of its problems. It is important, however, to rec- ognize that behavior analysis’s knowledge of “how behavior works,” even at the level of fundamental prin- ciples, is incomplete, as is the technology for changing socially significant behavior derived from those princi- ples. There are aspects of about which relatively little is known, and additional research, both basic and applied, is needed to clarify, extend, and fine-tune all existing knowledge (e.g., Critchfield & Kollins, 2001; Friman, Hayes, & Wilson, 1998; Murphy, McSweeny, Smith, & McComas, 2003; Stromer, McComas, & Rehfeldt, 2000).

Nevertheless, the still young science of applied be- havior analysis has contributed to a full range of areas in which human behavior is important. Even an informal, cursory survey of the research published in applied be- havior analysis reveals studies investigating virtually the full range of socially significant human behavior from A to Z and almost everywhere in between: AIDS prevention (e.g., DeVries, Burnette, & Redmona, 1991), conserva- tion of natural resources (e.g., Brothers, Krantz, & Mc- Clannahan, 1994), education (e.g., Heward et al., 2005), gerontology (e.g., Gallagher & Keenan, 2000), health and exercise (e.g., De Luca & Holborn, 1992), industrial safety (e.g., Fox, Hopkins, & Anger, 1987), language ac- quisition (e.g., Drasgow, Halle, & Ostrosky, 1998), lit- tering (e.g., Powers, Osborne, & Anderson, 1973), medical procedures (e.g., Hagopian & Thompson, 1999), parenting (e.g., Kuhn, Lerman, & Vorndran, 2003), seat- belt use (e.g., Van Houten, Malenfant, Austin, & Lebbon, 2005), sports (e.g., Brobst & Ward, 2002), and zoo man- agement and care of animals (e.g., Forthman & Ogden, 1992).

Applied behavior analysis provides an empirical basis for not only understanding human behavior but also improving it. Equally important, ABA continually tests and evaluates its methods.

Summary Some Basic Characteristics and a Definition of Science

1. Different types of scientific investigations yield knowl- edge that enables the description, prediction, and/or con- trol of the phenomena studied.

2. Descriptive studies yield a collection of facts about the observed events that can be quantified, classified, and ex- amined for possible relations with other known facts.

3. Knowledge gained from a study that finds the systematic covariation between two events—termed a correlation— can be used to predict the probability that one event will occur based on the occurrence of the other event.

4. Results of experiments that show that specific manipula- tions of one event (the independent variable) produce a re- liable change in another event (the dependent variable), and that the change in the dependent variable was unlikely the result of extraneous factors (confounding variables)— a finding known as a functional relation—can be used to control the phenomena under investigation.

5. The behavior of scientists in all fields is characterized by a common set of assumptions and attitudes:

• Determinism—the assumption that the universe is a law- ful and orderly place in which phenomena occur as a result of other events.

• Empiricism—the objective observation of the phenom- ena of interest.

• Experimentation—the controlled comparison of some measure of the phenomenon of interest (the dependent variable) under two or more different conditions in which only one factor at a time (the independent vari- able) differs from one condition to another.

• Replication—repeating experiments (and independent variable conditions within experiments) to determine the reliability and usefulness of findings

• Parsimony—simple, logical explanations must be ruled out, experimentally or conceptually, before more com- plex or abstract explanations are considered.

• Philosophic doubt—continually questioning the truth- fulness and validity of all scientific theory and knowledge.

A Brief History of the Development of Behavior Analysis

6. Behavior analysis consists of three major branches: be- haviorism, the experimental analysis of behavior (EAB), and applied behavior analysis (ABA).

7. Watson espoused an early form of behaviorism known as stimulus–response (S–R) psychology, which did not ac- count for behavior without obvious antecedent causes.

Definition and Characteristics of Applied Behavior Analysis

8. Skinner founded the experimental analysis of behavior (EAB), a natural science approach for discovering orderly and reliable relations between behavior and various types of environmental variables of which it is a function.

9. EAB is characterized by these methodological features:

• Rate of response is the most common dependent variable.

• Repeated or continuous measurement is made of care- fully defined response classes.

• Within-subject experimental comparisons are used in- stead of designs comparing the behavior of experimen- tal and control groups.

• The visual analysis of graphed data is preferred over statistical inference.

• A description of functional relations is valued over for- mal theory testing.

10. Through thousands of laboratory experiments, Skinner and his colleagues and students discovered and verified the basic principles of operant behavior that provide the empirical foundation for behavior analysis today.

11. Skinner wrote extensively about a philosophy for a sci- ence of behavior he called radical behaviorism. Radical behaviorism attempts to explain all behavior, including private events such as thinking and feeling.

12. Methodological behaviorism is a philosophical position that considers behavioral events that cannot be publicly observed to be outside the realm of the science.

13. Mentalism is an approach to understanding behavior that assumes that a mental, or “inner,” dimension exists that differs from a behavioral dimension and that phenomena in this dimension either directly cause or at least mediate some forms of behavior; it relies on hypothetical constructs and explanatory fictions.

14. The first published report of the application of operant conditioning with a human subject was a study by Fuller (1949), in which an arm-raising response was conditioned in an adolescent with profound retardation.

15. The formal beginnings of applied behavior analysis can be traced to 1959 and the publication of Ayllon and Michael’s article, ”The Psychiatric Nurse as a Behavioral Engineer.”

16. Contemporary applied behavior analysis (ABA) began in 1968 with the publication of the first issue of the Journal of Applied Behavior Analysis (JABA).

Defining Characteristics of Applied Behavior Analysis

17. Baer, Wolf, and Risley (1968, 1987) stated that a research study or behavior change program should meet seven defining dimensions to be considered applied behavior analysis:

• Applied—investigates socially significant behaviors with immediate importance to the subject(s).

• Behavioral—entails precise measurement of the actual behavior in need of improvement and documents that it was the subject’s behavior that changed.

• Analytic—demonstrates experimental control over the occurrence and nonoccurrence of the behavior—that is, if a functional relation is demonstrated.

• Technological—the written description of all procedures used in the study is sufficiently complete and detailed to enable others to replicate it.

• Conceptually systematic—behavior change interven- tions are derived from basic principles of behavior.

• Effective—improves behavior sufficiently to produce practical results for the participant/client.

• Generality—produces behavior changes that last over time, appear in other environments, or spread to other behaviors.

18. ABA offers society an approach toward solving many of its problems that is accountable, public, doable, empow- ering, and optimistic.

A Definition of Applied Behavior Analysis

19. Applied behavior analysis is the science in which tactics derived from the principles of behavior are applied sys- tematically to improve socially significant behavior and experimentation is used to identify the variables respon- sible for behavior change.

20. Behavior analysts work in one or more of four interrelated domains: behaviorism (theoretical and philosophical is- sues), the experimental analysis of behavior (basic re- search), applied behavior analysis (applied research), and professional practice (providing behavior analytic services to consumers).

21. ABA’s natural science approach to discovering environ- mental variables that reliably influence socially signifi- cant behavior and developing a technology to take practical advantage of those discoveries offers humankind its best hope for solving many of its problems.

Definition and Characteristics of Applied Behavior Analysis

antecedent automaticity of reinforcement aversive stimulus behavior behavior change tactic conditioned punisher conditioned reflex conditioned reinforcer conditioned stimulus consequence contingency contingent deprivation discriminated operant discriminative stimulus (SD) environment extinction

habituation higher order conditioning history of reinforcement motivating operation negative reinforcement neutral stimulus ontogeny operant behavior operant conditioning phylogeny positive reinforcement principle of behavior punisher punishment reflex reinforcement reinforcer

repertoire respondent behavior respondent conditioning respondent extinction response response class satiation selection by consequences stimulus stimulus class stimulus control stimulus–stimulus pairing three-term contingency unconditioned punisher unconditioned reinforcer unconditioned stimulus

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes, and Concepts

3-1 Define and provide examples of behavior/response/response class.

3-2 Define and provide examples of stimulus and stimulus class.

3-3 Define and provide examples of positive and negative reinforcement.

3-4 Define and provide examples of conditioned and unconditioned reinforcement.

3-5 Define and provide examples of positive and negative punishment.

3-6 Define and provide examples of conditioned and unconditioned punishment.

3-7 Define and provide examples of stimulus control.

3-8 Define and provide examples of establishing operations.

3-9 Define and provide examples of behavioral contingencies.

(continued )

Basic Concepts

Key Terms

This chapter defines the basic elements involved in a scientific analysis of behavior and intro- duces several principles that have been discov-

ered through such an analysis. The first concept we examine—behavior—is the most fundamental of all. Be- cause the controlling variables of primary importance in applied behavior analysis are located in the environment, the concepts of environment and stimulus are defined next. We then introduce several essential findings that the scientific study of behavior–environment relations has discovered. Two functionally distinct types of behavior— respondent and operant—are described, and the basic ways the environment influences each type of behavior— respondent conditioning and operant conditioning—are introduced. The three-term contingency—a concept for expressing and organizing the temporal and functional relations between operant behavior and environment— and its importance as a focal point in applied behavior analysis are then explained.1 The chapter’s final section recognizes the incredible complexity of human behav- ior, reminds us that behavior analysts possess an incom- plete knowledge, and identifies some of the obstacles and challenges faced by those who strive to change behavior in applied settings.

Behavior What, exactly, is behavior? Behavior is the activity of liv- ing organisms. Human behavior is everything people do, including how they move and what they say, think, and feel. Tearing open a bag of peanuts is behavior, and so is thinking how good the peanuts will taste once the bag is open. Reading this sentence is behavior, and if you’re holding the book, so is feeling its weight and shape in your hands.

Although words such as activity and movement ade- quately communicate the general notion of behavior, a more precise definition is needed for scientific purposes. How a scientific discipline defines its subject matter ex- erts profound influence on the methods of measurement, experimentation, and theoretical analysis that are appro- priate and possible.

Building on Skinner’s (1938) definition of behavior as “the movement of an organism or of its parts in a frame of reference provided by the organism or by various ex- ternal objects or fields” (p. 6), Johnston and Pennypacker (1980, 1993a) articulated the most conceptually sound and empirically complete definition of behavior to date.

The behavior of an organism is that portion of an organ- ism’s interaction with its environment that is character- ized by detectable displacement in space through time of some part of the organism and that results in a mea- surable change in at least one aspect of the environ- ment. (p. 23)

Johnston and Pennypacker (1993a) discussed the major elements of each part of this definition. The phrase behavior of an organism restricts the subject matter to the activity of living organisms, leaving notions such as the “behavior” of the stock market outside the realm of the scientific use of the term.

The phrase portion of the organism’s interaction with the environment specifies “the necessary and sufficient conditions for the occurrence of behavior as (a) the exis- tence of two separate entities, organism and environment, and (b) the existence of a relation between them” (John- ston & Pennypacker, 1993a, p. 24). The authors elabo- rated on this part of the definition as follows:

Behavior is not a property or attribute of the organism. It happens only when there is an interactive condition be- tween an organism and its surroundings, which include its own body. This means that independent states of the organism, whether real or hypothetical, are not behav- ioral events, because there is no interactive process. Being hungry or being anxious are examples of states that are sometimes confused with the behavior that they are supposed to explain. Neither phrase specifies an en- vironmental agent with which the hungry or anxious or- ganism interacts, so no behavior is implied.

Basic Concepts

1The reader should not be overwhelmed by the many technical terms and concepts contained in this chapter. With the exception of the material on respondent behavior, all of the concepts introduced in this chapter are ex- plained in greater detail in subsequent chapters. This initial overview of basic concepts is intended to provide background information that will fa- cilitate understanding those portions of the text that precede the more de- tailed explanations.

Content Area 3: Principles, Processes, and Concepts (continued )

3-13 Describe and provide examples of the respondent conditioning paradigm.

3-14 Describe and provide examples of the operant conditioning paradigm.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and ques- tions about this document must be submitted directly to the BACB.

Basic Concepts

Similarly, independent conditions or changes in the environment do not define behavioral occurrences be- cause no interaction is specified. Someone walking in the rain gets wet, but “getting wet” is not an instance of behavior. A child may receive tokens for correctly work- ing math problems, but “receiving a token” is not be- havior. Receiving a token implies changes in the environment but does not suggest or require change in the child’s movement. In contrast, both doing math problems and putting the token in a pocket are behav- ioral events because the environment both prompts the child’s actions and is then changed by them. (Johnston & Pennypacker, 1993a, p. 24)

Behavior is movement, regardless of scale; hence the phrase displacement in space through time. In addition to excluding static states of the organism, the definition does not include bodily movements produced by the ac- tion of independent physical forces as behavioral events. For example, being blown over by a strong gust of wind is not behavior; given sufficient wind, nonliving objects and organisms move similarly. Behavior can be accom- plished only by living organisms. A useful way to tell whether movement is behavior is to apply the dead man test: “If a dead man can do it, it ain’t behavior. And if a dead man can’t do, then it is behavior” (Malott & Trojan Suarez, 2004, p. 9). So, although being knocked down by strong wind is not behavior (a dead man would also be blown over), moving arms and hands in front of one’s face, tucking and rolling, and yelling “Whoa!” as one is being blown over are behaviors.2

The displacement in space through time phrase also highlights the properties of behavior most amenable to measurement. Johnston and Pennypacker (1993a) re- ferred to these fundamental properties by which behav- ior can be measured as temporal locus (when in time a specified behavior occurs), temporal extent (the duration of a given behavioral event), and repeatability (the fre- quency with which a specified behavior occurs over time).

Acknowledging that the last phrase of the defini- tion—that results in a measurable change in at least one aspect of the environment—is somewhat redundant, John- ston and Pennypacker (1993a) noted that it emphasizes an important qualifier for the scientific study of behavior.

Because the organism cannot be separated from an envi- ronment and because behavior is the relation between organism and environment, it is impossible for a behav- ioral event not to influence the environment in some way. . . . This is an important methodological point be-

2Odgen Lindsley originated the dead man test in the mid-1960s as a way to help teachers determine whether they were targeting real behaviors for measurement and change as opposed to inanimate states such as “being quiet.”

3Most behavior analysts use the word behavior both as a mass noun to refer to the subject matter of the field in general or a certain type or class of behavior (e.g., operant behavior, study behavior) and as a count noun to refer to specific instances (e.g., two aggressive behaviors). The word behavior is often implied and unnecessary to state. We agree with Friman’s (2004) recommendation that, “If the object of our interest is hitting and spitting, let’s just say ‘hitting’ and ‘spitting.’ Subsequently, when we are gathering our thoughts with a collective term, we can call them behaviors” (p. 105).

cause it says that behavior must be detected and mea- sured in terms of its effects on the environment. (p. 27)

As Skinner (1969) wrote, “To be observed, a re- sponse must affect the environment—it must have an ef- fect upon an observer or upon an instrument which in turn can affect an observer. This is as true of the con- traction of a small group of muscle fibers as of pressing a lever or pacing a figure 8” (p. 130).

The word behavior is usually used in reference to a larger set or class of responses that share certain physi- cal dimensions (e.g., hand-flapping behavior) or func- tions (e.g., study behavior).3 The term response refers to a specific instance of behavior. A good technical defini- tion of response is an “action of an organism’s effector. An effector is an organ at the end of an efferent nerve fiber that is specialized for altering its environment me- chanically, chemically, or in terms of other energy changes” (Michael, 2004, p. 8, italics in original). Human effectors include the striped muscles (i.e., skeletal mus- cles such as biceps and quadriceps), smooth muscles (e.g., stomach and bladder muscles), and glands (e.g., adrenal gland).

Like stimulus changes in the environment, behavior can be described by its form, or physical characteristics. Response topography refers to the physical shape or form of behavior. For example, the hand and finger movements used to open a bag of peanuts can be described by their topographical elements. However, careful observation will reveal that the topography differs somewhat each time a person opens a bag of snacks. The difference may be significant or slight, but each “bag opening response” will vary somewhat from all others.

Although it is sometimes useful to describe behavior by its topography, behavior analysis is characterized by a functional analysis of the effects of behavior on the en- vironment. A group of responses with the same function (that is, each response in the group produces the same effect on the environment) is called a response class. Membership in some response classes is open to re- sponses of widely varying form (e.g., there are many ways to open a bag of peanuts), whereas the topograph- ical variation among members of other response classes is limited (e.g., a person’s signature, grip on a golf club).

Another reason underscoring the importance of a functional analysis of behavior over a structural or

Basic Concepts

topographical description is that two responses of the same topography can be vastly different behaviors de- pending on the controlling variables. For example, say- ing the word fire while looking the letters, f-i-r-e, is a vastly different behavior from yelling “Fire!” when smelling smoke or seeing flames in a crowded theatre.

Behavior analysts use the term repertoire in at least two ways. Repertoire is sometimes used to refer to all of the behaviors that a person can do. More often the term denotes a set or collection of knowledge and skills a per- son has learned that are relevant to particular settings or tasks. In the latter sense, each person has acquired or learned multiple repertoires. For example, each of us has a repertoire of behaviors appropriate for informal social situations that differs somewhat (or a lot) from the be- haviors we use to navigate formal situations. And each person has repertoires with respect to language skills, academic tasks, everyday routines, recreation, and so on. When you complete your study of this text, your reper- toire of knowledge and skills in applied behavior analy- sis will be enriched.

Environment All behavior occurs within an environmental context; be- havior cannot be emitted in an environmental void or vac- uum. Johnston and Pennypacker (1993a) offered the following definition of environment and two critical im- plications of that definition for a science of behavior:

“Environment” refers to the conglomerate of real circum- stances in which the organism or referenced part of the organism exists. A simple way to summarize its coverage is as “everything except the moving parts of the organism involved in the behavior.” One important implication . . . is that only real physical events are included.

Another very important consequence of this con- ception of the behaviorally relevant environment is that it can include other aspects of the organism. That is, the environment for a particular behavior can in- clude not only the organism’s external features but physical events inside its skin. For instance, scratch- ing our skin is presumably under control of the exter- nal visual stimulus provided by your body, particularly that part being scratched, as well as the stimulation that we call itching, which lies inside the skin. In fact, both types of stimulation very often contribute to be- havioral control. This means that the skin is not an especially important boundary in the understanding of behavioral laws, although it can certainly provide observational challenges to discovering those laws. (p. 28)

The environment is a complex, dynamic universe of events that differs from instance to instance. When be-

4Although the concepts of stimulus and response have proven useful for conceptual, experimental, and applied analyses of behavior, it is important to recognize that stimuli and responses do not exist as discrete events in na- ture. Stimuli and responses are detectable “slices” of the continuous and ever-changing interaction between an organism and its environment cho- sen by scientists and practitioners because they have proven useful in un- derstanding and changing behavior. However, the slices imposed by the behavior analyst may not parallel naturally occurring divisions. 5Respondent conditioning and the operant principles mentioned here are in- troduced later in this chapter.

havior analysts describe particular aspects of the envi- ronment, they talk in terms of stimulus conditions or events.4 A good definition of stimulus is “an energy change that affects an organism through its receptor cells” (Michael, 2004, p. 7). Humans have receptor sys- tems that detect stimulus changes occurring outside and inside the body. Exteroceptors are sense organs that de- tect external stimuli and enable vision, hearing, olfac- tion, taste, and cutaneous touch. Two types of sense organs sensitive to stimulus changes within the body are interoceptors, which are sensitive to stimuli originating in the viscera (e.g., feeling a stomach ache), and pro- prioceptors, which enable the kinesthetic and vestibu- lar senses of movement and balance. Applied behavior analysts most often study the effects of stimulus changes that occur outside the body. External stimulus conditions and events are not only more accessible to observation and manipulation than are internal conditions, but also they are key features of the physical and social world in which people live.

The environment influences behavior primarily by stimulus change and not static stimulus conditions. As Michael (2004) noted, when behavior analysts speak of the presentation or occurrence of a stimulus, they usu- ally mean stimulus change.

For example, in respondent conditioning the conditioned stimulus may be referred to as a tone. However, the rele- vant event is actually a change from the absence of tone to the tone sounding . . . , and although this is usually understood without having to be mentioned, it can be overlooked in the analysis of more complex phenomena. Operant discriminative stimuli, conditioned reinforcers, conditioned punishers, and conditioned motivative vari- ables are also usually important as stimulus changes, not static conditions (Michael, 2004, pp. 7–8).5

Stimulus events can be described formally (by their physical features), temporally (by when they occur with respect to a behavior of interest), and functionally (by their effects on behavior). Behavior analysts used the term stimulus class to refer to any group of stimuli sharing a predetermined set of common elements in one or more of these dimensions.

Basic Concepts

Formal Dimensions of Stimuli

Behavior analysts often describe, measure, and manipu- late stimuli according to their formal dimensions, such as size, color, intensity, weight, and spatial position rel- ative to other objects. Stimuli can be nonsocial (e.g., a red light, a high-pitched sound) or social (e.g., a friend asking, “Want some more peanuts?”).

Temporal Loci of Stimuli

Because behavior and the environmental conditions that influence it occur within and across time, the temporal lo- cation of stimulus changes is important. In particular, be- havior is affected by stimulus changes that occur prior to and immediately after the behavior. The term antecedent refers to environmental conditions or stimulus changes that exist or occur prior to the behavior of interest.

Because behavior cannot occur in an environmental void or vacuum, every response takes place in the context of a particular situation or set of antecedent conditions. These antecedent events play a critical part in learning and motivation, and they do so irrespective of whether the learner or someone in the role of behavior analyst or teacher has planned or is even aware of them.

For example, just some of the functionally relevant an- tecedents for a student’s performance on a timed math test might include the following: the amount of sleep the student had the night before; the temperature, lighting, and seating arrangements in the classroom; the teacher reminding the class that students who beat their personal best scores on the test will get a free homework pass; and the specific type, format, and sequence of math problems on the test. Each of those antecedent variables (and others) has the potential to exert a great deal, a lit- tle, or no noticeable effect on performance as a function of the student’s experiences with respect to a particular antecedent. (Heward & Silvestri, 2005, p. 1135)

A consequence is a stimulus change that follows a behavior of interest. Some consequences, especially those

that are immediate and relevant to current motivational states, have significant influence on future behavior; other consequences have little effect. Consequences combine with antecedent conditions to determine what is learned. Again, this is true whether the individual or someone try- ing to change his behavior is aware of or systematically plans the consequences.

Like antecedent stimulus events, consequences may also be social or nonsocial events. Table 1 shows ex- amples of various combinations of social and nonsocial antecedent and consequent events for four behaviors.

Behavioral Functions of Stimulus Changes

Some stimulus changes exert immediate and powerful control over behavior, whereas others have delayed ef- fects, or no apparent effect. Even though we can and often do describe stimuli by their physical characteristics (e.g., the pitch and decibel level of a tone, the topography of a person’s hand and arm movements), stimulus changes are understood best through a functional analysis of their effects on behavior. For example, the same decibel tone that functions in one environment and set of conditions as a prompt for checking the clothes in the dryer may function as a warning signal to fasten a seat belt in an- other setting or situation; the same hand and arm motion that produces a smile and a “Hi” from another person in one set of conditions receives a scowl and obscene ges- ture in another.

Stimulus changes can have one or both of two basic kinds of functions or effects on behavior: (a) an imme- diate but temporary effect of increasing or decreasing the current frequency of the behavior, and/or (b) a delayed but relatively permanent effect in terms of the frequency of that type of behavior in the future (Michael, 1995). For example, a sudden downpour on a cloudy day is likely to increase immediately the frequency of all behavior that has resulted in the person successfully escaping rain in the

Table 1 Antecedent (Situation) and Consequent Events Can Be Nonsocial (Italicized), Social (Boldface), or a Combination of Social and Nonsocial

Situation Response Consequence

Drink machine Deposit coins Cold drink

Five cups on table “One-two-three-four-five cups” Teacher nods and smiles

Friend says “turn left” Turn left Arrive at destination

Friend asks “What time is it?” “Six-fifteen” Friend says “Thanks”

Basic Concepts

past, such as running for cover under an awning or pulling her jacket over her head. If the person had decided not to carry her umbrella just before leaving the house, the downpour may decrease the frequency of that behavior on cloudy days in the future.

Respondent Behavior All intact organisms enter the world able to respond in predictable ways to certain stimuli; no learning is re- quired. These ready-made behaviors protect against harmful stimuli (e.g., eyes watering and blinking to re- move particles on the cornea), help regulate the internal balance and economy of the organism (e.g., changes in heart rate and respiration in response to changes in tem- perature and activity levels), and promote reproduction (e.g., sexual arousal). Each of these stimulus–response relations, called a reflex, is part of the organism’s genetic endowment, a product of natural evolution because of its

survival value to the species. Each member of a given species comes equipped with the same repertoire of un- conditioned (or unlearned) reflexes. Reflexes provide the organism with a set of built-in responses to specific stim- uli; these are behaviors the individual organism would not have time to learn. Table 2 shows examples of re- flexes common to humans.

The response component of the stimulus–response reflex is called respondent behavior. Respondent be- havior is defined as behavior that is elicited by antecedent stimuli. Respondent behavior is induced, or brought out, by a stimulus that precedes the behavior; nothing else is required for the response to occur. For example, bright light in the eyes (antecedent stimulus) will elicit pupil contraction (respondent). If the relevant body parts (i.e., receptors and effectors) are intact, pupil contraction will occur every time. However, if the eliciting stimulus is presented repeatedly over a short span of time, the strength or magnitude of the response will diminish, and in some cases the response may not occur at all. This

Table 2 Examples of Unconditioned Human Reflexes Susceptible to Respondent Conditioning

Unconditioned stimulus Unconditioned response Type of effector

Loud sound or touch to cornea Eye blink (lid closes) Striped muscle

Tactile stimulus under lid or chemical Lachrimal gland secretion Gland (duct) irritant (smoke) (eyes watering)

Irritation to nasal mucosa Sneezing Striped and smooth muscle

Irritation to throat Coughing Striped and smooth muscle

Low temperature Shivering, surface vasoconstriction Striped and smooth muscle

High temperature Sweating, surface vasodilation Gland, smooth muscle

Loud sound Contraction of tensor tympani and Striped muscles stapedius muscles (reduces amplitude of ear drum vibrations)

Food in mouth Salivation Gland

Undigestible food in stomach Vomiting Striped and smooth muscle

Pain stimulus to hand or foot Hand or foot withdrawal Striped muscle

A single stimulus that is painful or very Activation syndrome—all of the following: intense or very unusual Heart rate increase Cardiac muscle

Adrenaline secretion Gland (ductless)

Liver release of sugar into bloodstream Gland (duct)

Constriction of visceral blood vessels Smooth muscle

Dilation of blood vessels in skeletal Smooth muscle muscles

Galvanic skin response (GSR) Gland (duct)

Pupillary dilation (and many more) Smooth muscle

Basic Concepts

process of gradually diminishing response strength is known as habituation.

Respondent Conditioning

New stimuli can acquire the ability to elicit respondents. Called respondent conditioning, this type of learning is associated most with the Russian physiologist Ivan Petrovich Pavlov (1849–1936).6 While studying the di- gestive system of dogs, Pavlov noticed that the animals salivated every time his laboratory assistant opened the cage door to feed them. Dogs do not naturally salivate at the sight of someone in a lab coat, but in Pavlov’s labo- ratory they consistently salivated when the door was opened. His curiosity aroused, Pavlov (1927) designed and conducted an historic series of experiments. The re- sult of this work was the experimental demonstration of respondent conditioning.

Pavlov started a metronome just an instant be- fore feeding the dogs. Prior to being exposed to this stimulus–stimulus pairing procedure, food in the mouth, an unconditioned stimulus (US), elicited salivation, but the sound of the metronome, a neutral stimulus (NS), did not. After experiencing several trials consisting of the sound of the metronome followed by the presentation of food, the dogs began salivating in response to the sound of the metronome. The metronome had thus become a conditioned stimulus (CS), and a conditioned reflex was established.7 Respondent conditioning is most ef- fective when the NS is presented immediately before or simultaneous with the US. However, some conditioning effects can sometimes be achieved with considerable delay between the onset of the NS and the onset of the US, and even with backward conditioning in which the US precedes the NS.

Respondent Extinction

Pavlov also discovered that once a conditioned reflex was established, it would weaken and eventually cease altogether if the conditioned stimulus was presented

6Respondent conditioning is also referred to as classical or Pavlovian con- ditioning. Pavlov was not the first to study reflexes; like virtually all sci- entists, his work was an extension of others, most notably Ivan Sechenov (1829–1905) (Kazdin, 1978). See Gray (1979) and Rescorla (1988) for ex- cellent and interesting descriptions of Pavlov’s research. 7Unconditioned stimulus and conditioned stimulus are the most commonly used terms to denote the stimulus component of respondent relations. How- ever, because the terms ambiguously refer to both the immediate evocative (eliciting) effect of the stimulus change and its somewhat permanent and delayed function-altering effect (the conditioning effect on other stimuli), Michael (1995) recommended that the terms unconditioned elicitor (UE) and conditioned elicitor (CE) be used when referring to the evocative func- tion of these variables.

repeatedly in the absence of the unconditioned stimu- lus. For example, if the sound of the metronome was presented repeatedly without being accompanied or fol- lowed by food, it would gradually lose its ability to elicit salivation. The procedure of repeatedly present- ing a conditioned stimulus without the unconditioned stimulus until the conditioned stimulus no longer elic- its the conditioned response is called respondent extinction.

Figure 1 shows schematic representations of re- spondent conditioning and respondent extinction. In this example, a puff of air produced by a glaucoma-testing machine is the US for the eye blink reflex. The opthal- mologist’s finger pressing the button of the machine makes a faint clicking sound. But prior to conditioning, the clicking sound is an NS: It has no effect on eye blink- ing. After being paired with the air puff just a few times, the finger-on-the-button sound becomes a CS: It elicits eye blinking as a conditioned reflex.

Conditioned reflexes can also be established by stim- ulus–stimulus pairing of an NS with a CS. This form of respondent conditioning is called higher order (or sec- ondary) conditioning. For example, secondary respon- dent conditioning could occur in a patient who has learned to blink at the clicking sound of the button dur- ing the glaucoma-testing situation as follows. The patient detects a slight movement of the ophthalmologist’s fin- ger (NS) just before it contacts the button that makes the clicking sound (CS). After several NS–CS pairings, movement of the ophthalmologist’s finger may become a CS capable of eliciting blinking.

The form, or topography, of respondent behaviors changes little, if at all, during a person’s lifetime. There are two exceptions: (a) Certain reflexes disappear with maturity, such as that of grasping an object placed in the palm of the hand, a reflex usually not seen after the age of 3 months (Bijou & Baer, 1965); and (b) sev- eral unconditioned reflexes first appear later in life, such as those related to sexual arousal and reproduc- tion. However, during a person’s lifetime an infinite range of stimuli that were previously neutral (e.g., the high-pitched whine of the dentist’s drill) can come to elicit respondents (i.e., increased heartbeat and perspiration).

Respondents make up a small percentage of the behaviors typically of interest to the applied behavior analyst. As Skinner (1953) pointed out, “Reflexes, con- ditioned or otherwise, are mainly concerned with the in- ternal physiology of the organism. We are most often interested, however, in behavior which has some effect upon the surrounding world” (p. 59). It is this latter type of behavior, and the process by which it is learned, that we will now examine.

Basic Concepts

Before Conditioning US (air puff)

UR (eye blink)

NS (clicking sound)

no eye blink

Respondent Conditioning NS + US (click & air puff)

UR (eye blink)

NS + US (click & air puff)

(more trials)

UR (eye blink)

Product of Respondent Conditioning

US (air puff)

UR (eye blink)

CS (clicking sound)

CR (eye blink)

Respondent Extinction CS (clicking sound)

CR (eye blink)

CS (clicking sound) CR (eye blink)

CS (clicking sound)

(more trials)

CR (eye blink)

Results of Respondent Extinction

US (air puff)

UR (eye blink)

NS (clicking sound)

no eye blink

⇓

Figure 1 Schematic representation of respondent conditioning and respondent extinction. The top panel shows an uncondi- tioned reflex: a puff of air (unconditioned stimulus, or US) elicits an eye blink (an unconditioned response, or UR). Before conditioning, a clicking sound (a neutral stimulus, or NS) has no effect on eye blinking. Respondent conditioning consists of a stimulus–stimulus pairing procedure in which the clicking sound is presented repeatedly just prior to, or simultaneously with, the air puff. The product of respondent conditioning is a conditioned reflex (CR): In this case the clicking sound has become a conditioned stimulus (CS) that elicits an eye blink when presented alone. The bottom two panels illustrate the procedure and outcome of respondent extinction: Repeated presen- tations of the CS alone gradually weaken its ability to elicit eye blinking to the point where the CS eventually becomes an NS again. The unconditioned reflex remains unchanged before, during, and after respon- dent conditioning.

Operant Behavior A baby in a crib moves her hands and arms, setting in motion a mobile dangling above. The baby is literally op- erating on her environment, and the mobile’s movement and musical sounds—stimulus changes produced by the baby’s batting at the toy with her hands—are immediate consequences of her behavior. Her movements are con- tinuously changing as a result of those consequences.

Members of a species whose only way of interacting with the world is a genetically determined fixed set of responses would find it difficult to survive, let alone thrive, in a complex environment that differed from the environment in which their distant ancestors evolved. Al- though respondent behavior comprises a critically im- portant set of “hardwired” responses, respondent behavior does not provide an organism with the ability to learn from the consequences of its actions. An organism whose behavior is unchanged by its effects on the environment will be unable to adapt to a changing one.

8The verb emit is used in conjunction with operant behavior. Its use fits in well with the definition of operant behavior, allowing reference to the con- sequences of behavior as the major controlling variables. The verb elicit is inappropriate to use with operant behavior because it implies that an an- tecedent stimulus has primary control of the behavior.

Fortunately, in addition to her repertoire of geneti- cally inherited respondent behaviors, our baby entered her world with some uncommitted behavior that is highly mal- leable and susceptible to change through its consequences. This type of behavior, called operant behavior, enables the baby over the course of her life to learn novel, in- creasingly complex responses to an ever-changing world.8

Operant behavior is any behavior whose future fre- quency is determined primarily by its history of conse- quences. Unlike respondent behavior, which is elicited by antecedent events, operant behavior is selected, shaped, and maintained by the consequences that have followed it in the past.

Basic Concepts

Unlike respondent behaviors, whose topography and basic functions are predetermined, operant behaviors can take a virtually unlimited range of forms. The form and function of respondent behaviors are constant and can be identified by their topography (e.g., the basic form and function of salivation is always the same). By compari- son, however, the “meaning” of operant behavior cannot be determined by its topography. Operants are defined functionally, by their effects. Not only does the same op- erant often include responses of widely different topogra- phies (e.g., a diner may obtain a glass of water by nodding his head, pointing to a glass of water, or saying yes to a waiter), but also, as Skinner (1969) explained, the same movements comprise different operants under different conditions.

Allowing water to pass over one’s hands can perhaps be adequately described as topography, but “washing one’s hands” is an “operant” defined by the fact that, when one has behaved this way in the past, one’s hands have become clean—a condition which has become reinforc- ing because, say, it has minimized a threat of criticism or contagion. Behavior of precisely the same topography would be part of another operant if the reinforcement had consisted of simple stimulation (e.g., “tickling”) of the hands or the evocation of imitative behavior in a child whom one is teaching to wash his hands. (p. 127)

Table 3 compares and contrasts defining features and key characteristics of respondent behavior and oper- ant behavior.

Selection by Consequences Human behavior is the joint product of (i) the contingen- cies of survival responsible for the natural selection of the species and (ii) the contingencies of reinforcement responsible for the repertoires acquired by its members, including (iii) the special contingencies maintained by the social environment. [Ultimately, of course, it is all a matter of natural selection, since operant conditioning is an evolved process, of which cultural practices are spe- cial applications.]

—B. F. Skinner (1981, p. 502)

Skinner’s discovery and subsequent elucidation of oper- ant selection by consequences have rightly been called “revolutionary” and “the bedrock on which other behav- ioral principles rest” (Glenn, 2004, p. 134). Selection by consequences “anchors a new paradigm in the life sci- ences known as selectionism. A basic tenet of this posi- tion is that all forms of life, from single cells to complex cultures, evolve as a result of selection with respect to function” (Pennypacker, 1994, pp. 12–13).

Selection by consequences operates during the life- time of the individual organism (ontogeny) and is a con-

ceptual parallel to Darwin’s (1872/1958) natural selec- tion in the evolutionary history of a species (phylogeny). In response to the question, “Why do giraffes have long necks?” Baum (1994) gave this excellent description of natural selection:

Darwin’s great contribution was to see that a relatively simple mechanism could help explain why phylogeny followed the particular course it did. The explanation about giraffes’ necks requires reference to the births, lives, and deaths of countless giraffes and giraffe an- cestors over many millions of years. . . . Within any population of organisms, individuals vary. They vary partly because of environmental factors (e.g., nutri- tion), and also because of genetic inheritance. Among the giraffe ancestors that lived in what is now the Serengeti Plain, for instance, variation in genes meant that some had shorter necks and some had longer necks. As the climate gradually changed however, new, taller types of vegetation became more frequent. The giraffe ancestors that had longer necks, being able to reach higher, got a little more to eat, on the average. As a result, they were a little healthier, resisted disease a little better, evaded predators a little better—on the av- erage. Any one individual with a longer neck may have died without offspring, but on the average longer- necked individuals had more offspring, which tended on the average to survive a little better and produce more offspring. As longer necks became more fre- quent, new genetic combinations occurred, with the re- sult that some offspring had still longer necks than those before, and they did still better. As the longer- necked giraffes continued to out-reproduce the shorter- necked ones, the average neck length of the whole population grew. (p. 52)

Just as natural selection requires a population of individual organisms with varied physical features (e.g., giraffes with necks of different lengths), operant se- lection by consequences requires variation in behav- ior. Those behaviors that produce the most favorable outcomes are selected and “survive,” which leads to a more adaptive repertoire. Natural selection has en- dowed humans with an initial population of uncom- mitted behavior (e.g., babies babbling and moving their limbs about) that is highly malleable and susceptible to the influence of the consequences that follow it. As Glenn (2004) noted,

By outfitting humans with a largely uncommitted behav- ioral repertoire, natural selection gave our species a long leash for local behavioral adaptations. But the uncom- mitted repertoire of humans would be lethal without the . . . susceptibility of human behavior to operant se- lection. Although this behavioral characteristic is shared by many species, humans appear to be most exquisitely sensitive to behavioral contingencies of selection. (Schwartz, 1974, p. 139)

Basic Concepts

Table 3 Comparing and Contrasting Defining Features and Key Characteristics of Respondent and Operant Behavior

Characteristics or features Respondent behavior Operant behavior

Definition Behavior elicited by antecedent stimuli. Behavior selected by its consequences.

Basic unit Reflex: an antecedent stimulus elicits a par- Operant response class: A group of responses ticular response (S–R). all of which produce the same effect on the

environment; described by three-term contin- gency relation of antecedent stimulus condi- tions, behavior, and consequence (A–B–C).

Examples Newborn’s grasping and suckling to touch; pupil Talking, walking, playing the piano, riding a constriction to bright light; cough/gag to irritation bike, counting change, baking a pie, hitting a in throat; salivation at smell of food; withdrawing curveball, laughing at a joke, thinking about a hand from painful stimulus; sexual arousal to grandparent, reading this book. stimulation.

Body parts (effectors) that Primarily smooth muscles and glands (adrena- Primarily striated (skeletal) muscles; sometimes most often produce the line squirt); sometimes striated (skeletal) mus- smooth muscles and glands. response (not a defining cles (e.g., knee-jerk to tap just below patella). feature)

Function or usefulness for Maintains internal economy of the organism; Enables effective interaction and adaptation in individual organism provides a set of “ready-made” survival re- an ever-changing environment that could not be

sponses the organism would not have time anticipated by evolution. to learn.

Function or usefulness Promotes continuation of species indirectly Individuals whose behavior is most sensitive to for species (protective reflexes help individuals survive to consequences are more likely to survive and

reproductive age) and directly (reflexes related reproduce. to reproduction).

Conditioning process Respondent (also called, classical or Pavlovian) Operant conditioning: Some stimulus changes conditioning: Through a stimulus–stimulus pair- immediately following a response increase (re- ing procedure in which a neutral stimulus (NS) inforcement) or decrease (punishment) the presented just prior to or simultaneous with an future frequency of similar responses under unconditioned (US) or conditioned (CS) eliciting similar conditions. Previously neutral stim- stimulus, the NS becomes a CS that elicits the ulus changes become conditioned reinforcers response and a conditioned reflex is created. or punishers as result of stimulus–stimulus (See Figure 1) pairing with other reinforcers or punishers.

Repertoire limits Topography and function of respondents deter- Topography and function of each person’s mined by natural evolution of species (phylogeny). repertoire of operant behaviors are selected by All biologically intact members of a species consequences during the individual’s lifetime possess the same set of unconditioned reflexes. (ontogeny). New and more complex operant re- Although new forms of respondent behavior are sponse classes can emerge. Response prod- not learned, an infinite number of conditioned ucts of some human operants (e.g., airplanes) reflexes may emerge in an individual’s repertoire enable some behaviors not possible by anatom- depending on the stimulus–stimulus pairing he ical structure alone (e.g., flying). has experienced (ontogeny).

Operant Conditioning Operant conditioning may be seen everywhere in the multifarious activities of human beings from birth until death. . . . It is present in our most delicate discrimina- tions and our subtlest skills; in our earliest crude habits and the highest refinements of creative thought. —Keller and Schoenfeld (1950, p. 64)

Operant conditioning refers to the process and se- lective effects of consequences on behavior.9 From an operant conditioning perspective a functional conse- quence is a stimulus change that follows a given behavior

9Unless otherwise noted, the term behavior will refer to operant behavior throughout the remainder of the text.

Basic Concepts

in a relatively immediate temporal sequence and alters the frequency of that type of behavior in the future. “In operant conditioning we ‘strengthen’ an operant in the sense of making a response more probable or, in actual fact, more frequent” (Skinner, 1953, p. 65). If the move- ment and sounds produced by the baby’s batting at the mobile with her hands increase the frequency of hand movements in the direction of the toy, operant condi- tioning has occurred.

When operant conditioning consists of an increase in response frequency, reinforcement has taken place, and the consequence responsible, in this case the movement and sound of the mobile, would be called a reinforcer.10

Although operant conditioning is used most often to refer to the “strengthening” effects of reinforcement, as Skin- ner described earlier, it also encompasses the principle of punishment. If the mobile’s movement and musical sounds resulted in a decrease in the baby’s frequency of moving it with her hands, punishment has occurred, and the mobile’s movement and sound would be called puni- shers. Before we examine the principles of reinforce- ment and punishment further, it is important to identify several important qualifications concerning how conse- quences affect behavior.

Consequences Can Affect Only Future Behavior

Consequences affect only future behavior. Specifically, a behavioral consequence affects the relative frequency with which similar responses will be emitted in the future under similar stimulus conditions. This point may seem too obvious to merit mention because it is both logically and physically impossible for a consequent event to af- fect a behavior that preceded it, when that behavior is over before the consequent event occurs. Nevertheless, the statement “behavior is controlled by its conse- quences” raises the question. (See Box 1 for further discussion of this apparent logical fallacy.)

Consequences Select Response Classes, Not Individual Responses

Responses emitted because of the effects of reinforce- ment of previous responses will differ slightly from the previous responses but will share enough common ele-

10Skinner (1966) used rate of responding as the fundamental datum for his research. To strengthen an operant is to make it more frequent. However, rate (or frequency) is not the only measurable and malleable dimension of behavior. As we will see in chapters entitled “Selecting and Defining Tar- get Behaviors” and “Measuring Behavior,” sometimes the duration, la- tency, magnitude, and/or topography of behavior changes are of pragmatic importance.

ments with the former responses to produce the same consequence.

Reinforcement strengthens responses which differ in topography from the response reinforced. When we rein- force pressing a lever, for example, or saying Hello, re- sponses differing quite widely in topography grow more probable. This is a characteristic of behavior which has strong survival value . . . , since it would be very hard for an organism to acquire an effective repertoire if rein- forcement strengthened only identical responses. (Skin- ner, 1969, p. 131)

These topographically different, but functionally similar, responses comprise an operant response class. Indeed, “an operant is a class of acts all of which have the same environmental effect” (Baum, 1994, p. 75). It is the response class that is strengthened or weakened by op- erant conditioning. The concept of response class is “im- plied when it is said that reinforcement increases the future frequency of the type of behavior that immedi- ately preceded the reinforcement” (Michael, 2004, p. 9). And the concept of response class is a key to the devel- opment and elaboration of new behavior.

If consequences (or natural evolution) selected only a very narrow range of responses (or genotypes), the ef- fect would “tend toward uniformity and a perfection of sorts” (Moxley, 2004, p. 110) that would place the be- havior (or species) at risk of extinction should the envi- ronment change. For example, if the mobile’s movement and sound reinforced only arm and hand movements that fell within an exact and narrow range of motion and no similar movements survived, the baby would be unable to contact that reinforcement if one day her mother mounted the mobile in a different location above the crib.

Immediate Consequences Have the Greatest Effect

Behavior is most sensitive to stimulus changes that occur immediately after, or within a few seconds of, the responses.

It is essential to emphasize the importance of the imme- diacy of reinforcement. Events that are delayed more than a few seconds after the response do not directly in- crease its future frequency. When human behavior is ap- parently affected by long-delayed consequences, the change is accomplished by virtue of the human’s com- plex social and verbal history, and should not be thought of as an instance of the simple strengthening of behavior by reinforcement. . . . [As with reinforcement,] the longer the time delay between the occurrence of the re- sponse and the occurrence of the stimulus change (be- tween R and SP), the less effective the punishment will be in changing the relevant response frequency, but not

Basic Concepts

The professor was ready to move on to his next point, but a raised hand in the front row caught his attention.

Professor: Yes?

Student: You say that operant behavior, like talking, writing, running, reading, driving a car, most every- thing we do—you say all of those behaviors are con- trolled by their consequences, by things that happen after the response was emitted?

Professor: Yes, I said that. Yes.

Student: Well, I have a hard time with that. When my telephone rings and I pick up the receiver, that’s an operant response, right? I mean, answering the phone when it rings certainly didn’t evolve geneti- cally as a reflex to help our species survive. So, we are talking about operant behavior, correct?

Professor: Correct.

Student: All right then. How can we say that my pick- ing up my telephone is controlled by its conse- quence? I pick up the phone because it is ringing. So does everybody else. Ringing controls the re- sponse. And ringing can’t be a consequence because it comes before the response.

The professor hesitated with his reply just long enough for the student to believe himself the hero, nailing a professor for pontificating about some theoretical concept with little or no relevance to the everyday real world. Simultaneously sensing victory, other students began to pile on with their comments.

Another Student: How about stepping on the brake when you see a stop sign? The sign controls the braking response, and that’s not a consequence either.

A Student from the Back of the Room: And take a com- mon classroom example. When a kid sees the prob- lem 2 + 2 on his worksheet and he writes 4, the response of writing 4 has to be controlled by the writ- ten problem itself. Otherwise, how could anyone learn the correct answers to any question or problem?

Most of the Class: Yah, that’s right!

Professor: (with a wry smile) All of you are cor- rect. . . . So too am I.

Someone Else in the Class: What do you mean?

Professor: That was exactly my next point, and I was hoping you would pick up on it. (The professor smiled a thank you at the student who had started the discussion and went on.) All around us, every day, we are exposed to thousands of changing stim- ulus conditions. All of the situations you’ve de- scribed are excellent examples of what behavior analysts call stimulus control. When the frequency of a given behavior is higher in the presence of a given stimulus than when that stimulus is absent, we say that stimulus control is at work. Stimulus control is a very important and useful principle in behavior analysis, and it will be the subject of much discus- sion this semester.

But, and here’s the important point: A dis- criminative stimulus, the antecedent event that comes before the response of interest, acquires its ability to control a particular response class because it has been associated with certain consequences in the past. So it is not just the sound of the phone’s ring that causes you to pick up the receiver. It is the fact that in the past answering the phone when it was ringing was followed by a person’s voice. It’s that person talking to you, the consequence of pick- ing up the receiver, that really controlled the be- havior in the first place, but you pick up the phone only when you hear it ringing. Why? Because you have learned that there’s someone on the other end only when the phone’s ringing. So we can still speak of consequences as having the ultimate control in terms of controlling operant behavior, but by being paired with differential consequences, antecedent stimuli can indicate what kind of consequence is likely. This concept is called the three-term contin- gency, and its understanding, analysis, and manip- ulation is central to applied behavior analysis.

Box 1 When the Phone Rings:

A Dialogue about Stimulus Control

Basic Concepts

much is known about upper limits. (Michael, 2004, p. 110, 36 emphasis in original, words in brackets added)

Consequences Select Any Behavior

Reinforcement and punishment are “equal opportunity” selectors. No logical or healthy or (in the long run) adap- tive connection between a behavior and the consequence that functions to strengthen or weaken it is necessary. Any behavior that immediately precedes reinforcement (or punishment) will be increased (or decreased).

It is the temporal relation between behavior and con- sequence that is functional, not the topographical or log- ical ones. “So far as the organism is concerned, the only important property of the contingency is temporal. The reinforcer simply follows the response. How this is brought about does not matter” (Skinner, 1953, p. 85, emphasis in original). The arbitrary nature of which be- haviors are reinforced (or punished) in operant condi- tioning is exemplified by the appearance of idiosyncratic behaviors that have no apparent purpose or function. An example is the superstitious routine of a poker player who taps and arranges his cards in a peculiar fashion be- cause similar movements in the past were followed by winning hands.

Operant Conditioning Occurs Automatically

Operant conditioning does not require a person’s aware- ness. “A reinforcing connection need not be obvious to the individual [whose behavior is] reinforced” (Skinner, 1953, p. 75, words in brackets added). This statement refers to the automaticity of reinforcement; that is, be- havior is modified by its consequences regardless of whether the individual is aware that she is being re- inforced.11 A person does not have to understand or verbalize the relation between her behavior and a consequence, or even know that a consequence has oc- curred, for reinforcement to “work.”

Reinforcement

Reinforcement is the most important principle of behav- ior and a key element of most behavior change programs designed by behavior analysts (Flora, 2004; Northup, Vollmer, & Serrett, 1993). If a behavior is followed

11Automaticity of reinforcement is a different concept from that of automatic reinforcement, which refers to responses producing their “own” reinforce- ment (e.g., scratching an insect bite). Automatic reinforcement is described in chapter entitled “Positive Reinforcement.”

12The basic effect of reinforcement is often described as increasing the probability or strength of the behavior, and at times we use these phrases also. In most instances, however, we use frequency when referring to the basic effect of operant conditioning, following Michael’s (1995) rationale: “I use frequency to refer to number of responses per unit time, or number of response occurrences relative to the number of opportunities for a re- sponse. In this way I can avoid such terms as probability, likelihood and strength when referring to behavior. The controlling variables for these terms are problematic, and because of this, their use encourages a language of intervening variables, or an implied reference to something other than an observable aspect of behavior” (p. 274). 13Malott and Trojan Suarez (2004) referred to these two operations as “stimulus addition” and “stimulus subtraction.”

closely in time by a stimulus event and as a result the fu- ture frequency of that type of behavior increases in sim- ilar conditions, reinforcement has taken place.12

Sometimes the delivery of just one reinforcer results in significant behavior change, although most often several responses must be followed by reinforcement before sig- nificant conditioning will occur.

Most stimulus changes that function as reinforcers can be described operationally as either (a) a new stim- ulus added to the environment (or increased in intensity), or (b) an already present stimulus removed from the en- vironment (or reduced in intensity).13 These two opera- tions provide for two forms of reinforcement, called positive and negative (see Figure 2).

Positive reinforcement occurs when a behavior is followed immediately by the presentation of a stimulus and, as a result, occurs more often in the future. Our baby’s increased frequency of batting the mobile with her hands, when doing so produces movement and music, is an example of positive reinforcement. Likewise, a child’s independent play is reinforced when it increases as a result of his parent’s giving praise and attention when he plays.

When the frequency of a behavior increases because past responses have resulted in the withdrawal or termi- nation of a stimulus, the operation is called negative re- inforcement. Skinner (1953) used the term aversive stimulus to refer to, among other things, stimulus con- ditions whose termination functioned as reinforcement. Let us assume now that a parent programs the mobile to automatically play music for a period of time. Let us also assume that if the baby bats the mobile with hands or feet, the music immediately stops for a few seconds. If the baby bats the mobile more frequently when doing so ter- minates the music, negative reinforcement is at work, and the music can be called aversive.

Negative reinforcement is characterized by escape or avoidance contingencies. The baby escaped the music by striking the mobile with her hand. A person who jumps out of the shower when water suddenly becomes too hot

Basic Concepts

Positive Reinforcement Negative Reinforcement

Positive Punishment Negative Punishment

Present or Increase Intensity of Stimulus

Withdraw or Decrease Intensity of Stimulus

E ff

ec t

o n

F u

tu re

F re

q u

en cy

o f

B eh

av io

r Type of Stimulus Change Figure 2 Positive and negative re-

inforcement and positive and negative punishment are defined by the type of stimulus change operation that immedi- ately follows a behavior and the effect that operation has on the future frequency of that type of behavior.

escapes the overly hot water. Likewise, when the fre- quency of a student’s disruptive behavior increases as a result of being sent to the principal’s office, negative re- inforcement has occurred. By acting out, the misbehav- ing student escapes (or avoids altogether, depending on the timing of his misbehavior) the aversive (to him) class- room activity.

The concept of negative reinforcement has confused many students of behavior analysis. Much of the confu- sion can be traced to the inconsistent early history and development of the term and to psychology and education textbooks and professors who have used the term inac- curately.14 The most common mistake is equating nega- tive reinforcement with punishment. To help avoid the error, Michael (2004) suggested the following:

Think about how you would respond if someone asked you (1) whether or not you like negative reinforcement; also if you were asked (2) which you prefer, positive or negative reinforcement. Your answer to the first question should be that you do indeed like negative reinforce- ment, which consists of the removal or termination of an aversive condition that is already present. The term negative reinforcement refers only to the termination of the stimulus. In a laboratory procedure the stimulus must, of course, be turned on and then its termination can be made contingent upon the critical response. No one wants an aversive stimulus turned on, but once it is on, its termination is usually desirable. Your answer to the second question should be that you cannot choose without knowing the specifics of the positive and nega- tive reinforcement involved. The common error is to

14For examples and discussions of the implications of inaccurate repre- sentations of principles of behavior and behaviorism in psychology and education textbooks, see Cameron (2005), Cooke (1984), Heward (2005), Heward and Cooper (1992), and Todd and Morris (1983, 1992).

choose positive reinforcement, but removal of a very se- vere pain would certainly be preferred over the presenta- tion of a small monetary reward or an edible, unless the food deprivation was very severe. (p. 32, italics and bold type in original)

Remembering that the term reinforcement always means an increase in response rate and that the modifiers positive and negative describe the type of stimulus change operation that best characterizes the consequence (i.e., adding or withdrawing a stimulus) should facilitate the discrimination of the principles and application of posi- tive and negative reinforcement.

After a behavior has been established with rein- forcement, it need not be reinforced each time it occurs. Many behaviors are maintained at high levels by sched- ules of intermittent reinforcement. However, if rein- forcement is withheld for all members of a previously reinforced response class, a procedure based on the prin- ciple of extinction, the frequency of the behavior will gradually decrease to its prereinforcement level or cease to occur altogether.

Punishment

Punishment, like reinforcement, is defined functionally. When a behavior is followed by a stimulus change that decreases the future frequency of that type of behavior in similar conditions, punishment has taken place. Also, like reinforcement, punishment can be accomplished by either of two types of stimulus change operations. (See the bottom two boxes of Figure 2.)

Basic Concepts

Although most behavior analysts support the defi- nition of punishment as a consequence that decreases the future frequency of the behavior it follows (Azrin & Holz, 1966), a wide variety of terms have been used in the literature to refer to the two types of consequence operations that fit the definition. For example, the Be- havior Analyst Certification Board (BACB, 2005) and textbook authors (e.g., Miltenberger, 2004) use the terms positive punishment and negative punishment, paralleling the terms positive reinforcement and negative reinforcement. As with reinforcement, the modifiers positive and negative used with punishment connote nei- ther the intention nor the desirability of the behavior change produced; they only specify how the stimulus change that served as the punishing consequence was affected—whether it was presented (positive) or with- drawn (negative).

Although the terms positive punishment and negative punishment are consistent with the terms used to differentiate the two reinforcement operations, they are less clear than the descriptive terms for the two pun- ishment operations—punishment by contingent stimu- lation and punishment by contingent withdrawal of a positive reinforcer—first introduced by Whaley and Malott (1971) in their classic text, Elementary Princi- ples of Behavior. These terms highlight the procedural difference between the two forms of punishment. Dif- ferences in procedure as well as in the type of stimulus change involved—reinforcer or punisher—hold im- portant implications for application when a punish- ment-based behavior-reduction technique is indicated. Foxx (1982) introduced the terms Type I punishment and Type II punishment for punishment by contingent stimulation and punishment by contingent withdrawal of a stimulus, respectively. Many behavior analysts and teachers continue to use Foxx’s terminology today. Other terms such as penalty principle have also been used to refer to negative punishment. (Malott & Tro- jan Suarez, 2004). However, it should be remembered that these terms are simply brief substitutes for the more complete terminology introduced by Whaley and Malott.

As with positive and negative reinforcement, nu- merous behavior change procedures incorporate the two basic punishment operations. Although some textbooks reserve the term punishment for procedures involving positive (or Type I) punishment and describe time-out from positive reinforcement and response cost as sepa- rate “principles” or types of punishment, both the meth- ods for reducing behavior are derivatives of negative (or Type II) punishment. Therefore, time-out and response cost should be considered behavior change tactics and not basic principles of behavior.

15Michael (1975) and Baron and Galizio (2005) present cogent arguments for why positive and negative reinforcement are examples of the same fun- damental operant relation. This issue is discussed further in chapter enti- tled “Negative Reinforcement.” 16Some authors use the modifiers primary or unlearned to identify un- conditioned reinforcers and unconditioned punishers.

Reinforcement and punishment can each be accom- plished by either of two different operations, depending on whether the consequence consists of presenting a new stimulus (or increasing the intensity of a current stimu- lus) or withdrawing (or decreasing the intensity of ) a cur- rently present stimulus in the environment (Morse & Kelleher, 1977; Skinner, 1953). Some behavior analysts argue that from a functional and theoretical standpoint only two principles are required to describe the basic ef- fects of behavioral consequences—reinforcement and punishment.15 However, from a procedural perspective (a critical factor for the applied behavior analyst), a num- ber of behavior change tactics are derived from each of the four operations represented in Figure 2.

Most behavior change procedures involve several principles of behavior (see Box 2). It is critical for the behavior analyst to have a solid conceptual understand- ing of the basic principles of behavior. Such knowledge permits better analysis of current controlling variables as well as more effective design and assessment of behav- ioral interventions that recognize the role various princi- ples may be playing in a given situation.

Stimulus Changes That Function as Reinforcers and Punishers

Because operant conditioning involves the consequences of behavior, it follows that anyone interested in using op- erant conditioning to change behavior must identify and control the occurrence of relevant consequences. For the applied behavior analyst, therefore, an important ques- tion becomes, What kinds of stimulus changes function as reinforcers and punishers?

Unconditioned Reinforcement and Punishment

Some stimulus changes function as reinforcement even though the organism has had no particular learning his- tory with those stimuli. A stimulus change that can in- crease the future frequency of behavior without prior pairing with any other form of reinforcement is called an unconditioned reinforcer.16 For example, stimuli such as food, water, and sexual stimulation that support the biological maintenance of the organism and survival of the species often function as unconditioned reinforcers. The words can and often in the two previous sentences

Basic Concepts

A principle of behavior describes a basic behavior–en- vironment relation that has been demonstrated repeat- edly in hundreds, even thousands, of experiments. A principle of behavior describes a functional relation between behavior and one or more of its controlling vari- ables (in the form of y = fx) that has thorough generality across individual organisms, species, settings, and be- haviors. A principle of behavior is an empirical general- ization inferred from many experiments. Principles describe how behavior works. Some examples of prin- ciples are reinforcement, punishment, and extinction.

In general, a behavior change tactic is a method for operationalizing, or putting into practice, the knowledge provided by one or more principles of behavior. A be- havior change tactic is a research-based, technologi- cally consistent method for changing behavior that has been derived from one or more basic principles of be- havior and that possesses sufficient generality across sub- jects, settings, and/or behaviors to warrant its codification and dissemination. Behavior change tactics constitute the technological aspect of applied behavior analysis. Ex- amples of behavior change procedures include backward

chaining, differential reinforcement of other behavior, shaping, response cost, and time-out.

So, principles describe how behavior works, and be- havior change tactics are how applied behavior analysts put the principles to work to help people learn and use so- cially significant behaviors. There are relatively few prin- ciples of behavior, but there are many derivative behavior change tactics. To illustrate further, reinforcement is a behavioral principle because it describes a lawful rela- tion between behavior, an immediate consequence, and an increased frequency of the behavior in the future under similar conditions. However, the issuance of checkmarks in a token economy or the use of contingent social praise are behavior change tactics derived from the principle of reinforcement. To cite another example, punishment is a principle behavior because it describes the established relations between the presentation of a consequence and the decreased frequency of similar behavior in the fu- ture. Response cost and time-out, on the other hand, are methods for changing behavior; they are two different tactics used by practitioners to operationalize the princi- ple of punishment.

Box 2 Distinguishing between Principles

of Behavior and Behavior Change Tactics

recognize the important qualification that the momentary effectiveness of an unconditioned reinforcer is a function of current motivating operations. For example, a cer- tain level of food deprivation is necessary for the pre- sentation of food to function as a reinforcer. However, food is unlikely to function as reinforcement for a person who has recently eaten a lot of food (a condition of sat- iation).

Similarly, an unconditioned punisher is a stimulus change that can decrease the future frequency of any be- havior that precedes it without prior pairing with any other form of punishment. Unconditioned punishers in- clude painful stimulation that can cause tissue damage (i.e., harm body cells). However, virtually any stimulus to which an organism’s receptors are sensitive—light, sound, and temperature, to name a few—can be intensi- fied to the point that its delivery will suppress behavior even though the stimulus is below levels that actually cause tissue damage (Bijou & Baer, 1965).

Events that function as unconditioned reinforcers and punishers are the product of the natural evolution of the species (phylogeny). Malott, Tillema, and Glenn (1978)

17In addition to using aversive stimulus as a synonym for a negative rein- forcer, Skinner (1953) also used the term to refer to stimuli whose onset or presentation functions as punishment, a practice continued by many be- havior analysts (e.g., Alberto & Troutman, 2006; Malott & Trojan Suarez, 2004; Miltenberger, 2004). The term aversive stimulus (and aversive con- trol when speaking of behavior change techniques involving such stimuli) is used widely in the behavior analysis literature to refer to one or more of three different behavioral functions: an aversive stimulus may be (a) a neg- ative reinforcer if its termination increases behavior, (b) a punisher if its pre- sentation decreases behavior, and/or (c) a motivating operation if its presentation increases the current frequency of behaviors that have termi- nated it in the past (see chapter entitled “Motivating Operations”). When speaking or writing technically, behavior analysts must be careful that their use of omnibus terms such as aversive does not imply unintended func- tions (Michael, 1995).

described the natural selection of “rewards” and “aver- sives” as follows:17

Some rewards and aversives control our actions because of the way our species evolved; we call these unlearned rewards or aversives. We inherit a biological structure that causes some stimuli to be rewarding or aversive. This structure evolved because rewards helped our an- cestors survive, while aversives hurt their survival. Some of these unlearned rewards, such as food and fluid, help us survive by strengthening our body cells. Others help

Basic Concepts

our species survive by causing us to produce and care for our offspring—these stimuli include the rewarding stimulation resulting from copulation and nursing. And many unlearned aversives harm our survival by damag- ing our body cells; such aversives include burns, cuts and bruises. (p. 9)

While unconditioned reinforcers and punishers are critically important and necessary for survival, relatively few behaviors that comprise the everyday routines of peo- ple as they go about working, playing, and socializing are directly controlled by such events. For example, al- though going to work each day earns the money that buys food, eating that food is far too delayed for it to exert any direct operant control over the behavior that earned it. Remember: Behavior is most affected by its immediate consequences.

Conditioned Reinforcers and Punishers

Stimulus events or conditions that are present or that occur just before or simultaneous with the occurrence of other reinforcers (or punishers) may acquire the ability to reinforce (or punish) behavior when they later occur on their own as consequences. Called conditioned rein- forcers and conditioned punishers, these stimulus changes function as reinforcers and punishers only be- cause of their prior pairing with other reinforcers or pun- ishers.18 The stimulus–stimulus pairing procedure responsible for the creation of conditioned reinforcers or punishers is the same as that used for respondent condi- tioning except that the “outcome is a stimulus that func- tions as a reinforcer [or punisher] rather than a stimulus that will elicit a response” (Michael, 2004, p. 66, words in brackets added).

Conditioned reinforcers and punishers are not related to any biological need or anatomical structure; their abil- ity to modify behavior is a result of each person’s unique history of interactions with his or her environment (on- togeny). Because no two people experience the world in exactly the same way, the roster of events that can serve as conditioned reinforcers and punishers at any particu- lar time (given a relevant motivating operation) is idio- syncratic to each individual and always changing. On the other hand, to the extent that two people have had simi- lar experiences (e.g., schooling, profession, the culture in general), they are likely to be affected in similar ways to many similar events. Social praise and attention are examples of widely effective conditioned reinforcers in our culture. Because social attention and approval (as well as disapproval) are often paired with so many other reinforcers (and punishers), they exert powerful con-

18Some authors use the modifiers secondary or learned to identify conditioned reinforcers and conditioned punishers.

trol over human behavior and will be featured when spe- cific tactics for changing behavior are presented.

Because people who live in a common culture share similar histories, it is not unreasonable for a practitioner to search for potential reinforcers and punishers for a given client among classes of stimuli that have proven effective with other similar clients. However, in an effort to help the reader establish a fundamental understanding of the nature of operant conditioning, we have purposely avoided presenting a list of stimuli that may function as reinforcers and punishers. Morse and Kelleher (1977) made this important point very well.

Reinforcers and punishers, as environmental “things,” appear to have a greater reality than orderly temporal changes in ongoing behavior. Such a view is deceptive. There is no concept that predicts reliably when events will be reinforcers or punishers; the defining character- istics of reinforcers and punishers are how they change behavior [italics added). Events that increase or de- crease the subsequent occurrence of one response may not modify other responses in the same way.

In characterizing reinforcement as the presentation of a reinforcer contingent upon a response, the tendency is to emphasize the event and to ignore the importance of both the contingent relations and the antecedent and subsequent behavior. It is how [italics added] they change behavior that defines the terms reinforcer and punisher; thus it is the orderly change in behavior that is the key to these definitions. It is not [italics added] appropriate to presume that particular environmental events such as the presentation of food or electric shock are reinforcers or punishers until a change in the rate of responding has occurred when the event is scheduled in relation to specified responses.

A stimulus paired with a reinforcer is said to have become a conditioned reinforcer, but actually it is the behaving subject that has changed, not the stimulus. . . . It is, of course, useful shorthand to speak of conditioned reinforcers . . . just as it is convenient to speak about a reinforcer rather than speaking about an event that has followed an instance of a specific response and resulted in a subsequent increase in the occurrence of similar re- sponses. The latter may be cumbersome, but it has the advantage of empirical referents. Because many differ- ent responses can be shaped by consequent events, and because a given consequent event is often effective in modifying the behavior of different individuals, it be- comes common practice to refer to reinforcers without specifying the behavior that is being modified. These common practices have unfortunate consequences. They lead to erroneous views that responses are arbitrary and that the reinforcing or punishing effect of an event is a specific property of the event itself. (pp. 176–177, 180)

The point made by Morse and Kelleher (1977) is of paramount importance to understanding behavior–

Basic Concepts

environment relations. Reinforcement and punishment are not simply the products of certain stimulus events, which are then called reinforcers and punishers without reference to a given behavior and environmental condi- tions. There are no inherent or standard physical proper- ties of stimuli that determine their permanent status as reinforcers and punishers. In fact, a stimulus can func- tion as a positive reinforcer under one set of conditions and a negative reinforcer under different conditions. Just as positive reinforcers are not defined with terms such as pleasant or satisfying, aversive stimuli should not be de- fined with terms such as annoying or unpleasant. The terms reinforcer and punisher should not to be used on the basis of a stimulus event’s assumed effect on behavior or on any inherent property of the stimulus event itself. Morse and Kelleher (1977) continued:

When the borders of the table are designated in terms of stimulus classes (positive–negative; pleasant–noxious) and experimental operations (stimulus presentation– stimulus withdrawal), the cells of the table are, by defin- ition, varieties of reinforcement and punishment. One problem is that the processes indicated in the cells have already been assumed in categorizing stimuli as positive or negative; a second is that there is a tacit assumption that the presentation or withdrawal of a particular stimu- lus will have an invariant effect. These relations are clearer if empirical operations are used to designate the border conditions. . . . The characterization of behavioral processes depends upon empirical observations. The same stimulus event, under different conditions, may in- crease behavior or decrease behavior. In the former case the process is called reinforcement and in the latter the process is called punishment. (p. 180)

At the risk of redundancy, we will state this impor- tant concept again. Reinforcers and punishers denote functional classes of stimulus events, the membership to which is not based on the physical nature of the stimulus changes or events themselves. Indeed, given a person’s in- dividual history and current motivational state, and the current environmental conditions, “any stimulus change can be a ‘reinforcer’ if the characteristics of the change, and the temporal relation of the change to the response under observation, are properly selected” (Schoenfeld, 1995, p. 184). Thus, the phrase “everything is relative” is thoroughly relevant to understanding functional behavior–environment relations.

The Discriminated Operant and Three-Term Contingency

We have discussed the role of consequences in influenc- ing the future frequency of behavior. But operant condi- tioning does much more than establish a functional relation between behavior and its consequences. Oper-

ant conditioning also establishes functional relations be- tween behavior and certain antecedent conditions.

In contrast to if-A-then-B formulations (such as S-R for- mulations), the AB-because-of-C formulation is a gen- eral statement that the relation between an event (B) and its context (A) is because of consequences (C). . . . Ap- plied to Skinner’s three-term contingency, the relation between (A) the setting and (B) behavior exists because of (C) consequences that occurred for previous AB (setting-behavior) relations. The idea [is] that reinforce- ment strengthens the setting-behavior relation rather than simply strengthening behavior. (Moxley, 2004, p. 111)

Reinforcement selects not just certain forms of be- havior; it also selects the environmental conditions that in the future will evoke (increase) instances of the response class. A behavior that occurs more frequently under some antecedent conditions than it does in others is called a discriminated operant. Because a discriminated operant occurs at a higher frequency in the presence of a given stimulus than it does in the absence of that stimulus, it is said to be under stimulus control. Answering the phone, one of the everyday behaviors discussed by the professor and his students in Box 1, is a discriminated operant. The telephone’s ring functions as a discriminative stim- ulus (SD) for answering the phone. We answer the phone when it is ringing, and we do not answer the phone when it is silent.

Just as reinforcers or punishers cannot be identified by their physical characteristics, stimuli possess no in- herent dimensions or properties that enable them to func- tion as discriminative stimuli. Operant conditioning brings behavior under the control of various properties or values of antecedent stimuli (e.g., size, shape, color, spatial relation to another stimulus), and what those fea- tures are cannot be determined a priori.

Any stimulus present when an operant is reinforced ac- quires control in the sense that the rate will be higher when it is present. Such a stimulus does not act as a goad; it does not elicit the response in the sense of forc- ing it to occur. It is simply an essential aspect of the oc- casion upon which a response is made and reinforced. The difference is made clear by calling it a discrimina- tive stimulus (or SD). An adequate formulation of the in- teraction between an organism and its environment must always specify three things: (1) the occasion upon which a response occurs; (2) the response itself; and (3) the re- inforcing consequences. The interrelationships among them are the ”contingencies of reinforcement.” (Skinner, 1969, p. 7)

The discriminated operant has its origin in the three- term contingency. The three-term contingency— antecedent, behavior, and consequence—is sometimes

Basic Concepts

Antecedent Stimulus Behavior Consequence

“Name a carnivorous dinosaur.” “Tyrannosaurus Rex.” “Well done!”

Foul smell under kitchen sink Take trash outside Foul smell is gone

Icy road Drive at normal speed Crash into car ahead

Popup box asks, “Warn when deleting unread

messages?” Click on “No”

Important e-mail message is lost

Future Frequency of Behavior in

Similar Conditions Operation

Positive Reinforcement

Negative Reinforcement

Positive Punishment

Negative Punishment

Figure 3 Three-term contingencies illustrating reinforcement and punishment operations.

called the ABCs of behavior analysis. Figure 3 shows examples of three-term contingencies for positive rein- forcement, negative reinforcement, positive punishment, and negative punishment.19 Most of what the science of behavior analysis has discovered about the prediction and control of human behavior involves the three-term con- tingency, which is “considered the basic unit of analysis in the analysis of operant behavior” (Glenn, Ellis, & Greenspoon, 1992, p. 1332).

The term contingency appears in behavior analysis literature with several meanings signifying various types of temporal and functional relations between behavior and antecedent and consequent variables (Lattal, 1995; Lattal & Shahan, 1997; Vollmer & Hackenberg, 2001). Perhaps the most common connotation of contingency refers to the dependency of a particular consequence on the occurrence of the behavior. When a reinforcer (or punisher) is said to be contingent on a particular behav- ior, the behavior must be emitted for the consequence to occur. For example, after saying, “Name a carnivorous

19Contingency diagrams, such as those shown in Figure 3, are an effec- tive way to illustrate temporal and functional relationships between be- havior and various environmental events. See Mattaini (1995) for examples of other types of contingency diagrams and suggestions for using them to teach and learn about behavior analysis. State notation is another means for visualizing complex contingency relations and experimental procedures (Mechner, 1959; Michael & Shafer, 1995).

20The phrase to make reinforcement contingent describes the behavior of the researcher or practitioner: delivering the reinforcer only after the tar- get behavior has occurred.

dinosaur,” a teacher’s “Well done!” depends on the stu- dent’s response, “Tyrannosaurus Rex” (or another di- nosaur of the same class).20

The term contingency is also used in reference to the temporal contiguity of behavior and its consequences. As stated previously, behavior is selected by the con- sequences that immediately follow it, irrespective of whether those consequences were produced by or de- pended on the behavior. This is the meaning of contin- gency in Skinner’s (1953) statement, “So far as the organism is concerned, the only important property of the contingency is temporal” (1953, p. 85).

Recognizing the Complexity of Human Behavior

Behavior—human or otherwise—remains an extremely difficult subject matter.

—B. F. Skinner (1969, p. 114)

The experimental analysis of behavior has discovered a number of basic principles—statements about how be- havior works as a function of environmental variables. These principles, several of which have been introduced

Basic Concepts

in this chapter, have been demonstrated, verified, and replicated in hundreds and even thousands of experi- ments; they are scientific facts.21 Tactics for changing be- havior derived for these principles have also been applied, in increasingly sophisticated and effective ways, to a wide range of human behaviors in natural settings. A summary of what has been learned from many of those applied be- havior analyses comprises the majority of this book.

The systematic application of behavior analysis tech- niques sometimes produces behavior changes of great magnitude and speed, even for clients whose behavior had been unaffected by other forms of treatment and ap- peared intractable. When such a happy (but not rare) out- come occurs, the neophyte behavior analyst must resist the tendency to believe that we know more than we do about the prediction and control of human behavior. Applied behavior analysis is a young science that has yet to achieve anything near a complete understanding and technological control of human behavior.

A major challenge facing applied behavior analysis lies in dealing with the complexity of human behavior, es- pecially in applied settings where laboratory controls are impossible, impractical, or unethical. Many of the fac- tors that contribute to the complexity of behavior stem from three general sources: the complexity of the human repertoire, the complexity of controlling variables, and individual differences.

Complexity of the Human Repertoire

Humans are capable of learning an incredible range of behaviors. Response sequences, sometimes of no appar- ent logical organization, contribute to the complexity of behavior (Skinner, 1953). In a response chain, effects produced by one response influence the emission of other responses. Returning a winter coat to the attic leads to rediscovering a scrapbook of old family photographs, which evokes a phone call to Aunt Helen, which sets the occasion for finding her recipe for apple pie, and so on.

Verbal behavior may be the most significant con- tributor to the complexity of human behavior (Donahoe & Palmer, 1994; Michael, 2003; Palmer, 1991; Skinner, 1957). Not only is a problem generated when the differ- ence between saying and doing is not recognized, but verbal behavior itself is often a controlling variable for many other verbal and nonverbal behaviors.

Operant learning does not always occur as a slow, gradual process. Sometimes new, complex, repertoires

21Like all scientific findings, these facts are subject to revision and even re- placement should future research reveal better ones.

appear quickly with little apparent direct conditioning (Epstein, 1991; Sidman, 1994). One type of rapid learn- ing has been called contingency adduction, a process whereby a behavior that was initially selected and shaped under one set of conditions is recruited by a different set of contingencies and takes on a new function in the person’s repertoire (Adronis, 1983; Layng & Adronis, 1984). Johnson and Layng (1992, 1994) described sev- eral examples of contingency adduction in which sim- ple (component) skills (e.g., addition, subtraction, and multiplication facts, isolating and solving for X in a sim- ple linear equation), when taught to fluency, combined without apparent instruction to form new complex (com- posite) patterns of behavior (e.g., factoring complex equations).

Intertwined lineages of different operants combine to form new complex operants (Glenn, 2004), which pro- duce response products that in turn make possible the ac- quisition of behaviors beyond the spatial and mechanical restraints of anatomical structure.

In the human case, the range of possibilities may be infi- nite, especially because the products of operant behavior have become increasingly complex in the context of evolving cultural practices. For example, anatomical constraints prevented operant flying from emerging in a human repertoire only until airplanes were constructed as behavioral products. Natural selection’s leash has been greatly relaxed in the ontogeny of operant units. (Glenn et al., 1992, p. 1332)

Complexity of Controlling Variables

Behavior is selected by its consequences. This megaprin- ciple of operant behavior sounds deceptively (and naively) simple. However, “Like other scientific principles, its simple form masks the complexity of the universe it de- scribes” (Glenn, 2004, p. 134). The environment and its effects on behavior are complex.

Skinner (1957) noted that, “(1) the strength of a sin- gle response may be, and usually is, a function of more than one variable and (2) a single variable usually affects more than one response” (p. 227). Although Skinner was writing in reference to verbal behavior, multiple causes and multiple effects are characteristics of many behavior– environment relations. Behavioral covariation illustrates one type of multiple effect. For example, Sprague and Horner (1992) found that blocking the emission of one problem behavior decreased the frequency of that be- havior but produced a collateral increase in other topogra- phies of problem behaviors in the same functional class. As another example of multiple effects, the presentation of an aversive stimulus may, in addition to suppressing the future occurrences of the behavior it follows, elicit

Basic Concepts

respondent behaviors and evoke escape and avoidance behaviors—three different effects from one event.

Many behaviors are the result of multiple causes. In a phenomena called joint control (Lowenkron, 2004), two discriminative stimuli can combine to evoke a common response class. Concurrent contingencies can also com- bine to make a behavior more or less likely to occur in a given situation. Perhaps we finally return our neighbor’s weed trimmer not just because he usually invites us in for a cup of coffee, but also because returning the tool re- duces the “guilt” we are feeling for keeping it for 2 weeks.

Concurrent contingencies often vie for control of in- compatible behaviors. We cannot watch “Baseball Tonight” and study (properly) for an upcoming exam. Al- though not a technical term in behavior analysis, algebraic summation is sometimes used to describe the effect of multiple, concurrent contingencies on behavior. The behavior that is emitted is thought to be the product of the competing contingencies “canceling portions of each other out” as in an equation in algebra.

Hierarchies of response classes within what was pre- sumed to be a single response class may be under multi- ple controlling variables. For example, Richman, Wacker, Asmus, Casey, and Andelman (1999) found that one topography of aggressive behavior was maintained by one type of reinforcement contingency while another form of aggression was controlled by a different contingency.

All of these complex, concurrent, interrelated con- tingencies make it difficult for behavior analysts to iden- tify and control relevant variables. It should not be surprising that the settings in which applied behavior an- alysts ply their trade are sometimes described as places where “reinforcement occurs in a noisy background” (Vollmer & Hackenberg, 2001, p. 251).

Consequently, as behavior analysts, we should rec- ognize that meaningful behavior change might take time and many trials and errors as we work to understand the interrelationships and complexities of the controlling vari- ables. Don Baer (1987) recognized that some of the larger problems that beset society (e.g., poverty, substance ad- diction, illiteracy), given our present level of technology, might be too difficult to solve. He identified three barri- ers to solving such complex problems:

(a) We are not empowered to solve these bigger remain- ing problems, (b) we have not yet made the analysis of how to empower ourselves to try them, and (c) we have not yet made the system-analytic task analyses that will prove crucial to solving those problems when we do em- power ourselves sufficiently to try them. . . . In my ex- perience, those projects that seem arduously long are arduous because (a) I do not have a strong interim rein- forcer compared to those in the existing system for sta- tus quo and must wait for opportunities when weak control may operate, even so, or (b) I do not yet have a

correct task analysis of the problem and must struggle through trials and errors. By contrast (c) when I have an effective interim reinforcer and I know the correct task analysis of this problem, long problems are simply those in which the task analysis requires a series of many be- havior changes, perhaps in many people, and although each of them is relatively easy and quick, the series of them requires not so much effort as time, and so it is not arduous but merely tedious. (pp. 335, 336–337)

Individual Differences

You did not need to read this textbook to know that peo- ple often respond very differently to the same set of en- vironmental conditions. The fact of individual differences is sometimes cited as evidence that principles of behav- ior based on environmental selection do not exist, at least not in a form that could provide the basis for a robust and reliable technology of behavior change. It is then argued that because people often respond differently to the same set of contingencies, control of behavior must come from within each person.

As each of us experiences varying contingencies of reinforcement (and punishment), some behaviors are strengthened (selected by the contingencies) and others are weakened. This is the nature of operant conditioning, which is to say, human nature. Because no two people ever experience the world in exactly the same way, each of us arrives at a given situation with a different history of reinforcement. The repertoire of behaviors each per- son brings to any situation has been selected, shaped, and maintained by his or her unique history of reinforcement. Each human’s unique repertoire defines him or her as a person. We are what we do, and we do what we have learned to do. “He begins as an organism and becomes a person or self as he acquires a repertoire of behavior” (Skinner, 1974, p. 231).

Individual differences in responding to current stim- ulus conditions, then, do not need to be attributed to dif- ferences in internal traits or tendencies, but to the orderly result of different histories of reinforcement. The behav- ior analyst must also consider people’s varying sensitiv- ities to stimuli (e.g., hearing loss, visual impairment) and differences in response mechanisms (e.g., cerebral palsy) and design program components to ensure that all par- ticipants have maximum contact with relevant contin- gencies (Heward, 2006).

Additional Obstacles to Controlling Behavior in Applied Settings

Compounding the difficulty of tackling the complexity of human behavior in the “noisy” applied settings where people live, work, and play, applied behavior anal- ysts are sometimes prevented from implementing an

Behavior

1. In general, behavior is the activity of living organisms.

2. Technically, behavior is “that portion of an organism’s in- teraction with its environment that is characterized by de- tectable displacement in space through time of some part of the organism and that results in a measurable change in at least one aspect of the environment” (Johnston & Pennypacker, 1993a, p. 23).

3. The term behavior is usually used in reference to a larger set or class of responses that share certain topographical di- mensions or functions.

4. Response refers to a specific instance of behavior.

5. Response topography refers to the physical shape or form of behavior.

6. A response class is a group of responses of varying topography, all of which produce the same effect on the environment.

7. Repertoire can refer to all of the behaviors a person can do or to a set of behaviors relevant to a particular setting or task.

Environment

8. Environment is the physical setting and circumstances in which the organism or referenced part of the organ- ism exists.

9. Stimulus is “an energy change that affects an organism through its receptor cells” (Michael, 2004, p. 7).

10. The environment influences behavior primarily by stimu- lus change, not static stimulus conditions.

11. Stimulus events can be described formally (by their phys- ical features), temporally (by when they occur), and func- tionally (by their effects on behavior).

12. A stimulus class is a group of stimuli that share specified common elements along formal, temporal, and/or func- tional dimensions.

13. Antecedent conditions or stimulus changes exist or occur prior to the behavior of interest.

14. Consequences are stimulus changes that follow a behav- ior of interest.

15. Stimulus changes can have one or both of two basic ef- fects on behavior: (a) an immediate but temporary effect of increasing or decreasing the current frequency of the behavior, and/or (b) a delayed but relatively permanent ef- fect in terms of the frequency of that type of behavior in the future.

Respondent Behavior

16. Respondent behavior is elicited by antecedent stimuli.

17. A reflex is a stimulus–response relation consisting of an antecedent stimulus and the respondent behavior it elicits (e.g., bright light–pupil contraction).

18. All healthy members of a given species are born with the same repertoire of unconditioned reflexes.

Basic Concepts

effective behavior change program due to logistical, fi- nancial, sociopolitical, legal, and/or ethical factors. Most applied behavior analysts work for agencies with lim- ited resources, which may make the data collection re- quired for a more complete analysis impossible. In addition, participants, parents, administrators, and even the general public may at times limit the behavior ana- lyst’s options for effective intervention (e.g., “We don’t want students working for tokens”). Legal or ethical con- siderations may also preclude determining experimen- tally the controlling variables for an important behavior.

Each of these practical complexities combines with the behavioral and environmental complexities previously mentioned to make the applied behavior analysis of so- cially important behavior a challenging task. However, the task need not be overwhelming, and few tasks are as re- warding or as important for the betterment of humankind.

It is sometimes expressed that a scientific account of behavior will somehow diminish the quality or enjoy- ment of the human experience. For example, will our in-

creasing knowledge of the variables responsible for cre- ative behavior lessen the feelings evoked by a powerful painting or a beautiful symphony, or reduce our appreci- ation of the artists who produced them? We think not, and we encourage you, as you read and study about the basic concepts introduced in this chapter and examined in more detail throughout the book, to consider Nevin’s (2005) response to how a scientific account of behavior adds immeasurably to the human experience:

At the end of Origin of Species (1859), Darwin invites us to contemplate a tangled bank, with its plants and its birds, its insects and its worms; to marvel at the com- plexity, diversity, and interdependence of its inhabitants; and to feel awe at the fact that all of it follows from the laws of reproduction, competition, and natural selection. Our delight in the tangled bank and our love for its in- habitants are not diminished by our knowledge of the laws of evolution; neither should our delight in the com- plex world of human activity and our love for its actors be diminished by our tentative but growing knowledge of the laws of behavior. (Tony Nevin, personal commu- nication, December 19, 2005)

Summary

Basic Concepts

19. An unconditioned stimulus (e.g., food) and the respondent behavior it elicits (e.g., salivation) are called unconditioned reflexes.

20. Conditioned reflexes are the product of respondent con- ditioning: a stimulus–stimulus pairing procedure in which a neutral stimulus is presented with an unconditioned stim- ulus until the neutral stimulus becomes a conditioned stim- ulus that elicits the conditioned response.

21. Pairing a neutral stimulus with a conditioned stimulus can also produce a conditioned reflex—a process called higher order (or secondary) respondent conditioning.

22. Respondent extinction occurs when a conditioned stimu- lus is presented repeatedly without the unconditioned stim- ulus until the conditioned stimulus no longer elicits the conditioned response.

Operant Behavior

23. Operant behavior is selected by its consequences.

24. Unlike respondent behavior, whose topography and basic functions are predetermined, operant behavior can take a virtually unlimited range of forms.

25. Selection of behavior by consequences operates during the lifetime of the individual organism (ontogeny) and is a conceptual parallel to Darwin’s natural selection in the evolutionary history of a species (phylogeny).

26. Operant conditioning, which encompasses reinforcement and punishment, refers to the process and selective effects of consequences on behavior:

• Consequences can affect only future behavior.

• Consequences select response classes, not individual responses.

• Immediate consequences have the greatest effect.

• Consequences select any behavior that precedes them.

• Operant conditioning occurs automatically.

27. Most stimulus changes that function as reinforcers or pun- ishers can be described as either (a) a new stimulus added to the environment, or (b) an already present stimulus re- moved from the environment.

28. Positive reinforcement occurs when a behavior is followed immediately by the presentation of a stimulus that in- creases the future frequency of the behavior.

29. Negative reinforcement occurs when a behavior is fol- lowed immediately by the withdrawal of a stimulus that in- creases the future frequency of the behavior.

30. The term aversive stimulus is often used to refer to stimulus conditions whose termination functions as reinforcement.

31. Extinction (withholding all reinforcement for a previously reinforced behavior) produces a decrease in response fre- quency to the behavior’s prereinforcement level.

32. Positive punishment occurs when a behavior is followed by the presentation of a stimulus that decreases the future frequency of the behavior.

33. Negative punishment occurs when a behavior is followed immediately by the withdrawal of a stimulus that decreases the future frequency of the behavior.

34. A principle of behavior describes a functional relation be- tween behavior and one or more of its controlling vari- ables that has thorough generality across organisms, species, settings, and behaviors.

35. A behavior change tactic is a technologically consistent method for changing behavior that has been derived from one or more basic principles of behavior.

36. Unconditioned reinforcers and punishers function irre- spective of any prior learning history.

37. Stimulus changes that function as conditioned reinforcers and punishers do so because of previous pairing with other reinforcers or punishers.

38. One important function of motivating operations is alter- ing the current value of stimulus changes as reinforcement or punishment. For example, deprivation and satiation are motivating operations that make food more or less effec- tive as reinforcement.

39. A discriminated operant occurs more frequently under some antecedent conditions than it does under others, an outcome called stimulus control.

40. Stimulus control refers to differential rates of operant re- sponding observed in the presence or absence of an- tecedent stimuli. Antecedent stimuli acquire the ability to control operant behavior by having been paired with cer- tain consequences in the past.

41. The three-term contingency—antecedent, behavior, and consequence—is the basic unit of analysis in the analysis of operant behavior.

42. If a reinforcer (or punisher) is contingent on a particular behavior, the behavior must be emitted for the conse- quence to occur.

43. All applied behavior analysis procedures involve ma- nipulation of one or more components of the three-term contingency.

Recognizing the Complexity of Human Behavior

44. Humans are capable of acquiring a huge repertoire of be- haviors. Response chains and verbal behavior also make human behavior extremely complex.

45. The variables that govern human behavior are often highly complex. Many behaviors have multiple causes.

46. Individual differences in histories of reinforcement and organic impairments also make the analysis and control of human behavior difficult.

47. Applied behavior analysts are sometimes prevented from conducting an effective analysis of behavior because of practical, logistical, financial, sociopolitical, legal, and/or ethical reasons.

Selecting and Defining Target Behaviors

Key Terms ABC recording anecdotal observation behavior checklist behavioral assessment behavioral cusp

ecological assessment function-based definition habilitation normalization pivotal behavior

reactivity relevance of behavior rule social validity target behavior topography-based definition

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List ©, Third Edition

Content Area 1: Ethical Considerations

1-5 Assist the client with identifying lifestyle or systems change goals and targets for behavior change that are consistent with:

(a) the applied dimension of applied behavior analysis.

(b) applicable laws.

Content Area 8: Selecting Intervention Outcomes and Strategies

8-2 Make recommendations to the client regarding target outcomes based on such factors as client preferences, task analysis, current repertoires, supporting environments, constraints, social validity, assessment results, and best available scientific evidence.

8-3 State target intervention outcomes in observable and measurable terms.

8-5 Make recommendations to the client regarding behaviors that must be estab- lished, strengthened, and/or weakened to attain the stated intervention outcomes.

Content Area 6: Measurement of Behavior

6-2 Define behavior in observable and measurable terms.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and questions about this document must be submitted directly to the BACB.

Applied behavior analysis is concerned with producing predictable and replicable improve- ments in behavior. However, not just any behav-

ior will do: Applied behavior analysts improve socially important behaviors that have immediate and long-lasting meaning for the person and for those who interact with that person. For instance, applied behavior analysts de- velop language, social, motor, and academic skills that make contact with reinforcers and avoid punishers. An important preliminary step involves choosing the right be- haviors to target for measurement and change.

This chapter describes the role of assessment in ap- plied behavior analysis, including preassessment con- siderations, assessment methods used by behavior analysts, issues in determining the social significance of potential target behaviors, considerations for prioritizing target behaviors, and the criteria and dimensions by which selected behaviors should be defined to enable accurate and reliable measurement.

Role of Assessment in Applied Behavior Analysis Assessment is considered a linchpin in a four-phase sys- tematic intervention model that includes: assessment, planning, implementation, and evaluation (Taylor, 2006).

Definition and Purpose of Behavioral Assessment

Traditional psychological and educational assessments typically involve a series of norm- and/or criterion-refer- enced standardized tests to determine a person’s strengths and weaknesses within cognitive, academic, social, and/or psychomotor domains. Behavioral assessment involves a variety of methods including direct observations, inter- views, checklists, and tests to identify and define targets for behavior change. In addition to identifying behav- ior(s) to change, comprehensive behavioral assessment will discover resources, assets, significant others, com- peting contingencies, maintenance and generalization fac- tors, and potential reinforcers and/or punishers that may inform or be included in intervention plans to change the target behavior (Snell & Brown, 2006).1

Linehan (1977) offered a succinct and accurate de- scription of the purpose of behavioral assessment: “To figure out what the client’s problem is and how to change

1The behavioral assessment of problem behaviors sometimes includes a three-step process, called functional behavior assessment, for identifying and systematically manipulating antecedents and/or consequences that may be functioning as controlling variables for problems. Chapter entitled “Functional Behavior Assessment” describes this process in detail.

2See O’Neill and colleagues (1997) for their discussion of competing be- havior analysis as a bridge between functional assessment and subsequent intervention programs.

it for the better” (p. 31). Implicit in Linehan’s statement is the idea that behavioral assessment is more than an ex- ercise in describing and classifying behavioral abilities and deficiencies. Behavioral assessment goes beyond try- ing to obtain a psychometric score, grade equivalent data, or rating measure, as worthy as such findings might be for other purposes. Behavioral assessment seeks to dis- cover the function that behavior serves in the person’s en- vironment (e.g., positive reinforcement by social attention, negative reinforcement by escape from a task). Results of a comprehensive behavioral assessment give the be- havior analyst a picture of variables that increase, de- crease, maintain, or generalize the behavior of interest. A well-constructed and thorough behavioral assessment provides a roadmap from which the variables controlling the behavior can be identified and understood. Conse- quently, subsequent interventions can be aimed more directly and have a much better chance of success. As Bourret, Vollmer, and Rapp (2004) pointed out: “The critical test of . . . assessment is the degree to which it differentially indicates an effective teaching strategy” (p. 140).

Phases of Behavioral Assessment

Hawkins (1979) conceptualized behavioral assessment as funnel shaped, with an initial broad scope leading to an eventual narrow and constant focus. He described five phases or functions of behavioral assessment: (a) screen- ing and general disposition, (b) defining and generally quantifying problems or desired achievement criteria, (c) pinpointing the target behavior(s) to be treated, (d) mon- itoring progress, and (e) following up. Although the five phases form a general chronological sequence, there is often overlap. This chapter is concerned with the prein- tervention functions of assessment, the selection and de- finition of a target behavior—the specific behavior selected for change.

To serve competently, applied behavior analysts must know what constitutes socially important behavior, have the technical skills to use appropriate assessment meth- ods and instruments, and be able to match assessment data with an intervention strategy.2 For instance, the re- medial reading specialist must understand the critical be- haviors of a competent reader, be able to determine which of those skills a beginning or struggling reader lacks, and deliver an appropriate and effective instruction as inter- vention. Likewise, the behaviorally trained marriage and family therapist must be knowledgeable about the range

Selecting and Defining Target Behaviors

of behaviors that constitutes a functional family, be able to assess family dynamics accurately, and provide so- cially acceptable interventions that reduce dysfunctional interactions. In short, any analyst must be knowledge- able about the context of the target behavior.

A Preassessment Consideration

Before conducting an informal or formal behavioral as- sessment for the purpose of pinpointing a target behav- ior, the analyst must address the fundamental question, Who has the authority, permission, resources, and skills to complete an assessment and intervene with the be- havior? If a practitioner does not have authority or per- mission, then his role in assessment and intervention is restricted. For example, suppose that a behavior analyst is standing in a checkout line near a parent attempting to manage an extremely disruptive child. Does the behavior analyst have the authority or permission to assess the problem or suggest an intervention to the parent? No. However, if the same episode occurred after the parent had requested assistance with such problems, the behav- ior analyst could offer assessment and advice. In effect, applied behavior analysts must not only recognize the role of assessment in the assessment–intervention con- tinuum, but also recognize those situations in which using their knowledge and skills to assess and change behavior is appropriate.3

Assessment Methods Used by Behavior Analysts Four major methods for obtaining assessment information are (a) interviews, (b) checklists, (c) tests, and (d) direct observation. Interviews and checklists are indirect as- sessment approaches because the data obtained from these measures are derived from recollections, recon- structions, or subjective ratings of events. Tests and direct observation are considered direct assessment approaches because they provide information about a person’s behavior as it occurs (Miltenberger, 2004). Although in- direct assessment methods often provide useful infor- mation, direct assessment methods are preferred because they provide objective data on the person’s actual per- formance, not an interpretation, ranking, or qualitative index of that performance (Hawkins, Mathews, & Ham- dan, 1999; Heward, 2003). In addition to being aware of these four major assessment methods, analysts can provide further assistance to those they serve by increas- ing their skills relative to the ecological implications of assessment.

3“Ethical Considerations for Applied Behavior Analysts,” examines this important issue in detail.

Interviews

Assessment interviews can be conducted with the target person and/or with people who come into daily or regu- lar contact with the individual (e.g., teachers, parents, care providers).

Interviewing the Person

A behavioral interview is often a first and important step in identifying a list of potential target behaviors, which can be verified or rejected by subsequent direct observa- tion. The interview can be considered a direct assessment method when the person’s verbal behavior is of interest as a potential target behavior (Hawkins, 1975).

A behavioral interview differs from a traditional in- terview by the type of questions asked and the level of in- formation sought. Behavior analysts rely primarily on what and when questions that focus on the environmen- tal conditions that exist before, during, and after a be- havioral episode, instead of why questions, which tend to evoke mentalistic explanations that are of little value in understanding the problem.

Asking the clients why they do something presumes they know the answer and is often frustrating to clients, because they probably do not know and it seems that they should (Kadushin, 1972).

“Why” questions encourage the offering of “motiva- tional” reasons that are usually uninformative such as “I’m just lazy.” Instead, the client could be asked “What happens when . . . ?” One looks closely at what actually happens in the natural environment. Attention is directed toward behavior by questions that focus on it, such as, “Can you give me an example of what [you do]?” When one example is gained, then another can be requested until it seems that the set of behaviors to which the client refers when he employs a given word have been identified. (Gambrill, 1977, p. 153)

Figure 1 provides examples of the kinds of what and when questions that can be applied during a behav- ioral assessment interview. This sequence of questions was developed by a behavioral consultant in response to a teacher who wanted to reduce the frequency of her neg- ative reactions to acting out and disruptive students. Sim- ilar questions could be generated to address situations in homes or community settings (Sugai & Tindal, 1993).

The main purpose of the behavioral assessment in- terview questions in Figure 1 is to identify variables that occur before, during, and/or after the occurrence of negative teacher attending behavior. Identifying envi- ronmental events that correlate with the behavior pro- vides valuable information for formulating hypotheses about the controlling function of these variables and for

Selecting and Defining Target Behaviors

Figure 1 Sample behavioral interview questions.

Problem Identification Interview Form Reason for referral: The teacher requested help reducing her negative attention to acting out and disruptive students who were yelling and noncompliant.

1. In your own words, can you define the problem behaviors that prompted your request?

2. Are there any other teacher-based behaviors that concern you at this time? 3. When you engage in negative teacher attention (i.e., when you attend to yelling or

noncompliant behavior), what usually happens immediately before the negative teacher attending behavior occurs?

4. What usually happens after the negative teacher attention behavior occurs? 5. What are the students’ reactions when you yell or attend to their noncompliant

behavior? 6. What behaviors would the students need to perform so that you would be less likely

to attend to them in a negative way? 7. Have you tried other interventions? What has been their effect?

planning interventions. Hypothesis generation leads to experimental manipulation and the discovery of func- tional relations.

As an outgrowth of being interviewed, clients may be asked to complete questionnaires or so-called needs assessment surveys. Questionnaires and needs assessment surveys have been developed in many human services areas to refine or extend the interview process (Altschuld & Witkin, 2000). Sometimes as a result of an initial in- terview, the client is asked to self-monitor her behavior in particular situations. Self-monitoring can entail writ- ten or tape-recorded accounts of specific events.4 Client- collected data can be useful in selecting and defining target behaviors for further assessment or for intervention. For example, a client seeking behavioral treatment to quit smoking might self-record the number of cigarettes he smokes each day and the conditions when he smokes (e.g., morning coffee break, after dinner, stuck in a traffic jam). These client-collected data may shed light on antecedent conditions correlated with the target behavior.

Interviewing Significant Others

Sometimes the behavior analyst either cannot interview the client personally or needs information from others who are important in the client’s life (e.g., parents, teach- ers, coworkers). In such cases, the analyst will interview one or more of these significant others. When asked to de- scribe a behavioral problem or deficit, significant others often begin with general terms that do not identify spe- cific behaviors to change and often imply causal factors intrinsic to the client (e.g., she is afraid, aggressive, un- motivated or lazy, withdrawn). By asking variations of what, when, and how questions, the behavior analyst can

4Procedures for self-monitoring, a major component of many self- management interventions, are described in Chapter entitled “Self- Management.”

help significant others describe the problem in terms of specific behaviors and environmental conditions and events associated with those behaviors. For example, the following questions could be used in interviewing par- ents who have asked the behavior analyst for help be- cause their child is “noncompliant” and “immature.”

• What is Derek doing when you would be most likely to call him immature or noncompliant?

• During what time of day does Derek seem most im- mature (or noncompliant)? What does he do then?

• Are there certain situations or places where Derek is noncompliant or acts immature? If so, where, and what does he do?

• How many different ways does Derek act imma- ture (or noncompliant)?

• What’s the most frequent noncompliant thing that Derek does?

• How do you and other family members respond when Derek does these things?

• If Derek were to be more mature and independent as you would like, what would he do differently than he does now?

Figure 2 shows a form that parents or significant others can use to begin to identify target behaviors.

In addition to seeking help from significant others in identifying target behaviors and possible controlling vari- ables that can help inform an intervention plan, the be- havior analyst can sometimes use the interview to determine the extent to which significant others are will- ing and able to help implement an intervention to change the behavior. Without the assistance of parents, siblings, teacher aides, and staff, many behavior change programs cannot be successful.

Selecting and Defining Target Behaviors

Figure 2 Form that parents or significant others in a person’s life can use to generate starter lists of possible target behaviors.

The 5 + 5 Behavior List

Child’s name: ____________________________________________________________

Person completing this list: __________________________________________________

Listmaker’s relationship to child: ____________________________________________

5 good things _____ does now 5 things I’d like to see _____ learn to do more (or less) often

1. ______________________________ 1. ________________________________

2. ______________________________ 2. ________________________________

3. ______________________________ 3. ________________________________

4. ______________________________ 4. ________________________________

5. ______________________________ 5. ________________________________

Directions: Begin by listing in the left-hand column 5 desirable behaviors your child (or student) does regularly now; things that you want him or her to continue doing. Next, list in the right-hand column 5 behaviors you would like to see your child do more often (things that your child does sometimes but should do with more regularity) and/or undesirable behaviors that you want him or her to do less often (or not at all). You may list more than 5 behaviors in either column, but try to identify at least 5 in each.

Checklists

Behavior checklists and rating scales can be used alone or in combination with interviews to identify potential target behaviors. A behavior checklist provides de- scriptions of specific behaviors (usually in hierarchical order) and the conditions under which each behavior should occur. Situation- or program-specific checklists can be created to assess one particular behavior (e.g., tooth brushing) or a specific skill area (e.g., a social skill), but most practitioners use published checklists to rate a wide range of areas (e.g., The Functional Assessment Checklist for Teachers and Staff [March et al., 2000]).

Usually a Likert scale is used that includes informa- tion about antecedent and consequence events that may affect the frequency, intensity, or duration of behaviors. For example, the Child Behavior Checklist (CBCL) comes in teacher report, parent report, and child report

forms and can be used with children ages 5 through 18 (Achenbach & Edelbrock, 1991). The teacher’s form in- cludes 112 behaviors (e.g., “cries a lot,” “not liked by other pupils”) that are rated on a 3-point scale: “not true,” “somewhat or sometimes true,” or “very true or often true.” The CBCL also includes items representing social com- petencies and adaptive functioning such as getting along with others and acting happy.

The Adaptive Behavior Scale—School (ABS-S) (Lambert, Nihira, & Leland, 1993) is another frequently used checklist for assessing children’s adaptive behav- ior. Part 1 of the ABS-S contains 10 domains related to in- dependent functioning and daily living skills (e.g., eating, toilet use, money handling, numbers, time); Part 2 as- sesses the child’s level of maladaptive (inappropriate) be- havior in seven areas (e.g., trustworthiness, self-abusive behavior, social engagement). Another version of the Adaptive Behavior Scale, the ABS-RC, assesses adaptive

Selecting and Defining Target Behaviors

behavior in residential and community settings (Nihira, Leland, & Lambert, 1993).

Information obtained from good behavior checklists (i.e., those with objectively stated items of relevance to the client’s life) can help identify behaviors worthy of more direct and intensive assessment.

Standardized Tests

Literally thousands of standardized tests and assessment devices have been developed to assess behavior (cf., Spies & Plake, 2005). Each time a standardized test is admin- istered, the same questions and tasks are presented in a specified way and the same scoring criteria and proce- dures are used. Some standardized tests yield norm- referenced scores. When a norm-referenced test is being developed, it is administered to a large sample of people selected at random from the population for whom the test is intended. Test scores of people in the norming sample are then used to represent how scores on the test are gen- erally distributed throughout the population.

The majority of standardized tests on the market, however, are not conducive to behavioral assessment be- cause the results cannot be translated directly into target behaviors for instruction or treatment. For example, re- sults from standardized tests commonly used in the schools, such as the Iowa Tests of Basic Skills (Hoover, Hieronymus, Dunbar, & Frisbie, 1996), the Peabody In- dividual Achievement Test–R/NU (Markwardt, 2005), and the Wide Range Achievement Test—3 (WRAT) (Wilkin- son, 1994), might indicate that a fourth-grader is per- forming at the third-grade level in mathematics and at the first-grade level in reading. Such information might be useful in determining how the student performs in these subjects compared to students in general, but it nei- ther indicates the specific math or reading skills the stu- dent has mastered nor provides sufficient direct context with which to launch an enrichment or remedial pro- gram. Further, behavior analysts may not be able to ac- tually administer a given test because of licensing requirements. For instance, only a licensed psychologist can administer some types of intelligence tests and per- sonality inventories.

Tests are most useful as behavioral assessment de- vices when they provide a direct measure of the person’s performance of the behaviors of interest. In recent years, an increasing number of behaviorally oriented teachers have recognized the value of criterion-referenced and curriculum-based assessments to indicate exactly which skills students need to learn and, equally important, which skills they have mastered (Browder, 2001; Howell, 1998). Curriculum-based assessments can be considered direct measures of student performance because the data that

are obtained bear specifically on the daily tasks that the student performs (Overton, 2006).

Direct Observation

Direct and repeated observations of the client’s behavior in the natural environment are the preferred method for determining which behaviors to target for change. A basic form of direct continuous observation, first described by Bijou, Peterson, and Ault (1968), is called anecdotal ob- servation, or ABC recording. With anecdotal observa- tion the observer records a descriptive, temporally sequenced account of all behavior(s) of interest and the antecedent conditions and consequences for those be- haviors as those events occur in the client’s natural envi- ronment (Cooper, 1981). This technique produces behavioral assessment data that can be used to identify potential target behaviors.

Rather than providing data on the frequency of a spe- cific behavior, anecdotal observation yields an overall description of a client’s behavior patterns. This detailed record of the client’s behavior within its natural context provides accountability to the individual and to others in- volved in the behavior change plan and is extremely help- ful in designing interventions (Hawkins et al., 1999).

Accurately describing behavioral episodes as they occur in real time is aided by using a form to record rel- evant antecedents, behaviors, and consequences in tem- poral sequence. For example, Lo (2003) used the form shown in Figure 3 to record anecdotal observations of a fourth-grade special education student whose teacher had complained that the boy’s frequent talk-outs and out- of-seat behavior were impairing his learning and often disrupted the entire class. (ABC observations can also be recorded on a checklist of specific antecedents, behav- iors, and consequent events individually created for the client based on information from interviews and/or initial observations.)

ABC recording requires the observer to commit full attention to the person being observed. A classroom teacher, for example, could not use this assessment pro- cedure while engaging in other activities, such as man- aging a reading group, demonstrating a mathematics problem on the chalkboard, or grading papers. Anecdo- tal observation is usually conducted for continuous peri- ods of 20 to 30 minutes, and when responsibilities can be shifted temporarily (e.g., during team teaching). Fol- lowing are some additional guidelines and suggestions for conducting anecdotal direct observations:

• Write down everything the client does and says and everything that happens to the client.

Selecting and Defining Target Behaviors

Figure 3 Example of an anecdotal ABC recording form.

Student: Student 4 Date: 3/10/03 Setting: SED resource room (math period)

Observer: Experimenter Starting time: 2:40 P.M. Ending time: 3:00 P.M.

Time Antecedents (A) Behavior (B) Consequences (C)

2:40 T tells students to work Walks around the room T says,“Everyone is working, quietly on their math and looks at other students but you. I don’t need to

worksheets tell you what to do.”

✓ Sits down, makes funny noises A female peer says:“Would with mouth you please be quiet?”

✓ Says to the peer,“What? Me?” Peer continues to work Stops making noises

2:41 Math worksheet Sits in his seat and works No one pays attention quietly to him

Math worksheet Pounds on desk with hands SED aide asks him to stop

2:45 Math worksheet Makes vocal noises Ignored by others

Math worksheet Yells-out T’s name three times T helps him with the and walks to her with her questions

worksheet

2:47 Everyone is working quietly Gets up and leaves his seat T asks him to sit down and work

✓ Sits down and works Ignored by others

Everyone is working quietly Gets up and talks to a peer T asks him to sit down and work

✓ Comes back to his seat Ignored by others and works

2:55 Math worksheet, no one is Hand grabs a male peer and Peer refuses attending to him asks him to help on the

worksheet

✓ Asks another male peer to Peer helps him help him

2:58 ✓ Tells T he’s finished the work T asks him to turn in his and it’s his turn to work on work and tells him that it’s

computer not his turn on the computer

✓ Whines about why it’s not T explains to him that other his turn students are still working

on the computer and he needs to find a book to read

✓ Stands behind a peer who Ignored by T is playing computer and

watches him play the game

Adapted from Functional Assessment and Individualized Intervention Plans: Increasing the Behavior Adjustment of Urban Learners in General and Special Education Settings (p. 317) by Y. Lo. Unpublished doctoral dissertation. Columbus, OH:The Ohio State University. Used by permission.

• Use homemade shorthand or abbreviations to make recording more efficient, but be sure the notes can be and are accurately expanded immediately after the observation session.

• Record only actions that are seen or heard, not in- terpretations of those actions.

• Record the temporal sequence of each response of interest by writing down what happened just before and just after it.

• Record the estimated duration of each instance of the client’s behavior. Mark the beginning and end- ing time of each behavioral episode.

• Be aware that continuous anecdotal observation is often an obtrusive recording method. Most people behave differently when they see someone with a pencil and clipboard staring at them. Knowing this, observers should be as unobtrusive as possible (e.g., stay a reasonable distance away from the subject).

Selecting and Defining Target Behaviors

• Carry out the observations over a period of several days so that the novelty of having someone observe the client will lessen and the repeated observations can produce a valid picture of day-to-day behavior.

Ecological Assessment

Behavior analysts understand that human behavior is a function of multiple events and that many events have multiple effects on behavior (cf., Michael, 1995). An eco- logical approach to assessment recognizes the complex interrelationships between environment and behavior. In an ecological assessment a great deal of information is gathered about the person and the various environments in which that person lives and works. Among the many factors that can affect a person’s behavior are physiolog- ical conditions, physical aspects of the environment (e.g., lighting, seating arrangements, noise level), interactions with others, home environment, and past reinforcement history. Each of these factors represents a potential area for assessment.

Although a thorough ecological assessment will pro- vide a tremendous amount of descriptive data, the basic purpose of assessment—to identify the most pressing be- havior problem and possible ways to alleviate it—should not be forgotten. It is easy to go overboard with the eco- logical approach, gathering far more information than necessary. Ecological assessment can be costly in terms of professional and client time, and it may raise ethical and perhaps legal questions regarding confidentiality (Koocher & Keith-Spiegel, 1998). Ultimately, good judg- ment must be used in determining how much assessment information is necessary. Writing about the role of eco- logical assessment for special education teachers, Heron and Heward (1988) suggested that

The key to using an ecological assessment is to know when to use it. Full-scale ecological assessments for their own sake are not recommended for teachers charged with imparting a great number of important skills to many children in a limited amount of time. In most cases, the time and effort spent conducting an ex- haustive ecological assessment would be better used in direct instruction. While the results of an ecological as- sessment might prove interesting, they do not always change the course of a planned intervention. Under what conditions then will an ecological assessment yield data that will significantly affect the course of treatment? Herein lies the challenge. Educators must strive to be- come keen discriminators of: (1) situations in which a planned intervention has the potential for affecting stu- dent behaviors other than the behavior of concern; and (2) situations in which an intervention, estimated to be effective if the target behavior were viewed in isolation, may be ineffective because other ecological variables

5Reactive effects of assessment are not necessarily negative. Self monitor- ing has become as much a treatment procedure as it is an assessment pro- cedure; see Chapter entitled “Self-Management.”

come into play. Regardless of the amount and range of information available concerning the student, a teacher must still make instructional decisions based on an em- pirical analysis of the target behavior. Ultimately, this careful analysis (i.e., direct and daily measurement) of the behavior of interest may be ineffective because other ecological variables come into play. (p. 231)

Reactive Effects of Direct Assessment

Reactivity refers to the effects of an assessment proce- dure on the behavior being assessed (Kazdin, 1979). Re- activity is most likely when observation is obtrusive—that is, the person being observed is aware of the observer’s presence and purpose (Kazdin, 2001). Numerous stud- ies have demonstrated that the presence of observers in applied settings can influence a subject’s behavior (Mer- catoris & Craighead, 1974; Surratt, Ulrich, & Hawkins, 1969; White, 1977). Perhaps the most obtrusive assess- ment procedures are those that require the subject to mon- itor and record her own behavior. Research on self-monitoring shows that the procedure commonly af- fects the behavior under assessment (Kirby, Fowler, & Baer, 1991).5

Although research suggests that even when the pres- ence of an observer alters the behavior of the people being observed, the reactive effects are usually temporary (e.g., Haynes & Horn, 1982; Kazdin, 1982). Nevertheless, be- havior analysts should use assessment methods that are as unobtrusive as possible, repeat observations until ap- parent reactive effects subside, and take possible reactive effects into account when interpreting the results of ob- servations.

Assessing the Social Significance of Potential Target Behaviors In the past, when a teacher, therapist, or other human ser- vices professional determined that a client’s behavior should be changed, few questions were asked. It was as- sumed that the change would be beneficial to the person. This presumption of benevolence is no longer ethically acceptable (not that it ever was). Because behavior ana- lysts possess an effective technology to change behavior in predetermined directions, accountability must be served. Both the goals and the rationale supporting

Selecting and Defining Target Behaviors

behavior change programs must be open to critical ex- amination by the consumers (clients and their families) and by others who may be affected (society) by the be- havior analyst’s work. In selecting target behaviors, prac- titioners should consider whose behavior is being assessed—and changed—and why.

Target behaviors should not be selected for the pri- mary benefit of others (e.g., “Be still, be quiet, be docile,” in Winett & Winkler, 1972), to simply maintain the sta- tus quo (Budd & Baer, 1976; Holland, 1978), or because they pique the interest of someone in a position to change the behaviors, as illustrated in the following incident:

A bright, conscientious graduate student was interested in doing his thesis in a program for seriously malad- justed children. He wanted to teach cursive writing to a . . . child [who] could not read (except his name), print, or even reliably identify all the letters of the alphabet. I asked “Who decided that was the problem to work on next?” (Hawkins, 1975, p. 195)

Judgments about which behaviors to change are dif- ficult to make. Still, practitioners are not without direc- tion when choosing target behaviors. Numerous authors have suggested guidelines and criteria for selecting tar- get behaviors (e.g., Ayllon & Azrin, 1968; Bailey & Lessen, 1984; Bosch & Fuqua, 2001; Hawkins, 1984; Komaki, 1998; Rosales-Ruiz & Baer, 1997), all of which revolve around the central question, To what extent will the proposed behavior change improve the person’s life experience?

A Definition of Habilitation

Hawkins (1984) suggested that the potential meaning- fulness of any behavior change should be judged within the context of habilitation, which he defined as follows:

Habilitation (adjustment) is the degree to which the per- son’s repertoire maximizes short and long term rein- forcers for that individual and for others, and minimizes short and long term punishers. (p. 284)

Hawkins (1986) cited several advantages of the de- finition in that it (a) is conceptually familiar to behavior analysts, (b) defines treatment using measurable out- comes, (c) is applicable to a wide range of habilitative activities, (d) deals with individual and societal needs in a nonjudgmental way, (e) treats adjustment along a con- tinuum of adaptive behavior and is not deficit driven, and (f) is culturally and situationally relative.

Judgments about how much a particular behavior change will contribute to a person’s overall habilitation (adjustment, competence) are difficult to make. In many cases we simply do not know how useful or functional a given behavior change will prove to be (Baer, 1981, 1982),

even when its short-term utility can be predicted. Applied behavior analysts, however, must place the highest im- portance on the selection of target behaviors that are truly useful and habilitative (Hawkins, 1991). In effect, if a po- tential target behavior meets the habilitation standard, then the individual is much more likely to acquire additional re- inforcement in the future and avoid punishment.

From both ethical and pragmatic perspectives, any behavior targeted for change must benefit the person ei- ther directly or indirectly. Examining potential target be- haviors according to the 10 questions and considerations described in the following sections should help clarify their relative social significance and habilitative value. Figure 4 summarizes these considerations in a work- sheet format that can be used in evaluating the social sig- nificance of potential target behaviors.

Is This Behavior Likely to Produce Reinforcement in the Client’s Natural Environment After Treatment Ends?

To determine whether a particular target behavior is func- tional for the client, the behavior analyst, significant oth- ers, and whenever possible the client should ask whether the proposed behavior change will be reinforced in the person’s daily life. Ayllon and Azrin (1968) called this the relevance of behavior rule; it means that a target be- havior should be selected only when it can be determined that the behavior is likely to produce reinforcement in the person’s natural environment. The likelihood that a new behavior will result in reinforcement after the be- havior change program is terminated is the primary determinant of whether the new behavior will be main- tained, thereby having the possibility of long-term ben- efits for that person.

Judging whether occurrences of the target behavior will be reinforced in the absence of intervention can also help to clarify whether the proposed behavior change is primarily for the individual’s benefit or for someone else. For instance, despite parental wishes or pressure, it would be of little value to try to teach math skills to a student with severe developmental disabilities with pervasive deficits in communication and social skills. Teaching communication skills that would enable the student to have more effective interactions in her current environ- ment should take precedence over skills that she might be able to use in the future (e.g., making change in the gro- cery store). Sometimes target behaviors are selected ap- propriately not because of their direct benefit to the person, but because of an important indirect benefit. In- direct benefits can occur in several different ways as de- scribed by the three questions that follow.

Selecting and Defining Target Behaviors

Figure 4 Worksheet for evaluating the social significance of potential target behaviors.

Client’s/Student’s name: __________________________________ Date: ______________

Person completing worksheet: ________________________________________________

Rater’s relationship to client/student: ____________________________________________

Behavior: __________________________________________________________________

Considerations Assessment Rationale/Comments

Is this behavior likely to Yes No Not sure produce reinforcement in the client’s natural envi- ronment after intervention ends?

Is this behavior a neces- Yes No Not sure sary prerequisite for a more complex and functional skill?

Will this behavior in- Yes No Not sure crease the client’s access to environments in which other important behaviors can be acquired and used?

Will changing this be- Yes No Not sure havior predispose others to interact with the client in a more appropriate and supportive manner?

Is this behavior a pivotal Yes No Not sure behavior or behavioral cusp?

Is this an age-appropriate Yes No Not sure behavior?

If this behavior is to be Yes No Not sure reduced or eliminated from the client’s repertoire, has an adaptive and functional behavior been selected to replace it?

Does this behavior repre- Yes No Not sure sent the actual problem/ goal, or is it only indirectly related?

Is this “just talk,” or is it Yes No Not sure the real behavior of interest?

If the goal itself is not a Yes No Not sure specific behavior (e.g., losing 20 lbs.), will this behavior help achieve it?

Summary notes/comments: __________________________________________________

Selecting and Defining Target Behaviors

Is This Behavior a Necessary Prerequisite for a Useful Skill?

Some behaviors that, in and of themselves, are not im- portant are targeted for instruction because they are nec- essary prerequisites to learning other functional behaviors. For example, advances in reading research have demonstrated that teaching phonemic awareness skills (e.g., sound isolation: What is the first sound in nose?; phoneme segmentation: What sounds do you hear in the word fat?; odd word out: What word starts with a different sound: cat, couch, fine, cake?) to nonreaders has positive effects on their acquisition of reading skills (National Reading Panel, 2000).6

Will This Behavior Increase the Client’s Access to Environments in Which Other Important Behaviors Can Be Learned and Used?

Hawkins (1986) described the targeting of “access be- haviors” as a means of producing indirect benefits to clients. For example, special education students are some- times taught to complete their workbook pages neatly, interact politely with the general education classroom teacher, and stay in their seats during the teacher’s pre- sentation. These behaviors are taught with the expectation that they will increase acceptance into a general educa- tion classroom, thereby increasing access to general ed- ucation and instructional programs.

Will Changing This Behavior Predispose Others to Interact with the Client in a More Appropriate and Supportive Manner?

Another type of indirect benefit occurs when a behavior change is of primary interest to a significant other in the person’s life. The behavior change may enable the sig- nificant other to behave in a manner more beneficial to the person. For example, suppose a teacher wants the par- ents of his students to implement a home-based instruc- tion program, believing that the students’ language skills would improve considerably if their parents spent just 10 minutes per night playing a vocabulary game with them. In meeting with one student’s parents, however, the

6A target behavior’s indirect benefit as a necessary prerequisite for another important behavior should not be confused with indirect teaching. Indirect teaching involves selecting a target behavior different from the true purpose of the behavior because of a belief that they are related (e.g., having stu- dents with poor reading skills practice shape discrimination or balance beam walking). The importance of directness in target behavior selection is discussed later in this section.

teacher realizes that although the parents are also con- cerned about poor language skills, they have other and, in their opinion, more pressing needs—the parents want their child to clean her room and help with the dinner dishes. Even though the teacher believes that straighten- ing up a bedroom and washing dishes are not as important to the child’s ultimate welfare as language development, these tasks may indeed be important target behaviors if a sloppy room and a sink full of dirty dishes impede pos- itive parent–child interactions (including playing the teacher’s vocabulary-building games). In this case, the daily chores might be selected as the target behaviors for the direct, immediate benefit of the parents, with the expectation that the parents will be more likely to help their daughter with school-related activities if they are happier with her because she straightens her bedroom and helps with the dishes.

Is This Behavior a Behavioral Cusp or a Pivotal Behavior?

Behavior analysts often use a building block method to develop repertoires for clients. For example, in teaching a complex skill (e.g., two-digit multiplication), simpler and more easily attainable skills are taught first (e.g., ad- dition, regrouping, single-digit multiplication), or with shoe tying, crossing laces, making bows, and tying knots are taught systematically. As skill elements are mastered, they are linked (i.e., chained) into increasingly complex repertoires. At any point along this developmental skill continuum when the person performs the skill to crite- rion, reinforcement follows, and the practitioner makes the determination to advance to the next skill level. As systematic and methodical as this approach has proven to be, analysts are researching ways to improve the effi- ciency of developing new behavior. Choosing target be- haviors that are behavioral cusps and pivotal behaviors may increase this efficiency.

Behavioral Cusps.

Rosales-Ruiz and Baer (1997) defined a behavioral cusp as:

a behavior that has consequences beyond the change it- self, some of which may be considered important. . . . What makes a behavior change a cusp is that it exposes the individual’s repertoire to new environments, espe- cially new reinforcers and punishers, new contingen- cies, new responses, new stimulus controls, and new communities of maintaining or destructive contingen- cies. When some or all of those events happen, the individual’s repertoire expands; it encounters a differen- tially selective maintenance of the new as well as some

Selecting and Defining Target Behaviors

old repertoires, and perhaps that leads to some further cusps. (p. 534)

Rosales-Ruiz and Baer (1997) cited as examples of possible behavioral cusps behaviors such as crawling, reading, and generalized imitation, because they “suddenly open the child’s world to new contingencies that will de- velop many new, important behaviors” (p. 535, emphasis added). Cusps differ from component or prerequisite be- haviors. For an infant, specific arm, head, leg, or posi- tional movements would be component behaviors for crawling, but crawling is the cusp because it enables the infant to contact new environments and stimuli as sources of motivation and reinforcement (e.g., toys, parents), which in turn opens a new world of contingencies that can further shape and select other adaptive behaviors.

The importance of cusps is judged by (a) the extent of the behavior changes they systematically enable, (b) whether they systematically expose behavior to new cusps, and (c) the audience’s view of whether these changes are important for the organism, which in turn is often controlled by societal norms and expectations of what behaviors should develop in children and when that development should happen. (Rosales-Ruiz & Baer, 1997, p. 537)

Bosch and Fuqua (2001) suggested that clarifying “a priori dimensions for determining if a behavior might be a cusp is an important step in realizing the potential of the cusp concept” (p. 125). They stated that a behavior might be a cusp if it meets one or more of five criteria: “(a) ac- cess to new reinforcers, contingencies, and environments; (b) social validity; (c) generativeness; (d) competition with inappropriate responses; and (e) number and the rel- ative importance of people affected” (p. 123). The more criteria that a behavior satisfies, the stronger the case that the behavior is a cusp. By identifying and assessing tar- get behaviors based on their cusp value, practitioners may indeed open “a new world” of far-reaching potential for those they serve.

Pivotal Behavior

Pivotal behavior has emerged as an interesting and promising concept in behavioral research, especially as it relates to the treatment of people with autism and de- velopmental disabilities. R. L. Koegel and L. K. Koegel and their colleagues have examined pivotal behavior as- sessment and treatment approaches across a wide range of areas (e.g., social skills, communication ability, dis- ruptive behaviors) (Koegel & Frea, 1993; Koegel & Koegel, 1988; Koegel, Koegel, & Schreibman, 1991). Briefly stated, a pivotal behavior is a behavior that, once

learned, produces corresponding modifications or co- variations in other adaptive untrained behaviors. For in- stance, Koegel, Carter, and Koegel (2003) indicated that teaching children with autism to “self-initiate” (e.g., ap- proach others) may be a pivotal behavior. The “longitu- dinal outcome data from children with autism suggest that the presence of initiations may be a prognostic in- dicator of more favorable long-term outcomes and there- fore may be ‘pivotal’ in that they appear to result in widespread positive changes in a number of areas” (p. 134). That is, improvement in self-initiations may be pivotal for the emergence of untrained response classes, such as asking questions and increased production and diversity of talking.

Assessing and targeting pivotal behaviors can be ad- vantageous for both the practitioner and the client. From the practitioner’s perspective, it might be possible to as- sess and then train pivotal behaviors within relatively few sessions that would later be emitted in untrained settings or across untrained responses (Koegel et al., 2003). From the client’s perspective, learning a pivotal behavior would shorten intervention, provide the person with a new reper- toire with which to interact with his environment, im- prove the efficiency of learning, and increase the chances of coming into contact with reinforcers. As Koegel and colleagues concluded: “The use of procedures that teach the child with disabilities to evoke language learning op- portunities in the natural environment may be particu- larly useful for speech and language specialists or other special educators who desire ongoing learning outside of language teaching sessions” (p. 143).

Is This an Age-Appropriate Behavior?

A number of years ago it was common to see adults with developmental disabilities being taught behaviors that a nondisabled adult would seldom, if ever, do. It was thought—perhaps as a by-product of the concept of men- tal age—that a 35-year-old woman with the verbal skills of a 10-year-old should play with dolls. Not only is the selection of such target behaviors demeaning, but their occurrence lessens the probability that other people in the person’s environment will set the occasion for and reinforce more desirable, adaptive behaviors, which could lead to a more normal and rewarding life.

The principle of normalization refers to the use of progressively more typical environments, expectations, and procedures “to establish and/or maintain personal behaviors which are as culturally normal as possible” (Wolfensberger, 1972, p. 28). Normalization is not a sin- gle technique, but a philosophical position that holds the goal of achieving the greatest possible physical and so- cial integration of people with disabilities into the main- stream of society.

Selecting and Defining Target Behaviors

In addition to the philosophical and ethical reasons for selecting age- and setting-appropriate target behav- iors, it should be reemphasized that adaptive, indepen- dent, and social behaviors that come into contact with reinforcement are more likely to be maintained than are behaviors that do not. For example, instruction in leisure- time skills such as sports, hobbies, and music-related ac- tivities would be more functional for a 17-year-old boy than teaching him to play with toy trucks and building blocks. An adolescent with those behaviors—even in an adapted way—has a better chance of interacting in a typ- ical fashion with his peer group, which may help to ensure the maintenance of his newly learned skills and provide opportunities for learning other adaptive behaviors.

If the Proposed Target Behavior Is to Be Reduced or Eliminated, What Adaptive Behavior Will Replace It?

A practitioner should never plan to reduce or eliminate a behavior from a person’s repertoire without (a) determin- ing an adaptive behavior that will take its place and (b) designing the intervention plan to ensure that the replace- ment behavior is learned. Teachers and other human ser- vices professionals should be in the business of building positive, adaptive repertoires, not merely reacting to and eliminating behaviors they find troublesome (Snell & Brown, 2006). Even though a child’s maladaptive behav- iors may be exceedingly annoying to others, or even dam- aging physically, those undesirable responses have proven functional for the child. That is, the maladaptive behavior has worked for the child in the past by producing rein- forcers and/or helping the child avoid or escape punishers. A program that only denies that avenue of reinforcement is a nonconstructive approach. It does not teach adaptive behaviors to replace the inappropriate behavior.

Some of the most effective and recommended meth- ods for eliminating unwanted behavior focus primarily on the development of desirable replacement behaviors. Goldiamond (1974) recommended that a “constructional” approach—as opposed to an eliminative approach—be used for the analysis of and intervention into behavioral problems. Under the constructional approach the “solu- tion to problems is the construction of repertoires (or their reinstatement or transfer to new situations) rather than the elimination of repertoires” (Goldiamond, 1974, p. 14).

If a strong case cannot be made for specific, positive replacement behaviors, then a compelling case has not been made for eliminating the undesirable target behav- ior. The classroom teacher, for example, who wants a be- havior change program to maintain students staying in their seats during reading period must go beyond the sim- ple notion that ”they need to be in their seats to do the

work.” The teacher must select materials and design con- tingencies that facilitate that goal and motivate the stu- dents to accomplish their work.

Does This Behavior Represent the Actual Problem or Goal, or Is It Only Indirectly Related?

An all-too-common error in education is teaching a re- lated behavior rather than the behavior of interest. Many behavior change programs have been designed to increase on-task behaviors when the primary objective should have been to increase production or work output. On-task be- haviors are chosen because people who are productive also tend to be on task. However, as on task is usually defined, it is quite possible for a student to be on task (i.e., in her seat, quiet, and oriented toward or handling academic materials) yet produce little or no work.

Targeting needed prerequisite skills should not be confused with selecting target behaviors that do not di- rectly represent or fulfill the primary reasons for the be- havior analysis effort. Prerequisite skills are not taught as terminal behaviors for their own sake, but as neces- sary elements of the desired terminal behavior. Related, but indirect, behaviors are not necessary to perform the true objective of the program, nor are they really intended outcomes of the program by themselves. In attempting to detect indirectness, behavior analysts should ask two questions: Is this behavior a necessary prerequisite to the intended terminal behavior? Is this behavior what the in- structional program is really all about? If either question can be answered affirmatively, the behavior is eligible for target behavior status.

Is This Just Talk, or Is It the Real Behavior of Interest?

Many nonbehavioral therapies rely heavily on what peo- ple say about what they do and why they do it. The client’s verbal behavior is considered important because it is believed to reflect the client’s inner state and the mental processes that govern the client’s behavior. There- fore, getting a person to talk differently about himself (e.g., in a more healthful, positive, and less self-effacing way) is viewed as a significant step in solving the per- son’s problem. Indeed, this change in attitude is consid- ered by some to be the primary goal of therapy.

Behavior analysts, on the other hand, distinguish be- tween what people say and what they do (Skinner, 1953). Knowing and doing are not the same. Getting someone to understand his maladaptive behavior by being able to talk logically about it does not necessarily mean that his behavior will change in more constructive directions. The gambler may know that compulsive betting is ruining his

Selecting and Defining Target Behaviors

life and that his losses would cease if he simply stopped placing bets. He may even be able to verbalize these facts to a therapist and state quite convincingly that he will not gamble in the future. Still, he may continue to bet.

Because verbal behavior can be descriptive of what people do, it is sometimes confused with the performance itself. A teacher at a school for juvenile offenders intro- duced a new math program that included instructional games, group drills, timed tests, and self-graphing. The students responded with many negative comments: “This is stupid,” “Man, I’m not writin’ down what I do,” “I’m not even going to try on these tests.” If the teacher had at- tended only to the students’ talk about the program, it would probably have been discarded on the first day. But the teacher was aware that negative comments about school and work were expected in the peer group of ado- lescent delinquents and that many of her students’ nega- tive remarks had enabled them in the past to avoid tasks they thought they would not enjoy. Consequently, the teacher ignored the negative comments and attended to and rewarded her students for accuracy and rate of math computation when they participated in the program. In one week’s time the negative talk had virtually ceased, and the students’ math production was at an all-time high.

There are, of course, situations in which the behav- ior of interest is what the client says. Helping a person re- duce the number of self-effacing comments he makes and increase the frequency of positive self-descriptions is an example of a program in which talk should be the target behavior—not because the self-effacing comments are indicative of a poor self-concept, but because the client’s verbal behavior is the problem.

In every case, a determination must be made of ex- actly which behavior is the desired functional outcome of the program: Is it a skill or motor performance, or is it verbal behavior? In some instances, doing and talking behaviors might be important. A trainee applying for a lawn mower repair position may be more likely to get a job if he can describe verbally how he would fix a cranky starter on a mower. However, it is possible that, once hired, he can hold his job if he is skilled and efficient in repairing lawn mowers and does not talk about what he does. However, it is highly unlikely that a person will last very long on the job if he talks about how he would fix a lawn mower but is not able to do so. Target behaviors must be functional.

What If the Goal of the Behavior Change Program Is Not a Behavior?

Some of the important changes people want to make in their lives are not behaviors, but are instead the result or product of certain other behaviors. Weight loss is an ex- ample. On the surface it might appear that target behav-

ior selection is obvious and straightforward—losing weight. The number of pounds can be measured accu- rately; but weight, or more precisely losing weight, is not a behavior. Losing weight is not a specific response that can be defined and performed; it is the product or result of other behaviors—most notably reduced food con- sumption and/or increased exercise. Eating and exercise are behaviors and can be specifically defined and mea- sured in precise units.

Some otherwise well-designed weight loss programs have not been successful because behavior change con- tingencies were placed on the goal (reduced weight) and not on the behaviors necessary to produce the goal. Tar- get behaviors in a weight loss program should be mea- sures of food consumption and exercise level, with intervention strategies designed to address those behav- iors (e.g., De Luca & Holborn, 1992; McGuire, Wing, Klem, & Hill, 1999). Weight should be measured and charted during a weight loss program, not because it is the target behavior of interest, but because weight loss shows the positive effects of increased exercise or decreased food consumption.

There are numerous other examples of important goals that are not behaviors, but are the end products of behavior. Earning good grades, for example, is a goal that must be analyzed to determine what behaviors pro- duce better grades (e.g., solving math problems via guided and independent practice). Behavior analysts can better help clients achieve their goals by selecting target behaviors that are the most directly and functionally re- lated to those goals.

Some goals expressed by and for clients are not the di- rect product of a specific target behavior, but broader, more general goals: to be more successful, to have more friends, to be creative, to learn good sportsmanship, to develop an improved self-concept. Clearly, none of these goals are de- fined by specific behaviors, and all are more complex in terms of their behavioral components than losing weight or getting a better grade in math. Goals such as being suc- cessful represent a class of related behaviors or a general pattern of responding. They are labels that are used to de- scribe people who behave in certain ways. Selecting target behaviors that will help clients or students attain these kinds of goals is even more difficult than their complexity sug- gests because the goals themselves often mean different things to different people. Being a success can entail a wide variety of behaviors. One person may view success in terms of income and job title. For another, success might mean job satisfaction and good use of leisure time. An important role of the behavior analyst during assessment and target behavior identification is to help the client select and de- fine personal behaviors, the sum of which will result in the client and others evaluating her repertoire in the intended fashion.

Selecting and Defining Target Behaviors

Prioritizing Target Behaviors Once a “pool” of eligible target behaviors has been iden- tified, decisions must be made about their relative prior- ity. Sometimes the information obtained from behavioral assessment points to one particular aspect of the person’s repertoire in need of improvement more so than another. More often, though, assessment reveals a constellation of related, and sometimes not-so-related, behaviors in need of change. Direct observations, along with a be- havioral interview and needs assessment, may produce a long list of important behaviors to change. When more than one eligible target behavior remains after careful evaluation of the considerations described in the previous section, the question becomes, Which behavior should be changed first? Judging each potential target behavior in light of the following nine questions may help deter- mine which behavior deserves attention first, and the rel- ative order in which the remaining behaviors will be addressed.

1. Does this behavior pose any danger to the client or to others? Behaviors that cause harm or pose a serious threat to the client’s or to others’ personal safety or health must receive first priority.

2. How many opportunities will the person have to use this new behavior? or How often does this problem behavior occur? A student who consis- tently writes reversed letters presents more of a problem than does a child who reverses letters only occasionally. If the choice is between first teaching a prevocational student to pack his lunch or to learn how to plan his two-week vacation each year, the former skill takes precedence be- cause the employee-to-be may need to pack his lunch every workday.

3. How long-standing is the problem or skill deficit? A chronic behavior problem (e.g., bullying) or skill deficit (e.g., lack of social interaction skills) should take precedence over problems that appear sporadi- cally or that have just recently surfaced.

4. Will changing this behavior produce higher rates of reinforcement for the person? If all other con- siderations are equal, a behavior that results in higher, sustained levels of reinforcement should take precedence over a behavior that produces little additional reinforcement for the client.

5. What will be the relative importance of this target behavior to future skill development and indepen- dent functioning? Each target behavior should be judged in terms of its relation (i.e., prerequisite or

supportive) to other critical behaviors needed for optimal learning and development and maximum levels of independent functioning in the future.

6. Will changing this behavior reduce negative or un- wanted attention from others? Some behaviors are not maladaptive because of anything inherent in the behavior itself, but because of the unnecessary problems the behavior causes the client. Some peo- ple with developmental and motoric disabilities may have difficulty at mealtimes with using uten- sils and napkins appropriately, thus reducing op- portunities for positive interaction in public. Granted, public education and awareness are war- ranted as well, but it would be naive not to con- sider the negative effects of public reaction. Also, not teaching more appropriate mealtime skills may be a disservice to the person. Idiosyncratic public displays or mannerisms may be high-priority target behaviors if their modification is likely to provide access to more normalized settings or important learning environments.

7. Will this new behavior produce reinforcement for significant others? Even though a person’s behav- ior should seldom, if ever, be changed simply for the convenience of others or for maintenance of the status quo, neither should the effect of a person’s behavior change on the significant others in his life be overlooked. This question is usually answered best by the significant others themselves because people not directly involved in the person’s life would often have

no idea how rewarding it is to see your retarded 19 year old acquire the skill of toilet flushing on com- mand or pointing to food when she wants a second helping. I suspect that the average taxpayer would not consider it “meaningful” to him or her for Kar- rie to acquire such skills. And, although we cannot readily say how much Karrie’s being able to flush the toilet enhances her personal reinforcement/pun- ishment ratio, I can testify that it enhances mine as a parent. (Hawkins, 1984, p. 285)

8. How likely is success in changing this target be- havior? Some behaviors are more difficult to change than others. At least three sources of infor- mation can help assess the level of difficulty or, more precisely, predict the ease or degree of suc- cess in changing a particular behavior. First, what does the literature say about attempts to change this behavior? Many of the target behaviors that confront applied behavior analysts have been stud- ied. Practitioners should stay abreast of published research reports in their areas of application. Not

Selecting and Defining Target Behaviors

only is such knowledge likely to improve the selec- tion of proven and efficient techniques for behavior change, but also it may help to predict the level of difficulty or chance of success.

Second, how experienced is the practitioner? The practitioner’s own competencies and experi- ences with the target behavior in question should be considered. A teacher who has worked success- fully for many years with acting-out, aggressive children may have an array of effective behavior management strategies ready to employ and might predict success with even the most challenging child. However, that same teacher might decide that he is less able to improve a student’s written language skills.

Third, to what extent can important variables in the client’s environment be controlled effec- tively? Whether a certain behavior can be changed is not the question. In an applied setting, however, identifying and then consistently manipulating the controlling variables for a given target behavior will determine whether the behavior will be changed.

Fourth, are the resources available to imple- ment and maintain the intervention at a level of fi- delity and intensity long enough that is likely to achieve the desired outcomes? No matter how ex- pertly designed a treatment plan, implementing it without the personnel and other resources needed to carry out the intervention properly is likely to yield disappointing results.

9. How much will it cost to change this behavior? Cost should be considered before implementing any sys- tematic behavior change program. However, a cost–benefit analysis of several potential target be- haviors does not mean that if a teaching program is expensive, it should not be implemented. Major courts have ruled that the lack of public funds may not be used as an excuse for not providing an appro- priate education to all children regardless of the severity of their disability (cf., Yell & Drasgow, 2000). The cost of a behavior change program cannot be determined by simply adding dollar amounts that might be expended on equipment, materials, trans- portation, staff salaries, and the like. Consideration should also be given to how much of the client’s time the behavior change program will demand. If, for ex- ample, teaching a fine motor skill to a child with se- vere disabilities would consume so much of the child’s day that there would be little time remaining for her to learn other important behaviors—such as communication, leisure, and self-help skills—or sim- ply to have some free time, the fine motor skill objec- tive may be too costly.

Developing and Using a Target Behavior Ranking Matrix

Assigning a numerical rating to each of a list of potential target behaviors can produce a priority ranking of those behaviors. One such ranking matrix is shown in Figure 5; it is an adaptation of a system described by Dardig and Heward (1981) for prioritizing and selecting learning goals for students with disabilities. Each behavior is given a number representing the behavior’s value on each of the prioritizing variables (e.g., 0 to 4 with 0 representing no value or contribution and 4 representing maximum value or benefit).

Professionals involved in planning behavior change programs for certain student or client populations will usually want to weigh some of the variables differentially, require a maximum rating on certain selection variables, and/or add other variables that are of particular impor- tance to their overall goals. For example, professionals planning behavior change programs for senior citizens would probably insist that target behaviors with imme- diate benefits receive high priority. Educators serving secondary students with disabilities would likely advocate for factors such as the relative importance of a target behavior for future skill development and independent functioning.

Sometimes the behavior analyst, the client, and/or significant others have conflicting goals. Parents may want their teenage daughter in the house by 10:30 P.M. on weekends, but the daughter may want to stay out later. The school may want a behavior analyst to develop a pro- gram to increase students’ adherence to dress and social codes. The behavior analyst may believe that these codes are outdated and are not in the purview of the school. Who decides what is best for whom?

One way to minimize and work through conflicts is to obtain client, parent, and staff/administration partici- pation in the goal determination process. For example, the active participation of parents and, when possible, the student in the selection of short- and long-term goals and treatment procedures is required by law in planning spe- cial education services for students with disabilities (In- dividuals with Disabilities Education Improvement Act of 2004). Such participation by all of the significant parties can avoid and resolve goal conflicts, not to mention the invaluable information the participants can provide rela- tive to other aspects of program planning (e.g., identifi- cation of likely reinforcers). Reviewing the results of assessment efforts and allowing each participant to pro- vide input on the relative merits of each proposed goal or target behavior can often produce consensus on the best direction. Program planners should not commit a priori that whatever behavior is ranked first will necessarily be considered the highest priority target behavior. However,

Selecting and Defining Target Behaviors

Figure 5 Worksheet for prioritizing potential target behaviors.

Client’s/Student’s name:________________________________________ Date:__________

Person completing worksheet:__________________________________________________

Rater’s relationship to client/student: ____________________________________________

Directions: Use the key below to rank each potential target behavior by the extent to which it meets or fulfills each prioritization criteria. Add each team member’s ranking of each potential target behavior. The behavior(s) with the highest total scores would presumably be the highest priority for intervention. Other criteria relevant to a particular program or individual’s situation can be added, and the criteria can be differentially weighted.

Key: 0 = No or Never; 1 = Rarely; 2 = Maybe or Sometimes; 3 = Probably or Usually; 4 = Yes or Always

Potential Target Behaviors

(1) _______ (2) _______ (3) _______ (4) _______

Prioritization Criteria

Does this behavior pose danger 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 to the person or to others?

How many opportunities will the 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 person have to use this new skill in the natural environment? or How often does the problem behavior occur?

How long-standing is the 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 problem or skill deficit?

Will changing this behavior pro- 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 duce a higher rate of reinforce- ment for the person?

What is the relative importance 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 of this target behavior to future skill development and indepen- dent functioning?

Will changing this behavior re- 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 duce negative or unwanted attention from others?

Will changing this behavior pro- 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 duce reinforcement for signifi- cant others?

How likely is success in chang- 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 ing this behavior?

How much will it cost to change 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 this behavior?

Totals _____ _____ _____ _____

Selecting and Defining Target Behaviors

if the important people involved in a person’s life go through a ranking process such as the one shown in Figure 5, they are likely to identify areas of agreement and disagreement, which can lead to further discussions of target behavior selection and concentration on the crit- ical concerns of those involved.

Defining Target Behaviors Before a behavior can undergo analysis, it must be de- fined in a clear, objective, and concise manner. In con- structing target behavior definitions, applied behavior analysts must consider the functional and topographical implications of their definitions.

Role and Importance of Target Behavior Definitions in Applied Behavior Analysis

Applied behavior analysis derives its validity from its systematic approach to seeking and organizing knowl- edge about human behavior. Validity of scientific knowl- edge in its most basic form implies replication. When predicted behavioral effects can be reproduced, princi- ples of behavior are confirmed and methods of practice developed. If applied behavior analysts employ defini- tions of behavior not available to other scientists, repli- cation is less likely. Without replication, the usefulness or meaningfulness of data cannot be determined beyond the specific participants themselves, thereby limiting the or- derly development of the discipline as a useful technol- ogy (Baer, Wolf, & Risley, 1968). Without explicit, well-written definitions of target behaviors, researchers would be unable to accurately and reliably measure the same response classes within and across studies; or to aggregate, compare, and interpret their data.7

Explicit, well-written definitions of target behavior are also necessary for the practitioner, who may not be so concerned with replication by others or development of the field. Most behavior analysis programs are not con- ducted primarily for the advancement of the field; they are implemented by educators, clinicians, and other human services professionals to improve the lives of their clients. However, implicit in the application of behavior analysis is an accurate, ongoing evaluation of the target behavior, for which an explicit definition of behavior is a must.

A practitioner concerned only with evaluating his ef- forts in order to provide optimum service to his clients, might ask, “As long as I know what I mean by [name of

7Procedures for measuring behavior accurately and reliably are discussed in Chapter entitled “Measuring Behavior.”

target behavior], why must I write down a specific defi- nition?” First, a good behavioral definition is operational. It provides the opportunity to obtain complete informa- tion about the behavior’s occurrence and nonoccurrence, and it enables the practitioner to apply procedures in a consistently accurate and timely fashion. Second, a good definition increases the likelihood of an accurate and be- lievable evaluation of the program’s effectiveness. Not only does an evaluation need to be accurate to guide on- going program decisions, but also the data must be be- lievable to those with a vested interest in the program’s effectiveness. Thus, even though the practitioner may not be interested in demonstrating an analysis to the field at large, she must always be concerned with demonstrating effectiveness (i.e., accountability) to clients, parents, and administrators.

Two Types of Target Behavior Definitions

Target behaviors can be defined functionally or topo- graphically.

Function-Based Definitions

A function-based definition designates responses as members of the targeted response class solely by their common effect on the environment. For example, Irvin, Thompson, Turner, and Williams (1998) defined hand mouthing as any behavior that resulted in “contact of the fingers, hand, or wrist with the mouth, lips, or tongue (p. 377). Figure 6 shows several examples of function- based definitions.

Applied behavior analysts should use function-based definitions of target behaviors whenever possible for the following reasons:

• A function-based definition encompasses all rele- vant forms of the response class. However, target behavior definitions based on a list of specific topographies might omit some relevant members of the response class and/or include irrelevant re- sponse topographies. For example, defining chil- dren’s offers to play with peers in terms of specific things the children say and do might omit re- sponses to which peers respond with reciprocal play and/or include behaviors that peers reject.

• The outcome, or function, of behavior is most im- portant. This holds true even for target behaviors for which form or aesthetics is central to their being valued as socially significant. For example, the flowing strokes of the calligrapher’s pen and the gymnast’s elegant movements during a floor

Selecting and Defining Target Behaviors

Figure 6 Function-based definitions of various target behaviors.

Creativity in Children’s Blockbuilding The child behaviors of blockbuilding were defined according to their products, block forms. The researchers created a list of 20 arbitrary, but frequently seen forms, including:

Arch—any placement of a block atop two lower blocks not in contiguity.

Ramp—a block leaned against another, or a triangular block placed contiguous to another, to simulate a ramp.

Story—two or more blocks placed one atop another, the upper block(s) resting solely upon the lower.

Tower—any story of two or more blocks in which the lowest block is at least twice as tall as it is wide. (Goetz & Baer, 1973, pp. 210–211).

Exercise by Obese Boys Riding a stationary bicycle—each wheel revolution constituted a response, which was automatically recorded by magnetic counters (DeLuca & Holborn, 1992, p. 672).

Compliance at Stop Signs by Motorists Coming to a complete stop—observers scored a vehicle as coming to a complete stop if the tires stopped rolling prior to the vehicle entering the intersection (Van Houten & Retting, 2001, p. 187).

Recycling by Office Employees Recycling office paper—number of pounds and ounces of recyclable office paper found in recycling and trash containers. All types of paper accepted as recyclable were identified as well as examples of nonrecyclable paper (Brothers, Krantz, & McClannahan, 1994, p. 155).

Safety Skills to Prevent Gun Play by Children Touching the firearm—the child making contact with the firearm with any part of his or her body or with any object (e.g., a toy) resulting in the displacement of the firearm. Leaving the area—the child removing himself or herself from the room in which the firearm was located within 10 seconds of seeing the firearm (Himle, Miltenberger, Flessner, & Gatheridge, 2004, p. 3).

routine are important (i.e., have been selected) be- cause of their effects or function on others (e.g., praise from the calligraphy teacher, high scores from gymnastics judges).

• Functional definitions are often simpler and more concise than topography-based definitions, which leads to easier and more accurate and reliable mea- surement and sets the occasion for the consistent application of intervention. For example, in their study on skill execution by college football play- ers, Ward and Carnes (2002) recorded a correct tackle according to the clear and simple definition, “if the offensive ball carrier was stopped” (p. 3).

Function-based definitions can also be used in some situations in which the behavior analyst does not have direct and reliable access to the natural outcome of the target behavior, or cannot use the natural outcome of the target behavior for ethical or safety reasons. In such cases, a function-based definition by proxy can be considered.

For example, the natural outcome of elopement (i.e., run- ning or walking away from a caregiver without consent) is a lost child. By defining elopement as “any movement away from the therapist more than 1.5 m without per- mission” (p. 240), Tarbox, Wallace, and Williams (2003) were able to measure and treat this socially significant target behavior in a safe and meaningful manner.

Topography-Based Definitions

A topography-based definition identifies instances of the target behavior by the shape or form of the behavior. Topography-based definitions should be used when the behavior analyst (a) does not have direct, reliable, or easy access to the functional outcome of the target behavior, and/or (b) cannot rely on the function of the behavior be- cause each instance of the target behavior does not pro- duce the relevant outcome in the natural environment or the outcome might be produced by other events. For ex- ample, Silvestri (2004) defined and measured two classes

Selecting and Defining Target Behaviors

of positive teacher statements according to the words that made up the statements, not according to whether the comments produced specific outcomes (see Figure 7).

Topography-based definitions can also be used for target behaviors for which the relevant outcome is some- times produced in the natural environment by undesir- able variations of the response class. For example, because a duffer’s very poor swing of a golf club some- times produces a good outcome (i.e., the ball lands on the green), it is better to define a correct swing by the po- sition and movement of the golf club and the golfer’s feet, hips, head, and hands.

A topography-based definition should encompass all response forms that would typically produce the relevant outcome in the natural environment. Although topogra- phy provides an important element for defining target be- haviors, the applied behavior analyst must be especially careful not to select target behaviors solely on the basis of topography (see Box 1).

Writing Target Behavior Definitions

A good definition of a target behavior provides an accu- rate, complete, and concise description of the behavior to be changed (and therefore measured). It also states what is not included in the behavioral definition. Asking to be excused from the dinner table is an observable and measurable behavior that can be counted. By compari- son, “exercising good manners” is not a description of any particular behavior; it merely implies a general re- sponse class of polite and socially acceptable behaviors. Hawkins and Dobes (1977) described three characteris- tics of a good definition:

1. The definition should be objective, referring only to observable characteristics of the behavior (and environment, if needed) or translating any inferen- tial terms (such as “expressing hostile feelings,” “intended to help,” or “showing interest in”) into more objective ones.

Figure 7 Topography-based definitions for two types of teacher statements.

Generic Positive Statements Generic positive statements were defined as audible statements by the teacher that referred to one or more student’s behavior or work products as desirable or commendable (e.g., “I’m proud of you!”, “Great job, everyone.”). Statements made to other adults in the room were recorded if they were loud enough to be heard by the students and made direct reference to student behavior or work products (e.g., “Aren’t you impressed at how quietly my students are working today?”). A series of positive comments that specified neither student names nor behaviors with less than 2 seconds between comments was recorded as one statement. For example, if the teacher said, “Good, good, good. I’m so impressed” when reviewing three or four students’ work, it was recorded as one statement.

Teacher utterances not recorded as generic positive statements included (a) statements that referred to specific behavior or student names, (b) neutral statements indicating only that an academic response was correct (e.g., “Okay”, “Correct”), (c) positive statements not related to student behavior (e.g., saying “Thanks for dropping off my attendance forms” to a colleague), and (d) incomprehensible or inaudible statements.

Behavior-Specific Positive Statements Behavior-specific positive statements made explicit reference to an observable behavior (e.g., “Thank you for putting your pencil away”). Specific positive statements could refer to general classroom behavior (e.g., “You did a great job walking back to your seat quietly”) or academic performance (e.g., “That was a super smart answer!”). To be recorded as separate responses, specific positive statements were separated from one another by 2 seconds or by differentiation of the behavior praised. In other words, if a teacher named a desirable behavior and then listed multiple students who were demonstrating the behavior, this would be recorded as one statement (e.g., “Marissa, Tony, and Mark, you did a great job of returning your materials when you were finished with them”). However, a teacher’s positive comment noting several different behaviors would be recorded as multiple statements regardless of the interval between the end of one comment and the start of the next. For example, “Jade, you did a great job cleaning up so quickly; Charles, thanks for putting the workbooks away; and class, I appreciate that you lined up quietly ” would be recorded as three positive statements.

Adapted from The Effects of Self-Scoring on Teachers’ Positive Statements during Classroom Instruction (pp. 48–49) by S. M. Silvestri. Unpublished doctoral dissertation. Columbus, OH: The Ohio State University. Used by permission.

Suppose you are a behavior analyst in a position to de- sign and help implement an intervention to change the following four behaviors:

1. A child repeatedly raises her arm, extending and retracting her fingers toward her palm in a grip- ping/releasing type of motion.

2. An adult with developmental disabilities pushes his hand hard against his eye, making a fist and rubbing his eye rapidly with his knuckles.

3. Several times each day a high school student rhythmically drums her fingers up and down, sometimes in bursts of 10 to 15 minutes in dura- tion.

4. A person repeatedly grabs at and squeezes another person’s arms and legs so hard that the other per- son winces and says “Ouch!”

How much of a problem does the behavior pose for the person or for others who share his or her current and fu- ture environments? Would you rate each behavior as a mild, moderate, or serious problem? How important do you think it would be to target each of these behaviors for reduction or elimination from the repertoires of the four individuals?

Appropriate answers to these questions cannot be found in topographical descriptions alone. The meaning and relative importance of any operant behavior can be determined only in the context of the environmental an- tecedents and consequences that define the behavior. Here is what each of the four people in the previous ex- amples were actually doing:

1. An infant learning to wave “bye-bye.”

2. A man with allergies rubbing his eye to relieve the itching.

3. A student typing unpredictable text to increase her keyboarding fluency and endurance.

4. A massage therapist giving a relaxing, deep- muscle massage to a grateful and happy customer.

Applied behavior analysts must remember that the meaning of any behavior is determined by its function, not its form. Behaviors should not be targeted for change on the basis of topography alone.

Note: Examples 1 and 2 are adapted from Meyer and Evans, 1989, p. 53.

Box 1 How Serious Are These Behavior Problems?

2. The definition should be clear in that it should be readable and unambiguous so that experienced observers could read it and readily paraphrase it accurately.

3. The definition should be complete, delineating the “boundaries” of what is to be included as an in- stance of the response and what is to be excluded, thereby directing the observers in all situations that are likely to occur and leaving little to their judg- ment. (p. 169)

Stated succinctly, a good definition must be objec- tive, assuring that specific instances of the defined target behavior can be observed and recorded reliably. An ob- jective definition increases the likelihood of an accurate and believable evaluation of program effectiveness. Sec- ond, a clear definition is technological, meaning that it en- ables others to use and replicate it (Baer et al., 1968). A clear definition therefore becomes operational for pre- sent and future purposes. Finally, a complete definition discriminates between what is and what is not an instance of the target behavior. A complete definition allows oth-

ers to record an occurrence of a target behavior, but not record instances of nonoccurrence, in a standard fashion. A complete definition is a precise and concise description of the behavior of interest. Note how the target behavior definitions in Figures 6 and 7 meet the standard for being objective, clear, and complete.

Morris (1985) suggested testing the definition of a target behavior by asking three questions:

1. Can you count the number of times that the behav- ior occurs in, for example, a 15-minute period, a 1-hour period, or one day? Or, can you count the number of minutes that it takes for the child to per- form the behavior? That is, can you tell someone that the behavior occurred “x” number of times of “x” number of minutes today? (Your answer should be “yes.”)

2. Will a stranger know exactly what to look for when you tell him/her the target behavior you are plan- ning to modify? That is, can you actually see the child performing the behavior when it occurs? (Your answer should be “yes.”)

Selecting and Defining Target Behaviors

3. Can you break down the target behavior into smaller behavioral components, each of which is more specific and observable than the original tar- get behavior? (Your answer should be “no.”).

In responding to the suggestion that perhaps a sourcebook of standard target behavior definitions be de- veloped because it would increase the likelihood of exact replications among applied researchers and would save the considerable time spent in developing and testing situation-specific definitions, Baer (1985) offered the fol- lowing perspectives. Applied behavior analysis programs are implemented because someone (e.g., teacher, parent, individual himself) has “complained” that a behavior needs to be changed. A behavioral definition has validity in applied behavior analysis only if it enables observers to capture every aspect of the behavior that the com- plainer is concerned with and none other. Thus, to be valid from an applied perspective, definitions of target behaviors should be situation specific. Attempts to stan- dardize behavior definitions assume an unlikely similar- ity across all situations.

Setting Criteria for Behavior Change Target behaviors are selected for study in applied behav- ior analysis because of their importance to the people in- volved. Applied behavior analysts attempt to increase, maintain, and generalize adaptive, desirable behaviors and decrease the occurrence of maladaptive, undesirable behaviors. Behavior analysis efforts that not only target

important behaviors but also change those behaviors to an extent that a person’s life is changed in a positive and meaningful way, are said to have social validity.8 But how much does a target behavior need to change before it makes a meaningful difference in the person’s life?

Van Houten (1979) made a case for specifying the desired outcome criteria before efforts to modify the tar- get behavior begin.

This step [specifying outcome criteria] becomes as im- portant as the previous step [selecting socially important target behaviors] if one considers that for most behav- iors there exists a range of responding within which per- formance is most adaptive. When the limits of this range are unknown for a particular behavior, it is possible that one could terminate treatment when performance is above or below these limits. Hence, the behavior would not be occurring within its optimal range. . . .

In order to know when to initiate and terminate a treatment, practitioners require socially validated stan- dards for which they can aim. (pp. 582, 583)

Van Houten (1979) suggested two basic approaches to determining socially valid goals: (a) Assess the perfor- mance of people judged to be highly competent, and (b) experimentally manipulate different levels of performance to determine empirically which produces optimal results.

Regardless of the method used, specifying treatment goals before intervention begins provides a guideline for continuing or terminating a treatment. Further, setting objective, predetermined goals helps to eliminate dis- agreements or biases among those involved in evaluat- ing a program’s effectiveness.

Role of Assessment in Behavior Analysis

1. Behavioral assessment involves a full range of inquiry meth- ods including direct observations, interviews, checklists, and tests to identify and define targets for behavior change.

2. Behavioral assessment can be conceptualized as funnel shaped, with an initial broad scope leading to an eventual narrow and constant focus.

3. Behavioral assessment consists of five phases or functions: (a) screening, (b) defining and quantifying problems or goals, (c) pinpointing the target behavior(s) to be treated (d) monitoring progress, and (e) following up.

4. Before conducting a behavioral assessment, the behav- ior analyst must determine whether he has the authority and permission, resources, and skills to assess and change the behavior.

Summary

8A third component of social validity concerns the social acceptability of the treatment methods and procedures employed to change the behavior. The importance of social validity in evaluating applied behavior analysis will be discussed in Chapter entitled “Planning and Evaluating Applied Behavior Analysis Research.”

Assessment Methods Used by Behavior Analysts

5. Four major methods for obtaining assessment information are (a) interviews, (b) checklists, (c) tests, and (d) direct observations.

6. The client interview is used to determine the client’s de- scription of problem behaviors or achievement goals. What, when, and where questions are emphasized, focus- ing on the actual behavior of the client and the responses of significant others to that behavior.

Selecting and Defining Target Behaviors

7. Questionnaires and needs assessment surveys are some- times completed by the client to supplement the informa- tion gathered in the interview.

8. Clients are sometimes asked to self-monitor certain situ- ations or behaviors. Self-collected data may be useful in selecting and defining target behaviors.

9. Significant others can also be interviewed to gather assess- ment information and, in some cases, to find out whether they will be willing and able to assist in an intervention.

10. Direct observation with a behavior checklist that contains specific descriptions of various skills can indicate possi- ble target behaviors.

11. Anecdotal observation, also called ABC recording, yields a descriptive, temporally sequenced account of all behav- ior(s) of interest and the antecedent conditions and con- sequences for those behaviors as those events occur in the client’s natural environment.

12. Ecological assessment entails gathering a large amount of information about the person and the environments in which that person lives and works (e.g., physiological con- ditions, physical aspects of the environment, interactions with others, past reinforcement history). A complete eco- logical assessment is neither necessary nor warranted for most applied behavior analysis programs.

13. Reactivity, the effects of an assessment procedure on the behavior being assessed, is most likely when the person being observed is aware of the observer’s presence and purpose. Behavior analysts should use assessment meth- ods that are as unobtrusive as possible, repeat observations until apparent reactive effects subside, and take possible re- active effects into account when interpreting the results of observations.

Assessing the Social Significance of Potential Target Behaviors

14. Target behaviors in applied behavior analysis must be so- cially significant behaviors that will increase a person’s habilitation (adjustment, competence).

15. The relative social significance and habilitative value of a potential target behavior can be clarified by viewing it in light of the following considerations:

• Will the behavior be reinforced in the person’s daily life? The relevance of behavior rule requires that a tar- get behavior produce reinforcement for the person in the postintervention environment.

• Is the behavior a necessary prerequisite for a useful skill?

• Will the behavior increase the person’s access to envi- ronments in which other important behaviors can be learned or used?

• Will the behavior predispose others to interact with the person in a more appropriate and supportive manner?

• Is the behavior a cusp or pivotal behavior? Behavioral cusps have sudden and dramatic consequences that ex- tend well beyond the idiosyncratic change itself be- cause they expose the person to new environments, reinforcers, contingencies, responses, and stimulus con- trols. Learning a pivotal behavior produces corre- sponding modifications or covariations in other untrained behaviors.

• Is the behavior age appropriate?

• Whenever a behavior is targeted for reduction or elim- ination, a desirable, adaptive behavior must be selected to replace it.

• Does the behavior represent the actual problem or achievement goal, or is it only indirectly related?

• A person’s verbal behavior should not be confused with the actual behavior of interest. However, in some situ- ations the client’s verbal behavior should be selected as the target behavior because it is the behavior of interest.

• If a person’s goal is not a specific behavior, a target be- havior(s) must be selected that will produce the desired results or state.

Prioritizing Target Behaviors

16. Assessment often reveals more than one possible behav- ior or skill area for targeting. Prioritization can be ac- complished by rating potential target behavior against key questions related to their relative danger, frequency, long- standing existence, potential for reinforcement, relevance for future skill development and independent functioning, reduced negative attention from others, likelihood of suc- cess, and cost.

17. Participation by the person whose behavior is to be changed, parents and/or other important family members, staff, and administration in identifying and prioritizing tar- get behaviors can help reduce goal conflicts.

Defining Target Behaviors

18. Explicit, well-written target behavior definitions are nec- essary for researchers to accurately and reliably measure the same response classes within and across studies or to aggregate, compare, and interpret their data.

19. Good target behaviors definitions are necessary for prac- titioners to collect accurate and believable data to guide ongoing program decisions, apply procedures consis- tently, and provide accountability to clients, parents, and administrators.

20. Function-based definitions designate responses as mem- bers of the targeted response class solely by their common effect on the environment.

21. Topography-based definitions define instances of the tar- geted response class behavior by the shape or form of the behavior.

Selecting and Defining Target Behaviors

22. A good definition must be objective, clear, and complete, and must discriminate between what is and what is not an instance of the target behavior.

23. A target behavior definition is valid if it enables observers to capture every aspect of the behavior that the “com- plainer” is concerned with and none other.

Setting Criteria for Behavior Change

24. A behavior change has social validity if it changes some aspect of the person’s life in an important way.

25. Outcome criteria specifying the extent of behavior change desired or needed should be determined before efforts to modify the target behavior begin.

26. Two approaches to determining socially validated perfor- mance criteria are (a) assessing the performance of people judged to be highly competent and (b) experimentally ma- nipulating different levels of performance to determine which produces optimal results.

artifact celeration celeration time period celeration trend line count discrete trial duration event recording free operant

frequency interresponse time (IRT) magnitude measurement measurement by permanent product momentary time sampling partial-interval recording percentage planned activity check (PLACHECK)

rate repeatability response latency temporal extent temporal locus time sampling topography trials-to-criterion whole-interval recording

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 6: Measurement of Behavior

6-1 Identify the measurable dimensions of behavior (e.g., rate, duration, latency, or interresponse times).

6-3 State the advantages and disadvantages of using continuous measurement procedures and sampling techniques (e.g., partial- and whole-interval record- ing, momentary time sampling).

6-4 Select the appropriate measurement procedure given the dimensions of the behavior and the logistics of observing and recording.

6-6 Use frequency (i.e., count).

6-7 Use rate (i.e., count per unit of time).

6-8 Use duration.

6-9 Use latency.

6-10 Use interresponse time (IRT).

6-11 Use percentage of occurrence.

6-12 Use trials-to-criterion.

6-13 Use interval recording methods.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and questions about this document must be submitted directly to the BACB.

Measuring Behavior

Key Terms

Lord Kelvin, the British mathematician and physicist, supposedly said, “Until you can ex- press what you are talking about in numbers and

can measure it, your knowledge is meager and unsatis- factory.” Measurement (applying quantitative labels to describe and differentiate natural events) provides the basis for all scientific discoveries and for the develop- ment and successful application of technologies derived from those discoveries. Direct and frequent measurement provides the foundation for applied behavior analysis. Applied behavior analysts use measurement to detect and compare the effects of various environmental arrange- ments on the acquisition, maintenance, and generalization of socially significant behaviors.

But what is it about behavior that applied behavior analysts can and should measure? How should those measures be obtained? And, what should we do with these measures once we have obtained them? This chap- ter identifies the dimensions by which behavior can be measured and describes the methods behavior analysts commonly use to measure them. But first, we examine the definition and functions of measurement in applied behavior analysis.

Definition and Functions of Measurement in Applied Behavior Analysis Measurement is “the process of assigning numbers and units to particular features of objects or events. . . . [It] in- volves attaching a number representing the observed ex- tent of a dimensional quantity to an appropriate unit. The number and the unit together constitute the measure of the object or event” (Johnston & Pennypacker, 1993a, pp. 91, 95). A dimensional quantity is the particular feature of an object or event that is measured. For example, a single occurrence of a phenomenon—say, the response of “c-a-t” to the question, “How do you spell cat?”—would be assigned the number 1 and the unit label correct. Ob- serving another instance of the same response class would change the label to 2 correct. Other labels could also be applied based on accepted nomenclature. For example, if 8 of 10 observed responses met a standard definition of “correct,” accuracy of responding could be described with the label, 80% correct. If the 8 correct responses were emitted in 1 minute, a rate or frequency label of 8 per minute would apply.

Bloom, Fischer, and Orme (2003)—who described measurement as the act or process of applying quantita- tive or qualitative labels to events, phenomena, or ob- served properties using a standard set of consensus-based rules by which to apply labels to those occurrences—

pointed out that the concept of measurement also includes the characteristics of what is being measured, the quality and appropriateness of the measurement tools, the tech- nical skill of the measurer, and how the measures obtained are used. In the end, measurement gives re- searchers, practitioners, and consumers a common means for describing and comparing behavior with a set of la- bels that convey a common meaning.

Researchers Need Measurement

Measurement is how scientists operationalize empiri- cism. Objective measurement enables (indeed, it re- quires) scientists to describe the phenomena they observe in precise, consistent, and publicly verifiable ways. With- out measurement, all three levels of scientific knowl- edge—description, prediction, and control—would be relegated to guesswork subject to the “individual preju- dices, tastes, and private opinions of the scientist” (Zuriff, 1985, p. 9). We would live in a world in which the alchemist’s suppositions about a life-prolonging elixir would prevail over the chemist’s propositions de- rived from experimentation.

Applied behavior analysts measure behavior to ob- tain answers to questions about the existence and nature of functional relations between socially significant be- havior and environmental variables. Measurement en- ables comparisons of a person’s behavior within and between different environmental conditions, thereby af- fording the possibility of drawing empirically based con- clusions about the effects of those conditions on behavior. For example, to learn whether allowing students with emotional and behavioral challenges to choose the aca- demic tasks they will work on would influence their en- gagement with those tasks and disruptive behavior, Dunlap and colleagues (1994) measured students’ task engagement and disruptive behaviors during choice and no-choice conditions. Measurement revealed the level of both target behaviors during each condition, whether and how much the behaviors changed when choice was in- troduced or withdrawn, and how variable or stable the behaviors were during each condition.

The researcher’s ability to achieve a scientific un- derstanding of behavior depends on her ability to measure it. Measurement makes possible the detection and verifi- cation of virtually everything that has been discovered about the selective effects of the environment on behav- ior. The empirical databases of the basic and applied branches of behavior analysis consist of organized col- lections of behavioral measurements. Virtually every graph in the Journal of the Experimental Analysis of Be- havior and the Journal of Applied Behavior Analysis dis- plays an ongoing record or summary of behavioral measurement. In short, measurement provides the very

Measuring Behavior

basis for learning and talking about behavior in scientif- ically meaningful ways.1

Practitioners Need Measurement

Behavioral practitioners are dedicated to improving the lives of the clients they serve by changing socially sig- nificant behaviors. Practitioners measure behavior ini- tially to determine the current level of a target behavior and whether that level meets a threshold for further in- tervention. If intervention is warranted, the practitioner measures the extent to which his efforts are successful. Practitioners measure behavior to find out whether and when it has changed; the extent and duration of behavior changes; the variability or stability of behavior before, during, and after treatment; and whether important be- havior changes have occurred in other settings or situa- tions and spread to other behaviors.

Practitioners compare measurements of the target be- havior before and after treatment (sometimes including pre- and posttreatment measures obtained in nontreat- ment settings or situations) to evaluate the overall effects of behavior change programs (summative evaluation). Frequent measures of behavior during treatment (formative assessment) enable dynamic, data-based de- cision making concerning the continuation, modification, or termination of treatment.

The practitioner who does not obtain and attend to frequent measures of the behavior targeted for interven- tion is vulnerable to committing two kinds of preventable mistakes: (a) continuing an ineffective treatment when no real behavior change has occurred, or (b) discontinu- ing an effective treatment because subjective judgment detects no improvement (e.g., without measurement, a teacher would be unlikely to know that a student’s oral reading has increased from 70 words per minute to 80 word per minute). Thus, direct and frequent measurement enables practitioners to detect their successes and, equally important, their failures so they can make changes to change failure to success (Bushell & Baer, 1994; Green- wood & Maheady, 1997).

Our technology of behavior change is also a technology of behavior measurement and of experimental design; it developed as that package, and as long as it stays in that package, it is a self-evaluating enterprise. Its successes are successes of known magnitude; its failures are al- most immediately detected as failures; and whatever its outcomes, they are attributable to known inputs and pro- cedures rather than to chance events or coincidences. (D. M. Baer, personal communication, October 21, 1982)

1Measurement is necessary but not sufficient to the attainment of scien- tific understanding. See chapter entitled “Analyzing Behavior Change: Basic Assumptions and Strategies.”

In addition to enabling ongoing program monitoring and data-based decision making, frequent measurement provides other important benefits to practitioners and the clients they serve:

• Measurement helps practitioners optimize their ef- fectiveness. To be optimally effective, a practitioner must maximize behavior change efficiency in terms of time and resources. Only by maintaining close, continual con- tact with relevant outcome data can a practitioner hope to achieve optimal effectiveness and efficiency (Bushell & Baer, 1994). Commenting on the critical role of direct and frequent measurement on maximizing the effective- ness of classroom practice, Sidman (2000) noted that teachers “must remain attuned to the pupil’s messages and be ready to try and to evaluate modifications [in in- structional methods]. Teaching, then, is not just a matter of changing the behavior of pupils; it is an interactive so- cial process” (p. 23, words in brackets added). Direct and frequent measurement is the process by which practi- tioners hear their clients’ messages.

• Measurement enables practitioners to verify the le- gitimacy of treatments touted as “evidence based.” Prac- titioners are increasingly expected, and in some fields mandated by law, to use evidence-based interventions. An evidence-based practice is a treatment or intervention method that has been demonstrated to be effective through substantial, high-quality scientific research. For example, the federal No Child Left Behind Act of 2001 requires that all public school districts ensure that all children are taught by “highly qualified” teachers using curriculum and instructional methods validated by rigorous scientific research. Although guidelines and quality indicators re- garding the type (e.g., randomized clinical trials, single- subject studies) and quantity (e.g., number of studies published in peer-reviewed journals, minimum number of participants) of research needed to qualify a treatment method as an evidence-based practice have been pro- posed, the likelihood of complete consensus is slim (e.g., Horner, Carr, Halle, McGee, Odom, & Wolery, 2005; U.S. Department of Education, 2003). When implementing any treatment, regardless of the type or amount of re- search evidence to support it, practitioners can and should use direct and frequent measurement to verify its effec- tiveness with the students or clients they serve.

• Measurement helps practitioners identify and end the use of treatments based on pseudoscience, fad, fash- ion, or ideology. Numerous educational methods and treatments have been claimed by their advocates as break- throughs. Many controversial treatments and proposed cures for people with developmental disabilities and autism (e.g., facilitated communication, holding therapy, megadoses of vitamins, strange diets, weighted vests,

Measuring Behavior

dolphin-assisted therapy) have been promoted in the ab- sence of sound scientific evidence of effectiveness (Heflin & Simpson, 2002; Jacobson, Foxx, & Mulick, 2005). The use of some of these so-called breakthrough therapies has led to disappointment and loss of precious time at best for many people and their families, and in some cases, to disastrous consequences (Maurice, 1993). Even though well-controlled studies have shown many of these methods to be ineffective, and even though these pro- grams are not justified because they lacked sound, sci- entific evidence of effects, risks, and benefits, parents and practitioners are still bombarded with sincere and well-meaning testimonials. In the quest to find and ver- ify effective treatments and root out those whose strongest support is in the form of testimonials and slick adver- tisements on the Internet, measurement is the practi- tioner’s best ally. Practitioners should maintain a healthy skepticism regarding claims for effectiveness.

Using Plato’s “Allegory of the Cave” as a metaphor for teachers and practitioners who use untested and pseudo-instructional ideas, Heron, Tincani, Peterson, and Miller (2002) argued that practitioners would be better served if they adopted a scientific approach and cast aside pseudo-educational theories and philosophies. Practi- tioners who insist on direct and frequent measurement of all intervention and treatment programs will have em- pirical support to defend against political or social pres- sures to adopt unproven treatments. In a real sense they will be armed with what Carl Sagan (1996) called a “baloney detection kit.”

• Measurement enables practitioners to be ac- countable to clients, consumers, employers, and society. Measuring the outcomes of their efforts directly and fre- quently helps practitioners answer confidently questions from parents and other caregivers about the effects of their efforts.

• Measurement helps practitioners achieve ethical standards. Ethical codes of conduct for behavior ana- lytic practitioners require direct and frequent measure- ment of client behavior. Determining whether a client’s right to effective treatment or effective education is being honored requires measurement of the behavior(s) for which treatment was sought or intended (Nakano, 2004; Van Houten et al., 1988). A behavioral practitioner who does not measure the nature and extent of relevant be- havior changes of the clients she serves borders on mal- practice. Writing in the context of educators, Kauffman (2005) offered this perspective on the relationship be- tween measurement and ethical practice:

[T]he teacher who cannot or will not pinpoint and mea- sure the relevant behaviors of the students he or she is teaching is probably not going to be very effective. . . .

Not to define precisely and to measure these behavioral excesses and deficiencies, then, is a fundamental error; it is akin to the malpractice of a nurse who decides not to measure vital signs (heart rate, respiration rate, tempera- ture, and blood pressure), perhaps arguing that he or she is too busy, that subjective estimates of vital signs are quite adequate, that vital signs are only superficial esti- mates of the patient’s health, or that vital signs do not signify the nature of the underlying pathology. The teaching profession is dedicated to the task of changing behavior—changing behavior demonstrably for the bet- ter. What can one say, then, of educational practice that does not include precise definition and reliable measure- ment of the behavioral change induced by the teacher’s methodology? It is indefensible. (p. 439)

Measurable Dimensions of Behavior If a friend asked you to measure a coffee table, you would probably ask why he wants the table measured. In other words, what does he want measurement to tell him about the table? Does he need to know its height, width, and depth? Does he want to know how much the table weighs? Perhaps he is interested in the color of the table? Each of these reasons for measuring the table requires measuring a different dimensional quantity of the table (i.e., length, mass, and light reflection).

Behavior, like coffee tables and all entities in the physical world, also has features that can be measured (though length, weight, and color are not among them). Because behavior occurs within and across time, it has three fundamental properties, or dimensional quantities, that behavior analysts can measure. Johnston and Pen- nypacker (1993a) described these properties as follows:

• Repeatability (also called countability): Instances of a response class can occur repeatedly through time (i.e., behavior can be counted).

• Temporal extent: Every instance of behavior oc- curs during some amount of time (i.e., the duration of behavior can be measured).

• Temporal locus: Every instance of behavior occurs at a certain point in time with respect to other events (i.e., when behavior occurs can be measured).

Figure 1 shows a schematic representation of re- peatability, temporal extent, and temporal locus. Alone and in combination, these dimensional quantities provide the basic and derivative measures used by applied be- havior analysts. In the following pages these and two other measurable dimensions of behavior—its form and strength—will be discussed.

Measuring Behavior

R1 R2 R4

R3 Time

S1 S2

L L

Figure 1 Schematic representation of the dimensional quantities of repeatability, tem- poral extent, and temporal locus. Repeat- ability is shown by a count of four instances of a given response class (R1, R2, R3, and R4) within the observation period. The temporal extent (i.e., duration) of each response is shown by the raised and shaded portions of the time line. One aspect of the temporal locus (response latency) of two responses is shown by the elapsed time ( L ) between the onset of two antecedent stimulus events (S1 and S2) and the initiation of the responses that follow (R2 and R4).

➞➞

Measures Based on Repeatability

Count

Count is a simple tally of the number of occurrences of a behavior—for example, the number of correct and in- correct answers written on an arithmetic practice sheet, the number of words written correctly during a spelling test, the number of times an employee carpools to work, the number of class periods a student is tardy, the num- ber of widgets produced.

Although how often a behavior occurs is often of pri- mary interest, measures of count alone may not provide enough information to allow practitioners to make useful program decisions or analyses. For example, data show- ing that Katie wrote correct answers to 5, 10, and 15 long division problems over three consecutive math class pe- riods suggests improving performance. However, if the three measures of count were obtained in observation periods of 5 minutes, 20 minutes, and 60 minutes, re- spectively, a much different interpretation of Katie’s per- formance is suggested. Therefore, the observation period, or counting time, should always be noted when reporting measures of count.

Rate/Frequency

Combining observation time with count yields one of the most widely used measures in applied behavior analysis, rate (or frequency) of responding, defined as the num- ber of responses per unit of time.2 A rate or frequency measure is a ratio consisting of the dimensional quanti- ties of count (number of responses) and time (observation period in which the count was obtained).

2Although some technical distinctions exist between rate and frequency, the two terms are often used interchangeably in the behavior analysis literature. For discussion of various meanings of the two terms and examples of dif- ferent methods for calculating ratios combining count and observation time, see Johnston and Pennypacker (1993b, Reading 4: Describing Be- havior with Ratios of Count and Time).

Converting count to rate or frequency makes mea- surement more meaningful. For example, knowing that Yumi read 95 words correctly and 4 words incorrectly in 1 minute, that Lee wrote 250 words in 10 minutes, and that Joan’s self-injurious behavior occurred 17 times in 1 hour provides important information and context. Ex- pressing the three previously reported measures of Katie’s performance in math class as rate reveals that she correctly answered long division problems at rates of 1.0, 0.5, and 0.25 per minute over three consecutive class periods.

Researchers and practitioners typically report rate data as a count per 10 seconds, count per minute, count per day, count per week, count per month, or count per year. As long as the unit of time is standard within or across experiments, rate measures can be compared. It is possible to compare rates of responding from counts ob- tained during observation periods of different lengths. For example, a student who, over four daily class activ- ities of different durations, had 12 talk-outs in 20 minutes, 8 talk-outs in 12 minutes, 9 talk-outs in 15 minutes, and 12 talk-outs in 18 minutes had response rates of 0.60, 0.67, 0.60, and 0.67 per minute.

The six rules and guidelines described below will help researchers and practitioners obtain, describe, and interpret count and rate data most appropriately.

Always Reference the Counting Time. When re- porting rate-of-response data, researchers and practi- tioners must always include the duration of the observation time (i.e., the counting time). Comparing rate measures without reference to the counting time can lead to faulty interpretations of data. For example, if two students read from the same text at equal rates of 100 correct words per minute with no incorrect re- sponses, it would appear that they exhibited equal per- formances. Without knowing the counting time, however, an evaluation of these two performances cannot be con- ducted. Consider, for example, that Sally and Lillian each ran at a rate of 7 minutes per mile. We cannot compare their performances without reference to the

Measuring Behavior

distances they ran. Running 1 mile at a rate of 7 minutes per mile is a different class of behavior than running at a rate of 7 minutes per mile over a marathon distance.

The counting time used for each session needs to ac- company each rate measure when the counting time changes from session to session. For instance, rather than having a set counting time to answer arithmetic facts (e.g., a 1-minute timing), the teacher records the total time re- quired for the student to complete an assigned set of arith- metic problems during each class session. In this situation, the teacher could report the student’s correct and incorrect answers per minute for each session, and also report the counting times for each session because they changed from session to session.

Calculate Correct and Incorrect Rates of Response When Assessing Skill Development. When a participant has an opportunity to make a correct or an incorrect re- sponse, a rate of response for each behavior should be reported. Calculating rate correct and rate incorrect is crucial for evaluating skill development because an im- proving performance cannot be assessed by knowing only the correct rate. The rate of correct responses alone could show an improving performance, but if incorrect responding is also increasing, the improvement may be illusionary. Correct and incorrect rate measures together provide important information to help the teacher evalu- ate how well the student is progressing. Ideally, the cor- rect response rate accelerates toward a performance criterion or goal, and the incorrect response rate decel- erates to a low, stable level. Also, reporting rate correct and rate incorrect provides for an assessment of propor- tional accuracy while maintaining the dimensional quantities of the measurement (e.g., 20 correct and 5 in- correct responses per minute = 80% accuracy, or a mul- tiple of X4 proportional accuracy).

Correct and incorrect response rates provide essen- tial data for the assessment of fluent performance (i.e., proficiency) (Kubina, 2005). The assessment of fluency requires measurement of the number of correct and in- correct responses per unit of time (i.e., proportional ac- curacy). Analysts cannot assess fluency using only the correct rate because a fluenct performance must be ac- curate also.

Take Into Account the Varied Complexity of Re- sponses. Rate of responding is a sensitive and appro- priate measure of skill acquisition and the development of fluent performances only when the level of difficulty and complexity from one response to the next remains constant within and across observations. The rate-of- response measures previously discussed have been with whole units in which the response requirements are essentially the same from one response to the next. Many important behaviors, however, are composites of

two or more component behaviors, and different situa- tions call for varied sequences or combinations of the component behaviors.

One method for measuring rate of responding that takes varied complexity into account for multiple- component behaviors is to count the operations neces- sary to achieve a correct response. For example, in measuring students’ math calculation performance, in- stead of counting a two-digit plus three-digit addition problem with regrouping as correct or incorrect, a re- searcher might consider the number of steps that were completed in correct sequence within each problem. Hel- wig (1973) used the number of operations needed to pro- duce the answers to mathematics problems to calculate response rates. In each session the student was given 20 multiplication and division problems selected at random from a set of 120 problems. The teacher recorded the du- ration of time for each session. All the problems were of two types: a × b = c and a ÷ b = c. For each problem the student was asked to find one of the factors: the product, the dividend, the divisor, or the quotient. Depending on the problem, finding the missing factor required from one to five operations. For example, writing the answer 275 in response to the problem 55 × 5 = ? would be scored as four correct responses because finding the missing factor requires four operations:

1. Multiply the ones: 5 × 5 = 25.

2. Record the 5 ones and carry the 2 tens.

3. Multiply the tens: 5 × 5(0) = 25(0).

4. Add the 2 tens carried and record the sum (27).

When more than one way to find the answer was pos- sible, the mean number of operations was figured for that problem. For example, the answer to the problem, 4 × ? = 164, can be obtained by multiplication with two opera- tions and by division with four operations. The mean number of operations is three. Helwig counted the num- ber of operations completed correctly and incorrectly per set of 20 problems and reported correct and incorrect rates of response.

Use Rate of Responding to Measure Free Operants. Rate of response is a useful measure for all behaviors characterized as free operants. The term free operant refers to behaviors that have discrete beginning and end- ing points, require minimal displacement of the organism in time and space, can be emitted at nearly any time, do not require much time for completion, and can be emit- ted over a wide range of response rates (Baer & Fowler, 1984). Skinner (1966) used rate of response of free oper- ants as the primary dependent variable when develop- ing the experimental analysis of behavior. The bar press and key peck are typical free operant responses used in

Measuring Behavior

nonhuman animal laboratory studies. Human subjects in a basic laboratory experiment might depress keys on a keyboard. Many socially significant behaviors meet the definition of free operants: number of words read during a 1-minute counting period, number of head slaps per minute, number of letter strokes written in 3 minutes. A person can make these responses at almost any time, each discrete response does not use much time, and each response class can produce a wide range of rates.

Rate of response is a preferred measure for free oper- ants because it is sensitive to changes in behavior values (e.g., oral reading may occur at rates ranging from 0 to 250 or more correct words per minute), and because it offers clarity and precision by defining a count per unit of time.

Do Not Use Rate to Measure Behaviors that Occur within Discrete Trials. Rate of response is not a useful measure for behaviors that can occur only within lim- ited or restricted situations. For example, response rates of behaviors that occur within discrete trials are con- trolled by a given opportunity to emit the response. Typical discrete trials used in nonhuman animal labora- tory studies include moving from one end of a maze or shuttle box to another. Typical applied examples in- clude responses to a series of teacher-presented flash cards; answering a question prompted by the teacher; and, when presented with a sample color, pointing to a color from an array of three colors that matches the sample color. In each of these examples, the rate of re- sponse is controlled by the presentation of the an- tecedent stimulus. Because behaviors that occur within discrete trials are opportunity bound, measures such as percentage of response opportunities in which a re- sponse was emitted or trials-to-criterion should be em- ployed, but not rate measures.

Do Not Use Rate to Measure Continuous Behaviors that Occur for Extended Periods of Time. Rate is also a poor measure for continuous behaviors that occur for extended periods of time, such as participating in play- ground games or working at a play or activity center. Such behaviors are best measured by whether they are “on” or “off” at any given time, yielding data on duration or estimates of duration obtained by interval recording.

Celeration

Just like a car that accelerates when the driver presses the gas pedal and decelerates when the driver releases the pedal or steps on the brake, rates of response accel- erate and decelerate. Celeration is a measure of how rates of response change over time. Rate of response acceler- ates when a participant responds faster over successive counting periods and decelerates when responding slows over successive observations. Celeration incorporates

3Celeration, the root word of acceleration and deceleration, is a generic term without specific reference to accelerating or decelerating response rates. Practitioners and researchers should use acceleration or deceleration when describing increasing or decreasing rates of response. 4Methods for calculating and drawing trend lines are described in Chap- ter entitled “Constructing and Interpreting Graphic Displays of Behavioral Data.”

three dimensional quantities: count per unit time/per unit of time; or expressed another way, rate/per unit of time (Graf & Lindsley, 2002; Johnston & Pennypacker, 1993a). Celeration provides researchers and practitioners with a direct measure of dynamic patterns of behavior change such as transitions from one steady state of re- sponding to another and the acquisition of fluent levels of performance (Cooper, 2005).

The Standard Celeration Chart provides a standard format for displaying measures of celeration (Penny- packer, Gutierrez, & Lindsley, 2003).3 There are four Standard Celeration Charts, showing rate as count (a) per minute, (b) per week, (c) per month, and (d) per year. These four charts provide different levels of magnifica- tion for viewing and interpreting celeration. Response rate is displayed on the vertical, or y, axis of the charts, and successive calendar time in days, weeks, months, or years is presented on the horizontal, or x, axis. Teachers and other behavioral practitioners use the count per minute–successive calendar days chart most often.

Celeration is displayed on all Standard Celeration Charts with a celeration trend line. A trend line, a straight line drawn through a series of graphed data points, visu- ally represents the direction and degree of trend in the data.4 The celeration trend line shows a factor by which rate of response is multiplying (accelerating) or dividing (decelerating) across the celeration time period (e.g., rate per week, rate per month, rate per year, rate per decade). A celeration time period is 1/20th of the horizontal axis of all Standard Celeration Charts. For example, the cel- eration period for the successive calendar days chart is per week. The celeration period for the successive cal- endar weeks chart is per month.

A trend line drawn from the bottom left corner to the top right corner on all Standard Celeration Charts has a slope of 34°, and has an acceleration value of X2 (read as “times-2”; celerations are expressed as multiples or di- visors). It is this 34° angle of celeration, a linear mea- sure of behavior change across time, that makes a chart standard. For example, if the response rate were 20 per minute on Monday, 40 per minute the next Monday, and 80 per minute on the third Monday, the celeration line (i.e., the trend line) on the successive calendar days chart would show a X2 acceleration, a doubling in rate per week. A X2 acceleration on successive calendar weeks chart is a doubling in rate every week (e.g., 20 in Week 1, 40 in Week 2, and 80 in Week 3). In most cases, cel- eration values should not be calculated with fewer than seven data points.

Measuring Behavior

Measures Based on Temporal Extent

Duration

Duration, the amount of time in which behavior occurs, is the basic measure of temporal extent. Researchers and practitioners measure duration in standard units of time (e.g., Enrique worked cooperatively with his peer tutor for 12 minutes and 24 seconds today).

Duration is important when measuring the amount of time a person engages in the target behavior. Applied behavior analysts measure the duration of target behav- iors that a person has been engaging in for too long or for too short of a time period, such as a child with de- velopmental disabilities who tantrums for more than an hour at a time, or a student who sticks with an academic task for no more than 30 seconds at a time.

Duration is also an appropriate measure for behaviors that occur at very high rates (e.g., rocking; rapid jerks of the head, hands, legs) or task-oriented continuous be- haviors that occur for an extended time (e.g., coopera- tive play, on-task behavior, off-task behavior).

Behavioral researchers and practitioners commonly measure one or both of two kinds of duration measures: total duration per session or observation period and duration per occurrence.

Total Duration per Session. Total duration is a measure of the cumulative amount of time in which a person engages in the target behavior. Applied behavior analysts use two procedures for measuring and report- ing total duration. One method involves recording the cumulative amount of time a behavior occurs within a specified observation period. For example, a teacher concerned that a kindergarten child is spending too much time in solitary play could record the total time that the child is observed engaged in solitary play dur- ing daily 30-minute free-play periods. Procedurally, when the child engages in solitary play, the teacher ac- tivates a stopwatch. When solitary play ceases, the teacher stops the stopwatch but does not reset it. When the child drifts into solitary play again, the teacher starts the stopwatch again. Over the course of a 30- minute free-play period, the child might have engaged in a total of 18 minutes of solitary play. If the duration of the free-play periods varied from day to day, the teacher would report the total duration of solitary be- havior as a percentage of total time observed (i.e., total duration of solitary behavior ÷ duration of free-play pe- riod × 100 = % of solitary behavior in one free-play pe- riod). In this case, 18 minutes of cumulative solitary play within a 30-minute session yields 60%.

people with profound developmental disabilities. They used a stopwatch to record physical engagement with an item (i.e., contact with both hand and the item) during 2-minute trials. They reported total contact in seconds by summing the duration values across three 2-minute trials of each assessment. McCord, Iwata, Galensky, Elling- son, and Thomson (2001) started and stopped a stopwatch to measure the total duration in seconds that two adults with severe or profound mental retardation engaged in problem behavior.

The other measure of total duration recording is the amount of time a person spends engaged in an activity, or the time a person needs to complete a specific task, with- out specifying a minimum or maximum observation pe- riod. For example, a community planner concerned with the amount of time senior citizens attended a new recre- ation center could report the total number of minutes per day that each senior spends at the center.

Duration per Occurrence. Duration per occur- rence is a measure of the duration of time that each in- stance of the target behavior occurs. For example, assume that a student leaves his seat frequently and for varying amounts of time. Each time the student leaves his seat, the duration of his out-of-seat behavior could be recorded with a stopwatch. When he leaves his seat, the stopwatch is engaged. When he returns, the stop- watch is disengaged, the total time recorded, and the stopwatch reset to zero. When the student leaves his seat again, the process is repeated. The resulting mea- sures yield data on the duration of occurrences of each out-of-seat behavior over the observation period.

In a study evaluating an intervention for decreasing noisy disruptions by children on a school bus, Greene, Bailey, and Barber (1981) used a sound-recording device that automatically recorded both the number of times that outbursts of sound exceeded a specified threshold and the duration in seconds that each outburst remained above that threshold. The researchers used the mean duration per occurrence of noisy disruptions as one measure for evaluating the intervention’s effects.

Selecting and Combining Measures of Count and Duration. Measurements of count, total duration, and duration per occurrence provide different views of be- havior. Count and duration measure different dimen- sional quantities of behavior, and these differences provide the basis for selecting which dimension to mea- sure. Event recording meaures repeatability, whereas duration recording measures temporal extent. For in- stance, a teacher concerned about a student who is out of her seat “too much” could tally each time the student leaves her seat. The behavior is discrete and is unlikely

Zhou, Iwata, Goff, and Shore (2001) used total du- ration measurement to assess leisure-item preferences of

Measuring Behavior

to occur at such a high rate that counting the number of occurrences would be difficult. Because any instance of out-of-seat behavior has the potential to occur for an extended time and the amount of time the student is out of her seat is a socially significant aspect of the behavior, the teacher could also use total duration recording.

Using count to measure out-of-seat behavior pro- vides the number of times the student left her seat. A mea- sure of total duration will indicate the amount and proportion of time that the student was out of her seat during the observation period. Because of the relevance of temporal extent in this case, duration would provide a better measurement than count would. The teacher might observe that the student left her seat once in a 30-minute observation period. One occurrence of the behavior in 30 minutes might not be viewed as a problem. However, if the student remained out of her seat for 29 of the obser- vation period’s 30 minutes, a very different view of the behavior is obtained.

In this situation, duration per occurrence would make an even better measurement selection than either count or total duration recording would. That is because duration per occurrence measures the repeatability and the temporal extent of the behavior. A duration-per- occurrence measure would give the teacher information on the number of times the student was out of her seat and the duration of each occurrence. Duration per oc- currence is often preferable to total duration because it is sensitive to the number of instances of the target be- havior. Further, if a total duration measure is needed for other purposes, the individual durations of each of the counted and timed occurrences could be summed. How- ever, if behavior endurance (e.g., academic responding, motor movements) is the major consideration, then total duration recording may be sufficient (e.g., oral reading for 3 minutes, free writing for 10 minutes, running for 10 kilometers).

Measures Based on Temporal Locus

As stated previously, temporal locus refers to when an in- stance of behavior occurs with respect to other events of in- terest. The two types of events most often used by researchers as points of reference for measuring temporal locus are the onset of antecedent stimulus events and the cessation of the previous response. These two points of reference provide the context for measuring response latency and interresponse time, the two measures of temporal locus most frequently reported in the behavior analysis literature.

Response Latency

Response latency (or, more commonly, latency) is a mea- sure of the elapsed time between the onset of a stimulus

5Latency is most often used to describe the time between the onset of an antecedent stimulus change and the initiation of a response. However, the term can be used to refer to any measure of the temporal locus of a re- sponse with respect to any type of antecedent event. See Johnston and Pen- nypacker (1993b).

and the initiation of a subsequent response.5 Latency is an appropriate measure when the researcher or practitioner is interested in how much time occurs between an op- portunity to emit a behavior and when the behavior is ini- tiated. For example, a student might exhibit excessive delays in complying with teacher directions. Response latency would be the elapsed time from the end of the teacher’s direction and when the student begins to com- ply with the direction. Interest can also focus on latencies that are too short. A student may give incorrect answers because she does not wait for the teacher to complete the questions. An adolescent who, at the slightest provocation from a peer, immediately retaliates has no time to con- sider alternative behaviors that could defuse the situation and lead to improved interactions.

Researchers typically report response latency data by the mean, median and range of individual latencies measures per observation period. For example, Lerman, Kelley, Vorndran, Kuhn, and LaRue (2002) used a latency measure to assess the effects of different reinforcement magnitudes (i.e., 20 seconds, 60 seconds, or 300 seconds of access to a reinforcer) on postreinforcement pause. The researchers measured the number of seconds from the end of each reinforcer-access interval to the first in- stance of the target behavior (a communication response). They then calculated and graphed the mean, median, and range of response latencies measured during each ses- sion (see Lerman et al., 2002, p. 41).

Interresponse Time

Interresponse time (IRT) is the amount of time that elapses between two consecutive instances of a response class. Like response latency, IRT is a measure of tempo- ral locus because it identifies when a specific instance of behavior occurs with respect to another event (i.e., the previous response). Figure 2 shows a schematic repre- sentation of interresponse time.

Although it is a direct measure of temporal locus, IRT is functionally related to rate of response. Shorter IRTs coexist with higher rates of response, and longer IRTs are found within lower response rates. Applied behavior analysts measure IRT when the time between instances of a response class is important. IRT provides a basic measure for implementing and evaluating interventions using differential reinforcement of low rates (DRL), a procedure for using reinforcement to re- duce the rate of responding. Like latency data, IRT

100

Measuring Behavior

measures are typically reported and displayed graphically by mean (or median) and range per observation period.

Derivative Measures

Percentage and trials-to-criterion, two forms of data de- rived from direct measures of dimensional quantities of behavior, are frequently used in applied behavior analysis.

Percentage

A percentage is a ratio (i.e., a proportion) formed by combining the same dimensional quantities, such as count (i.e., number ÷ number) or time (i.e., duration ÷ duration; latency ÷ latency). A percentage expresses the propor- tional quantity of some event in terms of the number of times the event occurred per 100 opportunities that the event could have occurred. For example, if a student an- swered correctly 39 of 50 items on an exam, an accuracy percentage can be calculated by dividing the number of correct answers by the total number of test items and mul- tiplying that product by 100 (39 ÷ 50 = .78 × 100 = 78%).

Percentage is frequently used in applied behavior analysis to report the proportion of total correct responses. For example, Ward and Carnes (2002) used a percentage- of-correct-performances measure in their study evaluat- ing the effects of goal setting and public posting on skill execution of three defensive skills by linebackers on a college football team. The researchers recorded counts of correct and incorrect reads, drops, and tackles by each player and calculated accuracy percentages based on the number of opportunities for each type of play.

Percentage is also used frequently in applied behav- ior analysis to report the proportion of observation inter- vals in which the target behavior occurred. These measures are typically reported as a proportion of inter- vals within a session. Percentage can also be calculated for an entire observation session. In a study analyzing the differential effects of reinforcer rate, quality, imme- diacy, and response effort on the impulsive behavior of students with attention-deficit/ hyperactivity disorder, Neef, Bicard, and Endo (2001) reported the percentage of time each student allocated to two sets of concurrently available math problems (e.g., time allocated to math problems yielding high-quality delayed reinforcers ÷ total time possible 100 = %).

Percentages are used widely in education, psychol- ogy, and the popular media, and most people understand proportional relationships expressed as percentages. However, percentages are often used improperly and fre- quently misunderstood. Thus, we offer several notes of caution on the use and interpretation of percentages.

Percentages most accurately reflect the level of and changes in behavior when calculated with a divisor (or de- nominator) of 100 or more. However, most percentages used by behavioral researchers and practitioners are cal- culated with divisors much smaller than 100. Percentage measures based on small divisors are unduly affected by small changes in behavior. For example, a change in count of just 1 response per 10 opportunities changes the per- centage by 10%. Guilford (1965) cautioned that it is unwise to compute percentages with divisors smaller than 20. For research purposes, we recommend that whenever possi- ble, applied behavior analysts design measurement sys- tems in which resultant percentages are based on no fewer than 30 response opportunities or observation intervals.

Sometimes changes in percentage can erroneously suggest improving performance. For example, an accu- racy percentage could increase even though the frequency of incorrect responses remains the same or worsens. Con- sider a student whose accuracy in answering math prob- lems on Monday is 50% (5 of 10 problems answered correctly) and on Tuesday it is 60% (12 of 20 problems answered correctly). Even with the improved proportional accuracy, the number of errors increased (from 5 on Mon- day to 8 on Tuesday).

Although no other measure communicates propor- tional relationships better than percentage does, its use as a behavioral quantity is limited because a percentage has no dimensional quantities.6 For example, percentage cannot be used to assess the development of proficient or fluent behavior because an assessment of proficiency cannot occur without reference to count and time, but it can show the proportional accuracy of a targeted behav- ior during the development of proficiency.

R1 R2 R4R3

Time

IRT IRT IRT Figure 2 Schematic representation of three interresponse times (IRT). IRT, the elapsed time between the termination of one response and the initiation of the next response, is a commonly used measure of temporal locus.

6Because percentages are ratios based on the same dimensional quantity, the dimensional quantity is canceled out and no longer exists in the per- centage. For example, an accuracy percentage created by dividing the num- ber of correct responses by the number of response opportunities removes the actual count. However, ratios created from different dimensional quan- tities retain the dimensional quantities of each component. For example, rate retains a count per unit of time. See Johnston and Pennypacker (1993a) for further explication.

101

Measuring Behavior

Another limitation of percentage as a measure of be- havior change is that lower and upper limits are imposed on the data. For example, using percent correct to assess a student’s reading ability imposes an artificial ceiling on the measurement of performance. A learner who cor- rectly reads 100% of the words she is presented cannot improve in terms of the measure used.

Different percentages can be reported from the same data set, with each percentage suggesting significantly dif- ferent interpretations. For example, consider a student who scores 4 correct (20%) on a 20-item pretest and 16 correct (80%) when the same 20 items are given as a posttest. The most straightforward description of the student’s im- provement from pretest to posttest (60%) compares the two measures using the original basis or divisor (20 items). Because the student scored 12 more items correct on the posttest than he did on the pretest, his performance on the posttest could be reported as an increase (gain score) over his pretest performance of 60%. And given that the stu- dent’s posttest score represented a fourfold improvement in correct responses, some might report the posttest score as a 300% improvement of the pretest—a completely dif- ferent interpretation from an improvement of 40%.

Although percentages greater than 100% are some- times reported, strictly speaking, doing so is incorrect. Although a behavior change greater than 100% may seem impressive, it is a mathematical impossibility. A percent- age is a proportional measure of a total set, where x (the proportion) of y (the total) is expressed as 1 part in 100. A proportion of something cannot exceed the total of that something or be less than zero (i.e., there is no such thing as a negative percentage). Every coach’s favorite athlete who “always gives 110%” simply does not exist.7

Trials-to-Criterion

Trials-to-criterion is a measure of the number of re- sponse opportunities needed to achieve a predetermined level of performance. What constitutes a trial depends on the nature of the target behavior and the desired perfor- mance level. For a skill such as shoe tying, each oppor- tunity to tie a shoe could be considered a trial, and trials-to-criterion data are reported as the number of tri- als required for the learner to tie a shoe correctly without prompts or assistance. For behaviors involving problem solving or discriminations that must be applied across a large number of examples to be useful, a trial might con-

7When a person reports a percentage in excess of 100% (e.g., “Our mutual fund grew by 120% during the recent bear market”), he is probably using the misstated percentage in comparison to the previous base unit, not as a proportion of it. In this example, the mutual fund’s 20% increase makes its value 1.2 times greater than its value when the bear market began.

sist of a block, or series of response opportunities, in which each response opportunity involves the presenta- tion of a different exemplar of the problem or discrimi- nation. For example, a trial for discriminating between the short and long vowel sounds of the letter o could be a block of 10 consecutive opportunities to respond, in which each response opportunity is the presentation of a word containing the letter o, with short-vowel and long- vowel o words (e.g., hot, boat) presented in random order. Trials-to-criterion data could be reported as the number of blocks of 10 trials required for the learner to correctly pronounce the o sound in all 10 words. Count would be the basic measure from which the trials-to-criterion data would be derived.

Other basic measures, such as rate, duration, and la- tency, can also be used to determine trials-to-criterion data. For example, a trials-to-criterion measure for solv- ing two-digit minus two-digit subtraction problems re- quiring borrowing could be the number of practice sheets of 20 randomly generated and sequenced problems a learner completes before she is able to solve all 20 prob- lems on a single sheet in 3 minutes or less.

Trials-to-criterion data are often calculated and re- ported as an ex post facto measure of one important aspect of the “cost” of a treatment or instructional method. For example, Trask-Tyler, Grossi, and Heward (1994) reported the number of instructional trials needed by each of three students with developmental disabilities and visual im- pairments to prepare three food items from recipes with- out assistance on two consecutive times over two sessions. Each recipe entailed from 10 to 21 task-analyzed steps.

Trials-to-criterion data are used frequently to com- pare the relative efficiency of two or more treatments or instructional methods. For example, by comparing the number of practice trials needed for a student to master weekly sets of spelling words practiced in two different ways, a teacher could determine whether the student learns spelling words more efficiently with one method than with another. Sometimes trials-to-criterion data are supplemented by information on the number of minutes of instruction needed to reach predetermined performance criteria (e.g., Holcombe, Wolery, Werts, & Hrenkevich 1993; Repp, Karsh, Johnson, & VanLaarhoven, 1994).

Trials-to-criterion measures can also be collected and analyzed as a dependent variable throughout a study. For example, R. Baer (1987) recorded and graphed trials-to- criterion on a paired-associates memory task as a depen- dent variable in a study assessing the effects of caffeine on the behavior of preschool children.

Trials-to-criterion data can also be useful for assess- ing a learner’s increasing competence in acquiring a re- lated class of concepts. For instance, teaching a concept such as the color red to a child could consist of present- ing “red” and “not red” items to the child and providing

102

Measuring Behavior

differential reinforcement for correct responses. Trials- to-criterion data could be collected of the number of “red” and “not red” exemplars required before the child achieves a specified level of performance with the dis- crimination. The same instructional and data collection procedures could then be used in teaching other colors to the child. Data showing that the child achieves mastery of each newly introduced color in fewer instructional trials than it took to learn previous colors might be evidence of the child’s increasing agility in learning color concepts.

Definitional Measures

In addition to the basic and derived dimensions already discussed, behavior can also be defined and measured by its form and intensity. Neither form (i.e., topography) nor intensity (i.e., magnitude) of responding is a fundamen- tal dimensional quantity of behavior, but each is an im- portant quantitative parameter for defining and verifying the occurrence of many response classes. When behav- ioral researchers and practitioners measure the topogra- phy or magnitude of a response, they do so to determine whether the response represents an occurrence of the tar- get behavior. Occurrences of the target behavior that are verified on the basis of topography or magnitude are then measured by one or more aspects of count, temporal ex- tent, or temporal locus. In other words, measuring topog- raphy or magnitude is sometimes necessary to determine whether instances of the targeted response class have oc- curred, but the subsequent quantification of those re- sponses is recorded, reported, and analyzed in terms of the fundamental and derivative measures of count, rate, duration, latency, IRT, percentage, and trials-to-criterion.

Topography

Topography, which refers to the physical form or shape of a behavior, is both a measurable and malleable dimension of behavior. Topography is a measurable quantity of behavior because responses of varying form can be detected from one another. That topography is a malleable aspect of behavior is evidenced by the fact that responses of varying form are shaped and selected by their consequences.

A group of responses with widely different topogra- phies may serve the same function (i.e., form a response class). For example, each of the different ways of writing the word topography, shown in Figure 3, would pro- duce the same effect on most readers. Membership in some response classes, however, is limited to responses within a narrow range of topographies. Although each of the response topographies in Figure 3 would meet the functional requirements of most written communications, none would meet the standards expected of an advanced calligraphy student.

Topography is of obvious and primary importance in performance areas in which form, style, or artfulness of behavior is valued in its own right (e.g., painting, sculpt- ing, dancing, gymnastics). Measuring and providing dif- ferential consequences for responses of varied topographies is also important when the functional outcomes of the be- havior correlate highly with specific topographies. A stu- dent who sits with good posture and looks at the teacher is more likely to receive positive attention and opportunities to participate academically than is a student who slouches with her head on the desk (Schwarz & Hawkins, 1970). Basketball players who execute foul shots with a certain form make a higher percentage of shots than they do when they shoot idiosyncratically (Kladopoulos & McComas, 2001).

Trap, Milner-Davis, Joseph, and Cooper (1978) mea- sured the topography of cursive handwriting by first- grade students. Plastic transparent overlays were used to

Figure 3 Topography, the physical form or shape of behavior, is a measurable dimension of behavior.

103

Measuring Behavior

Figure 4 Examples of outlines on a transparent overlay used to measure inside and outside boundaries of manuscript letters and an illustration of using the transparent overlay to measure the letter m. Because the vertical stroke of the letter m extended beyond the confines of the outline, it did not meet the topographical criteria for a correct response.

From “The Measurement of Manuscript Letter Strokes” by J. J. Helwig, J. C. Johns, J. E. Norman, J. O. Cooper, 1976. Journal of Applied Behavior Analysis, 9, p. 231. Copyright 1976 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

detect deviations from model letters in lower- and upper- case letters written by the children (see Figure 4). The researchers counted the number of correct letter strokes— those that met all of the specified topographical criteria (e.g., all letter strokes contained within the 2-mm para- meters of the overlay, connected, complete, sufficient length)—and used the percentage correct of all strokes written by each student to assess the effects of visual and verbal feedback and a certificate of achievement on the children’s acquisition of cursive handwriting skills.

Magnitude

Magnitude refers to the force or intensity with which a response is emitted. The desired outcomes of some be- haviors are contingent on responding at or above (or below) a certain intensity or force. A screwdriver must be turned with sufficient force to insert or remove screws; a pencil must be applied to paper with enough force to leave legible marks. On the other hand, applying too much torque to a misaligned screw or bolt is likely to

strip the threads, and pushing too hard with a pencil will break its point.

The magnitude of speech or other vocalizations that were considered too loud or too soft has been mea- sured in several studies. Schwarz and Hawkins (1970) measured the voice loudness of Karen, a sixth-grade girl who spoke so softly in class that her voice was usually inaudible to others. Karen’s voice was recorded on videotape during two class periods each day. (The videotape was also used to obtain data on two other behaviors: face touching and the amount of time Karen sat in a slouched position). The researchers then played the videotape into an audiotape recorder with a loud- ness indicator and counted the number of times the needle went above a specified level on the loudness meter. Schwarz and Hawkins used the number (pro- portion) of needle inflections per 100 words spoken by Karen as the primary measure for evaluating the ef- fects of an intervention on increasing her voice vol- ume during class.

Greene and colleagues (1981) used an automated sound-recording device to measure the magnitude of noisy disruptions by middle school students on a school bus. The recording device could be adjusted so that only sound levels above a predetermined threshold activated it. The device automatically recorded both the number of times outbursts of sound exceeded a specified threshold (93 dB) and the total duration in seconds that the sound remained above that threshold. When the noise level ex- ceeded the specified threshold, a light on a panel that all students could see was activated automatically. When the light was off, students listened to music during the bus ride; when the number of noisy disruptions was below a criterion, they participated in a raffle for prizes. This in- tervention drastically reduced the outbursts and other problem behaviors as well. Greene and colleagues re- ported both the number and the mean duration per oc- currence of noisy disruptions as measures of evaluating the intervention’s effects.8

Table 1 summarizes the measurable dimensions of behavior and considerations for their use.

8Researchers sometimes manipulate and control response topography and magnitude to assess their possible effects as independent variables. Piazza, Roane, Kenney, Boney, and Abt (2002) analyzed the effects of different response topographies on the frequency of pica (e.g., ingesting nonnutri- tive matter that may threaten one’s life) by three females. Pica items were located in various places, which required the subjects to respond in differ- ent ways (e.g., reach, bend over, get on the floor, open a container) to ob- tain them. Pica decreased when more elaborate response topographies were required to obtain pica items. Van Houten (1993) reported that when a boy with a long history of intense and high-frequency face slapping wore 1.5- pound wrist weights, face slapping immediately dropped to zero. Studies such as these suggest that problem behaviors may be reduced when en- gaging in those behaviors requires more effortful responses in terms of topography or magnitude.

104

Ta b

le 1

F un

da m

en ta

l, de

riv ed

, a nd

d ef

in iti

on al

d im

en si

on s

by w

hi ch

b eh

av io

r ca

n be

m ea

su re

d an

d de

sc rib

ed .

Fu nd

am en

ta l m

ea su

re s

H ow

c al

cu la

te d

C on

si de

ra tio

C ou

nt :T

he n

um be

r of

r es

po ns

es em

itt ed

d ur

in g

an o

bs er

va tio

n pe

rio d.

S im

pl e

ta lly

o f t

he n

um be

r of

r es

po ns

es o

bs er

ve d.

Ju da

h co

nt rib

ut ed

5 c

om m

en ts

d ur

in g

th e

10 -m

in ut

e cl

as s

di sc

us si

on .

• O

bs er

va tio

n tim

e in

w hi

ch c

ou nt

w as

r ec

or de

d sh

ou ld

b e

re fe

re nc

ed .

• M

os t u

se fu

l f or

c om

pa ris

on w

he n

ob se

rv at

io n

tim e

(c ou

nt in

g tim

e) is

c on

st an

t a cr

os s

ob se

rv at

io ns

• U

se d

in c

al cu

la tin

g ra

te /fr

eq ue

nc y,

c el

er at

io n,

p er

ce nt

ag es

, a nd

tr ia

ls -t

o- cr

ite rio

n m

ea su

re s.

R at

e/ fre

qu en

cy :A

ra tio

o f c

ou nt

pe r

ob se

rv at

io n

tim e;

of te

n ex

pr es

se d

as c

ou nt

p er

s ta

nd ar

d un

it of

ti m

e (e

.g .,

pe r

m in

ut e,

p er

ho ur

, p er

d ay

R ep

or t n

um be

r of

r es

po ns

es r

ec or

de d

pe r

tim e

in w

hi ch

ob se

rv at

io ns

w er

e co

nd uc

te d.

If Ju

da h’

s co

m m

en ts

w er

e co

un te

d du

rin g

a 10

-m in

ut e

cl as

s di

sc us

si on

, h is

ra te

o f r

es po

nd in

g w

ou ld

b e

5 co

m m

en ts

p er

1 0

m in

ut es

O fte

n ca

lc ul

at ed

b y

di vi

di ng

th e

nu m

be r

of r

es po

ns es

r ec

or de

d by

nu m

be r

of s

ta nd

ar d

un its

o f t

im e

in w

hi ch

o bs

er va

tio ns

w er

e co

nd uc

te d.

Ju da

h m

ad e

co m

m en

ts a

t a ra

te o

f 0 .5

p er

m in

ut e.

• If

ob se

rv at

io n

tim e

va rie

s ac

ro ss

m ea

su re

m en

ts , c

al cu

la te

w ith

st an

da rd

u ni

ts o

f t im

• M

in im

iz e

fa ul

ty in

te rp

re ta

tio ns

b y

re po

rt in

g co

un tin

g tim

• E

va lu

at in

g sk

ill d

ev el

op m

en t a

nd fl

ue nc

y re

qu ire

s m

ea su

re m

en t o

f co

rr ec

t a nd

in co

rr ec

t r es

po ns

e ra

te s.

• A

cc ou

nt fo

r va

rie d

co m

pl ex

ity a

nd d

iff ic

ul ty

w he

n ca

lc ul

at in

g re

sp on

se ra

te s.

• R

at e

is th

e m

os t s

en si

tiv e

m ea

su re

o f c

ha ng

es in

r ep

ea ta

bi lit

• P

re fe

rr ed

m ea

su re

fo r

fr ee

o pe

ra nt

• P

oo r

m ea

su re

fo r

be ha

vi or

s th

at o

cc ur

w ith

in d

is cr

et e

tr ia

ls o

r fo

r be

ha vi

or s

th at

o cc

ur fo

r ex

te nd

ed d

ur at

io ns

• M

os t s

en si

tiv e

m ea

su re

o f b

eh av

io r

re pe

at ab

ili ty

D er

iv ed

m ea

su re

s H

ow c

al cu

la te

d C

on si

de ra

tio ns

C el

er at

io n:

T he

c ha

ng e

(a cc

el er

at io

n or

d ec

el er

at io

n) in

ra te

o f r

es po

nd in

g ov

er ti

m e.

B as

ed o

n co

un t p

er u

ni t o

f t im

e (r

at e)

;e xp

re ss

ed a

s fa

ct or

b y

w hi

ch re

sp on

di ng

is a

cc el

er at

in g/

de ce

le ra

tin g

(m ul

tip ly

in g

or d

iv id

in g)

A tr

en d

lin e

co nn

ec tin

g Ju

da h’

s m

ea n

ra te

s of

c om

m en

tin g

ov er

4 w

ee ks

o f 0

.1 , 0

.2 , 0

.4 , a

nd 0

.8 c

om m

en ts

p er

m in

ut e,

r es

pe ct

iv el

y, w

ou ld

s ho

w a

n ac

ce le

ra tio

n of

b y

a fa

ct or

o f t

im es

-2 p

er w

ee k.

• R

ev ea

ls d

yn am

ic p

at te

rn s

of b

eh av

io r

ch an

ge s

uc h

as tr

an si

tio ns

fr om

o ne

s te

ad y

st at

e to

a no

th er

a nd

a cq

ui si

tio n

of fl

ue nc

• D

is pl

ay ed

w ith

a tr

en d

lin e

on a

S ta

nd ar

d C

el er

at io

n C

ha rt

• M

in im

um o

f s ev

en m

ea su

re s

of ra

te r

ec om

m en

de d

fo r

ca lc

ul at

in g.

( c on

tin ue

105

R es

po ns

e la

te nc

y: T

he p

oi nt

in tim

e w

he n

a re

sp on

se o

cc ur

s w

ith r

es pe

ct to

th e

oc cu

rr en

ce o

f an

a nt

ec ed

en t s

tim ul

us .

R ec

or d

tim e

el ap

se d

fr om

th e

on se

t o f t

he a

nt ec

ed en

t s tim

ul us

e ve

nt an

d th

e be

gi nn

in g

of th

e re

sp on

se ;o

fte n

re po

rt ed

b y

m ea

n or

m ed

ia n

an d

ra ng

e of

la te

nc ie

s pe

r se

ss io

Ju da

h’ s

co m

m en

ts to

da y

ha d

a m

ea n

la te

nc y

of 3

0 se

co nd

s fo

llo w

in g

a pe

er s’

co m

m en

t ( ra

ng e,

5 to

9 0

se co

nd s)

• Im

po rt

an t m

ea su

re w

he n

ta rg

et b

eh av

io r

is a

p ro

bl em

b ec

au se

it is

e m

itt ed

w ith

la te

nc ie

s th

at a

re to

o lo

ng o

r to

o sh

or t.

• D

ec re

as in

g la

te nc

ie s

ca n

re ve

al p

er so

n’ s

in cr

ea si

ng m

as te

ry o

f so

m e

sk ill

R ec

or d

tim e

el ap

se d

fr om

th e

en d

of p

re vi

ou s

re sp

on se

a nd

th e

be gi

nn in

g of

n ex

t r es

po ns

e; of

te n

re po

rt ed

b y

m ea

n or

m ed

ia n

an d

ra ng

e of

IR T

s pe

r se

ss io

Ju da

h’ s

co m

m en

ts to

da y

ha d

a m

ed ia

n IR

T o

f 2 m

in ut

es a

nd a

ra ng

e of

1 0

se co

nd s

to 5

m in

ut es

• Im

po rt

an t m

ea su

re w

he n

th e

tim e

be tw

ee n

re sp

on se

s, o

r pa

ci ng

of b

eh av

io r,

is th

e fo

cu s.

• A

lth ou

gh a

m ea

su re

o f t

em po

ra l l

oc us

, I R

T is

c or

re la

te d

w ith

ra te

of r

es po

nd in

• A

n im

po rt

an t m

ea su

re w

he n

im pl

em en

tin g

an d

ev al

ua tin

g D

R L.

In te

rr es

po ns

e tim

e (IR

T) :T

he po

in t i

n tim

e w

he n

a re

sp on

se oc

cu rs

w ith

r es

pe ct

to th

e oc

cu rr

en ce

o f t

he p

re vi

ou s

re sp

on se

D iv

id e

nu m

be r

of r

es po

ns es

m ee

tin g

sp ec

ifi ed

c rit

er ia

( e.

g. , c

or re

ct re

sp on

se s,

r es

po ns

es w

ith m

in im

um IR

T, r

es po

ns es

o f p

ar tic

ul ar

to po

gr ap

hy )

by th

e to

ta l n

um be

r of

r es

po ns

es e

m itt

ed (

or r

es po

ns e

op po

rt un

iti es

) an

d m

ul tip

ly b

y 10

S ev

en ty

p er

ce nt

o f J

ud ah

’s c

om m

en ts

to da

y w

er e

re le

va nt

to th

e di

sc us

si on

to pi

• P

er ce

nt ag

es b

as ed

o n

di vi

so rs

s m

al le

r th

an 2

0 ar

e un

du ly

in flu

- en

ce d

by s

m al

l c ha

ng es

in b

eh av

io r.

M in

im um

o f 3

0 ob

se rv

at io

n in

te rv

al s

or r

es po

ns e

op po

rt un

iti es

r ec

om m

en de

d fo

r re

se ar

ch .

• C

ha ng

e in

p er

ce nt

ag e

m ay

e rr

on eo

us ly

s ug

ge st

im pr

ov ed

pe rfo

rm an

ce .

• A

lw ay

s re

po rt

th e

di vi

so r

on w

hi ch

p er

ce nt

ag e

m ea

su re

s ar

e ba

se d.

• C

an no

t b e

us ed

to a

ss es

s pr

of ic

ie nc

y or

fl ue

nc y.

Pe rc

en ta

ge :A

p ro

po rt

io n,

ex pr

es se

d as

a n

um be

r of

p ar

ts pe

r 10

0; ty

pi ca

lly e

xp re

ss ed

a s

a ra

tio o

f t he

n um

be r

of r

es po

ns es

of a

c er

ta in

ty pe

p er

to ta

l n um

be r

of r

es po

ns es

( or

o pp

or tu

ni tie

s or

in te

rv al

s in

w hi

ch s

uc h

a re

sp on

se c

ou ld

h av

e oc

cu rr

ed ).

Ta b

le 1

( co

n ti

n u

ed )

Fu nd

am en

ta l m

ea su

re s

H ow

c al

cu la

te d

C on

si de

ra tio

D ur

at io

n: T

he a

m ou

nt o

f t im

e th

at a

be ha

vi or

o cc

ur s.

To ta

l d ur

at io

n: Tw

o m

et ho

ds :(

a) A

dd th

e in

di vi

du al

a m

ou nt

s of

ti m

e fo

r ea

ch r

es po

ns e

du rin

g an

o bs

er va

tio n

pe rio

d; or

( b)

r ec

or d

to ta

l tim

e in

di vi

du al

is e

ng ag

ed in

a n

ac tiv

ity , o

r ne

ed s

to c

om pl

et e

a ta

sk ,

w ith

ou t a

m in

im um

o r

m ax

im um

o bs

er va

tio n

pe rio

Ju da

h sp

en t 1

.5 m

in ut

es c

om m

en tin

g in

c la

ss to

da y.

D ur

at io

n pe

r o cc

ur re

nc e:

R ec

or d

du ra

tio n

of ti

m e

fo r

ea ch

in st

an ce

of th

e be

ha vi

or ;o

fte n

re po

rt ed

b y

m ea

n or

m ed

ia n

an d

ra ng

e of

du ra

tio ns

p er

s es

si on

Ju da

h’ s

6 co

m m

en ts

to da

y ha

d a

m ea

n du

ra tio

n of

1 1

se co

nd s,

w ith

a ra

ng e

of 3

to 2

4 se

co nd

• Im

po rt

an t m

ea su

re w

he n

ta rg

et b

eh av

io r

is p

ro bl

em at

ic b

ec au

se it

oc cu

rs fo

r du

ra tio

ns th

at a

re to

o lo

ng o

r to

o sh

or t.

• U

se fu

l m ea

su re

fo r

be ha

vi or

s th

at o

cc ur

a t v

er y

hi gh

ra te

s an

d fo

r w

hi ch

a cc

ur at

e ev

en t r

ec od

in g

is d

iff ic

ul t (

e. g.

, f in

ge r

fli ck

in g)

• U

se fu

l m ea

su re

fo r

be ha

vi or

s th

at d

o no

t h av

e di

sc re

te b

eg in

ni ng

s an

d fo

r w

hi ch

e ve

nt r

ec od

in g

is d

iff ic

ul t (

e. g.

, h um

m in

g) .

• U

se fu

l m ea

su re

fo r

ta sk

-o rie

nt ed

o r

co nt

in uo

us b

eh av

io rs

( e.

g. ,

co op

er at

iv e

pl ay

• D

ur at

io n

pe r

oc cu

rr en

ce o

fte n

pr ef

er re

d ov

er to

ta l d

ur at

io n

be ca

us e

it in

cl ud

es d

at a

on c

ou nt

a nd

to ta

l d ur

at io

• U

se to

ta l d

ur at

io n

w he

n in

cr ea

si ng

th e

en du

ra nc

e of

b eh

av io

r is

th e

go al

• M

ea su

rin g

du ra

tio n

pe r

oc cu

rr en

ce e

nt ai

ls c

ou nt

in g

re sp

on se

s, w

hi ch

c an

b e

us ed

to c

al cu

la te

ra te

o f r

es po

nd in

D er

iv ed

m ea

su re

s H

ow c

al cu

la te

d C

on si

de ra

tio ns

106

U se

d to

d et

er m

in e

w he

th er

r es

po ns

es m

ee t t

op og

ra ph

ic al

c rit

er ia

; re

sp on

se s

m ee

tin g

th os

e cr

ite ria

a re

m ea

su re

d an

d re

po rt

ed b

y on

e or

m or

e fu

nd am

en ta

l o r

de riv

at iv

e m

ea su

re s

(e .g

., pe

rc en

ta ge

o f

re sp

on se

s m

ee tin

g to

po gr

ap hi

ca l c

rit er

ia ).

T he

p la

ne o

f t he

g ol

f c lu

b re

m ai

ne d

w ith

in p

lu s

or m

in us

2 d

eg re

es fr

om b

ac ks

w in

g to

fo llo

w -t

hr ou

gh o

n 85

% o

f A m

an da

’s s

w in

gs .

• Im

po rt

an t m

ea su

re w

he n

de si

re d

ou tc

om es

o f b

eh av

io r

ar e

co nt

in ge

nt o

n re

sp on

se s

m ee

tin g

ce rt

ai n

to po

gr ap

hi es

• Im

po rt

an t m

ea su

re fo

r pe

rfo rm

an ce

a re

as in

w hi

ch fo

rm , s

ty le

, o r

ar tfu

ln es

s is

v al

ue d.

To po

gr ap

hy :T

he fo

rm o

r sh

ap e

of b

eh av

io r.

A dd

n um

be r

of r

es po

ns es

o r

pr ac

tic e

tr ia

ls n

ec es

sa ry

fo r

le ar

ne r

to ac

hi ev

e sp

ec ifi

ed c

rit er

io n.

D ur

in g

cl as

s di

sc us

si on

tr ai

ni ng

s es

si on

s co

nd uc

te d

in th

e re

so ur

ce ro

om , 1

4 bl

oc ks

o f 1

0 op

po rt

un iti

es to

c om

m en

t w er

e ne

ed ed

fo r

Ju da

h to

a ch

ie ve

th e

cr ite

rio n

of 8

o n-

to pi

c co

m m

en ts

p er

1 0

op po

rt un

iti es

• P

ro vi

de s

an e

x po

st fa

ct o

de sc

rip tio

n of

th e

“c os

t” of

a tr

ea tm

en t o

r in

st ru

ct io

na l m

et ho

• U

se fu

l f or

c om

pa rin

g re

la tiv

e ef

fic ie

nc y

of d

iff er

en t m

et ho

ds o

f in

st ru

ct io

n or

tr ai

ni ng

• U

se fu

l i n

as se

ss in

g ch

an ge

s in

th e

ra te

a t w

hi ch

a le

ar ne

r m

as te

rs ne

w s

ki lls

( ag

ili ty

Tr ia

ls to

c rit

er io

n: N

um be

r of

re sp

on se

s, in

st ru

ct io

na l t

ria ls

, o r

pr ac

tic e

op po

rt un

iti es

n ee

de d

to re

ac h

a pr

ed et

er m

in ed

pe rfo

rm an

ce c

rit er

io n.

U se

d to

d et

er m

in e

w he

th er

r es

po ns

es m

ee t m

ag ni

tu de

c rit

er ia

; re

sp on

se s

m ee

tin g

th os

e cr

ite ria

a re

m ea

su re

d an

d re

po rt

ed b

y on

e or

m or

e fu

nd am

en ta

l o r

de riv

at iv

e m

ea su

re s

(e .g

., co

un t o

f re

sp on

se s

m ee

tin g

m ag

ni tu

de c

rit er

ia ).

Ji ll

be nc

h pr

es se

d a

60 -p

ou nd

b ar

2 0

tim es

• Im

po rt

an t m

ea su

re w

he n

de si

re d

ou tc

om es

o f b

eh av

io r

ar e

co nt

in ge

nt o

n re

sp on

se s

w ith

in c

er ta

in ra

ng e

of m

ag ni

tu de

s. M

ag ni

tu de

:T he

s tr

en gt

h, in

te ns

ity , o

r fo

rc e

of b

eh av

io r.

• Im

po se

s up

pe r

an d

lo w

er li

m its

o n

pe rfo

rm an

ce (

i.e .,

ca nn

ot ex

ce ed

1 00

% )

• W

id el

y di

ffe re

nt p

er ce

nt ag

es c

an b

e re

po rt

ed fr

om th

e sa

m e

da ta

se t.

• To

c al

cu la

te a

n ov

er al

l p er

ce nt

ag e

fr om

p er

ce nt

ag es

b as

ed o

n di

ffe re

nt d

en om

in at

or s

(e .g

., 90

% [9

/1 0]

, 8 7.

5% [7

/8 ],

33 %

[1 /3

], 10

0% [1

/1 ],

di vi

de th

e to

ta l n

um er

at or

s of

th e

co m

po ne

nt pe

rc en

ta ge

s (e

.g .,

18 )

by th

e to

ta l d

en om

in at

or s

(e .g

., 18

/ 22

= 81

.8 %

). A

m ea

n of

th e

pe rc

en ta

ge s

th em

se lv

es y

ie ld

s a

di ffe

re nt

ou tc

om e

(e .g

., 90

% +

8 7.

5% +

3 3%

+ 1

00 %

/ 4

= 77

.6 %

D ef

in iti

on al

m ea

su re

s H

ow c

al cu

la te

d C

on si

de ra

tio ns

107

Measuring Behavior

Procedures for Measuring Behavior Procedures for measuring behavior used most often by applied behavior analysts involve one or a combination of the following: event recording, timing, and various time sampling methods.

Event Recording

Event recording encompasses a wide variety of proce- dures for detecting and recording the number of times a behavior of interest occurs. For example, Cuvo, Lerch, Leurquin, Gaffaney, and Poppen (1998) used event recording while analyzing the effects of work require- ments and reinforcement schedules on the choice behav- ior of adults with mental retardation and preschool children while they were engaged in age-appropriate tasks (e.g., adults sorting silverware, children tossing beanbags or jumping hurdles). The researchers recorded each piece of silverware sorted, each beanbag tossed, and each hur- dle jumped.

Devices for Event Recording

Although pencil and paper are sufficient for making event recordings, the following devices and procedures may facilitate the counting process.

• Wrist counters. Wrist counters are useful for tallying student behaviors. Golfers use these counters to tally strokes. Most wrist counters record from 0 to 99 re- sponses. These counters can be purchased from sporting goods stores or large department stores.

• Hand-tally digital counters. These digital counters are similar to wrist counters. Hand-tally counters are frequently used in grocery chain stores, cafete- rias, military mess halls, and tollgates to tally the number of people served. These mechanical coun- ters are available in single or multiple channels and fit comfortably in the palm of the hand. With prac- tice, practitioners can operate the multiple-channel counters rapidly and reliably with just one hand. These digital counters can be obtained from office supply stores.

• Abacus wrist and shoestring counters. Landry and McGreevy (1984) described two types of abacus counters for measuring behavior. The abacus wrist counter is made from pipe cleaners and beads at- tached to a leather wristband to form an abacus with rows designated as ones and tens. An observer can tally from 1 to 99 occurrences of behavior by sliding the beads in abacus fashion. Responses are

tallied in the same way on an abacus shoestring counter except the beads slide on shoestrings at- tached to a key ring, which is attached to the ob- server’s belt, belt loop, or some other piece of clothing such as a buttonhole.

• Masking tape. Teachers can mark tallies on mask- ing tape attached to their wrists or desk.

• Pennies, buttons, paper clips. One item can be moved from one pocket to another each time the target behavior occurs.

• Pocket calculators. Pocket calculators can be used to tally events.

Event recording also applies to the measurement of discrete trial behaviors, in which the count for each trial or opportunity to respond is either 1 or 0, representing the occurrence or nonoccurrence of the behavior. Figure 5 shows a form used to record the occurrence of imita- tion responses by a preschooler with disabilities and his typically developing peer partner within a series of in- structional trials embedded into ongoing classroom ac- tivities (Valk, 2003). For each trial, the observer recorded the occurrence of a correct response, no response, an ap- proximation, or an inappropriate response by the target child and the peer by circling or marking a slash through letters representing each behavior. The form also allowed the observer to record whether the teacher prompted or praised the target child’s imitative behavior.

Considerations with Event Recording

Event recording is easy to do. Most people can tally dis- crete behaviors accurately, often on the first attempt. If the response rate is not too high, event recording does not interfere with other activities. A teacher can con- tinue with instruction while tallying occurrences of the target behavior.

Event recording provides a useful measurement for most behaviors. However, each instance of the target be- havior must have discrete beginning and ending points. Event recording is applicable for target behaviors such as students’ oral responses to questions, students’ written answers to math problems, and a parent praising his son or daughter. Behaviors such as humming are hard to mea- sure with event recording because an observer would have difficulty determining when one hum ends and another begins. Event recording is difficult for behaviors defined without specific discrete action or object relations, such as engagement with materials during free-play activity. Because engagement with materials does not present a specific discrete action or object relation, an observer may have difficulty judging when one engagement starts and ends, and then another engagement begins.

108

Measuring Behavior

Session date:

Target child:

Target behavior:

Session no:

Peer:

Condition:

Observer:

IOA day: YES NO

May 21 16

Jordan Ethan

Jennie

Place block on structure 5-sec time delay Code: C = Correct A = Approximation I = Inappropriate

Trial Target child's behavior Teacher behavior toward target child

Peer’s behavior Teacher praise

1 C N A I Prompt Praise C N A I Praise 2 C N A I Prompt Praise C N A I Praise 3 C N A I Prompt Praise C N A I Praise 4 C N A I Prompt Praise C N A I Praise 5 C N A I Prompt Praise C N A I Praise 6 C N A I Prompt Praise C N A I Praise 7 C N A I Prompt Praise C N A I Praise 8 C N A I Prompt Praise C N A I Praise 9 C N A I Prompt Praise C N A I Praise 10 C N A I Prompt Praise C N A I Praise

No. corrects by target child: No. corrects by peer:8 9

************************************************

Target behavior: Condition:Place sticker on paper 5-sec time delay

Trial Target child's behavior Teacher behavior toward target child

Peer’s behavior Teacher praise

No. corrects by target child: No. corrects by peer:4 8

N = No response

Figure 5 Data collection form for recording the behavior of two children and a teacher during a series of discrete trials. Adapted from The Effects of Embedded Instruction within the Context of a Small Group on the Acquisition of Imitation Skills of Young Children with Disabilities by J. E. Valk (2003), p. 167. Unpublished doctoral dissertation, The Ohio State University. Used by permission.

Another consideration with event recording is that the target behaviors should not occur at such high rates that an observer would have difficulty counting each dis- crete occurrence accurately. High-frequency behaviors that may be difficult to measure with event recording in- clude rapid talking, body rocks, and tapping objects.

Also, event recording does not produce accurate measures for target behaviors that occur for extended time periods, such as staying on task, listening, playing quietly alone, being out of one’s seat, or thumb sucking. Task-oriented or continuous behaviors (e.g., being “on task”) are examples of target behaviors for which event recording would not be indicated. Classes of continuous behaviors occurring across time are usually not a prime concern of applied behavior analysts. For example, read- ing per se is of less concern than the number of words read correctly and incorrectly, or the number of reading comprehension questions answered correctly and incor- rectly. Similarly, behaviors that demonstrate understand- ing are more important to measure than “listening behavior,” and the number of academic responses a stu- dent emits during an independent seatwork period is more important than being on task.

Timing

Researchers and practitioners use a variety of timing de- vices and procedures to measure duration, response la- tency, and interresponse time.

Timing the Duration of Behavior

Applied researchers often use semi-automated computer- driven systems for recording durations. Practitioners, however, will most likely use nonautomated instruments for recording duration. The most precise nonautomated instrument is a digital stopwatch. Practitioners can use wall clocks or wristwatches to measure duration, but the measures obtained will be less precise than those obtained with a stopwatch.

The procedure for recording the total duration of a target behavior per session with a stopwatch is to (a) ac- tivate the stopwatch as the behavior starts and (b) stop the watch at the end of the episode. Then, without reset- ting the stopwatch, the observer starts the stopwatch again at the beginning of the second occurrence of the behav- ior and stops the watch at the end of the second episode.

109

Measuring Behavior

The observer continues to accumulate the durations of time in this fashion until the end of the observation pe- riod, and then transfers the total duration of time show- ing on the stopwatch to a data sheet.

Gast and colleagues (2000) used the following pro- cedure for measuring duration with a foot switch and tape recorder that enables a practitioner to use both hands to present stimulus materials or interact with the participant in other ways throughout the session.

The duration of a student’s target behavior was recorded by using a cassette audio tape recorder with a blank tape in it that was activated by the instructor using her foot to activate a jellybean switch connected to the tape recorder. When the student engaged in the target behav- ior, the teacher activated the switch that started the tape running. When the student stopped engaging in the be- havior, the teacher stopped the tape by activating the switch again. At the end of the session the teacher said, “end” and stopped the tape recorder and rewound the tape. [After the session the teacher] then played the tape from the beginning until the “end” and timed the dura- tion with a stopwatch. (p. 398)

McEntee and Saunders (1997) recorded duration per occurrence using a bar code data collection system to measure (a) functional and stereotypic engagements with materials and (b) stereotypy without interaction with ma- terials and other aberrant behaviors. They created and arranged bar codes on a data sheet to record the behav- iors of four adolescents with severe to profound mental retardation. A computer with a bar code font and soft- ware read the bar code, recording the year, month, day, and time of particular events. The bar code data collec- tion system provided real-time duration measurements of engagements with materials and aberrant behaviors.

Timing Response Latency and Interresponse Time

Procedures for measuring latency and interresponse time (IRT) are similar to procedures used to measure duration. Measuring latency requires the precise detection and recording of the time that elapses from the onset of each occurrence of the antecedent stimulus event of interest and the onset of the target behavior. Measuring IRTs re- quires recording the precise time that elapses from the termination of each occurrence of the target behavior to the onset of the next. Wehby and Hollahan (2000) mea- sured the latency of compliance by an elementary school student with learning disabilities to instructions to begin math assignments. Using a laptop computer with soft- ware designed to detect latency (see Tapp, Wehby, & Ellis, 1995, for the MOOSES observation system), they mea- sured latency from the request to the onset of compliance.

9A variety of terms are used in the applied behavior analysis literature to describe measurement procedures that involve observing and recording behavior within or at the end of scheduled intervals. Some authors use the term time sampling to refer only to momentary time sampling. We include whole-interval and partial-interval recording under the rubric of time sam- pling because it is often conducted as a discontinuous measurement method to provide a representative “sample” of a person’s behavior during the ob- servation period interval.

Time Sampling

Time sampling refers to a variety of methods for ob- serving and recording behavior during intervals or at spe- cific moments in time. The basic procedure involves dividing the observation period into time intervals and then recording the presence or absence of behavior within or at the end of each interval.

Time sampling was developed originally by etholo- gists studying the behavior of animals in the field (Charlesworth & Spiker, 1975). Because it was not pos- sible or feasible to observe the animals continuously, these scientists arranged systematic schedules of relatively brief but frequent observation intervals. The measures obtained from these “samples” are considered to be representative of the behavior during the entire time period from which they were collected. For example, much of our knowl- edge about the behavior of gorillas is based on data col- lected by researchers such as Jane Goodall using time sampling observation methods.

Three forms of time sampling used often by applied behavior analysts are whole-interval recording, partial- interval recording, and momentary time sampling.9

Whole-Interval Recording

Whole-interval recording is often used to measure contin- uous behaviors (e.g., cooperative play) or behaviors that occur at such high rates that observers have difficulty dis- tinguishing one response from another (e.g., rocking, hum- ming) but can detect whether the behavior is occurring at any given time.With whole-interval recording, the observation period is divided into a series of brief time intervals (typi- cally from 5 to 10 seconds).At the end of each interval, the observer records whether the target behavior occurred throughout the interval. If a student’s on-task behavior is being measured via 10-second whole-interval recording, the student would need to meet the definition of being on task during an entire interval for the behavior to be recorded as occurring in that interval. The student who was on task for 9 of an interval’s 10 seconds would be scored as not being on task for that interval. Hence, data obtained with whole-interval recording usually underestimate the overall percentage of the observation period in which the behavior actually occurred.The longer the observation intervals, the greater the degree to which whole-interval recording will underestimate the actual occurrence of the behavior.

110

Measuring Behavior

Date: Group no: Session no:May 7 1 17 Observer: Robin

Experimental condition:

Obs. start time: 9:42

IOA session:

Baseline

Stop time: 10:12

Yes

On task

Productivity

10-sec intervals Student 1 Student 2 Student 3 Student 4 1 YES NO YES NO YES NO YES NO 2 YES NO YES NO YES NO YES NO 3 YES NO YES NO YES NO YES NO 4 YES NO YES NO YES NO YES NO 5 YES NO YES NO YES NO YES NO 6 YES NO YES NO YES NO YES NO 7 YES NO YES NO YES NO YES NO 8 YES NO YES NO YES NO YES NO 9 YES NO YES NO YES NO YES NO

10 YES NO YES NO YES NO YES NO 11 YES NO YES NO YES NO YES NO 12 YES NO YES NO YES NO YES NO 13 YES NO YES NO YES NO YES NO 14 YES NO YES NO YES NO YES NO 15 YES NO YES NO YES NO YES NO 16 YES NO YES NO YES NO YES NO 17 YES NO YES NO YES NO YES NO 18 YES NO YES NO YES NO YES NO 19 YES NO YES NO YES NO YES NO 20 YES NO YES NO YES NO YES NO 21 YES NO YES NO YES NO YES NO 22 YES NO YES NO YES NO YES NO 23 YES NO YES NO YES NO YES NO 24 YES NO YES NO YES NO YES NO 25 YES NO YES NO YES NO YES NO 26 YES NO YES NO YES NO YES NO 27 YES NO YES NO YES NO YES NO 28 YES NO YES NO YES NO YES NO 29 YES NO YES NO YES NO YES NO 30 YES NO YES NO YES NO YES NO

26 4 28 2 18 12 22 8

86.6% 93.3% 60.0% 73.3%

Totals % Intervals

on task

On-Task Recording Form

Yes = On-task No = Off-task

Figure 6 Observation form used for whole-interval recording of four students being on task during independent seatwork time. Adapted from Smiley faces and spinners: Effects of self- monitoring of productivity with an indiscriminable contingency of reinforcement on the on-task behavior and academic productivity by kindergarteners during independent seatwork by R. L. Ludwig, 2004, p. 101. Unpublished master’s thesis, The Ohio State University. Used by permission.

Data collected with whole-interval recording are re- ported as the percentage of total intervals in which the target behavior was recorded as occurring. Because they represent the proportion of the entire observation period that the person was engaged in the target behavior, whole- interval recording data yield an estimate of total dura- tion. For example, assume a whole-interval observation period consisting of six 10-second intervals (a one-minute time frame). If the target behavior was recorded as oc- curring for four of these whole intervals and not occur- ring for the remaining two intervals, it would yield a total duration estimate of 40 seconds.

Figure 6 shows an example of a whole-interval recording form used to measure the on-task behavior of four students during academic seatwork time (Ludwig, 2004). Each minute was divided into four 10-second ob- servation intervals; each observation interval was followed by 5 seconds in which the observer recorded the occur- rence or nonoccurrence of target behavior during the pre- ceding 10 seconds. The observer first watched Student 1 continuously for 10 seconds, and then she looked away

during the next 5 seconds and recorded whether Student 1 had been on task throughout the previous 10 seconds by circling YES or NO on the recording form. After the 5- second interval for recording Student 1’s behavior, the observer looked up and watched Student 2 continuously for 10 seconds, after which she recorded Student 2’s be- havior on the form. The same procedure for observing and recording was used for Students 3 and 4. In this way, the on-task behavior of each student was observed and recorded for one 10-second interval per minute.

Continuing the sequence of observation and record- ing intervals over a 30-minute observation period pro- vided thirty 10-second measures (i.e., samples) of each student’s on-task behavior. The data in Figure 6 show that the four students were judged by the observer to have been on task during Session 17 for 87%, 93%, 60%, and 73% of the intervals, respectively. Although the data are intended to represent the level of each student’s behavior throughout the observation period, it is important to re- member that each student was observed for a total of only 5 of the observation period’s 30 minutes.

111

Measuring Behavior

1 2 3 4

AStudent 1 T S D N A T S D N A T S D N A T S D N

AStudent 2 T S D N A T S D N A T S D N A T S D N

AStudent 3 T S D N A T S D N A T S D N A T S D N

Key: A = Academic response T = Talk-out S = Out-of-seat D = Other disruptive behavior N = No occurrences of target behaviors

Figure 7 Portion of a form used for partial- interval recording of four response classes by three students.

Observers using any form of time sampling should always make a recording response of some sort in every interval. For example, an observer using a form such as the one in Figure 6 would record the occurrence or nonoccurrence of the target behavior in each interval by circling YES or NO. Leaving unmarked intervals in- creases the likelihood of losing one’s place on the record- ing form and marking the result of an observation in the wrong interval space.

All time sampling methods require a timing device to signal the beginning and end of each observation and recording interval. Observers using pencil, paper, clip- board, and timers for interval measurement will often at- tach a stopwatch to a clipboard. However, observing and recording behavior while having to look simultaneously at a stopwatch is likely to have a negative impact on the accuracy of measurement. An effective solution to this problem is for the observer to listen by earphone to prerecorded audio cues signaling the observation and recording intervals. For example, observers using a whole-interval recording procedure such as the one just described could listen to an audio recording with a se- quence of prerecorded statements such as the following: “Observe Student 1,” 10 seconds later, “Record Stu- dent 1,” 5 seconds later, “Observe Student 2,” 10 seconds later, “Record Student 2,” and so on.

Tactile prompting devices can also be used to signal observation intervals. For example, the Gentle Re- minder (dan@gentlereminder.com) and MotivAider (www.habitchange.com) are small timing instruments that vibrate at the time intervals programmed by the user.

Partial-Interval Recording

When using partial-interval recording, the observer records whether the behavior occurred at any time during the interval. Partial-interval time sampling is not con- cerned with how many times the behavior occurred dur-

ing the interval or how long the behavior was present, just that it occurred at some point during the interval. If the target behavior occurs multiple times during the in- terval, it is still scored as occurring only once. An ob- server using partial-interval recording to measure a student’s disruptive behavior would mark an interval if disruptive behavior of any form meeting the target be- havior definition occurred for any amount of time dur- ing the interval. That is, an interval would be scored as disruptive behavior even if the student was disruptive for only 1 second of a 6-second interval. Because of this, data obtained via partial-interval recording often overes- timate the overall percentage of the observation period (i.e., total duration) that the behavior actually occurred.

Partial-interval data, like whole-interval data, are most often reported as a percentage of total intervals in which the target behavior was scored. Partial-interval data are used to represent the proportion of the entire obser- vation period in which the target behavior occurred, but unlike whole-interval recording, the results of partial- interval recording do not provide any information on du- ration per occurrence. That is because any instance of the target behavior, regardless of how brief its duration, will cause an interval to be scored.

If partial-interval recording with brief observation intervals is used to measure discrete responses of short duration per occurrence, the data obtained provide a crude estimate of the minimum rate of responding. For exam- ple, data showing that a behavior measured by partial- interval recording consisting of 6-second contiguous intervals (i.e., successive intervals are not separated by time in which the behavior is not observed) occurred in 50% of the total intervals indicates a minimum response rate of five responses per minute (on the average, at least one response occurred in 5 of the 10 intervals per minute). Although partial-interval recording often overestimates the total duration, it is likely to underestimate the rate of a high-frequency behavior. This is because an interval in which a person made eight nonverbal sounds would be

112

Measuring Behavior

scored the same as an interval in which the person made only one sound. When the evaluation and understanding of a target behavior requires an accurate and sensitive measure of response rate, event recording should be used.

Because an observer using partial-interval recording needs to record only that a behavior has occurred at any point during each interval (compared to having to watch the behavior throughout the entire interval with whole- interval), it is possible to measure multiple behaviors con- currently. Figure 7 shows a portion of a form for measuring four response classes by three students using partial-interval recording with 20-second intervals. The observer watches Student 1 throughout the first 20-second interval, Student 2 for the next 20 seconds, and Student 3 for the next 20 seconds. Each student is observed for 20 seconds out of each minute of the observation period. If a student engages in any of the behaviors being measured at any time during an observation interval, the observer marks the letter(s) corresponding to those behaviors. If a student engages in none of the behaviors being measured during an interval, the observer marks N to indicate no occurrences of the target behaviors. For example, during the first interval in which Student 1 was observed, he said, “Pacific Ocean” (an academic response). During the first interval in which Student 2 was observed, she left her seat and threw a pencil (a behavior within the re- sponse class “other disruptive behavior”). Student 3 emit- ted none of the four target behaviors during the first interval that he was observed.

Momentary Time Sampling

An observer using momentary time sampling records whether the target behavior is occurring at the moment that each time interval ends. If conducting momentary time sampling with 1-minute intervals, an observer would look at the person at the 1-minute mark of the observa- tion period, determine immediately whether the target behavior was occurring, and indicate that decision on the recording form. One minute later (i.e., 2 minutes into the observation period), the observer would look again at the person and then score the presence or absence of the target behavior. This procedure would continue until the end of the observation period.

As with interval recording methods, data from mo- mentary time sampling are typically reported as per- centages of the total intervals in which the behavior occurred and are used to estimate the proportion of the total observation period that the behavior occurred.

A major advantage of momentary time sampling is that the observer does not have to attend continuosly to measurement, whereas interval recording methods de- mand the undivided attention of the observer.

Because the person is observed for only a brief mo- ment, much behavior will be missed with momentary time sampling. Momentary time sampling is used pri- marily to measure continuous activity behaviors such as engagement with a task or activity, because such behav- iors are easily identified. Momentary time sampling is not recommended for measuring low-frequency, short- duration behaviors (Saudargas & Zanolli, 1990).

A number of studies have compared measures ob- tained by momentary time sampling using intervals of varying duration with measures of the same behavior obtained by continuous duration recording (e.g., Gunter, Venn, Patrick, Miller, & Kelly, 2003; Powell, Martin- dale, Kulp, Martindale, & Bauman, 1977; Powell, Mar- tindale, & Kulp, 1975; Simpson & Simpson, 1977; Saudargas and Zanolli 1990; Test & Heward, 1984). In general, this research has found that momentary time sampling both overestimates and underestimates the con- tinuous duration measure when time intervals are greater than 2 minutes. With intervals less than 2 minutes, the data obtained using momentary time sampling more closely matched that obtained using the continuous du- ration measure.

Results of a study by Gunter and colleagues (2003) are representative of this research. These researchers com- pared measures of on-task behavior of three elementary students with emotional/behavioral disorders over seven sessions obtained by momentary time sampling conducted at 2-, 4-, and 6-minute intervals with measures of the same behavior obtained by continuous duration recording. Mea- sures obtained using the 4-minute and 6-minute time sam- pling method produced data paths that were highly discrepant with the data obtained by continuous mea- surement, but data obtained using the 2-minute interval had a high degree of correspondence with the measures obtained using the continuous duration method. Figure 8 shows the results for one of the students.

Planned Activity Check

A variation of momentary time sampling, planned ac- tivity check (PLACHECK) uses head counts to mea- sure “group behavior.” A teacher using PLACHECK observes a group of students at the end of each time in- terval, counts the number of students engaged in the tar- geted activity, and records the tally with the total number of students in the group. For example, Doke and Risley (1972) used data obtained by PLACHECK measurement to compare group participation in required and optional before-school activities. Observers tallied the number of students in either the required or the optional activity area at the end of 3-minute intervals, and then the number of children actually participating in an activity in either area.

113

Measuring Behavior

100

P er

ce n

t o

f T

im e/

In te

rv al

1 2 3 4 5 6 7 Sessions

Harry

MTS 2

MTS 4

MTS 6

Continuous

Figure 8 Comparison of measures of the on-task behavior of an elemen- tary student obtained by 2-minute, 4-minute, and 6-minute momentary time sampling with measures of the same behavior obtained by continuous duration recording. From “Efficacy of using momentary time samples to determine on-task behavior of students with emotional/behavioral disorders” by P. L. Gunter, M. L. Venn, J. Patrick, K. A. Miller, and L. Kelly, 2003, Education and Treatment of Children, 26, p. 406. Used by permission.

1 2 3 4 5 6 8 9 107

− − + + − − + − −−

− + + + − + + + +−

− + + + − − + − +−

Duration

MTS

55%

30%

70%

50%

Consecutive Observed Intervals

M ea

su re

m en

t M

et h

o d

+ −

= Actual occurrence of behavior measured by continuous duration recording

= Behavior recorded as occurring during the interval = Behavior recorded as not occurring during the interval

WI = whole-interval recording PI = partial-interval recording MTS = momentary time sampling

Figure 9 Comparing measures of the same behavior obtained by three different time sampling methods with measure obtained by continuous duration recording.

They reported these data as separate percentages of chil- dren participating in required or optional activities.

Dyer, Schwartz, and Luce (1984) used a variation of PLACHECK to measure the percentage of students with disabilities living in a residential facility engaged in age- appropriate and functional activities. As students entered the observation area, they were observed individually for as long as it took to determine the activity in which they were engaged. The students were observed in a prede- termined order not exceeding 10 seconds per student.

Other variations of the PLACHECK measurement can be found in the literature, though they usually are called time sampling or momentary time sampling. For example, in a study examining the effects of response cards on the disruptive behavior of third-graders during daily math lessons, Armendariz and Umbreit (1999) recorded at 1-minute intervals whether each student in the class was being disruptive. By combining the PLACHECK data obtained across all of the no-response

cards (baseline) sessions, and graphing those results as the percentage of students who were disruptive at each 1- minute mark, and doing the same with the PLACHEK data from all response cards sessions. Armendariz and Umbreit created a clear and powerful picture of the dif- ferences in “group behavior” from the beginning to the end of a typical lesson in which response cards were used or not used.

Recognizing Artifactual Variability in Time Sampling Measures

As stated previously, all time sampling methods provide only an estimate of the actual occurrence of the behavior. Different measurement sampling procedures produce dif- ferent results, which can influence decisions and inter- pretations. Figure 9 illustrates just how different the results obtained by measuring the same behavior with

114

Measuring Behavior

different time sampling methods can be. The shaded bars indicate when the behavior was occurring within an ob- servation period divided into 10 contiguous intervals. The shaded bars reveal all three dimensional quantities of be- havior: repeatability (seven instances of the behavior), temporal extent (the duration of each response) and tem- poral locus (interresponse time is depicted by the space between the shaded bars).

Because the time sampling methods used in applied behavior analysis are most often viewed and interpreted as measures of the proportion of the total observation pe- riod in which the behavior occurred, it is important to compare the results of time sampling methods with those obtained by continuous measurement of duration. Con- tinuous measurement reveals that the behavior depicted in Figure 9 occurred 55% of the time during the ob- servation period. When the same behavior during the same observation period was recorded using whole in- terval recordings the measure obtained grossly underes- timated the actual occurrence of the behavior (i.e., 30% versus 55%), partial-interval recording grossly overesti- mated the actual occurrence (i.e., 70% versus 55%), and momentary time sampling yielded a fairly close estimate of actual occurrence of the behavior (50% versus 55%).

The fact that momentary time sampling resulted in a measure that most closely approximated the actual be- havior does not mean that it is always the preferred method. Different distributions of the behavior (i.e., tem- poral locus) during the observation period, even at the same overall frequency and duration as the session shown in Figure 9, would result in widely different outcomes from each of the three time sampling methods.

Discrepancies between measures obtained by differ- ent measurement methods are usually described in terms of the relative accuracy or inaccuracy of each method. However, accuracy is not the issue here. If the shaded bars in Figure 9 represent the true value of the behav- ior, then each of the time sampling methods was con- ducted with complete accuracy and the resulting data are what should be obtained by applying each method. An example of the inaccurate use of one of the measurement methods would be if the observer using whole-interval recording had marked the behavior as occurring in In- terval 2, when the behavior did not occur according to the rules of whole-interval recording.

But if the behavior actually occurred for 55% of the observation period, what should we call the wrong and misleading measures of 30% and 70% if not inaccurate? In this case, the misleading data are artifacts of the mea- surement procedures used to obtain them. An artifact is something that appears to exist because of the way it is examined or measured. The 30% measure obtained by whole-interval recording and the 70% measure obtained

measures were conducted. The fact that data obtained from whole-interval and partial-interval recording con- sistently underestimate and overestimate, respectively, actual occurrence of behavior as measured by continuous duration recording is an example of well-known artifacts.

It is clear that interval measurement and momentary time sampling result in some artifactual variability in the data, which must be considered carefully when interpre- tating results obtained with these measurement methods.

Measuring Behavior by Permanent Products Behavior can be measured in real time by observing a person’s actions and recording responses of interest as they occur. For example, a teacher can keep a tally of the number of times a student raises her hand during a class discussion. Some behaviors can be measured in real time by recording their effects on the environment as those ef- fects are produced. For example, a hitting instructor ad- vances a handheld counter each time a batter hits a pitch to the right-field side of second base.

Some behaviors can be measured after they have oc- curred. A behavior that produces consistent effects on the environment can be measured after it has occurred if the effects, or products left behind by the behavior, re- main unaltered until measurement is conducted. For ex- ample, if the flight of the baseballs hit by a batter during batting practice were not impeded, and the balls were left lying on the ground, a hitting instructor could collect data on the batter’s performance after the batter’s turn was completed by counting each ball found lying in fair ter- ritory on the right-field side of second base.

Measuring behavior after it has occurred by measur- ing the effects that the behavior produced on the environ- ment is known as measurement by permanent product. Measuring permanent products is an ex post facto method of data collection because measurement takes place after the behavior has occurred. A permanent product is a change in the environment produced by a behavior that lasts long enough for measurement to take place.

Although often described erroneously as a method for measuring behavior, measurement by permanent product does not refer to any particular measurement procedure or method. Instead, measurement by perma- nent product refers to the time of measurement (i.e., after the behavior has occurred) and the medium (i.e., the ef- fect of the behavior, not the behavior itself) by which the measurer comes in contact with (i.e., observes) the be- havior. All of the methods for measuring behavior de- scribed in this chapter—event recording, timing, and time

by partial-interval recording are artifacts of the way those

115

Measuring Behavior

sampling—can be applied to the measurement of per- manent products.

Permanent products can be natural or contrived out- comes of the behavior. Permanent products are natural and important outcomes of a wide range of socially sig- nificant behaviors in educational, vocational, domestic, and community environments. Examples in education in- clude compositions written (Dorow & Boyle, 1998), com- putations of math problems written (Skinner, Fletcher, Wildmon, & Belfiore, 1996), spelling words written (McGuffin, Martz, & Heron, 1997), worksheets com- pleted (Alber, Heward, & Hippler, 1999), homework as- signments turned in (Alber, Nelson, & Brennan, 2002), and test questions answered (e.g., Gardner, Heward, & Grossi, 1994). Behaviors such as mopping floors and dishwashing (Grossi & Heward, 1998), incontinence (Ad- kins & Matthews, 1997), drawing bathroom graffiti (Mueller, Moore, Doggett, & Tingstrom, 2000), recycling (Brothers, Krantz, & McClannahan, 1994), and picking up litter (Powers, Osborne, & Anderson, 1973) can also be measured by the natural and important changes they make on the environment.

Many socially significant behaviors have no direct effects on the physical environment. Reading orally, sit- ting with good posture, and repetitive hand flapping leave no natural products in typical environments. Neverthe- less, ex post facto measurement of such behaviors can often be accomplished via contrived permanent products. For example, by audiotaping students as they read out loud (Eckert, Ardoin, Daly, & Martens, 2002), videotap- ing a girl sitting in class (Schwarz & Hawkins, 1970), and videotaping a boy flapping his hands (Ahearn, Clark, Gar- denier, Chung, & Dube, 2003) researchers obtained con- trived permanent products for measuring these behaviors.

Contrived permanent products are sometimes useful in measuring behaviors that have natural permanent prod- ucts that are only temporary. For example, Goetz and Baer (1973) measured variations in the form of children’s block building from photographs taken of the construc- tions made by the children, and Twohig and Woods (2001) measured the length of fingernails from pho- tographs of nail-biters’ hands.

Advantages of Measurement by Permanent Product

Measurement by permanent product offers numerous ad- vantages to practitioners and researchers.

Practitioner Is Free to Do Other Tasks

Not having to observe and record behavior as it occurs en- ables the practitioner to do something else during the ob- servation period. For example, a teacher who uses an

audiotape recorder to record students’ questions, com- ments, and talk-outs during a class discussion can con- centrate on what her students are saying, provide individual help, and so on.

Makes Possible Measurement of Some Behaviors That Occur at Inconvenient or Inaccessible Times and Places

Many socially significant behaviors occur at times and places that are inconvenient or inaccessible to the re- searcher or practitioner. Measuring permanent products is indicated when observing the target behavior as it oc- curs would be difficult because the behavior occurs in- frequently, in various environments, or for extended periods of time. For example, a music teacher can have his guitar student make audio recordings of portions of his daily practice sessions at home.

Measurement May Be More Accurate, Complete, and Continuous

Although measuring behavior as it occurs provides the most immediate access to the data, it does not necessar- ily yield the most accurate, complete, and representative data. An observer measuring behavior from permanent products can take his time, rescore the worksheet, or lis- ten to and watch the videotape again. Videotape enables the observer to slow down, pause, and repeat portions of the session—to literally “hold still” the behavior so it can be examined and measured again and again if necessary. The observer might see or hear additional nuances and as- pects of the behavior, or other behaviors that she over- looked or missed altogether during live performances.

Measurement by permanent product enables data collection on more participants. An observer can look at a videotape once and measure one participant’s behav- ior, then replay the tape and measure a second partici- pant’s behavior.

Video- or audiotaping behavior provides data on the occurrence of all instances of a target behavior (Mil- tenberger, Rapp, & Long, 1999; Tapp & Walden, 2000). Having this permanent product of all instances of the be- havior lends itself to later scoring by using the built-in calibrated digital timer (e.g., on a VCR), set to zero sec- onds (or the first frame) at the beginning of the session, to note the exact time of behavior onset and offset. Fur- ther, software programs facilitate data collection and analysis based on exact timings. PROCORDER is a soft- ware system for facilitating the collection and analysis of videotaped behavior. According to Miltenberger and colleagues, “With a recording of the exact time of the onset and offset of the target behavior in the observation

116

Measuring Behavior

Generic Positive (transcribe first instance)

Excellent

Time index

0:17

Time index of repeats

1:37

4:00

7:45

10:22

3:36

4:15

9:11

10:34

Specific Positive (transcribe first instance)

Time index

Time index of repeats

2:10 3:28

I like how you helped her. Thank you for not talking.

1:05

1:57

Negative (transcribe first instance)

Time index

Time index of repeats

Beautiful 0:26

1:44

9:01

13:09

1:59

11:52

6:53

9:56

4:33

8:21Good—nice big word. You raised your hand, beautiful job.

2:45

3:37

Good 0:56

1:22

5:27

6:16

8:44

4:42

5:47

6:38

9:25

Good job, that's a new word. Thank you for paying attention.

3:46

4:56

Smart 5:14

7:06 11:59 Thank you for not writing.

7:50

High five 8:00

Transcribe positive and negative statements, corresponding time indexes, and time indexes of repeated statements in the boxes below.

Participant:

Observer:

Susan

Date of session:

Date of observation

4/23

Exp. condition:

Duration of observation:

Self-scoring generalization

15:00 Decimal: 15.0

Count: 28 % repeats: 83% Count: 13 % repeats 46% 0 % repeats:

No. per min. 1.9 All Positives: 41Count:

No. per min. 0.9 No. per min. 71%% repeats2.7

Figure 10 Data collection form for recording count and temporal locus of three classes of teacher statements from videotaped sessions.

From The effects of self-scoring on teachers’ positive statements during classroom instruction by S. M. Silvestri (2004), p. 124. Unpublished doctoral dissertation, The Ohio State University. Used by permission.

10Edwards and Christophersen (1993) described a time-lapse videotape recorder (TLVCR) that automatically records on one 2-hour tape time sam- ples of behavior over observation periods ranging from 2 to 400 hours. A TLVCR programmed to record over a 12-hour period would record for 0.10 of each second. Such a system can be useful for recoding very low fre- quency behaviors and for behaviors that occur over long periods of time (e.g., a child’s sleep behavior).

session, we are able to report the frequency (or rate) or the duration of the behavior” (p. 119).10

Facilitates Data Collection for Interobserver Agreement and Treatment Integrity

Video- or audiotapes assist with data collection tasks such as obtaining interobserver agreement data and assessing treatment integrity. Permanent products of behaviors make possible the repeated measurement of behaviors, eliminating the need to bring multiple observers into the research or treatment setting.

Enables Measurement of Complex Behaviors and Multiple Response Classes

Permanent products, especially videotaped records of be- havior, afford the opportunity to measure complex be- haviors and multiple response classes in busy social environments. Schwarz and Hawkins (1970) obtained measures of the posture, voice volume, and face touch- ing of an elementary student from videotapes taken dur- ing two class periods. The three behaviors were targeted as the result of operationalizing the girl’s “poor self- esteem.” The researchers were able to view the tapes re- peatedly and score them for different behaviors. In this study, the girl also watched and evaluated her behavior on the videotapes as part of the intervention.

Figure 10 shows an example of a recording form used by Silvestri (2004) to measure three types of state- ments by teachers—generic positive, specific positive,

117

Measuring Behavior

and negative—from audiotapes of classroom lessons. (See Figure 3.7 for definitions of these behaviors.) The movement, multiple voices, and general commotion that characterize any classrooms would combine to make it very difficult, if not impossible, for a live observer to con- sistently detect and accurately record these behaviors. Each teacher participant in the study wore a small wire- less microphone that transmitted a signal to a receiver that was connected to a cassette tape recorder.

Determining Whether Measurement by Permanent Product Is Appropriate

The advantages of measurement by permanent product are considerable, and it may seem as though permanent product measurement is always preferable to real-time measurement. Answering the following four questions will help practitioners and researchers determine whether measurement by permanent product is appropriate: Is real-time measurement needed? Can the behavior be mea- sured by permanent product? Will obtaining a contrived permanent product unduly affect the behavior? and How much will it cost?

Is Real-Time Measurement Needed?

Data-based decision making about treatment procedures and experimental conditions is a defining feature of ap- plied behavior analysis and one of its major strengths. Data-based decision making requires more than direct and frequent measurement of behavior; it also requires ongoing and timely access to the data those measures provide. Measuring behavior as it occurs provides the most immediate access to the data. Although real-time measurement by permanent product can be conducted in some situations (e.g., counting each batted ball that lands to the right of second base as a hitter takes batting prac- tice), measurement by permanent product will most often be conducted after the instructional or experimental ses- sion has ended.

Measures taken from video- or audiotapes cannot be obtained until the tapes are viewed after a session has ended. If treatment decisions are being made on a session- by-session basis, this behavior-to-measurement delay poses no problem as long as the data can be obtained from the tapes prior to the next session. However, when moment-to-moment treatment decisions must be made according to the participant’s behavior during the ses- sion, real-time measurement is necessary. Consider a be- havior analyst trying to reduce the rate of a person’s self-injurious behavior (SIB) by providing access to a preferred stimulus contingent on increasing durations of

time without SIB. Accurate implementation of the treat- ment protocol would require the real-time measurement of interresponse times (IRT).

Can the Behavior Be Measured by Permanent Product?

Not all behaviors are suitable for measurement by per- manent product. Some behaviors affect fairly permanent changes in the environment that are not reliable for the purposes of measurement. For example, self-injurious behavior (SIB) often produces long-lasting effects (bruises, welts, and even torn and bleeding skin) that could be measured after the occurrence of the behavior. But accurate measures of SIB could not be obtained by examining the client’s body on a regular basis. The pres- ence of discolored skin, abrasions, and other such marks would indicate that the person had been injured, but many important questions would be left unanswered. How many times did SIB occur? Were there acts of SIB that did not leave observable marks on the skin? Was each in- stance of tissue damage the result of SIB? These are im- portant questions for evaluating the effects of any treatment. Yet, they could not be answered with certainty because these permanent products are not precise enough for the measurement of the SIB. Behaviors suitable for measurement via permanent product must meet two rules.

Rule 1: Each occurrence of the target behavior must produce the same permanent product. The perma- nent product must be a result of every instance of the response class measured. All topographical variants of the target behavior and all responses of varying magni- tude that meet the definition of target behavior must produce the same permanent product. Measuring an employee’s work productivity by counting the number of correctly assembled widgets in his “completed work” bin conforms to this rule. An occurrence of the target behavior in this case is defined functionally as a correctly assembled widget. Measuring SIB by marks on the skin does not meet Rule 1 because some SIB re- sponses will not leave discernable marks.

Rule 2: The permanent product can only be pro- duced by the target behavior. This rule requires that the permanent product cannot result from (a) any behaviors by the participant other than the target behavior or (b) the behavior of any person other than the participant. Using the number of correctly assembled widgets in the employee’s “completed work” bin to measure his pro- ductivity conforms to Rule 2 if the observer can be as- sured that (a) the employee put no widgets in his bin that he did not assemble and (b) none of the assembled widgets in the employee’s bin were put there by anyone

118

Measuring Behavior

other than the employee. Using marks on the skin as a permanent product for measuring self-injurious behav- ior also fails to meet Rule 2. Marks on the skin could be produced by other behaviors by the person (e.g., run- ning too fast, which caused him to trip and hit his head, stepping in poison ivy) or by the behavior of other peo- ple (e.g., struck by another person).

Will Obtaining a Contrived Permanent Product Unduly Affect the Behavior?

Practitioners and researchers should always consider re- activity—the effects of the measurement procedure on the behavior being measured. Reactivity is most likely when observation and measurement procedures are ob- trusive. Obtrusive measurement alters the environment, which may in turn affect the behavior being measured. Permanent products obtained using recording equipment, the presence of which may cause a person to behave dif- ferently, are called contrived permanent products. For ex- ample, using an audiotape to record conversations might encourage the participant to talk less or more. But it should be recognized that reactivity to the presence of human observers is a common phenomenon, and reac- tive effects are usually temporary (e.g., Haynes & Horn, 1982; Kazdin, 1982, 2001). Even so, it is appropriate to anticipate the influence that equipment may have on the target behavior.

How Much Will It Cost to Obtain and Measure the Permanent Product?

The final issues to address in determining the appropri- ateness of measuring a target behavior by permanent product are availability, cost, and effort. If recording equipment is required to obtain a contrived product of the behavior, is it currently available? If not, how much will it cost to buy or rent the equipment? How much time will it take to learn to use the equipment initially? How difficult and time-consuming will it be to set up, store, and use the equipment for the duration of the study or behavior change program?

Computer-Assisted Measurement of Behavior Computer hardware and software systems for behavioral measurement and data analysis have become increasingly sophisticated and useful, especially for applied re- searchers. Developers have produced data collection and analysis software for observational measurement using laptops (Kahng & Iwata, 2000; Repp & Karsh, 1994),

11Descriptions of the characteristics and capabilities of a variety of com- puter-assisted behavioral measurement systems can be found in the fol- lowing sources: Emerson, Reever, & Felce (2000); Farrell (1991); Kahng & Iwata (1998, 2000); Repp, Harman, Felce, Vanacker, & Karsh (1989); Saunders, Saunders, & Saunders (1994); Tapp & Walden (2000); and Tapp and Wehby (2000).

handheld computers (Saudargas & Bunn, 1989), or per- sonal digital assistants (Emerson, Reever, & Felce, 2000). Some systems use bar-code scanners for data collection (e.g., McEntee & Saunders, 1997; Saunders, Saunders, & Saunders, 1994; Tapp & Wehby, 2000). Most computer software systems require a DOS or Windows operating system. A few software systems have been developed for the Mac OS. Regardless of the computer operating sys- tems, Kahng and Iwata (2000) stated:

These systems have the potential to facilitate the task of observation by improving the reliability and accuracy of recording relative to traditional but cumbersome meth- ods based on paper and pencil and to improve the effi- ciency of data calculation and graphing. (p. 35)

Developments in microchip technology have ad- vanced the measurement and data analysis capabilities of these systems, and have made the software increas- ingly easier to learn and apply. Many of these semiauto- mated systems can record a number of events, including discrete trials, the number of responses per unit of time, duration, latency, interresponse time (IRT), and fixed and variable intervals for time-sampling measurement. These systems can present data calculated as rate, duration, la- tency, IRT, percentage of intervals, percentage of trials, and conditional probabilities.

A distinct advantage of computer-based observation and measurement systems is that when aggregating rates, time-series analyses, conditional probabilities, sequen- tial dependencies, interrelationships, and combinations of events, these data can be clustered and analyzed. Be- cause these systems allow for the simultaneous recording of multiple behaviors across multiple dimensions, out- puts can be examined and analyzed from different per- spectives that would be difficult and time-consuming with paper-and-pencil methods.

In addition to recording behavior and calculating data, these systems provide analyses of interobserver agreement (e.g., smaller/larger, overall, occurrence, nonoccurrence) and measurement from audio and video documents. The semiautomated computer-driven sys- tems, as compared to the commonly used paper-and- pencil data recording and analysis methods, have the potential to improve interobserver agreements, the relia- bility of observational measurements, and the efficiency of data calculation (Kahng & Iwata, 1998).11

119

Measuring Behavior

The practical value derived from the greater effi- ciency and ease of use of computer-assisted measure- ment systems will likely increase their use by applied

Summary Definition and Functions of Measurement in Applied Behavior Analysis

1. Measurement is the process of applying quantitative labels to observed properties of events using a standard set of rules.

2. Measurement is how scientists operationalize empiricism.

3. Without measurement, all three levels of scientific knowl- edge—description, prediction, and control—would be rel- egated to guesswork and subjective opinions.

4. Applied behavior analysts measure behavior to obtain an- swers to questions about the existence and nature of func- tional relations between socially significant behavior and environmental variables.

5. Practitioners measure behavior before and after treatment to evaluate the overall effects of interventions (summa- tive evaluation) and frequent measures of behavior dur- ing treatment (formative assessment) to guide decisions concerning the continuation, modification, or termination of treatment.

6. Without frequent measures of the behavior targeted for in- tervention, practitioners may (a) continue an ineffective treatment when no real behavior change occurred, or (b) discontinue an effective treatment because subjective judg- ment detects no improvement.

7. Measurement also helps practitioners optimize their ef- fectiveness; verify the legitimacy of practices touted as “evidence based”; identify treatments based on pseudo- science, fad, fashion, or ideology; be accountable to clients, consumers, employers, and society; and achieve ethical standards.

Measureable Dimensions of Behavior

8. Because behavior occurs within and across time, it has three dimensional quantities: repeatability (i.e., count), temporal extent (i.e., duration), and temporal locus (i.e., when be- havior occurs). These properties, alone and in combination, provide the basic and derivative measures used by applied behavior analysts. (See Figure 5 for a detailed summary.)

9. Count is the number of responses emitted during an ob- servation period.

10. Rate, or frequency, is a ratio of count per observation pe- riod; it is often expressed as count per standard unit of time.

11. Celeration is a measure of the change (acceleration or de- celeration) in rate of responding per unit of time.

12. Duration is the amount of time in which behavior occurs.

13. Response latency is a measure of the elapsed time be- tween the onset of a stimulus and the initiation of a sub- sequent response.

14. Interresponse time (IRT) is the amount of time that elapses between two consecutive instances of a response class.

15. Percentage, a ratio formed by combining the same dimen- sional quantities, expresses the proportional quantity of an event in terms of the number of times the event occurred per 100 opportunities that the event could have occurred.

16. Trials-to-criterion is a measure of the number of response opportunities needed to achieve a predetermined level of performance.

17. Although form (i.e., topography) and intensity of re- sponding (i.e., magnitude) are not fundamental dimen- sional quantities of behavior, they are important quantitative parameters for defining and verifying the oc- currence of many response classes.

18. Topography refers to the physical form or shape of a be- havior.

19. Magnitude refers to the force or intensity with which a re- sponse is emitted.

Procedures for Measuring Behavior

20. Event recording encompasses a wide variety of procedures for detecting and recording the number of times a behav- ior of interest is observed.

21. A variety of timing devices and procedures are used to measure duration, response latency, and interresponse time.

22. Time sampling refers to a variety of methods for observ- ing and recording behavior during intervals or at specific moments in time.

23. Observers using whole-interval recording divide the ob- servation period into a series of equal time intervals. At the end of each interval, they record whether the target be- havior occurred throughout the entire interval.

24. Observers using partial-interval recording divide the ob- servation period into a series of equal time intervals. At the end of each interval, they record whether behavior occurred at any point during the interval.

25. Observers using momentary time sampling divide the ob- servation period into a series of time intervals. At the end of each interval, they record whether the target behavior is occurring at that specific moment.

researchers and practitioners who currently use me- chanical counters, timers, and paper and pencil for ob- servational recording.

120

Measuring Behavior

26. Planned activity check (PLACHECK) is a variation of mo- mentary time sampling in which the observer records whether each individual in a group is engaged in the tar- get behavior.

27. Measurement artifacts are common with time sampling.

Measuring Behavior by Permanent Products

28. Measuring behavior after it has occurred by measuring its effects on the environment is known as measurement by permanent product.

29. Measurement of many behaviors can be accomplished via contrived permanent products.

30. Measurement by permanent product offers numerous ad- vantages: The practitioner is free to do other tasks; it en- ables the measurement of behaviors that occur at inconvenient or inaccessible times and places; measure- ment may be more accurate, complete, and continuous; it facilitates the collection of interobserver agreement and treatment integrity data; and it enables the measurement of complex behaviors and multiple response classes.

31. If moment-to-moment treatment decisions must be made during the session, measurement by permanent product may not be warranted.

32. Behaviors suitable for measurement via permanent prod- ucts must meet two rules. Rule 1: Each occurrence of the target behavior must produce the same permanent product. Rule 2: The permanent product can only be produced by the target behavior.

Computer-Assisted Measurement of Behavior

33. Computer hardware and software systems for behavioral measurement and data analysis have become increasingly sophisticated and easier to use.

34. Developers have produced data collection and analysis software for observational measurement using laptops, handheld computers, personal digital assistants (PDAs), and desktop computers.

35. Some systems allow for the simultaneous recording of multiple behaviors across multiple dimensions. Outputs can be examined and analyzed from different perspectives that would be difficult and time-consuming with paper- and-pencil methods.

121

accuracy believability calibration continuous measurement direct measurement discontinuous measurement exact count-per-interval IOA indirect measurement interobserver agreement (IOA)

interval-by-interval IOA mean count-per-interval IOA mean duration-per-occurrence IOA measurement bias naive observer observed value observer drift observer reactivity reliability

scored-interval IOA total count IOA total duration IOA trial-by-trial IOA true value unscored-interval IOA validity

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List, © Third Edition

Content Area 6: Measurement of Behavior

6-4 Select the appropriate measurement procedure given the dimensions of the behavior and the logistics of observing and recording.

6-5 Select a schedule of observation and recording periods.

6-14 Use various methods of evaluating the outcomes of measurement procedures, such as interobserver agreement, accuracy, and reliability.

Improving and Assessing the Quality of Behavioral Measurement

Key Terms

122

The data obtained by measuring behavior are the primary material with which behavioral re- searchers and practitioners guide and evaluate

their work. Applied behavior analysts measure socially significant behaviors to help determine which behaviors need to be changed, to detect and compare the effects of various interventions on behaviors targeted for change, and to evaluate the acquisition, maintenance, and gener- alization of behavior changes.

Because so much of what the behavior analyst does either as a researcher or practitioner depends on mea- surement, concerns about the legitimacy of the data it produces must be paramount. Do the data meaningfully reflect the original reason(s) for measuring the behav- ior? Do the data represent the true extent of the behav- ior as it actually occurred? Do the data provide a consistent picture of the behavior? In other words, can the data be trusted?

This chapter focuses on improving and assessing the quality of behavioral measurement. We begin by defining the essential indicators of trustworthy measurement: va- lidity, accuracy, and reliability. Next, common threats to measurement are identified and suggestions for combat- ing these threats are presented. The chapter’s final sec- tions detail procedures for assessing the accuracy, reliability, and believability of behavioral measurement.

Indicators of Trustworthy Measurement

Three friends—John, Tim, and Bill—took a bicycle ride together. At the end of the ride John looked at his handlebar-mounted bike computer and said, “We rode 68 miles. Excellent!” “My computer shows 67.5 miles. Good ride, fellas!” Tim replied. As he dismounted and rubbed his backside, the third biker, Bill, said, “Gee whiz, I’m sore! We must’ve ridden 100 miles!” A few days later, the three friends completed the same route. After the second ride, John’s computer showed 68 miles, Tim’s computer read 70 miles, and Bill, because he wasn’t quite as sore as he was after the first ride, said they had ridden 90 miles. Following a third ride on the same country roads, John, Tim, and Bill reported dis- tances of 68, 65, and 80 miles, respectively.

How trustworthy were the measures reported by the three bicyclists? Which of the three friends’ data would be most usable for a scientific account of the miles they had ridden? To be most useful for science, measurement must be valid, accurate, and reliable. Were the three friends’ measurements characterized by validity, accu- racy, and reliability?

Validity

Measurement has validity when it yields data that are di- rectly relevant to the phenomenon measured and to the reason(s) for measuring it. Determining the validity of measurement revolves around this basic question: Was a relevant dimension of the behavior that is the focus of the investigation measured directly and legitimately?

Did the measurements of miles ridden by the three bicyclists have validity? Because the bikers wanted to know how far they had ridden each time, the number of miles ridden was a relevant, or valid, dimension of their riding behavior. Had the bikers’ primary interest been how long or how fast they had ridden, the number of miles ridden would not have been a valid measure. John and Tim’s use of their bike computers to measure di- rectly the miles they rode was a valid measure. Because Bill used an indirect measure (the relative tenderness of his backside) to determine the number of miles he had ridden, the validity of Bill’s mileage data is sus- pect. A direct measure of the actual behavior of interest will always possess more validity than an indirect mea- sure, because a direct measure does not require an in- ference about its relation to the behavior of interest, whereas an indirect measure always requires such an inference. Although soreness may be related to the dis- tance ridden, because it is also influenced by such fac- tors as the time on the bike saddle, the roughness of the road, riding speed, and how much (or little) the person has ridden recently, soreness as a measure of mileage has little validity.

Valid measurement in applied behavior analysis re- quires three equally important elements: (a) measuring directly a socially significant target behavior, (b) mea- suring a dimension (e.g., rate, duration) of the target be- havior relevant to the question or concern about the behavior, and (c) ensuring that the data are representa- tive of the behavior’s occurrence under conditions and during times that are most relevant to the question or con- cern about the behavior. When any of these elements are suspect or lacking—no matter how technically proficient (i.e., accurate and reliable) was the measurement that pro- duced the data—the validity of the resultant data are com- promised, perhaps to the point of being meaninglessness.

Accuracy

When used in the context of measurement, accuracy refers to the extent to which the observed value, the quantitative label produced by measuring an event, matches the true state, or true value, of the event as it ex- ists in nature (Johnston & Pennypacker, 1993a). In other words, measurement is accurate to the degree that it cor- responds to the true value of the thing measured. A true

Improving and Assessing the Quality of Behavioral Measurement

123

Improving and Assessing the Quality of Behavioral Measurement

value is a measure obtained by procedures that are independent of and different from the procedures that produced the data being evaluated and for which the re- searcher has taken “special or extraordinary precautions to insure that all possible sources of error have been avoided or removed” (p. 136).

How accurate were the three bikers’ measures of miles ridden? Because each biker obtained a different measure of the same event, all of their data could not be accurate. Skeptical of the training miles the three cyclists were claiming, a friend of theirs, Lee, drove the same country roads with a Department of Transportation odometer attached to the back bumper of his car. At the end of the route the odometer read 58 miles. Using the measure obtained by the DOT odometer as the true value of the route’s distance, Lee determined that none of the three cyclists’ measures were accurate. Each rider had overestimated the true mileage.

By comparing the mileage reported by John, Tim, and Bill with the true value of the route’s distance, Lee discovered not only that the riders’ data were inaccurate, but also that the data reported by all three riders were contaminated by a particular type of measurement error called measurement bias. Measurement bias refers to nonrandom measurement error; that is, error in mea- surement that is likely to be in one direction. When mea- surement error is random, it is just as likely to overestimate the true value of an event as it is to under- estimate it. Because John, Tim, and Bill consistently over- estimated the actual miles they had ridden, their data contained measurement bias.

Reliability

Reliability describes the extent to which a “measurement procedure yields the same value when brought into re- peated contact with the same state of nature” (Johnston & Pennypacker, 1993a, p. 138). In other words, reliable measurement is consistent measurement. Like validity and accuracy, reliability is a relative concept; it is a mat- ter of degree. The closer the values obtained by repeated measurement of the same event are to one another, the greater the reliability. Conversely, the more observed val- ues from repeated measurement of the same event differ from one another, the less the reliability.

How reliable were the bicyclists’ measurements? Be- cause John obtained the same value, 68 miles, each time he measured the same route, his measurement had com- plete reliability. Tim’s three measures of the same ride— 67.5, 70, and 65 miles—differed from one another by as much as 5 miles. Therefore, Tim’s measurement was less reliable than John’s. Bill’s measurement system was the least reliable of all, yielding values for the same route ranging from 80 to 100 miles.

Relative Importance of Validity, Accuracy, and Reliability

Behavioral measurement should provide legitimate data for evaluating behavior change and guiding research and treatment decisions. Data of the highest quality (i.e., data that are most useful and trustworthy for advancing sci- entific knowledge or for guiding data-based practice) are produced by measurement that is valid, accurate, and re- liable (see Figure 1). Validity, accuracy, and reliability are relative concepts; each can range from high to low.

Measurement must be both valid and accurate for the data to be trustworthy. If measurement is not valid, ac- curacy is moot. Accurately measuring a behavior that is not the focus of the investigation, accurately measuring an irrelevant dimension of the target behavior, or accu- rately measuring the behavior under circumstances or at times not representative of the conditions and times rel- evant to the analysis will yield invalid data. Conversely, the data obtained from measuring a meaningful dimen- sion of the right behavior under the relevant circum- stances and times is of little use if the observed values provide an inaccurate picture of the behavior. Inaccurate measurement renders invalid the data obtained by other- wise valid measurement.

Reliability should never be confused with accuracy. Although John’s bicycle computer provided totally reli- able measures, it was also totally inaccurate.

Concern about the reliability of data in the absence of a prior interest in their accuracy suggests that reliability is being mistaken for accuracy. The questions for a re- searcher or someone who is reading a published study is not, “Are the data reliable?” but “Are the data accurate?” (Johnston & Pennypacker, 1993a, p. 146)

If accuracy trumps reliability—and it does—why should researchers and practitioners be concerned with the reliability of measurement? Although high reliabil- ity does not mean high accuracy, poor reliability reveals problems with accuracy. Because Tim and Bill’s mea- surements were not reliable, we know that at least some of the data they reported could not be accurate, knowl- edge that could and should lead to checking the accuracy of their measurement tools and procedures.

Highly reliable measurement means that whatever degree of accuracy (or inaccuracy) exists in the mea- surement system will be revealed consistently in the data. If it can be determined that John’s computer reliably ob- tains observed values higher than the true values by a constant amount or proportion, the data could be adjusted to accommodate for that constant degree of inaccuracy.

The next two sections of the chapter describe meth- ods for combating common threats to the validity, accu- racy, and reliability of behavioral measurement.

124

Improving and Assessing the Quality of Behavioral Measurement

Figure 1 Measurement that is valid, accurate, and reliable yields the most trust- worthy and useful data for science and science-based practice.

Measurement that is . .

Valid Accurate Reliable . . . yields data that are . . .

Yes Yes Yes . . . most useful for advancing scientific knowledge and guiding data-based practice.

No Yes Yes . . . meaningless for the purposes for which measurement was conducted.

Yes No Yes . . . always wrong.1

Yes Yes No2 . . . sometimes wrong.3

1. If adjusted for consistent measurement error of standard size and direction, inaccurate data may still be usable.

2. If the accuracy of every datum in a data set can be confirmed, reliability is a moot point. In practice, however, that is seldom possible; therefore, knowing the consistency with which a valid and accurate measurement system has been applied contributes to the level of confidence in the overall trustworthiness of the data set.

3. User is unable to separate the good data from the bad.

Threats to Measurement Validity The validity of behavioral data is threatened when mea- surement is indirect, when the wrong dimension of the target behavior is measured, or when measurement is con- ducted in such a way that the data it produces are an ar- tifact of the actual events.

Indirect Measurement

Direct measurement occurs when “the phenomenon that is the focus of the experiment is exactly the same as the phenomenon being measured” (Johnston & Pennypacker, 1993a, p. 113). Conversely, indirect measurement oc- curs when “what is actually measured is in some way dif- ferent from” the target behavior of interest (Johnston & Pennypacker, 1993a, p. 113). Direct measurement of be- havior yields more valid data than will indirect measure- ment. This is because indirect measurement provides secondhand or “filtered” information (Komaki, 1998) that requires the researcher or practitioner to make inferences about the relationship between the event that was mea- sured and the actual behavior of interest.

Indirect measurement occurs when the researcher or practitioner measures a proxy, or stand-in, for the actual behavior of interest. An example of indirect measurement would be using children’s responses to a questionnaire as a measure of how often and well they get along with their classmates. It would be better to use a direct mea- sure of the number of positive and negative interactions among the children. Using a student’s score on a stan- dardized math achievement test as an indicator of her

1Strategies for increasing the accuracy of self-reports can be found in Critchfield, Tucker, and Vuchinich (1998) and Finney, Putnam, and Boyd (1998).

mastery of the math skills included in the school’s cur- riculum is another example of indirect measurement. Ac- cepting the student’s score on the achievement test as a valid reflection of her ability with the school’s curriculum would require an inference. By contrast, a student’s score on a properly constructed test consisting of math prob- lems from recently covered curriculum content is a di- rect measure requiring no inferences about what it means with respect to her performance in the curriculum.

Indirect measurement is usually not an issue in ap- plied behavior analysis because meeting the applied di- mension of ABA includes the targeting and meaningful (i.e., valid) measurement of socially significant behav- iors. Sometimes, however, the researcher or practitioner has no direct and reliable access to the behavior of inter- est and so must use some form of indirect measurement. For example, because researchers studying adherence to medical regimens cannot directly observe and measure patients’ behavior in their homes, they rely on self-reports for their data (e.g., La Greca & Schuman, 1995).1

Indirect measurement is sometimes used to make in- ferences about private events or affective states. For ex- ample, Green and Reid (1996) used direct measures of smiling to represent “happiness” by persons with pro- found multiple disabilities. However, research on private events does not necessarily involve indirect measurement. A research participant who has been trained to observe his own private events is measuring the behavior of interest directly (e.g., Kostewicz, Kubina, & Cooper, 2000; Ku- bina, Haertel, & Cooper, 1994).

125

Improving and Assessing the Quality of Behavioral Measurement

Whenever indirect measurement is used, it is the re- sponsibility of the researcher to provide evidence that the event measured directly reflects, in some reliable and meaningful way, something about the behavior for which the researcher wishes to draw conclusions (Johnston & Pennypacker, 1993a). In other words, it is incumbent upon the researcher to provide a convincing case for the validity of her data. Although it is sometimes attempted, the case for validity cannot be achieved by simply at- taching the name of the thing one claims to be measur- ing to the thing actually measured. With respect to that point, Marr (2003) recounted this anecdote about Abra- ham Lincoln:

“Sir, how many legs does this donkey have?” “Four, Mr. Lincoln.” “And how many tails does it have?” “One, Mr. Lincoln.” “Now, sir, what if we were to call a tail a leg; how many legs would the donkey have?” “Five, Mr. Lincoln.” “No sir, for you cannot make a tail into a leg by calling it one.” (pp. 66–67)

Measuring the Wrong Dimension of the Target Behavior

The validity of behavioral measurement is threatened much more often by measuring the wrong dimension of the behavior of interest than it is by indirect measure- ment. Valid measurement yields data that are relevant to the questions about the behavior one seeks to answer through measurement. Validity is compromised when measurement produces values for a dimension of the be- havior ill suited for, or irrelevant to, the reason for mea- suring the behavior.

Johnston and Pennypacker (1980) provided an ex- cellent example of the importance of measuring a di- mension that fits the reasons for measurement. “Sticking a ruler in a pot of water as the temperature is raised will yield highly reliable measures of the depth of the water but will tell us very little about the changing tempera- ture” (p. 192). While the units of measurement on a ruler are well suited for measuring length, or in this case, depth, they are not at all valid for measuring temperature. If the purpose of measuring the water is to determine whether it has reached the ideal temperature for making a pot of tea, a thermometer is the correct measurement tool.

If you are interested in measuring a student’s aca- demic endurance with oral reading, counting the number of correct and incorrect words read per minute without measuring and reporting the total time that the student read will not provide valid data on endurance. Number of words read per minute alone does not fit the reason for

measuring reading (i.e., academic endurance). To mea- sure endurance, the practitioner would need to report the duration of the reading period (e.g., 30 minutes). Simi- larly, measuring the percentage of trials on which a stu- dent makes a correct response will not provide valid data for answering questions about the student’s developing fluency with a skill, whereas measuring the number of correct responses per minute and the changing rates of responding (celeration) would.

Measurement Artifacts

Directly measuring a relevant dimension of a socially sig- nificant target behavior does not guarantee valid mea- surement. Validity is reduced when the data—no matter how accurate or reliable they are—do not give a mean- ingful (i.e., valid) representation of the behavior. When data give an unwarranted or misleading picture of the be- havior because of the way measurement was conducted, the data are called an artifact. A measurement artifact is something that appears to exist because of the way it is measured. Discontinuous measurement, poorly sched- uled measurement periods, and using insensitive or limiting measurement scales are common causes of mea- surement artifacts.

Discontinuous Measurement

Because behavior is a dynamic and continuous phenom- enon that occurs and changes over time, continuous mea- surement is the gold standard in behavioral research. Continuous measurement is measurement conducted in a manner such that all instances of the response class(es) of interest are detected during the observation period (Johnston & Pennypacker, 1993a). Discontinuous measurement describes any form of measurement in which some instances of the response class(es) of inter- est may not be detected. Discontinuous measurement— no matter how accurate and reliable—may yield data that are an artifact.

A study by Thomson, Holmber, and Baer (1974) pro- vides a good demonstration of the extent of artifactual variability in a data set that may be caused by discontin- uous measurement. A single, highly experienced observer used three different procedures for scheduling time sam- pling observations to measure the behavior of four sub- jects (two teachers and two children) in a preschool setting during 64-minute sessions. Thomson and colleagues called the three time sampling procedures contiguous, al- ternating, and sequential. With each time sampling pro- cedure, one-fourth of the observer’s time (i.e., 16 minutes) was assigned to each of the four subjects.

126

Improving and Assessing the Quality of Behavioral Measurement

When the contiguous observation scheduled was used, the observer recorded the behavior of Subject 1 throughout the first 16 minutes of the session, recorded the behavior Subject 2 during the second 16 minutes, and so on until all four students had been observed. In the al- ternating mode, Subjects 1 and 2 were observed in alter- nating intervals during the first half of the session, and Subjects 3 and 4 were observed in the same fashion dur- ing the last half of the session. Specifically, Student 1 was observed during the first 4 minutes, Subject 2 during the next 4 minutes, Subject 1 during the next 4 minutes, and so on until 32 minutes had expired. The same pro- cedure was then used for Students 3 and 4 during the last 32 minutes of the session. The sequential approach sys- tematically rotated the four subjects through 4-minute observations. Subject 1 was observed during the first 4 minutes, Subject 2 during the second 4 minutes, Subject 3 during the third 4 minutes, and Subject 4 during the fourth 4 minutes. This sequence was repeated four times to give the total of 64 minutes of observation.

To arrive at the percentage of artifactual variance in the data associated with each time sampling schedule, Thom- son and colleagues (1974) compared the observer’s data with “actual rates” for each subject produced by continu- ous measurement of each subject for the same 64-minute sessions. Results of the study showed clearly that the con- tiguous and alternating schedules produced the most un- representative (and therefore, less valid) measures of the target behaviors (often more than 50% variance from con- tinuous measurement), whereas sequential sampling pro- cedure produced results that more closely resembled the data obtained through continuous recording (from 4 to 11% variance from continuous measurement).

In spite of its inherent limitations, discontinuous measurement is used in many studies in applied behav- ior analysis in which individual observers measure the behavior of multiple subjects within the same session. Minimizing the threat to validity posed by discontinuous measurement requires careful consideration of when ob- servation and measurement periods should be scheduled. Infrequent measurement, no matter how accurate and re- liable it is, often yields results that are an artifact. Al- though a single measure reveals the presence or absence of the target behavior at a given point in time, it may not be representative of the typical value for the behavior.2 As a general rule, observations should be scheduled on a daily or frequent basis, even if for only brief periods.

Ideally, all occurrences of the behavior of interest should be recorded. However, when available resources

2Single measures, such as pretests and posttests, can provide valuable in- formation on a person’s knowledge and skills before and after instruction or treatment. The use of probes, occasional but systematic measures, to as- sess maintenance and generalization of behavior change is discussed in chapter entitled “Generalization and Maintenance of Behavior Change.”

preclude continuous measurement throughout an obser- vation period, the use of sampling procedures is neces- sary. A sampling procedure may be sufficient for decision making and analysis if the samples represent a valid ap- proximation of the true parameters of the behavior of interest. When measurement cannot be continuous throughout an observation period, it is generally prefer- able to sample the occurrence of the target behavior for numerous brief observation intervals that are evenly dis- tributed throughout the session than it is to use longer, less frequent intervals (Thomson et al., 1974; Thompson, Symons, & Felce, 2000). For example, measuring a sub- ject’s behavior in thirty 10-second intervals equally dis- tributed within a 30-minute session will likely yield more representative data than will observing the person for a single 5-minute period during the half hour.

Measuring behavior with observation intervals that are too short or too long may result in data that grossly over- or underestimate the true occurrence of behavior. For example, measuring off-task behavior by partial- interval recording with 10-minute intervals may produce data that make even the most diligent of students appear to be highly off task.

Poorly Scheduled Measurement Periods

The observation schedule should be standardized to pro- vide an equal opportunity for the occurrence or nonoc- currence of the behavior across sessions and consistent environmental conditions from one observation session to the next. When neither of these requirements is met, the resultant data may not be representative and may be in- valid. If observation periods are scheduled at times when and/or places where the frequency of behavior is atypi- cal, the data may not represent periods of high or low responding. For example, measuring students’ being on-task during only the first 5 minutes of each day’s 20-minute cooperative learning group activity may yield data that make on-task behavior appear higher than it ac- tually is over the entire activity.

When data will be used to assess the effects of an in- tervention or treatment, the most conservative observation times should be selected. That is, the target behavior should be measured during those times when their frequency of occurrence is most likely to be different from the desired or predicted outcomes of the treatment. Measurement of behaviors targeted for reduction should occur during times when those behaviors are most likely to occur at their high- est response rates. Conversely, behaviors targeted for increase should be measured when high-frequency re- sponding is least likely. If an intervention is not planned— as might be the case in a descriptive study—it is important to select the observation times most likely to yield data that are generally representative of the behavior.

127

Improving and Assessing the Quality of Behavioral Measurement

Insensitive and/or Limited Measurement Scales

Data that are artifacts may result from using measure- ment scales that cannot detect the full range of relevant values or that are insensitive to meaningful changes in behavior. Data obtained with a measurement scale that does not detect the full range of relevant performances may incorrectly imply that behavior cannot occur at lev- els below or above obtained measures because the scale has imposed an artificial floor or ceiling on performance. For example, measuring a student’s oral reading fluency by giving him a 100-word passage to read in 1 minute may yield data that suggest that his maximum perfor- mance is 100 wpm.

A measurement scale that is over- or undersensitive to relevant changes in behavior may produce data that show misleadingly that meaningful behavior change has (or has not) occurred. For example, using a percentage measure scaled in 10% increments to evaluate the effects of an intervention to improve quality control in a manu- facturing plant may not reveal important changes in per- formance if improvement in the percentage of correctly fabricated widgets from a baseline level of 92% to a range of 97 to 98% is the difference between unacceptable and acceptable (i.e., profitable) performance.

Threats to Measurement Accuracy and Reliability The biggest threat to the accuracy and reliability of data in applied behavior analysis is human error. Unlike the experimental analysis of behavior, in which measurement is typically automated and conducted by machines, most investigations in applied behavior analysis use human observers to measure behavior.3 Factors that contribute to human measurement error include poorly designed measurement systems, inadequate observer training, and expectations about what the data should look like.

Poorly Designed Measurement System

Unnecessarily cumbersome and difficult-to-use mea- surement systems create needless loss of accuracy and reliability. Collecting behavioral data in applied settings requires attention, keen judgment, and perseverance. The more taxing and difficult a measurement system is to use, the less likely an observer will be to consistently detect and record all instances of the target behavior. Simplify-

3We recommend using automatic data recording devices whenever possi- ble. For example, to measure the amount of exercise by boys on stationary bicycles, DeLuca and Holborn (1992) used magnetic counters that auto- matically recorded the number of wheel revolutions.

ing the measurement system as much as possible mini- mizes measurement errors.

The complexity of measurement includes such vari- ables as the number of individuals observed, the number of behaviors recorded, the duration of observation peri- ods, and/or the duration of the observation intervals, all of which may affect the quality of measurement. For in- stance, observing several individuals is more complex than observing one person; recording several behaviors is more complex than recording a single behavior; using contiguous 5-second observation intervals with no time between intervals to record the results of the observation is more difficult than a system in which time is reserved for recording data.

Specific recommendations concerning reducing com- plexity depend on the specific nature of the study. How- ever, when using time sampling measurements, applied behavior analysts can consider modifications such as de- creasing the number of simultaneously observed indi- viduals or behaviors, decreasing the duration of the observation sessions (e.g., from 30 minutes to 15 min- utes), and increasing the duration of time intervals (e.g., from 5 to 10 seconds). Requiring more practice during observer training, establishing a higher criterion for mas- tery of the observational code, and providing more fre- quent feedback to observers may also reduce the possible negative effects of complex measurement.

Inadequate Observer Training

Careful attention must be paid to the selection and train- ing of observers. Explicit and systematic training of ob- servers is essential for the collection of trustworthy data. Observation and coding systems require observers to dis- criminate the occurrence and nonoccurrence of specific classes of behaviors or events against an often complex and dynamic background of other behaviors or events and to record their observations onto a data sheet. Ob- servers must learn the definitions for each response class or event to be measured; a code or symbol notation sys- tem for each variable; a common set of recording proce- dures such as keystrokes or scan movements; and a method for correcting inadvertent handwritten, keystroke, or scan mistakes (e.g., writing a plus sign instead of a minus sign, hitting the F6 key instead of the F5 key, scan- ning an incorrect bar code).

Selecting Observers Carefully

Admittedly, applied researchers often scramble to find data collectors, but not all volunteers should be accepted into training. Potential observers should be interviewed to determine past experiences with observation and measurement activities, current schedule and upcoming

128

Improving and Assessing the Quality of Behavioral Measurement

commitments, work ethic and motivation, and overall social skills. The interview might include a pretest to determine current observation and skill levels. This can be accom- plished by having potential observers watch short video clips of behaviors similar to what they may be asked to ob- serve and noting their performance against a criterion.

Training Observers to an Objective Standard of Competency

Observer trainees should meet a specified criterion for recording before conducting observations in applied set- tings. During training, observers should practice record- ing numerous examples and nonexamples of the target behavior(s) and receive a critique and performance feed- back. Observers should have numerous practice sessions before actual data collection. Training should continue until a predetermined criterion is achieved (e.g., 95% ac- curacy for two or three consecutive sessions). For exam- ple, in training observers to measure the completion of preventive maintenance tasks of heavy equipment by mil- itary personnel, Komaki (1998) required three consecu- tive sessions of at least 90% agreement with a true value.

Various methods can be used to train observers. These include sample vignettes, narrative descriptions, video sequences, role playing, and practice sessions in the environment in which actual data will be collected. Practice sessions in natural settings are especially bene- ficial because they allow both observers and participants to adapt to each other’s presence and may reduce the re- active effects of the presence of observers on participants’ behavior. The following steps are an example of a sys- tematic approach for training observers.

Step 1 Trainees read the target behavior definitions and become familiar with data collection forms, pro- cedures for recording their observations, and the proper use of any measurement or recording devices (e.g., tape recorders, stopwatches, laptops, PDAs, bar code scanners).

Step 2 Trainees practice recording simplified narrative descriptions of behavioral vignettes until they obtain 100% accuracy over a predetermined number of instances.

Step 3 Trainees practice recording longer, more com- plex narrative descriptions of behavioral vignettes until they obtain 100% accuracy for a predetermined number of episodes.

Step 4 Trainees practice observing and recording data from videotaped or role-played vignettes depicting the target behavior(s) at the same speed and complexity as they will occur in the natural environment. Training vi- gnettes should be scripted and sequenced to provide trainees practice making increasingly difficult discrimi- nations between the occurrence and nonoccurrence of the

target behavior(s). Having trainees rescore the same se- ries of vignettes a second time and comparing the relia- bility of their measures provides an assessment of the consistency with which the trainees are applying the mea- surement system. Trainees remain at this step until their data reach preestablished accuracy and reliability criteria. (If the study involved collecting data from natural per- manent products such as compositions or academic work- sheets, Steps 2 through 4 should provide trainees with practice scoring increasingly extensive and more diffi- cult to score examples.)

Step 5 Practicing collecting data in the natural environ- ment is the final training step of observer training. An experienced observer accompanies the trainee and si- multaneously and independently measures the target behaviors. Each practice session ends with the trainee and experienced observer comparing their data sheets and discussing any questionable or heretofore unfore- seen instances. Training continues until a preestab- lished criterion of agreement between the experienced observer and the trainee is achieved (e.g., at least 90% for three consecutive sessions).

Providing Ongoing Training to Minimize Observer Drift

Over the course of a study, observers sometimes alter, often unknowingly, the way they apply a measurement system. Called observer drift, these unintended changes in the way data are collected may produce measurement error. Observer drift usually entails a shift in the ob- server’s interpretation of the definition of the target be- havior from that used in training. Observer drift occurs when observers expand or compress the original defini- tion of the target behavior. For example, observer drift might be responsible for the same behaviors by a child that were recorded by an observer as instances of non- compliance during the first week of a study being scored as instances of compliance during the study’s final week. Observers are usually unaware of the drift in their mea- surement.

Observer drift can be minimized by occasional ob- server retraining or booster sessions throughout the in- vestigation. Continued training provides the opportunity for observers to receive frequent feedback on the accuracy and reliability of measurement. Ongoing training can occur at regular, prescheduled intervals (e.g., every Fri- day morning) or randomly.

Unintended Influences on Observers

Ideally, data reported by observers have been influenced only by the actual occurrences and nonoccurrences of the target behavior(s) they have been trained to measure. In

129

Improving and Assessing the Quality of Behavioral Measurement

reality, however, a variety of unintended and undesired in- fluences on observers can threaten the accuracy and re- liability of the data they report. Common causes of this type of measurement error include presuppositions an observer may hold about the expected outcomes of the data and an observer’s awareness that others are measur- ing the same behavior.

Observer Expectations

Observer expectations that the target behavior should occur at a certain level under particular conditions, or change when a change in the environment has been made, pose a major threat to accurate measurement. For example, if an observer believes or predicts that a teacher’s implementa- tion of a token economy should decrease the frequency of inappropriate student behavior, she may record fewer in- appropriate behaviors during the token reinforcement con- dition than she would have recorded otherwise without holding that expectation. Data influenced by an observer’s expectations or efforts to obtain results that will please the researcher are characterized by measurement bias.

The surest way to minimize measurement bias caused by observer expectations is to use naive observers. A totally naive observer is a trained observer who is un- aware of the study’s purpose and/or the experimental con- ditions in effect during a given phase or observation period. Researchers should inform observer trainees that they will receive limited information about the study’s purpose and why that is. However, maintaining observers’ naiveté is often difficult and sometimes impossible.

When observers are aware of the purpose or hy- pothesized results of an investigation, measurement bias can be minimized by using target behavior definitions and recording procedures that will give a conservative picture of the behavior (e.g., whole-interval recording of on-task behavior with 10-second rather than 5-second in- tervals), frank and repeated discussion with observers about the importance of collecting accurate data, and fre- quent feedback to observers on the extent to which their data agree with true values or data obtained by observers who are naive. Observers should not receive feedback about the extent to which their data confirm or run counter to hypothesized results or treatment goals.

Observer Reactivity

Measurement error resulting from an observer’s aware- ness that others are evaluating the data he reports is called observer reactivity. Like reactivity that may occur when participants are aware that their behavior is being ob- served, the behavior of observers (i.e., the data they record and report) can be influenced by the knowledge that others are evaluating the data. For example, knowing

that the researcher or another observer is watching the same behavior at the same time, or will monitor the mea- surement through video- or audiotape later, may produce observer reactivity. If the observer anticipates that an- other observer will record the behavior in a certain way, his data may be influenced by what he anticipates the other observer may record.

Monitoring observers as unobtrusively as possible on an unpredictable schedule helps reduce observer re- activity. Separating multiple observers by distance or partition reduces the likelihood that their measures will be influenced by one another’s during an observation. One-way mirrors in some research and clinical settings eliminate visual contact between the primary and sec- ondary observers. If sessions are audiotaped or video- taped, the secondary observer can measure the behavior at a later time and the primary observer never has to come into contact with the secondary observer. In settings where one-way mirrors are not possible, and where audio- or videotaping may be intrusive, the secondary observer might begin measuring the behavior at a time unknown to the primary observer. For example, if the primary observer begins measuring behavior with the first interval, the secondary observer could start mea- suring behavior after 10 minutes have elapsed. The in- tervals used for comparisons would begin at the 10-minute mark, ignoring those intervals that the pri- mary observer recorded beforehand.

Assessing the Accuracy and Reliability of Behavioral Measurement After designing a measurement system that will produce a valid representation of the target behavior and training observers to use it in a manner that is likely to yield accurate and reliable data, the researcher’s next measurement-related tasks are evaluating the extent to which the data are, in fact, accurate and reliable. Essen- tially, all procedures for assessing the accuracy and reli- ability of behavioral data entail some form of “measuring the measurement system.”

Assessing the Accuracy of Measurement

Measurement is accurate when the observed values (i.e., the numbers obtained by measuring an event) match the true values of the event. The fundamental reason for de- termining the accuracy of data is obvious: No one wants to base research conclusions or make treatment decisions

130

Improving and Assessing the Quality of Behavioral Measurement

on faulty data. More specifically, conducting accuracy assessments serves four interrelated purposes. First, it is important to determine early in an analysis whether the data are good enough to serve as the basis for making ex- perimental or treatment decisions. The first person that the researcher or practitioner must try to convince that the data are accurate is herself. Second, accuracy assess- ments enable the discovery and correction of specific in- stances of measurement error. The two other approaches to assessing the quality of data to be discussed later in this chapter—reliability assessments and interobserver agreement—can alert the researcher to the likelihood of measurement errors, but neither approach identifies er- rors. Only the direct assessment of measurement accu- racy allows practitioners or applied researchers to detect and correct faulty data.

A third reason for conducting accuracy assessments is to reveal consistent patterns of measurement error, which can lead to the overall improvement or calibration of the measurement system. When measurement error is consistent in direction and value, the data can be adjusted to compensate for the error. For example, knowing that John’s bicycle computer reliably obtained a measure of 68 miles for a route with a true value of 58 miles led not only to the cyclists correcting the data in hand (in this case, confessing to one another and to their friend Lee that they had not ridden as many miles as previously claimed) but to their calibrating the measurement instru- ment so that future measures would be more accurate (in this case, adjusting the wheel circumference setting on John’s bike computer).

Calibrating any measurement tool, whether it is a mechanical device or human observer, entails compar- ing the data obtained by the tool against a true value. The measure obtained by the Department of Transportation’s wheel odometer served as the true value for calibrating John’s bike computer. Calibration of a timing device such as a stopwatch or countdown timer could be made against a known standard: the “atomic clock.”4 If no differences are detected when comparing the timing device against the atomic clock, or if the differences are tolerable for the intended purposes of measurement, then calibration is satisfied. If significant differences are found, the tim- ing device would need to be reset to the standard. We rec- ommend frequent accuracy assessments in the beginning stages of an analysis. Then, if the assessments have pro- duced high accuracy, less frequent assessments can be conducted to check the calibration of the recorders.

4The official time in the United States can be accessed through the Na- tional Bureau of Standards and the United States Naval Observatory atomic clock (actually 63 atomic clocks are averaged to determine official time): http://tycho.usno.navy.mil/what1.html. The atomic clock is accurate to 1 billionth of a second per day, or 1 second per 6 million years!

5The preferred spelling of a word may change (e.g., judgement becomes judgment), but in such cases a new true value is established.

A fourth reason for conducting accuracy assessments is to assure consumers that the data are accurate. Includ- ing the results of accuracy assessments in research re- ports helps readers judge the trustworthiness of the data being offered for interpretation.

Establishing True Values

“There is only one way to assess the accuracy of a set of measures—by comparing observed values to true val- ues. The comparison is relatively easy; the challenge is often obtaining measures of behavior that can legiti- mately be considered true values” (Johnston & Penny- packer, 1993a, p. 138). As defined previously, a true value is a measure obtained by procedures that are in- dependent of and different from the procedures that pro- duced the data being evaluated and for which the researcher has taken “special or extraordinary precau- tions to ensure that all possible sources of error have been avoided or removed” (p. 136).

True values for some behaviors are evident and uni- versally accepted. For example, obtaining the true values of correct responses in academic areas such math and spelling is straightforward. The correct response to the arithmetic problem 2 + 2 = ? has a true value of 4, and the Oxford English Dictionary is a source of true values for assessing the accuracy of measuring the spelling of English words.5 Although not universal, true values for many socially significant behaviors of interest to applied researchers and practitioners can be established condi- tionally on local context. For example, the correct re- sponse to the question “Name the three starches recommended as thickeners for pan gravy” on a quiz given to students in a culinary school has no universal true value. Nevertheless, a true value relevant to the stu- dents taking the quiz can be found in the instructor’s course materials.

True values for each of the preceding examples were obtained through sources independent of the measures to be evaluated. Establishing true values for many behav- iors studied by applied behavior analysts is difficult be- cause the process for determining a true value must be different from the measurement procedures used to obtain the data one wishes to compare to the true value. For ex- ample, determining true values for occurrences of a be- havior such as cooperative play between children is difficult because the only way to attach any values to the behavior is to measure it with the same observation pro- cedures used to produce the data in the first place.

It can be easy to mistake true values as values that only appear to be true values. For example, suppose that

131

Improving and Assessing the Quality of Behavioral Measurement

four well-trained and experienced observers view a video- tape of teacher and student interactions. Their task is to identify the true value of all instances of teacher praise contingent on academic accomplishments. Each observer views the tape independently and counts all occurrences of contingent teacher praise. After recording their re- spective observations, the four observers share their mea- surements, discuss disagreements, and suggest reasons for the disagreements. The observers independently record contingent praise a second time. Once again they share and discuss their results. After repeating the record- ing and sharing process several times, all observers agree that they have recorded every instance of teacher praise. However, the observers did not produce a true value of teacher praise for two reasons: (1) The observers could not calibrate their measurement of teacher praise to an independent standard of teacher praise, and (2) the process used to identify all instances of teacher praise may be biased (e.g., one of the observers may have con- vinced the others that her measures represented the true value). When true values cannot be established, re- searchers must rely on reliability assessments and mea- sures of interobserver agreement to evaluate the quality of their data.

Accuracy Assessment Procedures

Determining the accuracy of measurement is a straight- forward process of calculating the correspondence of each measure, or datum, assessed to its true value. For exam- ple, a researcher or practitioner assessing the accuracy of the score for a student’s performance on a 30-word spelling test reported by a grader would compare the grader’s scoring of each word on the test with the true value for that word found in a dictionary. Each word on the test that matched the correct letter sequence (i.e., or- thography) provided by the dictionary and was marked correct by the grader would be an accurate measure by the grader, as would each word marked incorrect by the grader that did not match the dictionary’s spelling. If the original grader’s scoring of 29 of the test’s 30 words cor- responded to the true values for those words, the grader’s measure would be 96.7% accurate.

Although an individual researcher or practitioner can assess the accuracy of the data she has collected, multi- ple independent observers are often used. Brown, Dunne, and Cooper (1996) described the procedures they used to assess the accuracy of measurement in a study of oral reading comprehension as follows:

An independent observer reviewed one student’s audio- tape of the delayed one-minute oral retell each day to assess our accuracy of measurement, providing an assessment of the extent that our counts of delayed retells approximated the true value of the audio-taped

correct and incorrect retells. The independent observer randomly selected each day’s audiotape by drawing a student’s name from a hat, then listened to the tape and scored correct and incorrect retells using the same defin- itions as the teacher. Observer scores were compared to teacher scores. If there was a discrepancy between these scores, the observer and the teacher reviewed the tape (i.e., the true value) together to identify the source of the discrepancy and corrected the counting error on the data sheet and the Standard Celeration Chart. The observer also used a stopwatch to time the duration of the audio- tape to ensure accuracy of the timings. We planned to have the teacher re-time the presentation or retell and re- calculate the frequency per minute for each timing dis- crepancy of more than 5 seconds. All timings, however, met the 5-second accuracy definition. (p. 392)

Reporting Accuracy Assessments

In addition to describing procedures used to assess the accuracy of the data, researchers should report the num- ber and percentage of measures that were checked for ac- curacy, the degree of accuracy found, the extent of measurement error detected, and whether those mea- surement errors were corrected in the data. Brown and colleagues (1996) used the following narrative to report the results of their accuracy assessment:

The independent observer and the teacher achieved 100% agreement on 23 of the 37 sessions checked. The teacher and the observer reviewed the tape together to identify the source of measurement errors for the 14 ses- sions containing measurement discrepancies and cor- rected the measurement errors. Accurate data from the 37 sessions rechecked were then displayed on the Stan- dard Celeration Charts. The magnitude of the measure- ment errors was very small, often a difference of 1 to 3 discrepancies. (p. 392)

A full description and reporting of the results of ac- curacy assessment helps readers of the study evaluate the accuracy of all of the data included in the report. For ex- ample, suppose a researcher reported that she conducted accuracy checks on a randomly selected 20% of the data, found those measures to be 97% accurate with the 3% error being nonbiased, and corrected the assessed data as needed. A reader of the study would know that 20% of the data are 100% accurate and be fairly confident that the remaining 80% of the data (i.e., all of the measures that were not checked for accuracy) is 97% accurate.

Assessing the Reliability of Measurement

Measurement is reliable when it yields the same values across repeated measures of the same event. Reliability is established when the same observer measures the same

132

Improving and Assessing the Quality of Behavioral Measurement

data set repeatedly from archived response products such as audiovisual products and other forms of permanent products. The more frequently a consistent pattern of ob- servation is produced, the more reliable the measurement (Thompson et al., 2000). Conversely, if similar observed values are not achieved with repeated observations, the data are considered unreliable. This leads to a concern about accuracy, which is the primary indicator of quality measurement.

But, as we have pointed out repeatedly, reliable data are not necessarily accurate data. As the three bicyclists discovered, totally reliable (i.e., consistent) measurement may be totally wrong. Relying on the reliability of mea- surement as the basis for determining the accuracy of measurement would be, as the philosopher Wittgenstein (1953) noted, “As if someone were to buy several copies of the morning paper to assure himself that what it said was true” (p. 94).

In many research studies and most practical appli- cations, however, checking the accuracy of every mea- sure is not possible or feasible. In other cases, true values for measures of the target behavior may be difficult to es- tablish. When confirming the accuracy of each datum is not possible or practical, or when true values are not available, knowing that a measurement system has been applied with a high degree of consistency contributes to confidence in the overall trustworthiness of the data. Al- though high reliability cannot confirm high accuracy, discovering a low level of reliability signals that the data are then suspect enough to be disregarded until prob- lems in the measurement system can be determined and repaired.

Assessing the reliability of behavioral measure- ment requires either a natural or contrived permanent product so the observer can remeasure the same events. For example, reliability of measurement of variables such as the number of adjectives or action verbs in students’ essays could be accomplished by having an observer rescore essays. Reliability of measurement of the number and type of response prompts and feed- back statements by parents to their children at the family dinner table could be assessed by having an ob- server replay and rescore videotapes of the family’s mealtime and compare the data obtained from the two measurements.

Observers should not remeasure the same permanent product soon after measuring it the first time. Doing so might result in the measures from the second scoring being influenced by what the observer remembered from the initial scoring. To avoid such unwanted influence, a researcher can insert several previously scored essays or videotapes randomly into the sequence of “new data” being recorded by observers.

Using Interobserver Agreement to Assess Behavioral Measurement Interobserver agreement is the most commonly used in- dicator of measurement quality in applied behavior analy- sis. Interobserver agreement (IOA) refers to the degree to which two or more independent observers report the same observed values after measuring the same events. There are numerous techniques for calculating IOA, each of which provides a somewhat different view of the ex- tent and nature of agreement and disagreement between observers (e.g., Hartmann, 1977; Hawkins & Dotson, 1975; Page & Iwata, 1986; Poling, Methot, & LeSage, 1995; Repp, Dietz, Boles, Dietz, & Repp, 1976).

Benefits and Uses of IOA

Obtaining and reporting interobserver agreement serves four distinct purposes. First, a certain level of IOA can be used as a basis for determining the competence of new observers. As noted earlier, a high degree of agreement between a newly trained observer and an experienced ob- server provides an objective index of the extent to which the new observer is measuring the behavior in the same way as experienced observers.

Second, systematic assessment of IOA over the course of a study can detect observer drift. When ob- servers who obtained the same, or nearly the same, ob- served values when measuring the same behavioral events at the beginning of a study (i.e., IOA was high) obtain different measures of the same events later in the study (i.e., IOA is now low), one of the observers may be using a definition of the target behavior that has drifted. Dete- riorating IOA assessments cannot indicate with assur- ance which of the observer’s data are being influenced by drift (or any other reason for disagreement), but the in- formation reveals the need for further evaluation of the data and/or for retraining and calibration of the observers.

Third, knowing that two or more observers consis- tently obtained similar data increases confidence that the definition of the target behavior was clear and unam- biguous and the measurement code and system not too difficult. Fourth, for studies that employ multiple ob- servers as data collectors, consistently high levels of IOA increase confidence that variability in the data is not a function of which observer(s) happened to be on duty for any given session, and therefore that changes in the data more likely reflect actual changes in the behavior.

The first two reasons for assessing IOA are proac- tive: They help researchers determine and describe the degree to which observers have met training criteria and detect possible drift in observers’ use of the measurement

133

Improving and Assessing the Quality of Behavioral Measurement

system. The second two purposes or benefits of IOA are as summative descriptors of the consistency of measure- ment across observers. By reporting the results of IOA as- sessments, researchers enable consumers to judge the relative believability of the data as trustworthy and de- serving of interpretation.

Requisites for Obtaining Valid IOA Measures

A valid assessment of IOA depends on three equally im- portant criteria. Although these criteria are perhaps ob- vious, it is nonetheless important to make them explicit. Two observers (usually two, but may be more) must (a) use the same observation code and measurement system, (b) observe and measure the same participant(s) and events, and (c) observe and record the behavior indepen- dent of any influence from one other.

Observers Must Use the Same Measurement System

Interobserver agreement assessments conducted for any of the four previously stated reasons require observers to use the same definitions of the target behavior, observa- tion procedures and codes, and measurement devices. Beyond using the same measurement system, all ob- servers participating in IOA measures used to assess the believability of data (as opposed to evaluating the ob- server trainees’ performance) should have received iden- tical training with the measurement system and achieved the same level of competence in using it.

Observers Must Measure the Same Events

The observers must be able to observe the same subject(s) at precisely the same observation intervals and periods. IOA for data obtained by real-time measurement requires that both observers be in the setting simultaneously. Real- time observers must be positioned such that each has a similar view of the subject(s) and environment. Two ob- servers sitting on opposite sides of a classroom, for ex- ample, might obtain different measures because the different vantage points enable only one observer to see or hear some occurrences of the target behavior.

Observers must begin and end the observation pe- riod at precisely the same time. Even a difference of a few seconds between observers may produce significant measurement disagreements. To remedy this situation, the timing devices could be started simultaneously and outside the observation setting, but before data collec- tion begins, with the understanding that the data collec- tion would actually start at a prearranged time (e.g., exactly at the beginning of the fifth minute). Alterna-

tively, but less desirably, one observer could signal the other at the exact moment the observation is to begin.

A common and effective procedure is for both ob- servers to listen by earphones to an audiotape of prere- corded cues signaling the beginning and end of each observation interval. An inexpensive splitter device that enables two earphones to be plugged into the same tape recorder allows observers to receive simultaneous cues unobtrusively and without depending on one another.

When assessing IOA for data obtained from perma- nent products, the two observers do not need to measure the behavior simultaneously. For example, the observers could each watch and record data from the same video- or audiotape at different times. Procedures must be in place, however, to ensure that each observer watched or listened to the same tapes and that they started and stopped their independent observations at precisely the same point(s) on the tapes. Ensuring that two observers measure the same events when the target behavior pro- duces natural permanent products, such as completed aca- demic assignments or widgets manufactured, would include procedures such as clearly marking the session number, date, condition, and subject’s name on the prod- uct and guarding the response products to ensure that they are not disturbed until the second observer has ob- tained his measure.

Observers Must Be Independent

The third essential ingredient for valid IOA assessment is ensuring that neither observer is influenced by the other’s measurements. Procedures must be in place to guarantee each observer’s independence. For example, observers conducting real-time measurement of behavior “must be situated so that they can neither see nor hear when the other observes and records a response” (Johnston & Pen- nypacker, 1993a, p. 147). Observers must not be seated or positioned so closely to one another that either ob- server can detect or be influenced by the other observer’s recordings.

Giving the second observer academic worksheets or written assignments that have already been marked by another observer would violate the observers’ indepen- dence. To maintain independence, the second observer must score photocopies of unadulterated and unmarked worksheets or assignments as completed by the subjects.

Methods for Calculating IOA

There are numerous methods for calculating IOA, each of which provides a somewhat different view of the extent and nature of agreement and disagreement between ob- servers (e.g., Hartmann, 1977; Hawkins & Dotson, 1975;

134

Improving and Assessing the Quality of Behavioral Measurement

Page & Iwata, 1986; Poling, Methot, & LeSage, 1995; Repp, Dietz, Boles, Dietz, & Repp, 1976). The following explanation of different IOA formats is organized by the three major methods for measuring behavioral data: event recording, timing, and interval recording or time sam- pling. Although other statistics are sometimes used, the percentage of agreement between observers is by far the most common convention for reporting IOA in applied behavior analysis.6 Therefore, we have provided the for- mula for calculating a percentage of agreement for each type of IOA.

IOA for Data Obtained by Event Recording

The various methods for calculating interobserver agree- ment for data obtained by event recording are based on comparing (a) the total count recorded by each observer per measurement period, (b) the counts tallied by each observer during each of a series of smaller intervals of time within the measurement period, or (c) each ob- server’s count of 1 or 0 on a trial-by-trial basis.

Total Count IOA.7 The simplest and crudest indi- cator of IOA for event recording data compares the total count recorded by each observer per measurement period. Total count IOA is expressed as a percentage of agreement between the total number of responses recorded by two observers and is calculated by dividing the smaller of the counts by the larger count and multi- plying by 100, as shown by this formula:

For example, suppose that a child care worker in a residential setting recorded that 9-year-old Mitchell used profane language 10 times during a 30-minute observa- tion period and that a second observer recorded that Mitchell swore 9 times during that same period. The total

6IOA can be calculated by product-moment correlations, which range from +1.0 to �1.0. However, expressing IOA by correlation coefficients has two major weaknesses: (a) High coefficients can be achieved if one observer consistently records more occurrences of the behavior than the other, and (b) correlation coefficients provide no assurance that the observers agreed on the occurrence of any given instance of behavior (Poling et al., 1995). Hartmann (1977) described the use of kappa (k) as an measure of IOA. The k statistic was developed by Cohen (1960) as a procedure for deter- mining the proportion of agreements between observers that would be ex- pected as a result of chance. However, the k statistic is seldom reported in the behavior analysis literature. 7Multiple terms are used in the applied behavior analysis literature for the same methods of calculating IOA, and the same terms are sometimes used with different meanings. We believe the IOA terms used here represent the discipline’s most used conventions. In an effort to point out and preserve some meaningful distinctions among variations of IOA measures, we have introduced several terms.

count IOA for the observation period would be 90% (i.e., 9 10 100 = 90%).

Great caution must be used in interpreting total count IOA because a high degree of agreement provides no as- surance that the two observers recorded the same in- stances of behavior. For example, the following is one of the countless ways that the data reported by the two observers who measured Mitchell’s use of profane lan- guage may not represent anywhere close to 90% agree- ment that they measured the same behaviors. The child care worker could have recorded all 10 occurrences of profane language on her data sheet during the first 15 minutes of the 30-minute observation period, a time when the second observer recorded just 4 of the 9 total responses he reported.

Mean Count-per-Interval IOA. The likelihood that significant agreement between observers’ count data means they measured the same events can be increased by (a) dividing the total observation period into a series of smaller counting times, (b) having the observers record the number of occurrences of the behavior within each interval, (c) calculating the agreement be- tween the two observers’ counts within each interval, and (d) using the agreements per interval as the basis for calculating the IOA for the total observation period. The hypothetical data shown in Figure 2 will be used to illustrate two methods for calculating count-per- interval IOA: mean count-per-interval and exact count- per-interval. During a 30-minute observation period, two observers independently tallied the number of times each witnessed an instance of a target behavior during each of six 5-minute intervals.

Even though each observer recorded a total of 15 re- sponses within the 30-minute period, their data sheets re- veal a high degree of disagreement within the observation period. Although the total count IOA for the entire obser- vation period was 100%, agreement between the two ob- servers within each 5-minute interval ranged from 0% to 100%, yielding a mean count-per-interval IOA of 65.3%.

Mean count-per-interval IOA is calculated by this formula:

Int 1 IOA + Int 2 IOA + Int N IOA —————————————————— n intervals = mean count-per-interval IOA %

Exact Count-per-Interval IOA. The most stringent description of IOA for most data sets obtained by event recording is obtained by computing the exact count- per-interval IOA—the percentage of total intervals in which two observers recorded the same count. The two observers whose data are shown in Figure 2 recorded

Smaller count

Larger count x 100 = total count IOA %

× ÷

135

Improving and Assessing the Quality of Behavioral Measurement

Figure 2 Two methods for computing interobserver agreement (IOA) for event recording data tallied within smaller time intervals.

Interval (Time) Observer 1 Observer 2 IOA per interval

1 (1:00–1:05) /// // 2/3 = 67%

2 (1:05–1:10) /// /// 3/3 = 100%

3 (1:10–1:15) / // 1/2 = 50%

4 (1:15–1:20) //// /// 3/4 = 75%

5 (1:20–1:25) 0 / 0/1 = 0%

6 (1:25–1:30) //// //// 4/4 = 100%

Total count Total count Mean count-per-interval IOA = 65.3% = 15 = 15 Exact count-per-interval IOA = 33%

the same number of responses in just two of the six in- tervals, an exact count-per-interval IOA of 33%.

The following formula is used to calculate exact count-per-interval IOA:

Trial-by-Trial IOA. The agreement between two ob- servers who measured the occurrence or nonoccurrence of discrete trial behaviors for which the count for each trial, or response opportunity, can only be 0 or 1 can be calculated by comparing the observers’ total counts or by comparing their counts on a trial-by-trial basis. Calculat- ing total count IOA for discrete trial data uses the same formula as total count IOA for free operant data: The smaller of the two counts reported by the observers is di- vided by the larger count and multiplied by 100, but in this case the number of trials for which each observer recorded the occurrence of the behavior is the count. Suppose, for example, that a researcher and a second ob- server independently measured the occurrence or nonoc- currence of a child’s smiling behavior during each of 20 trials that the researcher showed the child a funny pic- ture. The two observers compare data sheets at the end of the session and discover that they recorded smiles on 14 and 15 trials, respectively. The total count IOA for the session is 93% (i.e., 14 15 100 = 93.3%), which might lead an inexperienced researcher to conclude that the target behavior has been well defined and is being measured with consistency by both observers. Those conclusions, however, would not be warranted.

Total count IOA of discrete trial data is subject to the same limitations as total count IOA of free operant data:

It tends to overestimate the extent of actual agreement and does not indicate how many responses, or which re- sponses, trials, or items, posed agreement problems. Comparing the two observers’ counts of 14 and 15 trials suggests that they disagreed on the occurrence of smiling on only 1 of 20 trials. However, it is possible that any of the 6 trials scored as “no smile” by the experimenter was scored as a “smile” trial by the second observer and that any of the 5 trials recorded by the second observer as “no smile” was recorded as a “smile” by the experimenter. Thus, the total count IOA of 93% may vastly overesti- mate the actual consistency with which the two observers measured the child’s behavior during the session.

A more conservative and meaningful index of inter- observer agreement for discrete trial data is trial-by-trial IOA, which is calculated by the following formula:

The trial-by-trial IOA for the two observers’ smiling data, if calculated with the worst possible degree of agree- ment from the previous example—that is, if all 6 trials that the primary observer scored as “no smile” were recorded as “smile” trials by the second observer and all 5 trials marked by the second observer as “no smile” were recorded as “smile” trials by the experimenter—would be 45% (i.e., 9 trials scored in agreement divided by 20 trials 100).

IOA for Data Obtained by Timing

Interobserver agreement for data obtained by timing du- ration, response latency, or interresponse time (IRT) is obtained and calculated in essentially the same way as it

* 100 = exact count-per-interval IOA %

Number of intervals of 100% IOA

n intervals

* 100 = trial-by-trial IOA %

Number of trials (items) agreement

Total number of trials (items)

÷ × ×

136

Improving and Assessing the Quality of Behavioral Measurement

is for event recording data. Two observers independently time the duration, latency, or IRT of the target behavior, and IOA is based on comparing either the total time ob- tained by each observer for the session or the times recorded by each observer per occurrence of the behav- ior (for duration measures) or per response (for latency and IRT measures).

Total Duration IOA. Total duration IOA is com- puted by dividing the shorter of the two durations re- ported by the observers by the longer duration and multiplying by 100.

As with total count IOA for event recording data, high total duration IOA provides no assurance that the observers recorded the same durations for the same oc- currences of behavior. This is because a significant degree of disagreement between the observers’ timings of indi- vidual responses may be canceled out in the sum. For ex- ample, suppose two observers recorded the following durations in seconds for five occurrences of a behavior:

R1 R2 R3 R4 R5

Observer 1: 35 15 9 14 17 (total duration = 90 seconds)

Observer 2: 29 21 7 14 14 (total duration = 85 seconds)

Total duration IOA for these data is a perhaps com-

two observers obtained the same duration for only one of the five responses, and their timings of specific re- sponses varied by as much as 6 seconds. While recog- nizing this limitation of total duration IOA, when total duration is being recorded and analyzed as a dependent variable, reporting total duration IOA is appropriate. When possible, total duration IOA should be supple- mented with mean duration-per-occurrence IOA, which is described next.

Mean Duration-per-Occurrence IOA. Mean dura- tion-per-occurrence IOA should be calculated for dura- tion per occurrence data, and it is a more conservative and usually more meaningful assessment of IOA for total duration data. The formula for calculating mean duration-per-occurrence IOA is similar to the one used to determine mean count-per-interval IOA:

Using this formula to calculate the mean duration- per-occurrence IOA for the two observers’ timing data of the five responses just presented would entail the fol- lowing steps:

1. Calculate duration per occurrence IOA for each re-

2. Add the individual IOA percentages for each oc- currence: .83 + .71 + .78 + 1.00 + .82 = 4.14

3. Divide the sum of the individual IOAs per occur- rence by the total number of responses for which

4. Multiply by 100 and round to the nearest whole

This basic formula is also used to compute the mean latency-per-response IOA or mean IRT-per-response IOA for latency and IRT data. An observer’s timings of laten- cies or IRTs in a session should never be added and the total time compared to a similar total time obtained by

tency and IRT measures. In addition to reporting mean agreement per oc-

currence, IOA assessment for timing data can be en- hanced with information about the range of differences between observers’ timings and the percentage of re- sponses for which the two observers each obtained measures within a certain range of error. For example: Mean duration-per-occurrence IOA for Temple’s com- pliance was 87% (range across responses, 63 to 100%), and 96% of all timings obtained by the second observer were within +/–2 seconds of the primary observer’s measures.

IOA for Data Obtained by Interval Recording/Time Sampling

Three techniques commonly used by applied behavior analysts to calculate IOA for interval data are interval- by-interval IOA, scored-interval IOA, and unscored- interval IOA.

Interval-by-Interval IOA. When using an interval- by-interval IOA (sometimes referred to as the point-by- point and total interval method), the primary observer’s record for each interval is matched to the secondary ob- server’s record for the same interval. The formula for calculating interval-by-interval IOA is as follows:

forting 94% (i.e., 85 ÷ 90 × 100 = 94.4%). However, the

* 100 = mean duration-per-interval IOA %

Dur IOA R1 + Dur IOA R2 + Dur IOA Rn

n responses with Dur IOA

Shorter duration

Longer duration * 100 = total duration IOA %

sponse: R1, 29 ÷ 35 = .83; R2, 15 ÷ 21 = .71; R3, 7 ÷ 9 = .78; R4, 14 ÷ 14 = 1.0; and R5, 14 ÷ 17 = .82

two observers measured duration: 4.14 ÷ 5 = .828

number: .828 × 100 = 83%

* 100 = interval-by-interval IOA %

Number of intervals agreed

Number of intervals agreed + number of intervals disagreed

another observer as the basis for calculating IOA for la-

137

Improving and Assessing the Quality of Behavioral Measurement

Figure 3 When calculating interval-by-interval IOA, the number of intervals in which both observers agreed on the occurrence or the nonoccurrence of the behavior (shaded intervals) is divided by the total number of observation intervals. Interval-by- interval IOA for the data shown here is 70% (7/10).

Interval-by-Interval IOA

Interval no. 1 2 3 4 5 6 7 8 9 10

Observer 1 X X X 0 X X 0 X X 0

Observer 2 0 X X 0 X 0 0 0 X 0

X = behavior was recorded as occurring during interval

0 = behavior was recorded as not occurring during interval

Figure 4 Scored-interval IOA is calculated using only those intervals in which either observer recorded the occurrence of the behavior (shaded intervals). Scored- interval IOA for the data shown here is 33% (1/3).

Scored-Interval IOA

Interval no. 1 2 3 4 5 6 7 8 9 10

Observer 1 X 0 X 0 0 0 0 0 0 0

Observer 2 0 0 X 0 0 0 0 0 X 0

X = behavior was recorded as occurring during interval

0 = behavior was recorded as not occurring during interval

The hypothetical data in Figure 3 show the interval- by-interval method for calculating IOA based on the record of two observers who recorded the occurrence (X) and nonoccurrence (0) of behavior in each of 10 obser- vation intervals. The observers’ data sheets show that they agreed on the occurrence or the nonoccurrence of the be- havior for seven intervals (Intervals 2, 3, 4, 5, 7, 9, and 10). Interval-by-interval IOA for this data set is 70% (i.e.,

Interval-by-interval IOA is likely to overestimate the actual agreement between observers measuring behav- iors that occur at very low or very high rates. This is be- cause interval-by-interval IOA is subject to random or accidental agreement between observers. For example, with a behavior whose actual frequency of occurrence is only about 1 or 2 intervals per 10 observation intervals, even a poorly trained and unreliable observer who misses some of the few occurrences of the behavior and mis- takenly records the behavior as occurring in some inter- vals when the behavior did not occur is likely to mark most intervals as nonoccurrences. As a result of this chance agreement, interval-by-interval IOA is likely to be quite high. Two IOA methods that minimize the ef-

fects of chance agreements for interval data on behav- iors that occur at very low or very high rates are scored- interval IOA and unscored-interval IOA (Hawkins & Dotson, 1975).

Scored-Interval IOA. Only those intervals in which either or both observers recorded the occurrence of the target behavior are used in calculating scored- interval IOA. An agreement is counted when both ob- servers recorded that the behavior occurred in the same interval, and each interval in which one observer recorded the occurrence of the behavior and the other recorded its nonoccurrence is counted as a disagree- ment. For example, for the data shown in Figure 4, only Intervals 1, 3, and 9 would be used in calculating scored-interval IOA. Intervals 2, 4, 5, 6, 7, 8, and 10 would be ignored because both observers recorded that the behavior did not occur in those intervals. Because the two observers agreed that the behavior occurred in only one (Interval 3) of the three scored intervals, the scored-interval IOA measure is 33% (1 interval of agreement divided by the sum of 1 interval of agree- ment plus 2 intervals of disagreement 100 = 33%).

7 ÷ [7 + 3] × 100 = 70%).

→

138

Improving and Assessing the Quality of Behavioral Measurement

Figure 5 Unscored-interval IOA is calculated using only those intervals in which either observer recorded the nonoccurrence of the behavior (shaded intervals). Unscored interval IOA for the data shown here is 50% (2/4).

Unscored-Interval IOA

Interval no. 1 2 3 4 5 6 7 8 9 10

Observer 1 X X X 0 X X 0 X X 0

Observer 2 0 X X 0 X X 0 X X X

X = behavior was recorded as occurring during interval

0 = behavior was recorded as not occurring during interval

For behaviors that occur at low rates, scored-interval IOA is a more conservative measure of agreement than interval-by-interval IOA. This is because scored-interval IOA ignores the intervals in which agreement by chance is highly likely. For example, using the interval-by- interval method for calculating IOA for the data in Figure 4 would yield an agreement of 80%. To avoid overin- flated and possibly misleading IOA measures, we rec- ommend using scored-interval interobserver agreement for behaviors that occur at frequencies of approximately 30% or fewer intervals.

Unscored-Interval IOA. Only intervals in which either or both observers recorded the nonoccurrence of the target behavior are considered when calculating unscored-interval IOA. An agreement is counted when both observers recorded the nonoccurrence of the behavior in the same interval, and each interval in which one observer recorded the nonoccurrence of the behavior and the other recorded its occurrence is counted as a disagreement. For example, only Intervals 1, 4, 7, and 10 would be used in calculating the un- scored-interval IOA for the data in Figure 5 because at least one observer recorded the nonoccurrence of the behavior in each of those intervals. The two observers agreed that the behavior did not occur in Intervals 4 and 7. Therefore, the unscored-interval IOA in this example is 50% (2 intervals of agreement divided by the sum of 2 intervals of agreement plus 2 intervals of disagree- ment 100 = 50%).

For behaviors that occur at relatively high rates, unscored-interval IOA provides a more stringent assess- ment of interobserver agreement than does interval- by-interval IOA. To avoid overinflated and possibly mis- leading IOA measures, we recommend using unscored- interval interobserver agreement for behaviors that occur at frequencies of approximately 70% or more of intervals.

Considerations in Selecting, Obtaining, and Reporting Interobserver Agreement

The guidelines and recommendations that follow are or- ganized under a series of questions concerning the use of interobserver agreement to evaluate the quality of be- havioral measurement.

How Often and When Should IOA Be Obtained?

Interobserver agreement should be assessed during each condition and phase of a study and be distributed across days of the week, times of day, settings, and observers. Scheduling IOA assessments in this manner ensures that the results will provide a representative (i.e., valid) pic- ture of all data obtained in a study. Current practice and recommendations by authors of behavioral research methods texts suggest that IOA be obtained for a mini- mum of 20% of a study’s sessions, and preferably be- tween 25% and 33% of sessions (Kennedy, 2005; Poling et al., 1995). In general, studies using data obtained via real-time measurement will have IOA assessed for a higher percentage of sessions than studies with data ob- tained from permanent products.

The frequency with which data should be assessed via interobserver agreement will vary depending on the complexity of the measurement code, the number and ex- perience of observers, the number of conditions and phases, and the results of the IOA assessments themselves. More frequent IOA assessments are expected in studies that involve complex or new measurement systems, inex- perienced observers, and numerous conditions and phases. If appropriately conservative methods for obtaining and calculating IOA reveal high levels of agreement early in a study, the number and proportion of sessions in which IOA is assessed may decrease as the study progresses. For instance, IOA assessment might be conducted in each

→

139

Improving and Assessing the Quality of Behavioral Measurement

session at the beginning of an analysis, and then reduced to a schedule of once per four or five sessions.

For What Variables Should IOA Be Obtained and Reported?

In general, researchers should obtain and report IOA at the same levels at which they report and discuss the re- sults of their study. For example, a researcher analyzing the relative effects of two treatment conditions on two behaviors of four participants in two settings should re- port IOA outcomes on both behaviors for each partici- pant separated by treatment condition and setting. This would enable consumers of the research to judge the rel- ative believability of the data within each component of the experiment.

Which Method of Calculating IOA Should Be Used?

More stringent and conservative methods of calculating IOA should be used over methods that are likely to overestimate actual agreement as a result of chance. With event recording data used to evaluate the accu- racy of performance, we recommend reporting overall IOA on a trial-by-trial or item-by-item basis, perhaps supplemented with separate IOA calculations for cor- rect responses and incorrect responses. For data ob- tained by interval or time sampling measurement, we recommend supplementing interval-by-interval IOA with scored-interval IOA or unscored-interval IOA de- pending on the relative frequency of the behavior. In situations in which the primary observer scores the tar- get behavior as occurring in approximately 30% or fewer intervals, scored-interval IOA provides a con- servative supplement to interval-by-interval IOA. Con- versely, when the primary observer scores the target behavior as occurring in approximately 70% or more of the intervals, unscored-interval IOA should supplement interval-by-interval IOA. If the rate at which the target behavior occurs changes from very low to very high, or from very high to very low, across conditions or phases of a study, reporting both unscored-interval and scored- interval IOA may be warranted.

If in doubt about which form of IOA to report, cal- culating and presenting several variations will help read- ers make their own judgments regarding the believability of the data. However, if the acceptance of the data for in- terpretation or decision making rests on which formula for calculating IOA is chosen, serious concerns about the data’s trustworthiness exist that must be addressed.

What Are Acceptable Levels of IOA?

Carefully collected and conservatively computed IOA as- sessments increasingly enhance the believability of a data set as agreement approaches 100%. The usual conven- tion in applied behavior analysis is to expect indepen- dent observers to achieve a mean of no less than 80% agreement when using observational recording. However, as Kennedy (2005) pointed out, “There is no scientific justification for why 80% is necessary, only a long his- tory of researchers using this percentage as a benchmark of acceptability and being successful in their research ac- tivities” (p. 120).

Miller (1997) recommended that IOA should be 90% or greater for an established measure and at least 80% for a new variable. Various factors at work in a given situa- tion may make an 80% or 90% criterion too low or too high. Interobserver agreement of 90% on the number of words contained in student compositions should raise se- rious questions about the trustworthiness of the data. IOA near 100% is needed to enhance the believability of count data obtained from permanent products. However, some analysts might accept data with a mean IOA as low as 75% for the simultaneous measurement of multiple be- haviors by several subjects in a complex environment, es- pecially if it is based on a sufficient number of individual IOA assessments with a small range (e.g., 73 to 80%).

The degree of behavior change revealed by the data should also be considered when determining an accept- able level of interobserver agreement. When behavior change from one condition to another is small, the vari- ability in the data might represent inconsistent observa- tion more than actual change in the behavior. Therefore, the smaller the change in behavior across conditions, the higher the criterion should be for an acceptable IOA per- centage (Kennedy, 2005).

How Should IOA Be Reported?

IOA scores can be reported in narrative, table, and graphic form. Whichever format is chosen, it is important to note how, when, and how often interobserver agree- ment was assessed.

Narrative Description. The most common ap- proach for reporting IOA is a simple narrative descrip- tion of the mean and range of agreement percentages. For example, Craft, Alber, and Heward (1998) de- scribed the methods and results of IOA assessments in a study in which four dependent variables were mea- sured as follows:

Student recruiting and teacher praise. A second ob- server was present for 12 (30%) of the study’s 40

140

Improving and Assessing the Quality of Behavioral Measurement

Table 1 Interobserver Agreement Results for Each Dependent Variable by Participant and Experimental Condition

Range and Mean Percentage Interobserver Agreement on Scripted Interaction, Elaborations, and Unscripted Interaction by Child and Condition

Condition

Baseline Teaching New recipient Script fading New activities Type of interaction Range M Range M Range M Range M Range M

Scripted

David 88–100 94 100 100

Jeremiah 89–100 98 100 — a

Ben 80–100 98 90 — a

Elaborations

David 75–100 95 87–88 88 90–100 95

Jeremiah 83–100 95 92–100 96 — a

Ben 75–100 95 95 — a

Unscripted

David 100 100 87–88 88 97–100 98 98–100 99

Jeremiah 100 100 88–100 94 93–100 96 98

Ben 100 100 100 92–93 92 98–100 99

aNo data are available for scripted responses and elaborations in the script-fading condition, because interobserver agreement was obtained after scripts were removed (i.e., because scripts were absent, there could be only unscripted responses). From “Social Interaction Skills for Children with Autism: A Script-Fading Procedure for Beginning Readers,” by P. J. Krantz and L. E. McClannahan, 1998, Journal of Applied Behavior Analysis, 31, p. 196. Copyright 1998 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

sessions. The two observers independently and simulta- neously observed the 4 students, recording the number of recruiting responses they emitted and teacher praise they received. Descriptive narrative notes recorded by the observers enabled each recruiting episode to be iden- tified for agreement purposes. Interobserver agreement was calculated on an episode-by-episode basis by divid- ing the total number of agreements by the total number of agreements plus disagreements and multiplying by 100%. Agreement for frequency of student recruiting ranged across students from 88.2% to 100%; agreement for frequency of recruited teacher praise was 100% for all 4 students; agreement for frequency of nonrecruited teacher praise ranged from 93.3% to 100%.

Academic work completion and accuracy. A second observer independently recorded each student’s work completion and accuracy for 10 (25%) sessions. Interob- server agreement for both completion and accuracy on the spelling worksheets was 100% for all 4 students.

Table. An example of reporting interobserver agreement outcomes in table format is shown in Table 1. Krantz and McClannahan (1998) reported the range and mean IOA computed for three types of so-

cial interactions by three children across each experi- mental condition.

Graphic Display. Interobserver agreement can be represented visually by plotting the measures obtained by the secondary observer on a graph of the primary observer’s data as shown in Figure 6. Looking at both observers’ data on the same graph reveals the extent of agreement between the observers and the existence of observer drift or bias. The absence of observer drift is suggested in the hypothetical study shown in Figure 6 because the secondary observer’s measures changed in concert with the primary observer’s measures. Al- though the two observers obtained the same measure on only 2 of the 10 sessions in which IOA was assessed (Sessions 3 and 8), the fact that neither observer consis- tently reported measures that were higher or lower than the other suggests the absence of observer bias. An ab- sence of bias is usually indicated by a random pattern of overestimation and underestimation. In addition to revealing observer drift and bias, a third way that graphically displaying IOA assessments can enhance the believability of measurement is illustrated by the

141

0N u

m b

er o

f R

es p

o n

se s

p er

M in

u te

5 10 15 20 Sessions

Baseline Response Cost Baseline Response Cost

2nd Observer

Figure 6 Plotting measures obtained by a second observer on a graph of the primary observer’s data provide a visual representation of the extent and nature of interobserver agreement.

data in Figure 6. When the data reported by the pri- mary observer show clear change in the behavior be- tween conditions or phases and all of the measures reported by the secondary observer within each phase fall within the range of observed values obtained by the primary observer, confidence increases that the data represent actual changes in the behavior measured rather than changes in the primary observer’s behavior due to drift or extra-experimental contingencies.

Although published research reports in applied be- havior analysis seldom include graphic displays of IOA measures, creating and using such displays during a study is a simple and direct way for researchers to detect pat- terns in the consistency (or inconsistency) with which observers are measuring behavior that might be not be as evident in comparing a series of percentages.

Which Approach Should Be Used for Assessing the Quality of Measurement: Accuracy, Reliability, or Interobserver Agreement?

Assessments of the accuracy of measurement, the relia- bility of measurement, and the extent to which different observers obtain the same measures each provide differ- ent indications of data quality. Ultimately, the reason for conducting any type of assessment of measurement qual- ity is to obtain quantitative evidence that can be used for the dual purposes of improving measurement during the course of an investigation and judging and convincing others of the trustworthiness of the data.

After ensuring the validity of what they are measur- ing and how they are measuring it, applied behavior analysts should choose to assess the accuracy of measurement whenever possible rather than reliability or

interobserver agreement. If it can be determined that all measurements in a data set meet an acceptable accuracy criterion, questions regarding the reliability of measure- ment and interobserver agreement are moot. For data con- firmed to be accurate, conducting additional assessments of reliability or IOA is unnecessary.

When assessing the accuracy of measurement is not possible because true values are unavailable, an assess- ment of reliability provides the next best quality indica- tor. If natural or contrived permanent products can be archived, applied behavior analysts can assess the relia- bility of measurement, allowing consumers to know that observers have measured behavior consistently from ses- sion to session, condition to condition, and phase to phase.

When true values and permanent product archives are unavailable, interobserver agreement provides a level of believability for the data. Although IOA is not a direct indicator of the validity, accuracy, or reliability of mea- surement, it has proven to be a valuable and useful re- search tool in applied behavior analysis. Reporting interobserver agreement has been an expected and re- quired component of published research in applied be- havior analysis for several decades. In spite of its limitations, “the homely measures of observer agreement so widely used in the field are exactly relevant” (Baer, 1977, p. 119) to efforts to develop a robust technology of behavior change.

Percentage of agreement, in the interval-recording par- adigm, does have a direct and useful meaning: how often do two observers watching one subject, and equipped with the same definitions of behavior, see it occurring or not occurring at the same standard times? The two answers, “They agree about its occurrence X% of the relevant intervals, and about its nonoccurrence Y% of the relevant intervals,” are superbly useful. (Baer, 1977, p. 118)

There are no reasons to prevent researchers from using multiple assessment procedures to evaluate the same data set. When time and resources permit, it may even be desirable to include combinations of assessments. Applied behavior analysts can use any possible combi- nation of the assessment (e.g., accuracy plus IOA, relia- bility plus IOA). In addition, some aspects of the data set could be assessed for accuracy or reliability while other aspects are assessed with IOA. The previous example of accuracy assessment reported by Brown and colleagues (1996) included assessments for accuracy and IOA. In- dependent observers recorded correct and incorrect student-delayed retells. When IOA was less than 100%, data for that student and session were assessed for accu- racy. IOA was used as an assessment to enhance believ- ability, and also as a procedure for selecting data to be assessed for accuracy.

Improving and Assessing the Quality of Behavioral Measurement

142

Summary

Improving and Assessing the Quality of Behavioral Measurement

Indicators of Trustworthy Measurement

1. To be most useful for science, measurement must be valid, accurate, and reliable.

2. Valid measurement in ABA encompasses three equally im- portant elements: (a) measuring directly a socially signif- icant target behavior, (b) measuring a dimension of the target behavior relevant to the question or concern about the behavior, and (c) ensuring that the data are representa- tive of the behavior under conditions and during times most relevant to the reason(s) for measuring it.

3. Measurement is accurate when observed values, the data produced by measuring an event, match the true state, or true values, of the event.

4. Measurement is reliable when it yields the same values across repeated measurement of the same event.

Threats to Measurement Validity

5. Indirect measurement—measuring a behavior different from the behavior of interest—threatens validity because it requires that the researcher or practitioner make infer- ences about the relationship between the measures ob- tained and the actual behavior of interest.

6. A researcher who employs indirect measurement must provide evidence that the behavior measured directly re- flects, in some reliable and meaningful way, something about the behavior for which the researcher wishes to draw conclusions.

7. Measuring a dimension of the behavior that is ill suited for, or irrelevant to, the reason for measuring the behavior compromises validity.

8. Measurement artifacts are data that give an unwarranted or misleading picture of the behavior because of the way measurement was conducted. Discontinuous measure- ment, poorly scheduled observations, and insensitive or limiting measurement scales are common causes of mea- surement artifacts.

Threats to Measurement Accuracy and Reliability

9. Most investigations in applied behavior analysis use human observers to measure behavior, and human error is the biggest threat to the accuracy and reliability of data.

10. Factors that contribute to measurement error include poorly designed measurement systems, inadequate ob- server training, and expectations about what the data should look like.

11. Observers should receive systematic training and practice with the measurement system and meet predetermined ac- curacy and reliability criteria before collecting data.

12. Observer drift—unintended changes in the way an ob- server uses a measurement system over the course of an in-

vestigation—can be minimized by booster training ses- sions and feedback on the accuracy and reliability of measurement.

13. An observer’s expectations or knowledge about predicted or desired results can impair the accuracy and reliability of data.

14. Observers should not receive feedback about the extent to which their data confirm or run counter to hypothesized re- sults or treatment goals.

15. Measurement bias caused by observer expectations can be avoided by using naive observers.

16. Observer reactivity is measurement error caused by an observer’s awareness that others are evaluating the data he reports.

Assessing the Accuracy and Reliability of Behavioral Measurement

17. Researchers and practitioners who assess the accuracy of their data can (a) determine early in an analysis whether the data are usable for making experimental or treatment decisions, (b) discover and correct measurement errors, (c) detect consistent patterns of measurement error that can lead to the overall improvement or calibration of the measurement system, and (d) communicate to others the relative trustworthiness of the data.

18. Assessing the accuracy of measurement is a straightfor- ward process of calculating the correspondence of each measure, or datum, assessed to its true value.

19. True values for many behaviors of interest to applied be- havior analysts are evident and universally accepted or can be established conditionally by local context. True values for some behaviors (e.g., cooperative play) are difficult because the process for determining a true value must be different from the measurement procedures used to obtain the data one wishes to compare to the true value.

20. Assessing the extent to which observers are reliably ap- plying a valid and accurate measurement system provides a useful indicator of the overall trustworthiness of the data.

21. Assessing the reliability of measurement requires a nat- ural or contrived permanent product so the observer can re- measure the same behavioral events.

22. Although high reliability does not confirm high accuracy, discovering a low level of reliability signals that the data are suspect enough to be disregarded until problems in the mea- surement system can be determined and repaired.

Using Interobserver Agreement to Assess Behavioral Measurement

23. The most commonly used indicator of measurement qual- ity in ABA is interobserver agreement (IOA), the degree

143

to which two or more independent observers report the same observed values after measuring the same events.

24. Researchers and practitioners use measures of IOA to (a) determine the competence of new observers, (b) detect ob- server drift, (c) judge whether the definition of the target behavior is clear and the system not too difficult to use, and (d) convince others of the relative believability of the data.

25. Measuring IOA requires that two or more observers (a) use the same observation code and measurement system, (b) observe and measure the same participant(s) and events, and (c) observe and record the behavior indepen- dent of influence by other observers.

26. There are numerous techniques for calculating IOA, each of which provides a somewhat different view of the extent and nature of agreement and disagreement between observers.

27. Percentage of agreement between observers is the most common convention for reporting IOA in ABA.

28. IOA for data obtained by event recording can be calcu- lated by comparing (a) the total count recorded by each observer per measurement period, (b) the counts tallied by each observer during each of a series of smaller inter- vals of time within the measurement period, or (c) each observer’s count of 1 or 0 on a trial-by-trial basis.

29. Total count IOA is the simplest and crudest indicator of IOA for event recording data, and exact count-per-interval IOA is the most stringent for most data sets obtained by event recording.

30. IOA for data obtained by timing duration, response la- tency, or interresponse time (IRT) is calculated in essen- tially the same ways as for event recording data.

31. Total duration IOA is computed by dividing the shorter of the two durations reported by the observers by the longer duration. Mean duration-per-occurrence IOA is a more

conservative and usually more meaningful assessment of IOA for total duration data and should always be calculated for duration-per-occurrence data.

32. Three techniques commonly used to calculate IOA for in- terval data are interval-by-interval IOA, scored-interval IOA, and unscored-interval IOA.

33. Because it is subject to random or accidental agreement between observers, interval-by-interval IOA is likely to overestimate the degree of agreement between observers measuring behaviors that occur at very low or very high rates.

34. Scored-interval IOA is recommended for behaviors that occur at relatively low frequencies; unscored-interval IOA is recommended for behaviors that occur at relatively high frequencies.

35. IOA assessments should occur during each condition and phase of a study and be distributed across days of the week, times of day, settings, and observers.

36. Researchers should obtain and report IOA at the same lev- els at which they report and discuss the results of their study.

37. More stringent and conservative IOA methods should be used over methods that may overestimate agreement as a result of chance.

38. The convention for acceptable IOA has been a minimum of 80%, but there can be no set criterion. The nature of the behavior being measured and the degree of behavior change revealed by the data must be considered when de- termining an acceptable level of IOA.

39. IOA scores can be reported in narrative, table, and graphic form.

40. Researchers can use multiple indices to assess the quality of their data (e.g., accuracy plus IOA, reliability plus IOA).

Improving and Assessing the Quality of Behavioral Measurement

144

145

Constructing and Interpreting Graphic Displays of Behavioral Data

Key Terms

bar graph cumulative record cumulative recorder data data path dependent variable graph

independent variable level line graph local response rate overall response rate scatterplot semilogarithmic chart

split-middle line of progress Standard Celeration Chart trend variability visual analysis

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List ©, Third Edition

Content Area 7: Displaying and Interpreting Behavioral Data

7-1 Select a data display that effectively communicates quantitative relations.

7-2 Use equal-interval graphs.

7-3 Use Standard Celeration Charts (for BCBA only—excluded for BCABA).

7-4 Use a cumulative record to display data.

7-5 Use data displays that highlight patterns of behavior (e.g., scatterplot).

7-6 Interpret and base decision making on data displayed in various formats.

146

Applied behavior analysts document and quan- tify behavior change by direct and repeated measurement of behavior. The product of these

measurements, called data, is the medium with which behavior analysts work. In everyday usage the word data refers to a wide variety of often imprecise and subjec- tive information offered as facts. In scientific usage the word data means “the results of measurement, usually in quantified form” (Johnston & Pennypacker, 1993a, p. 365).1

Because behavior change is a dynamic and ongoing process, the behavior analyst—the practitioner and the researcher—must maintain direct and continuous con- tact with the behavior under investigation. The data ob- tained throughout a behavior change program or a research study are the means for that contact; they form the empirical basis for every important decision: to con- tinue with the present procedure, to try a different inter- vention, or to reinstitute a previous condition. But making valid and reliable decisions from the raw data themselves (a series of numbers) is difficult, if not impossible, and in- efficient. Inspecting a long row of numbers will reveal only very large changes in performance, or no change at all, and important features of behavior change can easily be overlooked.

Consider the three sets of data that follow; each con- sists of a series of numbers representing consecutive mea- sures of some target behavior. The first data set shows the results of successive measures of the number of responses emitted under two different conditions (A and B):

Condition A Condition B

120, 125, 115, 130, 114, 110, 115, 121,

126, 130, 123, 120, 110, 116, 107, 120,

120, 127 115, 112

Here are some data showing consecutive measures of the percentage of correct responses:

80, 82, 78, 85, 80, 90, 85, 85, 90, 92

The third data set consists of measures of responses per minute of a target behavior obtained on successive school days:

65, 72, 63, 60, 55, 68, 71, 65, 65, 62, 70, 75, 79, 63, 60

What do these numbers tell you? What conclusions can you draw from each data set? How long did it take you to reach your conclusions? How sure of them are you? What if the data sets contained many more mea- sures to interpret? How likely is it that others interested

1Although often used as a singular construction (e.g., “The data shows that . . .”), data is a plural noun of Latin origin and is correctly used with plural verbs (e.g., “These data are . . .”).

in the behavior change program or research study would reach the same conclusions? How could these data be di- rectly and effectively communicated to others?

Graphs—relatively simple formats for visually dis- playing relationships among and between a series of mea- surements and relevant variables—help people “make sense” of quantitative information. Graphs are the major device with which applied behavior analysts organize, store, interpret, and communicate the results of their work. Figure 1 includes a graph for each of the three data sets presented previously. The top graph reveals a lower level of responding during Condition B than during Condi- tion A. The middle graph clearly shows an upward trend

Trials

R es

p o

n se

s P

er M

in u

te F

re q

u en

Observations

School Days

P er

ce n

t C

o rr

ec t

Figure 1 Graphic displays of three sets of hypothet- ical data illustrating changes in the level of responding across conditions (top), trend (middle), and cyclical variability (bottom).

Constructing and Interpreting Graphic Displays of Behavioral Data

147

over time in the response measure. A variable pattern of responding, characterized by an increasing trend during the first part of each week and a decreasing trend toward the end of each week, is evident in the bottom graph. The graphs in Figure 1 illustrate three fundamental proper- ties of behavior change over time—level, trend, and vari- ability—each of which will be discussed in detail later in the chapter. The graphic display of behavioral data has proven an effective means of detecting, analyzing, and communicating these aspects of behavior change.

Purpose and Benefits of Graphic Displays of Behavioral Data Numerous authors have discussed the benefits of using graphs as the primary vehicle for interpreting and com- municating the results of behavioral treatments and re- search (e.g., Baer, 1977; Johnston & Pennypacker, 1993a; Michael, 1974; Parsonson, 2003; Parsonson & Baer, 1986, 1992; Sidman, 1960). Parsonson and Baer (1978) said it best:

In essence, the function of the graph is to communicate, in a readily assimilable and attractive manner, descrip- tions and summaries of data that enable rapid and accu- rate analysis of the facts. (p. 134)

There are at least six benefits of graphic display and visual analysis of behavioral data. First, plotting each measure of behavior on a graph right after the observa- tional period provides the practitioner or researcher with immediate access to an ongoing visual record of the par- ticipant’s behavior. Instead of waiting until the investi- gation or teaching program is completed, behavior change is evaluated continually, allowing treatment and experi- mental decisions to be responsive to the participant’s per- formance. Graphs provide the “close, continual contact with relevant outcome data” that can lead to “measurably superior instruction” (Bushell & Baer, 1994, p. 9).

Second, direct and continual contact with the data in a readily analyzable format enables the researcher as well as the practitioner to explore interesting variations in be- havior as they occur. Some of the most important research findings about behavior have been made because scien- tists followed the leads suggested by their data instead of following predetermined experimental plans (Sidman, 1960, 1994; Skinner, 1956).

Third, graphs, like statistical analyses of behavior change, are judgmental aids: devices that help the prac- titioner or experimenter interpret the results of a study or treatment (Michael, 1974). In contrast to the statistical tests of inference used in group comparison research, however, visual analysis of graphed data takes less time, is relatively easy to learn, imposes no predetermined or

2A comparison of the visual analysis of graphed data and inferences based on statistical tests of significance is presented in chapter entitled “Plan- ning and Evaluating Applied Behavior Analysis Research.” 3Graphs, like statistics, can also be manipulated to make certain interpre- tations of the data more or less likely. Unlike statistics, however, most forms of graphic displays used in behavior analysis provide direct access to the original data, which allows the inquisitive or doubtful reader to re- graph (i.e., manipulate) the data.

arbitrary level for determining the significance of be- havior change, and does not require the data to conform to certain mathematical properties or statistical assump- tions to be analyzed.

Fourth, visual analysis is a conservative method for determining the significance of behavior change. A be- havior change deemed statistically significant according to a test of mathematical probabilities may not look very impressive when the data are plotted on a graph that re- veals the range, variability, trends, and overlaps in the data within and across experimental or treatment condi- tions. Interventions that produce only weak or unstable effects are not likely to be reported as important findings in applied behavior analysis. Rather, weak or unstable effects are likely to lead to further experimentation in an effort to discover controlling variables that produce mean- ingful behavior change in a reliable and sustained man- ner. This screening out of weak variables in favor of robust interventions has enabled applied behavior ana- lysts to develop a useful technology of behavior change (Baer, 1977).2

Fifth, graphs enable and encourage independent judgments and interpretations of the meaning and sig- nificance of behavior change. Instead of having to rely on conclusions based on statistical manipulations of the data or on an author’s interpretations, readers of published re- ports of applied behavior analysis can (and should) con- duct their own visual analysis of the data to form independent conclusions.3

Sixth, in addition to their primary purpose of dis- playing relationships between behavior change (or lack thereof) and variables manipulated by the practitioner or researcher, graphs can also be effective sources of feed- back to the people whose behavior they represent (e.g., DeVries, Burnettte, & Redmon, 1991; Stack & Milan, 1993). Graphing one’s own performance has also been demonstrated to be an effective intervention for a vari- ety of academic and behavior change objectives (e.g., Fink & Carnine, 1975; Winette, Neale, & Grier, 1979).

Types of Graphs Used in Applied Behavior Analysis Visual formats for the graphic display of data most often used in applied behavior analysis are line graphs, bar graphs, cumulative records, semilogarithmic charts, and scatterplots.

Constructing and Interpreting Graphic Displays of Behavioral Data

148

Line Graphs

The simple line graph, or frequency polygon, is the most common graphic format for displaying data in applied behavior analysis. The line graph is based on a Cartesian plane, a two-dimensional area formed by the intersection of two perpendicular lines. Any point within the plane represents a specific relationship between the two di- mensions described by the intersecting lines. In applied behavior analysis, each point on a line graph shows the level of some quantifiable dimension of the target be- havior (i.e., the dependent variable) in relation to a spec- ified point in time and/or environmental condition (i.e., the independent variable) in effect when the measure was taken. Comparing points on the graph reveals the presence and extent of changes in level, trend, and/or variability within and across conditions.

Parts of a Basic Line Graph

Although graphs vary considerably in their final appear- ance, all properly constructed line graphs share certain elements. The basic parts of a simple line graph are shown in Figure 2 and described in the following sections.

1. Horizontal Axis. The horizontal axis, also called the x axis, or abscissa, is a straight horizontal line that most often represents the passage of time and the pres- ence, absence, and/or value of the independent variable. A defining characteristic of applied behavior analysis is the repeated measurement of behavior across time. Time is also the unavoidable dimension in which all manipulations of the independent variable occur. On

most line graphs the passage of time is marked in equal intervals on the horizontal axis. In Figure 2 succes- sive 10-minute sessions during which the number of property destruction responses (including attempts) was measured are marked on the horizontal axis. In this study, 8 to 10 sessions were conducted per day (Fisher, Lindauer, Alterson, & Thompson, 1998).

The horizontal axis on some graphs represents dif- ferent values of the independent variable instead of time. For example, Lalli, Mace, Livezey, and Kates (1998) scaled the horizontal axis on one graph in their study from less than 0.5 meters to 9.0 meters to show how the occurrence of self-injurious behavior by a girl with severe mental retardation decreased as the distance between the therapist and the girl increased.

2. Vertical Axis. The vertical axis, also called the y axis, or ordinate, is a vertical line drawn upward from the left-hand end of the horizontal axis. The vertical axis most often represents a range of values of the dependent variable, which in applied behavior analysis is always some quantifiable dimension of behavior. The intersec- tion of the horizontal and vertical axes is called the origin and usually, though not necessarily, represents the zero value of the dependent variable. Each succes- sive point upward on the vertical axis represents a greater value of the dependent variable. The most com- mon practice is to mark the vertical axis with an equal- interval scale. On an equal-interval vertical axis equal distances on the axis represent equal amounts of behav- ior. The vertical axis in Figure 2 represents the number of property destruction responses (and attempts) per minute with a range of 0 to 4 responses per minute.

P ro

p er

ty D

es tr

u ct

io n

a n

d A

tt em

p te

d P

ro p

er ty

D es

tr u

ct io

n p

er M

in u

5 10 15 20 25 30

Milo

6 5

Baseline Blocking Baseline Blocking Baseline Blocking

Sessions

Rates of property destruction (plus attempts) during baseline and the blocking condition for Milo.

Figure 2 The major parts of a simple line graph: (1) horizontal axis, (2) vertical axis, (3) condition change lines, (4) condi- tion labels, (5) data points, (6) data path, and (7) figure caption.

From “Assessment and Treatment of Destructive Behavior Maintained by Stereotypic Object Manipulation” by W. W. Fisher, S. E. Lindauer, C. J. Alterson, and R. H. Thompson, 1998, Journal of Applied Behavior Analysis, 31, p. 522. Copyright 1998 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Constructing and Interpreting Graphic Displays of Behavioral Data

149

3. Condition Change Lines. Condition change lines are vertical lines drawn upward from the horizontal axis to show points in time at which changes in the in- dependent variable occurred. The condition change lines in Figure 2 coincide with the introduction or withdrawal of an intervention the researchers called blocking. Condition change lines can be drawn as solid or dashed lines. When relatively minor changes occur within an overall condition, dashed vertical lines should be used to distinguish minor changes from major changes in conditions, which are shown by solid lines (see Figure 18).

4. Condition Labels. Condition labels, in the form of single words or brief descriptive phrases, are printed along the top of the graph and parallel to the horizontal axis. These labels identify the experimental conditions (i.e., the presence, absence, or some value of the inde- pendent variable) that are in effect during each phase of the study.4

5. Data Points. Each data point on a graph repre- sents two facts: (a) a quantifiable measure of the target behavior recorded during a given observation period and (b) the time and/or experimental conditions under which that particular measurement was conducted. Using two data points from Figure 2 as examples, we can see that during Session 5, the last session of the first baseline phase, property destruction and attempted property destruction responses occurred at a rate of ap- proximately 2 responses per minute; and in Session 9, the fourth session of the first blocking phase, 0 in- stances of the target behavior were recorded.

6. Data Path. Connecting successive data points within a given condition with a straight line creates a data path. The data path represents the level and trend of behavior between successive data points, and it is a primary focus of attention in the interpretation and analysis of graphed data. Because behavior is rarely observed and recorded continuously in applied behav- ior analysis, the data path represents an estimate of the actual course taken by the behavior during the time elapsed between the two measures. The more mea- surements and resultant data points per unit of time (given an accurate observation and recording system),

4The terms condition and phase are related but not synonymous. Properly used, condition indicates the environmental arrangements in effect at any given time; phase refers to a period of time within a study or behavior- change program. For example, the study shown in Figure 2 consisted of two conditions (baseline and blocking) and six phases.

the more confidence one can place in the story told by the data path.

7. Figure Caption. The figure caption is a concise statement that, in combination with the axis and condi- tion labels, provides the reader with sufficient informa- tion to identify the independent and dependent variables. The figure caption should explain any symbols or ob- served but unplanned events that may have affected the dependent variable (see Figure 6) and point out and clarify any potentially confusing features of the graph (see Figure 7).

Variations of the Simple Line Graph: Multiple Data Paths

The line graph is a remarkably versatile vehicle for dis- playing behavior change. Whereas Figure 2 is an ex- ample of the line graph in its simplest form (one data path showing a series of successive measures of behav- ior across time and experimental conditions) by the ad- dition of multiple data paths, the line graph can display more complex behavior–environment relations. Graphs with multiple data paths are used frequently in applied behavior analysis to show (a) two or more dimensions of the same behavior, (b) two or more different behaviors, (c) the same behavior under different and alternating ex- perimental conditions, (d) changes in target behavior rel- ative to the changing values of an independent variable, and (e) the behavior of two or more participants.

Two or More Dimensions of the Same Behavior. Showing multiple dimensions of the dependent variable on the same graph enables visual analysis of the ab- solute and relative effects of the independent variable on those dimensions. Figure 3 shows the results of a study of the effects of training three members of a women’s college basketball team proper foul shooting form (Kladopoulos & McComas, 2001). The data path created by connecting the open triangle data points shows changes in the percentage of foul shots executed with the proper form, whereas the data path connecting the solid data points reveals the percentage of foul shots made. Had the experimenters recorded and graphed only the players’ foul shooting form, they would not have known whether any improvements in the target behavior on which training was focused (correct foul shooting form) coincided with improvements in the be- havior by which the social significance of the study would ultimately be judged—foul shooting accuracy. By measuring and plotting both form and outcome on the same graph, the experimenters were able to analyze the effects of their treatment procedures on two critical dimensions of the dependent variable.

Constructing and Interpreting Graphic Displays of Behavioral Data

150

100

Baseline Form Training

Participant 1 foul shot data form data

100

0 Participant 2

100

0 Participant 3

Sessions 5 1510 2520

P er

ce n

ta ge

o f

Sh o

ts M

ad e

P er

ce n

ta ge

o f

C o

rr ec

t T

ri al

s Figure 1. Percentage of shots made (filled circles) and percentage of shots taken with correct form (open triangles) across sessions for each participant.

Figure 3 Graph using multiple data paths to show the effects of the independent variable (Form Training) on two dimensions (accuracy and topography) of the target behavior.

From “The Effects of Form Training on Foul- Shooting Performance in Members of a Women’s College Basketball Team” by C. N. Kladopoulos and J. J. McComas, 2001, Journal of Applied Behavior Analysis, 34, p. 331. Copyright 2001 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Two or More Different Behaviors. Multiple data paths are also used to facilitate the simultaneous com- parison of the effects of experimental manipulations on two or more different behaviors. Determining the co- variation of two behaviors as a function of changes in the independent variable is accomplished more easily if both can be displayed on the same set of axes. Figure 4 shows the percentage of intervals in which a boy with autism exhibited stereotypy (e.g., repetitive body movements, rocking) across three conditions and the number of times that he raised his hand for attention (in the attention condition), signed for a break (in the de- mand condition), and signed for access to preferred tan- gible stimuli (in the no-attention condition) in a study investigating a strategy called functional communica- tion training (Kennedy, Meyer, Knowles, & Shukla, 2000).5 By recording and graphing both stereotypic re- sponding and appropriate behavior, the investigators were able to determine whether increases in alternative communication responses (raising his hand and sign- ing) were accompanied by reductions in stereotypy. Note that a second vertical axis is used on Figure 4 to

5Functional communication training is described in chapter entitled “Antecedent Interventions.”

show the proper dimensional units and scaling for sign- ing frequency. Because of the differences in scale, read- ers of dual-vertical axis graphs must view them with care, particularly when assessing the magnitude of be- havior change.

Measures of the Same Behavior under Different Conditions. Multiple data paths are also used to repre- sent measures of the same behavior taken under differ- ent experimental conditions that alternate throughout an experimental phase. Figure 5 shows the number of self-injurious response per minute by a 6-year-old girl with developmental disabilities under four different conditions (Moore, Mueller, Dubard, Roberts, & Sterling- Turner, 2002). Graphing an individual’s behavior under multiple conditions on the same set of axes allows di- rect visual comparisons of differences in absolute levels of responding at any given time as well as relative changes in performance over time.

Changing Values of an Independent Variable. Multiple data path graphs are also used to show changes in the target behavior (shown on one data path) relative to changing values of the independent variable (repre- sented by a second data path). In each of the two graphs

Constructing and Interpreting Graphic Displays of Behavioral Data

151

100

Attention

100

Demand

Signing

Stereotypy

P er

ce n

ta ge

o f

In te

rv al

s o

f St

er eo

ty p

100

No Attention 30

0 5 10 15 20 25 30 35

BL Functional Communication Training

Sessions

Si gn

in g

F re

q u

en cy

Figure 2. Occurrence of stereotypy for James across attention, demand, and no-attention conditions. Data are arrayed as the percentage of intervals of stereotypy on the left y axis and number of signs per sessions on the right y axis.

Figure 4 Graph with multiple data paths showing two different behaviors by one participant during baseline and training across three different conditions. Note the different dimensions and scaling of the dual vertical axes.

From “Analyzing the Multiple Functions of Stereotypical Behavior for Students with Autism: Implications for Assessment and Treatment” by C. H. Kennedy, K. A. Meyer, T. Knowles, and S. Shukla, 2000, Journal of Applied Behavior Analysis, 33, p. 565. Copyright 2000 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

in Figure 6 one data path shows the duration of prob- lem behavior (plotted against the left-hand y axis scaled in seconds) relative to changes in noise level, which are depicted by the second data path (plotted against the right-hand y axis scaled in decibels) (McCord, Iwata, Galensky, Ellingson, & Thomson, 2001).

The Same Behavior of Two or More Participants. Multiple data paths are sometimes used to show the be- havior of two or more participants on the same graph.

Depending on the levels and variability of the data encompassed by each data path, a maximum of four dif- ferent data paths can be displayed effectively on one set of axes. However, there is no rule; Didden, Prinsen, and Sigafoos displayed five data paths in a single display (2000, p. 319). If too many data paths are displayed on the same graph, the benefits of making additional compar- isons may be outweighed by the distraction of too much visual “noise.” When more than four data paths must be

6A superb example of combining visual display techniques is Charles Mi- nard’s use of space-time-story graphics to illustrate the interrelations of six variables during Napoleon’s ill-fated Russian campaign of 1812–1813 (see Tufte, 1983, p. 41). Tufte called Minard’s graph perhaps “the best sta- tistical graphic ever drawn” (p. 40).

included on the same graph, other methods of display can be incorporated.6 For example, Gutowski and Stromer (2003) effectively used striped and shaded bars in com- bination with conventional data paths to display the num- ber of names spoken and the percentage of correct matching-to-sample responses by individuals with men- tal retardation (see Figure 7).

Bar Graphs

The bar graph, or histogram, is a simple and versatile format for graphically summarizing behavioral data. Like the line graph, the bar graph is based on the Cartesian

Constructing and Interpreting Graphic Displays of Behavioral Data

152

R es

p o

n se

s p

er M

in u

4 8 12 16 20 24 28 32 36 Sessions

Attention

Tangible

Play Demand

Figure 1. Rate of self-injurious behavior during the initial functional analysis.

Figure 5 Graph with multiple data paths showing the same behavior measured under four different conditions.

From “The Influence of Therapist Attention on Self-Injury during a Tangible Condition” by J. W. Moore, M. M. Mueller, M. Dubard, D. S. Roberts, and H. E. Sterling-Turner, 2002, Journal of Applied Behavior Analysis, 35, p. 285. Copyright 2002 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Noise

Problem behavior

Debbie

BL (EXT) EXT + Stimulus Fading

Sarah

20 40 60 80 100 120 140

DRO

A BF

Sessions

D u

ra ti

o n

o f

P ro

b le

m B

eh av

io r

(s ec

)

N o

is e

Le ve

l (d

B )

Figure 4. Results of Debbie’s and Sarah’s treatment evaluation. Sessions marked A and B near the end of treatment indicate two generalization probes in the natural environment; F indicates a follow-up probe.

A BF

Figure 6 Graph using two data paths to show the duration of problem behavior (dependent variable) by two adults with severe or profound mental retardation as noise level was increased gradually (independent variable).

From “Functional Analysis and Treatment of Problem Behavior Evoked by Noise” by B. E. McCord, B. A. Iwata, T. L. Galensky, S. A. Ellingson, and R. J. Thomson, 2001, Journal of Applied Behavior Analysis, 34, p. 457. Copyright 2001 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

plane and shares most of the line graph’s features with one primary difference: The bar graph does not have dis- tinct data points representing successive response mea- sures through time. Bar graphs can take a wide variety of forms to allow quick and easy comparisons of perfor- mance across participants and/or conditions.

Bar graphs serve two major functions in applied be- havior analysis. First, bar graphs are used for displaying and comparing discrete sets of data that are not related to one another by a common underlying dimension by which the horizontal axis can be scaled. For example, Gottschalk, Libby, and Graff (2000), in a study analyzing

Constructing and Interpreting Graphic Displays of Behavioral Data

153

100

60 2 10 17

Repeating Names

Matching to Two Dictated Names

Matching to Two Pictures

No Prompt Sm (1)

No Prompt 0–s Delay (Set 1)

No Prompt 1-s Delay (Set 1)

100

Prompt 1–s Delay (Set 1)

No Prompt 1–s Delay (Set 1)

Prompt 1–s (1)

No Prompt 1–s Delay (Set 1)

No Prompt 1–s Del (Set 2)

No Pt 1–s (3)

24 (58) 30 32 (50) 41 46 48

No Prompt Sm (3)

No Prompt 0–s Delay (Set 3)

Prompt 0–s Del (Set 3)

Olivia

Dan

2 (50) 9 (33)(58) (42 25)

(50 33)

100

25 28 37 40 44 47 49

P er

ce n

ta ge

C o

rr ec

t M

at ch

in g

to S

am p

Sessions

N u

m b

er o

f N

am es

N u

m b

er o

f N

am es

No Prompt 0–s Delay (Set 3)

Prompt 0–s (3)

No Prompt 0–s Delay (Set 3)

Prompt 0–s (3)

No Prompt 0–s (Set 3)

No Pt 0–s (2)

No Pt 0–s (1)

(58) (42)

Naming Pictures

Figure 4. Results for Olivia and Dan across simultaneous, delay, prompt, and no-prompt conditions: open circles and solid squares reflect percentages of correct matching. Striped bars and shaded bars reflect the number of names spoken on trials with two-name and two-picture samples, respectively. Bars with extended tic marks on the abscissa indicate that the number of names exceeded 25.

Figure 7 Graph using a combination of bars and data points to display changes in two response classes to two types of matching stimuli under different prompting conditions.

From “Delayed Matching to Two- Picture Samples by Individuals With and Without Disabilities: An Analysis of the Role of Naming” by S. J. Gutowski and Robert Stromer, 2003, Journal of Applied Behavior Analysis, 36, p. 498. Copyright 2003 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

the effects of establishing operations on preference as- sessments, used bar graphs to show the percentage of tri- als in which four children reached toward and picked up different items (see Figure 8).

Another common use of bar graphs is to give a visual summary of the performance of a participant or group of participants during the different conditions of an ex- periment. For example, Figure 9 shows the mean per- centage of spelling worksheet items completed and the mean percentage of completed items that were done cor- rectly by four students during baseline and combined generalization programming and maintenance conditions

that followed training each child how to recruit teacher attention while they were working (Craft, Alber, & Heward, 1998).

Bar graphs sacrifice the presentation of the variabil- ity and trends in behavior (which are apparent in a line graph) in exchange for the efficiency of summarizing and comparing large amounts of data in a simple, easy-to- interpret format. They should be viewed with the under- standing that they may mask important variability in the data. Although bar graphs are typically used to present a measure of central tendency, such as the mean or median score for each condition, the range of measures repre-

Constructing and Interpreting Graphic Displays of Behavioral Data

154

100

0 oyster cracker graham cracker gummi candy Twinkie

Satiation Control Deprivation

Ethan

P er

ce n

ta ge

o f

T ri

al s

w it

h A

p p

ro ac

h R

es p

o n

se s

Stimulus

100

0 jellybean popcorn Cheez-It cracker

Satiation Control Deprivation

Daniel

P er

ce n

ta ge

o f

T ri

al s

w it

h A

p p

ro ac

h R

es p

o n

se s

Stimulus

100

0 Skittles licorice jellybeans Reese's Pieces

Satiation Control Deprivation

Mark

P er

ce n

ta ge

o f

T ri

al s

w it

h A

p p

ro ac

h R

es p

o n

se s

Stimulus

100

0 jellybeans licorice Necco Wafers Reese's Pieces

Satiation Control Deprivation

Ashley

P er

ce n

ta ge

o f

T ri

al s

w it

h A

p p

ro ac

h R

es p

o n

se s

Stimulus Figure 1. Percentage of approach responses across conditions for Ethan, Daniel, Mark, and Ashley.

Figure 8 Bar graph used to summarize and display results of measurements taken under discrete conditions lacking an underlying dimension by which the horizontal axis could be scaled (e.g., time, duration of stimulus presentations).

From “The Effects of Establishing Operations on Preference Assessment Outcomes” by J. M. Gottschalk, M. E. Libby, and R. B Graff, 2000, Journal of Applied Behavior Analysis, 33, p. 87. Copyright 2000 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

sented by the mean can also be incorporated into the dis- play (e.g., see Figure 5 in Lerman, Kelley, Vorndran, Kuhn, & LaRue, 2002).

Cumulative Records

The cumulative record (or graph) was developed by Skinner as the primary means of data collection in the experimental analysis of behavior. A device called the cumulative recorder enables a subject to actually draw its own graph (see Figure 10). In a book cataloging 6 years of experimental research on schedules of rein- forcement, Ferster and Skinner (1957) described cumu- lative records in the following manner:

A graph showing the number of responses on the ordi- nate against time on the abscissa has proved to be the most convenient representation of the behavior observed in this research. Fortunately, such a “cumulative” record may be made directly at the time of the experiment. The record is raw data, but it also permits a direct inspection of rate and changes in rate not possible when the behav- ior is observed directly. . . . Each time the bird responds, the pen moves one step across the paper. At the same time, the paper feeds continuously. If the bird does not respond at all, a horizontal line is drawn in the direction of the paper feed. The faster the bird pecks, the steeper the line. (p. 23)

When cumulative records are plotted by hand or cre- ated with a computer graphing program, which is most

Constructing and Interpreting Graphic Displays of Behavioral Data

155

100

0 Baseline (6) Gen./Main. (24)

Latasha 100

0 Baseline (14) Gen./Main. (18)

Olivia

100

0 Baseline (16) Gen./Main. (16)

Octavian 100

0 Baseline (26) Gen./Main. (8)

Kenny% Completion

% Accuracy

P er

ce n

t C

o m

p le

ti o

n /A

cc u

ra cy

Experimental Conditions

Figure 4. Mean percentage of spelling worksheet items completed and mean percentage of accuracy by each student during baseline and combined generalization programming and maintenance conditions. Numbers in parentheses show total number of sessions per condition.

Figure 9 Bar graph comparing mean levels for two dimensions of participants’ performance between experimental conditions.

From “Teaching Elementary Students with Developmental Disabilities to Recruit Teacher Attention in a General Education Classroom: Effects on Teacher Praise and Academic Productivity” by M. A. Craft, S. R. Alber, and W. L. Heward, 1998, Journal of Applied Behavior Analysis, 31, p. 410. Copyright 1998 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Figure 10 Diagram of a cumulative recorder.

often the case in applied behavior analysis, the number of responses recorded during each observation period is added (thus the term cumulative) to the total number of responses recorded during all previous observation peri- ods. In a cumulative record the y axis value of any data point represents the total number of responses recorded since the beginning of data collection. The exception oc- curs when the total number of responses has exceeded the upper limit of the y axis scale, in which case the data path on a cumulative curve resets to the 0 value of the

y axis and begins its ascent again. Cumulative records are almost always used with frequency data, although other dimensions of behavior, such as duration and la- tency, can be displayed cumulatively.

Figure 11 is an example of a cumulative record from the applied behavior analysis literature (Neef, Iwata, & Page, 1980). It shows the cumulative number of spelling words mastered by a person with mental retar- dation during baseline and two training conditions. The graph shows that the individual mastered a total of 1 word during the 12 sessions of baseline (social praise for cor- rect spelling responses and rewriting incorrectly spelled words three times), a total of 22 words under the inter- spersal condition (baseline procedures plus the presen- tation of a previously learned word after each unknown word), and a total of 11 words under the high-density re- inforcement condition (baseline procedures plus social praise given after each trial for task-related behaviors such as paying attention and writing neatly).

In addition to the total number of responses recorded at any given point in time, cumulative records show the overall and local response rates. Rate is the number of responses emitted per unit of time, usually reported as responses per minute in applied behavior analysis. An overall response rate is the average rate of response over a given time period, such as during a specific session, phase, or condition of an experiment. Overall rates are

Constructing and Interpreting Graphic Displays of Behavioral Data

156

Sessions

C u

m u

la ti

ve N

u m

b er

o f

W o

rd s

M ea

su re

d Figure 11 Cumulative graph of number of spelling words learned by a man with mental retardation during baseline, interspersal training, and high-density reinforcement training. Points a–e have been added to illus- trate the differences between overall and local response rates.

From “The Effects of Interspersal Training Versus High Density Reinforcement on Spelling Acquisition and Retention” by N. A. Neef, B. A. Iwata, and T. J. Page, 1980, Journal of Applied Behavior Analysis, 13, p. 156. Copyright 1980 by the Society for the Experimental Analysis of Behavior, Inc. Adapted by permission.

calculated by dividing the total number of responses recorded during the period by the number of observation periods indicated on the horizontal axis. In Figure 11 the overall response rates are 0.46 and 0.23 words mas- tered per session for the interspersal and high-density re- inforcement conditions, respectively.7

On a cumulative record, the steeper the slope, the higher the response rate. To produce a visual representa- tion of an overall rate on a cumulative graph, the first and last data points of a given series of observations should be connected with a straight line. A straight line con- necting Points a and c in Figure 11 would represent the learner’s overall rate of mastering spelling words during the interspersal condition. A straight line connecting Points a and e represents the overall rate during the high- density reinforcement condition. Relative rates of re- sponse can be determined by visually comparing one slope to another; the steeper the slope, the higher the rate of response. A visual comparison of Slopes a–c and a–e shows that the interspersal condition produced the higher overall response rate.

Response rates often fluctuate within a given period. The term local response rate refers to the rate of re- sponse during periods of time smaller than that for which an overall rate has been given. Over the last four ses- sions of the study shown in Figure 11, the learner ex- hibited a local rate of responding during interspersal

7Technically, Figure 11 does not represent true rates of response because the number of words spelled correctly was measured and not the rate, or speed, at which they were spelled. However, the slope of each data path rep- resents the different ”rates” of mastering the spelling words in each session within the context of a total of 10 new words presented per session.

training (Slope b–c) that was considerably higher than his overall rate for that condition. At the same time his performance during the final four sessions of the high- density reinforcement condition (Slope d–e) shows a lower local response rate than his overall rate for that condition.

A legend giving the slopes of some representative rates can aid considerably in the determination and com- parison of relative response rates both within and across cumulative curves plotted on the same set of axes (e.g., see Kennedy & Souza, 1995, Figure 2). However, very high rates of responding are difficult to compare visually with one another on cumulative records.

Although the rate of responding is directly proportional to the slope of the curve, at slopes above 80 degrees small differences in angle represent very large differ- ences in rate; and although these can be measured ac- curately, they cannot be evaluated easily by [visual] inspection. (Ferster & Skinner, 1957, pp. 24–25)

Even though cumulative records derived from con- tinuous recording are the most directly descriptive dis- plays of behavioral data available, two other features of behavior, in addition to the comparison of very high rates, can be difficult to determine on some cumulative graphs. One, although the total number of responses since data collection began can be easily seen on a cumulative graph, the number of responses recorded for any given session can be hard to ascertain, given the number of data points and the scaling of the vertical axis. Two, gradual changes in slope from one rate to another can be hard to detect on cumulative graphs.

Constructing and Interpreting Graphic Displays of Behavioral Data

157

Days

C u

m u

la ti

ve N

u m

b er

o f

R es

p o

n se

Baseline1 Intervention1 Baseline2 Intervention2

Figure 12 Same set of hypothetical data plotted on noncumulative and cumulative graphs. Cumulative graphs more clearly reveal patterns of and changes in respond- ing for behaviors that can occur only once during each period of measurement.

From Working with Parents of Handicapped Children, p. 100, by W. L. Heward, J. C. Dardig, and A. Rossett, 1979, Columbus, OH: Charles E. Merrill. Copyright 1979 by Charles E. Merrill. Used by permission.

Four situations in which a cumulative graph may be preferable to a noncumulative line graph are as follows. First, cumulative records are desirable when the total number of responses made over time is important or when progress toward a specific goal can be measured in cu- mulative units of behavior. The number of new words learned, dollars saved, or miles trained for an upcoming marathon are examples. One look at the most recent data point on the graph reveals the total amount of behavior up to that point in time.

Second, a cumulative graph might also be more ef- fective than noncumulative graphs when the graph is used as a source of feedback for the participant. This is be- cause both total progress and relative rate of performance are easily detected by visual inspection (Weber, 2002).

Third, a cumulative record should be used when the target behavior is one that can occur or not occur only once per observation session. In these instances the ef- fects of any intervention are easier to detect on a cumula- tive graph than on a noncumulative graph. Figure 12 shows the same data plotted on a noncumulative graph and a cumulative graph. The cumulative graph clearly shows a relation between behavior and intervention, whereas the noncumulative graph gives the visual im- pression of greater variability in the data than really exists.

Fourth, cumulative records can “reveal the intricate relations between behavior and environmental variables” (Johnston & Pennypacker, 1993a, p. 317) Figure 13 is

an excellent example of how a cumulative graph enables a detailed analysis of behavior change (Hanley, Iwata, & Thompson, 2001). By plotting the data from single ses- sions cumulatively by 10-second intervals, the researchers revealed patterns of responding not shown by a graph of session-by-session data. Comparing the data paths for the three sessions for which the results were graphed cu- mulatively (Mult #106, Mixed #107, and Mixed #112) revealed two undesirable patterns of responding (in this case, pushing a switch that operated a voice output device that said “talk to me, please” as an alternative to self- injurious behavior and aggression) that are likely to occur during mixed schedules, and the benefits of including schedule-correlated stimuli (Mult #106).

Semilogarithmic Charts

All of the graphs discussed so far have been equal-inter- val graphs on which the distance between any two con- secutive points on each axis is always the same. On the x axis the distance between Session 1 and Session 2 is equal to the distance between Session 11 and Session 12; on the y axis, the distance between 10 and 20 responses per minute is equal to the distance between 35 and 45 re- sponses per minute. On an equal-interval graph equal ab- solute changes in behavior, whether an increase or decrease in performance, are expressed by equal distances on the y axis.

Constructing and Interpreting Graphic Displays of Behavioral Data

158

200

150

100

A lt

er n

at iv

e R

es p

o n

se s

(C u

m u

la ti

ve #

)

6 12 18 24 30 36 42 48 54 60

Mult #106

Mixed #107

Mixed #112

FR 1 EXT FR 1 EXT FR 1 EXT FR1

Julie

10-S Bins

Figure 4. Cumulative number of alternative responses across schedule components for three sessions from Julie’s assessment of schedule-correlated stimuli. Open symbols represent two mixed-schedule sessions; filled symbols represent a single multiple-schedule session.

Figure 13 Cumulative record used to make a detailed analysis and comparison of behavior across components of multiple- and mixed-reinforcement schedules within specific sessions of a study.

From “Reinforcement Schedule Thinning Following Treatment with Functional Com- munication Training” by G. P. Hanley, B. A. Iwata, and R. H. Thompson, 2001, Journal of Applied Behavior Analysis, 34, p. 33. Copy- right 2001 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Time Equal-Distance y Axis

Time Logarithmic y Axis

(Log Base 2)

R es

p o

n se

R at

R es

p o

n se

R at

Figure 14 Same set of data plotted on equal-interval arithmetic scale (left) and on equal-proportion ratio scale (right).

Another way of looking at behavior change is to ex- amine proportional or relative change. Logarithmic scales are well suited to display and communicate pro- portional change. On a logarithmic scale equal relative changes in the variable being measured are represented by equal distances. Because behavior is measured and charted over time, which progresses in equal intervals, the x axis is marked off in equal intervals and only the y axis is scaled logarithmically. Hence, the term semi- logarithmic chart refers to graphs in which only one axis is scaled proportionally.

On semilog charts all behavior changes of equal pro- portion are shown by equal vertical distances on the ver- tical axis, regardless of the absolute values of those changes. For example, a doubling of response rate from 4 to 8 per minute would appear on a semilogarithmic chart as the same amount of change as a doubling of 50 to 100 responses per minute. Likewise, a decrease in re- sponding from 75 to 50 responses per minute (a decrease

of one third) would occupy the same distance on the ver- tical axis as a change from 12 to 8 responses per minute (a decrease of one third).

Figure 14 shows the same data graphed on an equal-interval chart (sometimes called arithmetic or add- subtract charts) and on a semilogarithmic chart (some- times called ratio or multiply-divide charts). The behavior change that appears as an exponential curve on the arith- metic chart is a straight line when plotted on the semilog chart. The vertical axis in the semilog chart in Figure 14 is scaled by log-base-2 or X2 cycles, which means that each cycle going up on the y axis represents a times-2 in- crease (i.e., a doubling) of the cycle below it.

Standard Celeration Charts

In the 1960s, Ogden Lindsley developed the Standard Celeration Chart to provide a standardized means of charting and analyzing how frequency of behavior

Constructing and Interpreting Graphic Displays of Behavioral Data

159

x 16

x 4

x 2

x 1.4 x 1.0

per week 1000

100

2 Sep 01 Dy Mo Yr

30 Sep 01 Dy Mo Yr

28 Oct 01 Dy Mo Yr

25 Nov 01 Dy Mo Yr

23 Dec 01 Dy Mo Yr

20 Jan 02 Dy Mo Yr

Calendar Weeks

Likeness of Daily per minute Standard Celeration Chart

Actual Charts available from BEHAVIOR RESEARCH CO.

BOX 3351 - KANSAS CITY, KS 65103-3351 FAX ORDERS 913 362

5900

0 4 8 12 16 20

7. 10

5. 2.4.

Calendar synchronization

Phase change line

Deceleration target frequency

Acceleration target frequency

Charted day

Change indicator

Ignored day

Slice back to 50 basic addition problems

Celeration line

Aim star

Counting-time bar No-chance day Zero performance

Aim star

C o

u n

t p

er M

in u

The name of the person who sees the performer's chart on a monthly basis. The person may give advice to the Adviser or Manager.

The name of the person who advises the manager or performer on a weekly basis.

The name of the person who works with the performer on a daily basis.

The name of the person who charts the performer's counted behavior.

OPTIONAL: The age of the performer when the chart begins. If not filled in, draw a line through the space.

The name of the person whose performance appears on the chart.

A clear description of the performer's counted behavior. Use a learning channel and active verb/noun (e.g., see/say reading books).

0 14 28 42 56 70 84 98 112 126 140

Successive Calendar Days

SUPERVISOR

ORGANIZATION

ADVISOR

DIVISION

MANAGER

ROOM TIMER COUNTER

PERFORMER

CHARTER

COUNTED

Labelled Blanks (Adapted from Pennypacker, Gutierrez, & Lindsley. 2003)

The name of the organization where the counted behavior takes place.

The name of the division of the organization.

The room where the counting occurs.

The name of the person who times the performer.

The name of the person who counts the performer's behavior.

OPTIONAL: Any additional information relevant to the performer or chart. If not filled in, draw a line through the space.

Figure 15 Standard Celeration Chart showing basic charting conventions. See Table 1 for explanation.

changes over time (Lindsley, 1971; Pennypacker, Gutier- rez & Lindsley, 2003). The Standard Celeration Chart is a semilogarithmic chart with six X10 cycles on the ver- tical axis that can accommodate response rates as low as 1 per 24 hours (0.000695 per minute) or as high as 1,000 per minute.

There are four standard charts, differentiated from one another by the scaling on the horizontal axis: a daily chart with 140 calendar days, a weekly chart, a monthly chart, and a yearly chart. The daily chart shown in Figure 15 is used most often. Table 1 describes major parts of the Standard Celeration Chart and basic charting conventions.

The size of the chart and the consistent scaling of the y axis and x axis do not make the Standard Celeration Chart standard, as is commonly believed. What makes

the Standard Celeration Chart standard is its consistent display of celeration, a linear measure of frequency change across time, a factor by which frequency multi- plies or divides per unit of time. The terms acceleration and deceleration are used to describe accelerating per- formances or decelerating performances.

A line drawn from the bottom left corner to the top right corner has a slope of 34° on all Standard Celera- tion Charts. This slope has a celeration value of X2 (read as “times-2”; celerations are expressed with multiples or divisors). A X2 celeration is a doubling in frequency per celeration period. The celeration period for the daily chart is per week; it is per month for the weekly chart, per 6 months for the monthly chart, and per 5 years for the yearly chart.

Constructing and Interpreting Graphic Displays of Behavioral Data

160

Table 1 Basic Charting Conventions for the Daily Standard Celeration Chart (See also Figure 15)

Term Definition Convention

1. Charted day A day on which the behavior is recorded 1. Chart the behavior frequency on the chart on and charted. the appropriate day line.

2. Connect charted days except across phase change lines, no chance days, and ignored days.

a) Acceleration target Responses of the performer intended Chart a dot (•) on the appropriate day line. frequency to accelerate.

b) Deceleration target Responses of the performer intended Chart an (x) on the appropriate day line. frequency to decelerate.

2. No chance day A day on which the behavior had no chance Skip day on daily chart. to occur.

3. Ignored day A day on which the behavior could have occurred Skip day on daily chart. (Connect data across but no one recorded it. ignored days.)

4. Counting-time bar Designates on the chart the performer’s lowest Draw solid horizontal line from the Tuesday to (aka record floor) possible performance (other than zero) in a Thursday day lines on the chart at the “counting-

counting time. Always designated as “once per time bar.” counting time.”

5. Zero performance No performance recorded during the recording Chart on the line directly below the “counting- period. time bar.”

6. Phase change line A line drawn in the space between the last Draw a vertical line between the intervention charted day of one intervention phase and the phases. Draw the line from the top of the data to first charted day of a new intervention phase. the “counting-time bar.”

7. Change indicator Words, symbols, or phrases written on the chart Write word, symbol, and/or phrase. An arrow (➞) in the appropriate phase to indicate changes may be used to indicate the continuance of a during that phase. change into a new phase.

8. Aim star A symbol used to represent (a) the desired Place the point of the caret . . . frequency, and (b) the desired date to achieve ^ for acceleration data the frequency. v for deceleration data

. . . on the desired aim date. Place the horizontal bar on the desired frequency. The caret and horizontal line will create a “star.”

9. Calendar synchronize A standard time for starting all charts. It requires three charts to cover a full year. The Sunday before Labor Day begins the first week of the first chart. The twenty-first week after Labor Day begins the second chart. The forty-first week after Labor Day begins the third chart.

10. Celeration line A straight line drawn through 7–9 or more charted days. This line indicates the amount of improvement that has taken place in a given period of time. A new line is drawn for each phase for both acceleration and deceleration targets. (Note: For nonresearch projects it is acceptable to draw freehand celeration lines.) Acceleration target Deceleration target

Constructing and Interpreting Graphic Displays of Behavioral Data

161

x 16

x 4

x 2

x 1.4 x 1.0

per week 1000

100

.01

Dy Mo Yr Dy Mo Yr Dy Mo Yr Dy Mo Yr Dy Mo Yr Dy Mo Yr

Calendar Weeks

Likeness of Daily per minute Standard Celeration Chart

Actual Charts available from BEHAVIOR RESEARCH CO

BOX 3351. KANSAS CITY, KS 65103-3351 FAX ORDERS 913 362

5900

C o

u n

t p

er M

in u

0 14 28 42 56 70 84 98 112 126 140

Successive Calendar Days

SUPERVISOR

ORGANIZATION

ADVISER MANAGER

TIMER COUNTER CHARTER

PERFORMER

1. Frequency change = x 3.0

5.2.

18/min

6/min

Celeration calculation (Quarter- intersect method)

Quarter: 1st 2nd 3rd 4th

Frequency Change (FC) Celeration Change (CC) Bounce Change (BC)

Celeration change

calculation

Projection lines

Celeration Collection

High

Medium

Low

7 Celerations

Bounce change

Analysis Matrix

Total bounce = x 1.96

Total bounce = X 3.6

Outlier Total

bounce = X 1.69

Total bounce = x 2.25

FC = x 2.8

FCP = .001

CC = x 2.3

CCP = .003

BC = x 1.8

BCP = .005

Acceleration Target

FC = ÷ 2.8

FCP = .005

CC = ÷ 1.6

CCP = .0001

BC = ÷ 1.1

BCP = .01

Deceleration Target

Analysis Matrix

COUNTED

.001

Figure 16 Standard Celeration Chart showing advanced charting conventions. See Table 2 for explanation.

8Detailed descriptions and examples of precision teaching are provided by the Journal of Precision Teaching and Celeration; the Standard Celera- tion Society’s Web site (http://celeration.org/); Binder (1996); Kubina and Cooper (2001); Lindsley (1990, 1992, 1996); Potts, Eshleman, and Cooper (1993); West, Young, and Spooner (1990); and White and Haring (1980).

An instructional decision-making system, called precision teaching, has been developed for use with the Standard Celeration Chart.8 Precision teaching is predi- cated on the position that (a) learning is best measured as a change in response rate, (b) learning most often occurs through proportional changes in behavior, and (c) past changes in performance can project future learning.

Precision teaching focuses on celeration, not on the specific frequency of correct and incorrect responses as many believe. That frequency is not an emphasis on the Chart is clear because the Chart uses estimations for most frequency values. A practitioner or researcher might say, ”I don’t use the chart because I can’t tell by looking at the chart if the student emitted 24, 25, 26, or 27 responses.”

However, the purpose of the Chart makes such a fine dis- crimination irrelevant because celeration, not specific fre- quency, is the issue. A frequency of 24 or 27 will not change the line of progress—the celeration course.

Advanced charting conventions used by precision teachers are illustrated in Figure 16 and described in Table 2. Detailed explanations of the Standard Celera- tion Chart and its uses can be found in Cooper, Kubina, and Malanga (1998); Graf and Lindsley (2002); and Pen- nypacker, Gutierrez, and Lindsley (2003).

Scatterplots

A scatterplot is a graphic display that shows the relative distribution of individual measures in a data set with re- spect to the variables depicted by the x and y axes. Data points on a scatterplot are unconnected. Scatterplots show how much changes in the value of the variable depicted by one axis correlate with changes in the value of the variable represented by the other axis. Patterns of data

Constructing and Interpreting Graphic Displays of Behavioral Data

162

Table 2 Advanced Charting Conventions for the Daily Standard Celeration Chart (See also Figure 16)

Constructing and Interpreting Graphic Displays of Behavioral Data

Term Definition Convention

Frequency: 1. Frequency change (FC) The multiply “×” or divide “÷” value that compares the final fre- Place an “FC =” in the upper left (aka frequency jump up or quency of one phase to the beginning frequency in the next cell of the analysis matrix. jump down) phase. Compute this by comparing (1) the frequency where the Indicate the value with a “×”

celeration line crosses the last day of one phase to (2) the or “÷” sign (e.g., FC = × 3.0). frequency where the celeration line crosses the first day of the next phase (e.g., a frequency jump from 6/minute to 18/minute. FC = × 3.0).

Celeration: 2. Celeration calculation The process for graphically determining a celeration line (aka See advanced charting (quarter-intersect method) “the line of best fit”). (1) Divide the frequencies for each phase conventions sample chart.

into four equal quarters (include ignored and no chance days), (2) locate the median frequency for each half, and (3) draw a celeration line connecting the quarter intersect points.

3. Celeration finder A piece of mylar with standard celeration lines that can be used Buy commercially or copy and to compute celeration line values. cut out part of the vertical axis

on the Standard Celeration Chart.

4. Projection line A dashed line extending to the future from the celeration line. See advanced charting The projection offers a forecast that enables the calculation of conventions sample chart. the celeration change value.

5. Celeration change (CC) The multiply “×” or divide “÷” value that compares the celeration Place a “CC =” in the upper mid- (aka celeration turn up or of one phase to the celeration in the next phase (e.g., a celer- dle cell of the analysis matrix turn down) ation turn down from × 1.3 to ÷ 1.3. CC = ÷ 1.7). with the value indicated with a

“×” or “÷” sign (e.g., CC = ÷ 1.7).

6. Celeration collection A group of three or more celerations for different performers Numerically identify the high, relating to the same behavior over approximately the same middle, and low celeration in time period. the celeration collection and

indicate the total number of celerations in the collection.

7. Bounce change (BC) The multiply “×” or divide “÷” value that compares the bounce Place a “BC =” in the upper in one phase to the bounce in the next phase. Computed by right cell of the analysis comparing (1) the total bounce of one phase to (2) the total matrix with the value indicated bounce of the next phase (e.g., a bounce change from 5.0 to with a multiply “×” or divide “÷” × 1.4, BC = ÷ 3.6). symbol (e.g., BC = ÷ 3.6).

8. Analysis matrix The analysis matrix provides the numeric change information Place the analysis matrix be- regarding the effects of the independent variable(s) on tween the two phases being frequency, celeration, and bounce between two phases. compared. For acceleration tar-

gets place the matrix above the data. For deceleration targets place the matrix below the data.

Optional: 9. Frequency change The frequency change p-value is the probability that the noted Use “FCP =” and indicate the p-value (FCP) change in frequency would have occurred by chance. (Use the p-value in the lower left cell

Fisher exact probability formula to compute the p-value.) on the analysis matrix (e.g., FCP = .0001).

10. Celeration change The celeration change p-value is the probability that the change Use “CCP =” and indicate the p-value (CCP) noted in celeration would have occurred by chance. (Use the p-value in the lower middle cell

Fisher exact probability formula to compute the p-value.) of the matrix (e.g., CCP = .0001).

11. Bounce change The bounce change p-value is the probability that the change Use “BCP =” and indicate the p-value (BCP) noted in bounce would have occurred by chance. (Use the p-value in the lower right cell

Fisher exact probability formula to compute the p-value.) of the analysis matrix (e.g., BCP = .0001).

163

F o

ll o

w in

g D

is ta

n ce

i n

M et

er s

35 35 35 Speed in Miles per Hour

Safe

At-Risk

Young Males Young Females Middle Males Middle Females Older Males Older Females

Figure 17 Scatterplot showing how the behavior of individuals from different demo- graphic groups relates to a standard measure of safe driving.

From “A Technology to Measure Multiple Driving Behaviors without Self-Report or Participant Reactivity” by T. E. Boyce and E. S. Geller, 2001, Journal of Applied Behavior Analysis, 34, p. 49. Copyright 2001 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

points falling along lines on the plane or clusters suggest certain relationships.

Scatterplots can reveal relationships among differ- ent subsets of data. For example, Boyce and Geller (2001) created the scatterplot shown in Figure 17 to see how the behavior of individuals from different demographic groups related to a ratio of driving speed and following distance that represents one element of safe driving (e.g., the proportion of data points for young males falling in the at-risk area of the graph compared to the proportion of drivers from other groups). Each data point shows a single driver’s behavior in terms of speed and following distance and whether the speed and following distance combination is considered safe or at risk for accidents. Such data could be used to target interventions for certain demographic groups.

Applied behavior analysts sometimes use scatterplots to discover the temporal distribution of a target behavior (e.g., Kahng et al., 1998; Symons, McDonald, & Wehby, 1998; Touchette, MacDonald, & Langer, 1985). Touchette and colleagues described a procedure for observing and recording behavior that produces a scatterplot that graph- ically shows whether the behavior’s occurrence is typi- cally associated with certain time periods.

Constructing Line Graphs The skills required to construct effective, distortion-free graphic displays are as important as any in the behavior analyst’s repertoire. As applied behavior analysis has de- veloped, so have certain stylistic conventions and expec- tations regarding the construction of graphs. An effective graph presents the data accurately, completely, and clearly, and makes the viewer’s task of understanding the data as easy as possible. The graph maker must strive to fulfill each of these requirements while remaining alert to

features in the graph’s design or construction that might create distortion and bias—either the graph maker’s or that of a future viewer when interpreting the extent and nature of the behavior change depicted by the graph.

Despite the graph’s prominent role in applied be- havior analysis, relatively few detailed treatments of how to construct behavioral graphs have been published. No- table exceptions have been chapters by Parsonson and Baer (1978, 1986) and a discussion of graphic display tactics by Johnston and Pennypacker (1980, 1993a). Rec- ommendations from these excellent sources and others (Journal of Applied Behavior Analysis, 2000; American Psychological Association, 2001; Tufte, 1983, 1990) con- tributed to the preparation of this section. Additionally, hundreds of graphs published in the applied behavior analysis literature were examined in an effort to discover those features that communicate necessary information most clearly.

Although there are few hard-and-fast rules for con- structing graphs, adhering to the following conventions will result in clear, well-designed graphic displays con- sistent in format and appearance with current practice. Although most of the recommendations are illustrated by graphs presented throughout this text, Figures 18 and 19 have been designed to serve as models for most of the practices suggested here. The recommendations given here generally apply to all behavioral graphs. How- ever, each data set and the conditions under which the data were obtained present their own challenges to the graph maker.

Drawing, Scaling, and Labeling Axes

Ratio of the Vertical and Horizontal Axes

The relative length of the vertical axis to the horizontal axis, in combination with the scaling of the axes, deter- mines the degree to which a graph will accentuate or min-

Constructing and Interpreting Graphic Displays of Behavioral Data

164

Self-Record Self-Record Tokens Follow-Up

Figure 1. Percent of 10-second intervals in which an 8-year-old boy emitted appropriate and inappropriate study behaviors. Each interval was scored as appropriate, inappropriate, or neither, so the two behaviors do not always total 100%.

Baseline

School Days

P er

ce n

t o

f In

te rv

al s

Figure 18 Graph of hypothetical data illustrating a variety of conventions and guidelines for graphic display.

imize the variability in a given data set. The legibility of a graph is enhanced by a balanced ratio between the height and width so that data are neither too close together nor too spread apart. Recommendations in the behavioral literature for the ratio, or relative length, of the vertical axis to the horizontal axis range from 5:8 (Johnston & Pennypacker, 1980) to 3:4 (Katzenberg, 1975). Tufte (1983), whose book The Visual Display of Quantitative Information is a wonderful storehouse of guidelines and examples of effective graphing techniques, recommends a 1:1.6 ratio of vertical axis to horizontal axis.

A vertical axis that is approximately two-thirds the length of the horizontal axis works well for most be- havioral graphs. When multiple sets of axes will be pre- sented atop one another in a single figure and/or when the number of data points to be plotted on the horizon- tal axis is very large, the length of the vertical axis rel- ative to the horizontal axis can be reduced (as shown in Figures 3 and 7).

Scaling the Horizontal Axis

The horizontal axis should be marked in equal intervals, with each unit representing from left to right the chrono- logical succession of equal time periods or response op- portunities in which the behavior was (or will be) measured and from which an interpretation of behavior change is to be made (e.g., days, sessions, trials). When many data points are to be plotted, it is not necessary to

mark each point along the x axis. Instead, to avoid un- necessary clutter, regularly spaced points on the hori- zontal axis are indicated with tic marks numbered by 5s, 10s, or 20s.

When two or more sets of axes are stacked vertically and each horizontal axis represents the same time frame, it is not necessary to number the tic marks on the hori- zontal axes of the upper tiers. However, the hatch marks corresponding to those numbered on the bottom tier should be placed on each horizontal axis to facilitate com- parison of performance across tiers at any given point in time (see Figure 4).

Representing Discontinuities of Time on the Horizontal Axis

Behavior change, its measurement, and all manipulations of treatment or experimental variables occur within and across time. Therefore, time is a fundamental variable in all experiments that should not be distorted or arbitrarily represented in a graphic display. Each equally spaced unit on the horizontal axis should represent an equal passage of time. Discontinuities in the progression of time on the horizontal axis should be indicated by a scale break: an open spot in the axis with a squiggly line at each end. Scale breaks on the x axis can also be used to sig- nal periods of time when data were not collected or when regularly spaced data points represent consecu- tive measurements made at unequal intervals (see the

Constructing and Interpreting Graphic Displays of Behavioral Data

165

W o

rd s

R ea

d p

er M

in u

te St

u d

en t

St u

d en

Follow-Up

Sessions

Figure 1. Number of words read correctly and incorrectly during 1-minute probes following each session. Arrows under horizontal axes indicate sessions in which student used reading material brought from home. Break in data path for Student 2 was caused by 2 days’ absence.

Figure 19 Graph of hypothetical data illus- trating a variety of conventions and guidelines for graphic display.

numbering of school days for the follow-up condition in Figure 18).

When measurement occurs across consecutive ob- servations (e.g., stories read, meals, interactions) rather than standard units of time, the horizontal axis still serves as a visual representation of the progression of time be- cause the data plotted against it have been recorded one after the other. The text accompanying such a figure should indicate the real time in which the consecutive measurements were made (e.g., “Two or three peer tu- toring sessions were conducted each school week”), and discontinuities in that time context should be clearly marked with scale breaks (see Figure 19).

Labeling the Horizontal Axis

The dimension by which the horizontal axis is scaled should be identified in a brief label printed and centered below and parallel to the axis.

Scaling the Vertical Axis

On an equal-interval graph the scaling of the vertical axis is the most significant feature of the graph in terms of its portrayal of changes in level and variability in the data. Common practice is to mark the origin at 0 (on cumula- tive graphs the bottom on the vertical axis must be 0) and then to mark off the vertical axis so that the full range of values represented in the data set are accommodated. In- creasing the distance on the vertical axis between each unit of measurement magnifies the variability in the data, whereas contracting the units of measurement on the ver- tical axis minimizes the portrayal of variability in the data set. The graph maker should plot the data set against sev- eral different vertical axis scales, watching for distortion of the graphic display that might lead to inappropriate interpretations.

The social significance of various levels of behavior change for the behavior being graphed should be con- sidered in scaling the vertical axis. If relatively small

Constructing and Interpreting Graphic Displays of Behavioral Data

166

numerical changes in performance are socially signifi- cant, the y-axis scale should reflect a smaller range of values. For example, to display data most effectively from a training program in which an industrial em- ployee’s percentage of correctly executed steps in a safety checklist increased from an unsafe preinterven- tion range of 80% to 90% to an accident-free postinter- vention level of 100%, the vertical axis should focus on the 80% to 100% range. On the other hand, the scaling of the vertical axis should be contracted when small nu- merical changes in behavior are not socially important and the degree of variability obscured by the compressed scale is of little interest.

Horizontal numbering of regularly spaced tic marks on the vertical axis facilitates use of the scale. The verti- cal axis should not be extended beyond the hatch mark in- dicating the highest value on the axis scale.

When the data set includes several measures of 0, starting the vertical axis at a point slightly above the hor- izontal axis keeps data points from falling directly on the axis. This produces a neater graph and helps the viewer discriminate 0-value data points from those representing measures close to 0 (see Figure 18).

In most instances, scale breaks should not be used on the vertical axis, especially if a data path would cross the break. However, when two sets of data with widely different and nonoverlapping ranges are displayed against the same y axis, a scale break can be used to separate the range of measures encompassed by each data set (see Figure 19).

In multiple-tier graphs, equal distances on each ver- tical axis should represent equal changes in behavior to aid the comparison of data across tiers. Also, whenever possible, similar positions on each vertical axis of multiple-tier graphs should represent similar absolute val- ues of the dependent variable. When the differences in behavioral measures from one tier to another would re- sult in an overly long vertical axis, a scale break can be used to highlight the difference in absolute values, again aiding a point-to-point comparison of y axis positions.

Labeling the Vertical Axis

A brief label, printed and centered to the left and paral- lel to the vertical axis, should identify the dimension by which the axis is scaled. On multiple-tiered graphs, one label identifying the dimension portrayed on all of the vertical axes can be centered along the axes as a group. Additional labels identifying the different behaviors (or some other relevant aspect) graphed within each set of axes are sometimes printed to the left and parallel to each vertical axis. These individual tier labels should be printed to the right of and in smaller-sized font than the label

identifying the dimension by which all of the vertical axes are scaled.

Identifying Experimental Conditions

Condition Change Lines

Vertical lines extending upward from the horizontal axis indicate changes in treatment or experimental procedures. Condition change lines should be placed after (to the right of ) the data point representing the last measure prior to the change in conditions signified by the line and before (to the left of ) the data point representing the first mea- sure obtained after the change in procedure. In this way data points fall clearly on either side of change lines and never on the lines themselves. Drawing condition change lines to a height equal to the height of the vertical axis helps the viewer estimate the value of data points near the top of the vertical axis range.

Condition change lines can be drawn with either solid or dashed lines. However, when an experiment or a treatment program includes relatively minor changes within an ongoing condition, a combination of solid and dashed lines should be used to distinguish the major and minor changes in conditions. For example, the solid lines in Figure 18 change from baseline, to self-record, to self-record + tokens, to follow-up conditions, and dashed lines indicate changes in the schedule of reinforcement from CRF, to VR 5, to VR 10 within the self-record + tokens condition.

When the same manipulation of an independent vari- able occurs at different points along the horizontal axes of multiple-tiered graphs, a dog-leg connecting the con- dition change lines from one tier to the next makes it easy to follow the sequence and timing of events in the ex- periment (see Figure 19).

Unplanned events that occur during an experiment or treatment program, as well as minor changes in pro- cedure that do not warrant a condition change line, can be indicated by placing small arrows, asterisks, or other sym- bols next to the relevant data points (see Figure 6) or just under the x axis (see Figure 19). The figure caption should explain any special symbols.

Condition Labels

Labels identifying the conditions in effect during each period of an experiment are centered above the space de- lineated by the condition change lines. Whenever space permits, condition labels should be parallel to the horizontal axis. Labels should be brief but descriptive (e.g., Contin- gent Praise is preferable to Treatment), and the labels

Constructing and Interpreting Graphic Displays of Behavioral Data

167

should use the same terms or phrases used in the accom- panying text describing the condition. Abbreviations may be used when space or design limitations prohibit print- ing the complete label. A single condition label should be placed above and span across labels identifying minor changes within that condition (see Figure 18). Num- bers are sometimes added to condition labels to indicate the number of times the condition has been in effect dur- ing the study (e.g., Baseline 1, Baseline 2).

Plotting Data Points and Drawing Data Paths

Data Points

When graphing data by hand, behavior analysts must take great care to ensure that they plot each data point exactly on the coordinate of the horizontal and vertical axis values of the measurement it represents. The in- accurate placement of data points is an unnecessary source of error in graphic displays, which can lead to mistakes in clinical judgment and/or experimental method. Accurate placement is aided by careful selec- tion of graph paper with grid lines sized and spaced ap- propriately for the data to be plotted. When many different values must be plotted within a small distance on the vertical axis, a graph paper with many grid lines per inch should be used.9

Should a data point fall beyond the range of values described by the vertical axis scale, it is plotted just above the scale it transcends with the actual value of the mea- surement printed in parentheses next to the data point. Breaks in the data path leading to and from the off-the- scale data point also help to highlight its discrepancy (see Figure 19, Session 19).

Data points should be marked with bold symbols that are easily discriminated from the data path. When only one set of data is displayed on a graph, solid dots are most often used. When multiple data sets are plotted on the same set of axes, a different geometric symbol should be used for each set of data. The symbols for each data set should be selected so that the value of each data point can be determined when data points fall near or on the same coordinates on the graph (see Figure 18, Sessions 9–11).

9Although most graphs published in behavior analysis journals since the mid-1990s were constructed with computer software programs that en- sure precise placement of data points, knowing how to draw graphs by hand is still an important skill for applied behavior analysts, who often use hand-drawn graphs to make treatment decisions on a session-by-session basis.

Data Paths

Data paths are created by drawing a straight line from the center of each data point in a given data set to the center of the next data point in the same set. All data points in a given data set are connected in this manner with the following exceptions:

• Data points falling on either side of a condition change line are not connected.

• Data points should not be connected across a sig- nificant span of time in which behavior was not measured. To do so implies that the resultant data path represents the level and trend of the behavior during the span of time in which no measurement was conducted.

• Data points should not be connected across discon- tinuities of time in the horizontal axis (see Figure 18, 1-week school vacation).

• Data points on either side of a regularly scheduled measurement period in which data were not col- lected or were lost, destroyed, or otherwise not avail- able (e.g., participant’s absence, recording equipment failure) should not be joined together (see Figure 18, baseline condition of bottom graph).

• Follow-up or postcheck data points should not be connected with one another (see Figure 18) un- less they represent successive measures spaced in time in the same manner as measures obtained dur- ing the rest of the experiment (see Figure 19).

• If a data point falls beyond the values described by the vertical axis scale, breaks should be made in the data path connecting that data point with those that fall within the described range (Figure 19, top graph, Session 19 data point).

When multiple data paths are displayed on the same graph, different styles of lines, in addition to different symbols for the data points, may be used to help distin- guish one data path from another (see Figure 19). The behavior represented by each data path should be clearly identified, either by printed labels with arrows drawn to the data path (see Figures 18 and 19) or by a legend showing models of the symbols and line styles (see Figure 13). When two data sets travel the same path, their lines should be drawn close to and parallel with one another to help clarify the situation (see Figure 18, Sessions 9–11).

Writing the Figure Caption

Printed below the graph, the figure caption should give a concise but complete description of the figure. The cap- tion should also direct the viewer’s attention to any fea-

Constructing and Interpreting Graphic Displays of Behavioral Data

168

tures of the graph that might be overlooked (e.g., scale changes) and should explain the meaning of any added symbols representing special events.

Printing Graphs

Graphs should be printed in only one color—black. Al- though the use of color can enhance the attractiveness of a visual display and can effectively highlight certain fea- tures, it is discouraged in the scientific presentation of data. Every effort must be made to let the data stand on their own. The use of color can encourage perceptions of performance or experimental effects that differ from per- ceptions of the same data displayed in black. The fact that graphs and charts may be reproduced in journals and books is another reason for using black only.

Constructing Graphs with Computer Software

Software programs for producing computer-generated graphs have been available and are becoming both in- creasingly sophisticated and easier to use. Most of the graphs displayed throughout this book were constructed with computer software. Even though computer graphics programs offer a tremendous time savings over hand- plotted graphs, careful examination should be made of the range of scales available and the printer’s capability for both accurate data point placement and precise print- ing of data paths.

Carr and Burkholder (1998) provided an intro- duction to creating single-subject design graphs with Microsoft Excel. Silvestri (2005) wrote detailed, step-by-step instructions for creating behavioral graphs using Microsoft Excel. Her tutorial can be found on the companion Web site that accompanies this text, www.prenhall.com/cooper.

Interpreting Graphically Displayed Behavioral Data The effects of an intervention that produces dramatic, replicable changes in behavior that last over time are read- ily seen in a well-designed graphic display. People with little or no formal training in behavior analysis can read the graph correctly in such cases. Many times, however, be- havior changes are not so large, consistent, or durable. Be- havior sometimes changes in sporadic, temporary, delayed, or seemingly uncontrolled ways; and sometimes behavior may hardly change at all. Graphs displaying these kinds of data patterns often reveal equally important and interesting subtleties about behavior and its controlling variables.

Behavior analysts employ a systematic form of ex- amination known as visual analysis to interpret graphi- cally displayed data. Visual analysis of data from an applied behavior analysis study is conducted to answer two questions: (a) Did behavior change in a meaningful way, and (b) if so, to what extent can that change in be- havior be attributed to the independent variable? Although there are no formalized rules for visual analysis, the dy- namic nature of behavior, the scientific and technologi- cal necessity of discovering effective interventions, and the applied requirement of producing socially meaning- ful levels of performance all combine to focus the behavior analyst’s interpretive attention on certain fundamental properties common to all behavioral data: (a) the extent and type of variability in the data, (b) the level of the data, and (c) trends in the data. Visual analysis en- tails an examination of each of these characteristics both within and across the different conditions and phases of an experiment.

As Johnston and Pennypacker (1993b) so aptly noted, “It is impossible to interpret graphic data without being influenced by various characteristics of the graph itself” (p. 320). Therefore, before attempting to interpret the meaning of the data displayed in a graph, the viewer should carefully examine the graph’s overall construc- tion. First, the figure legend, axis labels, and all condi- tion labels should be read to determine a basic understanding of what the graph is about. The viewer should then look at the scaling of each axis, taking note of the location, numerical value, and relative significance of any scale breaks.

Next, a visual tracking of each data path should be made to determine whether data points are properly connected. Does each data point represent a single mea- surement or observation, or are the data “blocked” such that each data point represents an average or some other summary of multiple measurements? Do the data show the performance of an individual subject or the aver- age performance of a group of subjects? If blocked or group data are displayed, is a visual representation of the range or variation of scores provided (e.g., Armen- dariz & Umbreit, 1999; Epstein et al., 1981); or do the data themselves allow determination of the amount of variability that was collapsed in the graph? For exam- ple, if the horizontal axis is scaled in weeks and each data point represents a student’s average score for a week of daily five-word spelling tests, data points falling near 0 or at the top end of the closed scale, such as 4.8, pose little problem because they can be the re- sult of only minimal variability in the daily scores for that week. However, data points near the center of the scale, such as 2 to 3, can result from either stable or highly variable performance.

Constructing and Interpreting Graphic Displays of Behavioral Data

169

If the viewer suspects distortion produced by a graph’s construction, interpretive judgments of the data should be withheld until the data are replotted on a new set of axes. Distortion due to a loss of important data fea- tures in summarizing is not so easily remedied. The viewer must consider the report incomplete and forestall any in- terpretive conclusions until he has access to the raw data.

Only when the viewer is satisfied that the graph is properly constructed and does not visually distort the be- havioral and environmental events it represents, should the data themselves be examined. The data are then in- spected to find what they reveal about the behavior mea- sured during each condition of the study.

Visual Analysis within Conditions

Data within a given condition are examined to determine (a) the number of data points, (b) the nature and extent of variability in the data, (c) the absolute and relative level of the behavioral measure, and (d) the direction and de- gree of any trend(s) in the data.

Number of Data Points

First, the viewer should determine the quantity of data reported during each condition. This entails a simple counting of data points. As a general rule, the more mea- surements of the dependent variable per unit of time and the longer the period of time in which measurement oc- curred, the more confidence one can have in the data path’s estimation of the true course of behavior change (given, of course, a valid and accurate observation and measurement system).

The number of data points needed to provide a be- lievable record of behavior during a given condition also depends on how many times the same condition has been repeated during the study. As a rule, fewer data points are needed in subsequent replications of an experimen- tal condition if the data depict the same level and trend in performance that were noted in earlier applications of the condition.

The published literature of applied behavior analysis also plays a part in determining how many data points are sufficient. In general, less lengthy phases are required of experiments investigating relations between previously studied and well-established variables if the results are also similar to those of the previous studies. More data are needed to demonstrate new findings, whether or not new variables are under investigation.

There are other exceptions to the rule of the-more- data-the-better. Ethical concerns do not permit the repeated measurement of certain behaviors (e.g.,

self-injurious behavior) under an experimental condition in which there is little or no expectation for improvement (e.g., during a no-treatment baseline condition or a con- dition intended to reveal variables that exacerbate prob- lem behavior). Also, there is little purpose in repeated measurement in situations in which the subject cannot logically perform the behavior (e.g., measuring the num- ber of correct answers to long division problems when concurrent observations indicate that the student has not learned the necessary component skills of multiplication and subtraction). Nor are many data points required to demonstrate that behavior did not occur when in fact it had no opportunity to occur.

Familiarity with the response class measured and the conditions under which it was measured may be the graph viewer’s biggest aid in determining how many data points constitute believability. The quantity of data needed in a given condition is also partly determined by the analytic tactics employed in a given study.

Variability

How often and the extent to which multiple measures of behavior yield different outcomes is called variability. A high degree of variability within a given condition usu- ally indicates that the researcher or practitioner has achieved little control over the factors influencing the be- havior. (An important exception to this statement is when the purpose of an intervention is to produce a high de- gree of variability.) In general, the greater the variability within a given condition, the greater the number of data points that are necessary to establish a predictable pat- tern of performance. By contrast, fewer data points are re- quired to present a predictable pattern of performance when those data reveal relatively little variability.

Level

The value on the vertical axis scale around which a set of behavioral measures converge is called level. In the visual analysis of behavioral data, level is examined within a condition in terms of its absolute value (mean, median, and/or range) on the y-axis scale, the degree of stability or variability, and the extent of change from one level to another. The graphs in Figure 20 illustrate four differ- ent combinations of level and variability.

The mean level of a series of behavioral measures within a condition can be graphically illustrated by the ad- dition of a mean level line: a horizontal line drawn through a series of data points within a condition at that point on the vertical axis equaling the average value of the

Constructing and Interpreting Graphic Displays of Behavioral Data

170

Figure 20 Four data paths illustrating (A) a low, stable level of responding; (B) a high, variable level of responding; (C) an initially high, stable level of responding followed by a lower, more variable level of responding; and (D) an extremely variable pattern of responding not indicative of any overall level of responding. Dashed hori- zontal lines on graphs B, C, and D repre- sent the mean levels of responding.

series of measures (e.g., Gilbert, Williams, & McLaugh- lin, 1996). Although mean level lines provide an easy- to-see summary of average performance within a given condition or phase, they should be used and interpreted with caution. With highly stable data paths, mean level lines pose no serious drawbacks. However, the less vari- ability there is within a series of data points, the less need there is for a mean level line. For instance, a mean level line would serve little purpose in Graph A in Figure 20. And although mean level lines have been added to Graphs B, C, and D in Figure 20, Graph B is the only one of the three for which a mean level line provides an appropri- ate visual summary of level. The mean level line in Graph C is not representative of any measure of behavior taken during the phase. The data points in Graph C show a be- havior best characterized as occurring at two distinct lev- els during the condition and beg for an investigation of the factor(s) responsible for the clear change in levels. The mean level line in Graph D is also inappropriate because the variability in the data is so great that only 4 of the 12 data points fall close to the mean level line.

A median level line is another method for visually summarizing the overall level of behavior in a condition. Because a median level line represents the most typical performance within a condition, it is not so influenced by one or two measures that fall far outside the range of the remaining measures. Therefore, one should use a me- dian level line instead of a mean level line to graphically represent the central tendency of a series of data points that include several outliers, either high or low.

Change in level within a condition is determined by calculating the difference in absolute values on the y axis

between the first and last data points within the condi- tion. Another method, somewhat less influenced by vari- ability in the data, is to compare the difference between the median value of the first three data points in the con- dition with the median value of the final three data points in the condition (Koenig & Kunzelmann, 1980).

Trend

The overall direction taken by a data path is its trend. Trends are described in terms of their direction (increas- ing, decreasing, or zero trend), degree or magnitude, and extent of variability of data points around the trend. The graphs in Figure 21 illustrate a variety of trends. The di- rection and degree of trend in a series of graphed data points can be visually represented by a straight line drawn through the data called a trend line or line of progress. Several methods for calculating and fitting trends lines to a series of data have been developed. One can simply in- spect the graphed data and draw a straight line that visu- ally provides the best fit through the data. For this freehand method, Lindsley (1985) suggested ignoring one or two data points that fall well beyond the range of the remaining values in a data series and fitting the trend line to the remaining scores. Although the freehand method is the fastest way of drawing trend lines and can be useful for the viewer of a published graph, hand-drawn trend lines may not always result in an accurate repre- sentation of trend and are typically not found in graphs of published studies.

Trend lines can also be calculated using a mathe- matical formula called the ordinary least-squares linear

Constructing and Interpreting Graphic Displays of Behavioral Data

171

Figure 21 Data patterns indicating various combinations of trend direction, degree, and variability: (A) zero trend, high stability; (B) zero trend, high variability; (C) gradually increasing stable trend; (D) rapidly increasing variable trend; (E) rapidly decreasing stable trend; (F) gradually decreasing variable trend; (G) rapidly increasing trend followed by rapidly decreasing trend; (H) no meaningful trend, too much variability and missing data. Split- middle lines of progress have been added to Graphs C–F.

regression equation (McCain & McCleary, 1979; Par- sonson & Baer, 1978). Trend lines determined in this fashion have the advantage of complete reliability: The same trend line will always result from the same data set. The disadvantage of this method is the many mathemat- ical operations that must be performed to calculate the trend line. A computer program that can perform the equation can eliminate the time concern in calculating a least-squares trend line.

A method of calculating and drawing lines of progress that is more reliable than the freehand method and much less time-consuming than linear regression methods is the split-middle line of progress. The split- middle technique was developed by White (1971, 2005) for use with rate data plotted on semilogarithmic charts,

and it has proven a useful technique for predicting future behavior from such data. Split-middle lines of progress can also be drawn for data plotted against an equal- interval vertical axis, but it must be remembered that such a line is only an estimate that summarizes the overall trend (Bailey, 1984). Figure 22 provides a step-by-step illustration of how to draw split-middle lines of progress. A trend line cannot be drawn by any method through a se- ries of data points spanning a scale break in the vertical axis and generally should not be drawn across scale breaks in the horizontal axis.

The specific degree of acceleration or deceleration of trends in data plotted on semilogarithmic charts can be quantified in numerical terms. For example, on the daily Standard Celeration Chart a “times-2” celeration means

Constructing and Interpreting Graphic Displays of Behavioral Data

172

that the response rate is doubling each week, and a “times- 1.25” means that the response rate is accelerating by a fac- tor of one fourth each week. A “divide-by-2” celeration means that each week the response rate will be one half of what it was the week before, and a “divide-by-1.5” means that the frequency is decelerating by one third each week.

There is no direct way to determine visually from data plotted on equal-interval charts the specific rates at which trends are increasing or decreasing. But visual comparison of trend lines drawn through data on equal- interval charts can provide important information about the relative rates of behavior change.

First Half Second Half

Mid- Rate

M id

-D at

M id

-D at

Mid-Rate(s)

Quarter-In tersect Line of Progress

Split-M iddle Line of Progress

Figure 22 How to draw a split-middle line of progress through a series of graphically displayed data points.

Constructing and Interpreting Graphic Displays of Behavioral Data

173

Time

R es

p o

n se

M ea

su re

R es

p o

n se

M ea

su re

Time

Figure 23 Inappropriate use of mean level lines, encouraging interpretation of a higher overall level of responding in Condition B when extreme variability (top graph) and trends (bottom graph) warrant different conclusions.

Condition A, but the mean level lines suggest a clear change in behavior. Second, measures of central tendency can obscure important trends in the data that warrant in- terpretations other than those suggested by the central tendency indicators. Although a mean or median line ac- curately represents the average or typical performance, neither provides any indication of increasing or decreas- ing performance. In the bottom graph of Figure 23, for example, the mean line suggests a higher level of per- formance in Condition B than in Condition A, but an ex- amination of trend yields a very different picture of behavior change within and between Conditions A and B.

Those analyzing behavioral data should also note any changes in level that occur after a new condition has been in place for some time and any changes in level that occur early in a new condition but are later lost. Such de- layed or temporary effects can indicate that the indepen- dent variable must be in place for some time before behavior changes, or that the temporary level change was the result of an uncontrolled variable. Either case calls for further investigation in an effort to isolate and control relevant variables.

A trend may be highly stable with all of the data points falling on or near the trend line (see Figure 21, Graphs C and E). Data paths can also follow a trend even though a high degree of variability exists among the data points (see Figure 21, Graphs D and F).

Visual Analysis between Conditions

After inspection of the data within each condition or phase of a study, visual analysis proceeds with a com- parison of data between conditions. Drawing proper con- clusions entails comparing the previously discussed properties of behavioral data—level, trend, and stability/ variability—between different conditions and among sim- ilar conditions.

A condition change line indicates that an indepen- dent variable was manipulated at a given point in time. To determine whether an immediate change in behavior oc- curred at that point in time, one needs to examine the dif- ference between the last data point before the condition change line and the first data point in the new condition.

The data are also examined in terms of the overall level of performance between conditions. In general, when all data points in one condition fall outside the range of values for all data points in an adjacent condi- tion (that is, there is no overlap of data points between the highest values obtained in one condition and the lowest values obtained in the other condition), there is little doubt that behavior changed from one condition to the next. When many data points in adjacent conditions over- lap one another on the vertical axis, less confidence can be placed in the effect of the independent variable asso- ciated with the change in conditions.10

Mean or median level lines can be helpful in exam- ining the overall level between conditions. However, using mean or median level lines to summarize and com- pare the overall central tendency of data across condi- tions poses two serious problems. First, the viewer of such a visual display must guard against letting “appar- ently large differences among measures of central ten- dency visually overwhelm the presence of equally large amounts of uncontrolled variability” (Johnston & Pen- nypacker, 1980, p. 351). Emphasis on mean changes in performance in a graphic display can lead the viewer to believe that a greater degree of experimental control was obtained than is warranted by the data. In the top graph of Figure 23 half of the data points in Condition B fall within the range of values of the measures taken during

10Whether a documented change in behavior should be interpreted as a function of the independent variable depends on the experimental design used in the study. Strategies and tactics for designing experiments are pre- sented in chapters entitled “Analyzing Behavior Change: Basic Assump- tions and Strategies” through “Planning and Evaluating Applied Behavior Analysis Research.”

Constructing and Interpreting Graphic Displays of Behavioral Data

174

Figure 24 Stylized data paths illustrating the different combinations of change or lack of change in level and trend between two adjacent conditions: Graphs A and B show no change in either level or trend between the two conditions, Graphs C and D show changes in level and no change in trend, Graphs E and F depict no immediate change in level and a change in trend, and Graphs G and H reveal change in both level and trend.

From “Time-Series Analysis in Operant Research” by R. R. Jones, R. S. Vaught, and M. R. Weinrott, 1977, Journal of Applied Behavior Analysis, 10, p. 157. Copyright 1977 by the Society for the Experimental Analysis of Behavior, Inc. Adapted by permission.

Visual analysis of data between adjacent conditions includes an examination of the trends exhibited by the data in each condition to determine whether the trend found in the first condition changed in direction or slope during the subsequent condition. In practice, because each data point in a series contributes to level and trend, the two characteristics are viewed in conjunction with one another. Figure 24 presents stylized data paths illustrat- ing four basic combinations of change or lack of change in level and trend between adjacent conditions. Of course, many other data patterns could display the same charac- teristics. Idealized, straight-line data paths that eliminate the variability found in most repeated measures of be- havior have been used to highlight level and trend.

Visual analysis includes not only an examination and comparison of changes in level and trend between adja- cent conditions, but also an examination of performance across similar conditions. Interpreting what the data from an applied behavior analysis mean requires more than visual analysis and the identification and description of level, trend, and variability. When behavior change is demonstrated over the course of a treatment program or research study, the next question to be asked is, Was the change in behavior a function of the treatment or experi- mental variables?

Constructing and Interpreting Graphic Displays of Behavioral Data

175

Summary 1. Applied behavior analysts document and quantify behav-

ior change by direct and repeated measurement of behav- ior, and the product of those measurements called data.

2. Graphs are relatively simple formats for visually display- ing relationships among and between a series of mea- surements and relevant variables.

Purpose and Benefits of Graphic Display of Behavioral Data

3. Graphing each measure of behavior as it is collected pro- vides the practitioner or researcher with an immediate and ongoing visual record of the participant’s behavior, al- lowing treatment and experimental decisions to be re- sponsive to the participant’s performance.

4. Direct and continual contact with the data in a readily an- alyzable format enables the practitioner or researcher to identify and investigate interesting variations in behavior as they occur.

5. As a judgmental aid for interpreting experimental results, graphic display is a fast, relatively easy-to-learn method that imposes no arbitrary levels of significance for evalu- ating behavior change.

6. Visual analysis of graphed data is a conservative method for determining the significance of behavior change; only variables able to produce meaningful effects repeatedly are considered significant, and weak and unstable vari- ables are screened out.

7. Graphs enable and encourage independent judgments of the meaning and significance of behavior change by others.

8. Graphs can serve as effective sources of feedback to the people whose behavior they represent.

Types of Graphs Used in Applied Behavior Analysis

9. Line graphs, the most commonly used format for the graphic display of behavioral data, are based on the Carte- sian plane, a two-dimensional area formed by the inter- section of two perpendicular lines.

10. Major parts of the simple line graph are the horizontal axis (also called the x axis), the vertical axis (also called the y axis), condition change lines, condition labels, data points, the data path, and the figure caption.

11. Graphs with multiple data paths on the same set of axes are used in applied behavior analysis to show (a) two or more dimensions of the same behavior, (b) two or more differ- ent behaviors, (c) the same behavior under different and al- ternating experimental conditions, (d) changes in target behavior relative to the changing values of an independent variable, and (e) the behavior of two or more participants.

12. A second vertical axis, which is drawn on the right-hand side of the horizontal axis, is sometimes used to show dif- ferent scales for multiple data paths.

13. Bar graphs are used for two primary purposes: (a) to dis- play discrete data not related by an underlying dimension that can be used to scale the horizontal axis and (b) to sum- marize and enable easy comparison of the performance of a participant or group of participants during the different conditions of an experiment.

14. Each data point on a cumulative record represents the total number of responses emitted by the subject since mea- surement began. The steeper the slope of the data path on a cumulative graph, the higher the response rate.

15. Overall response rate refers to the average rate of re- sponse over a given time period; a local response rate refers to the rate of response during a smaller period of time within a larger period for which an overall response rate has been given.

16. Cumulative records are especially effective for displaying data when (a) the total number of responses made over time is important, (b) the graph is used as a source of feed- back to the subject, (c) the target behavior can occur only once per measurement period, and (d) a fine analysis of a single instance or portions of data from an experiment is desired.

17. Semilogarithmic charts use a logarithmic-scaled y axis so that changes in behavior that are of equal proportion (e.g., doublings of the response measure) are represented by equal distances on the vertical axis.

18. The Standard Celeration Chart is a six-cycle multiply- divide graph that enables the standardized charting of cel- eration, a linear measure of frequency change across time, a factor by which frequency multiplies or divides per unit of time.

19. A scatterplot shows the relative distribution of individual measures in a data set with respect to the variables de- picted by the x and y axes.

Constructing Line Graphs

20. The vertical axis is drawn to a length approximately two thirds that of the horizontal axis.

21. The horizontal axis is marked off in equal intervals, each representing from left to right the chronological succession of equal time periods within which behavior was measured.

22. Discontinuities of time are indicated on the horizontal axis by scale breaks.

23. The vertical axis is scaled relative to the dimension of be- havior measured, the range of values of the measures ob- tained, and the social significance of various levels of change in the target behavior.

24. Condition change lines indicate changes in the treatment program or manipulations of an independent variable and are drawn to the same height as the vertical axis.

Constructing and Interpreting Graphic Displays of Behavioral Data

176

25. A brief, descriptive label identifies each condition of an experiment or behavior change program.

26. Data points should be accurately placed with bold, solid dots. When multiple data paths are used, different geo- metric symbols are used to distinguish each data set.

27. Data paths are created by connecting successive data points with a straight line.

28. Successive data points should not be connected when (a) they fall on either side of a condition change line; (b) they span a significant period of time in which behavior was not measured; (c) they span discontinuities of time on the horizontal axis; (d) they fall on either side of a regularly scheduled measurement period in which data were not col- lected or were lost, destroyed, or otherwise not available; (e) they fall in a follow-up or postcheck period that is not regularly spaced in time in the same manner as the rest of the study; or (f) one member of the pair falls outside the range of values described by the vertical axis.

29. The figure caption provides a concise but complete de- scription of the graph, giving all of the information needed to interpret the display.

30. Graphs should be printed in black ink only.

Interpreting Graphically Displayed Behavioral Data

31. Visual analysis of graphed data attempts to answer two questions: (a) did a socially meaningful change in behav- ior take place, and (b) if so, can the behavior change be at- tributed to the independent variable?

32. Before beginning to evaluate the data displayed in a graph, a careful examination of the graph’s construction should be undertaken. If distortion is suspected from the features of the graph’s construction, the data should be replotted on a new set of axes before interpretation is attempted.

33. Blocked data and data representing the average perfor- mance of a group of subjects should be viewed with the un- derstanding that significant variability may have been lost in the display.

34. Visual analysis of data within a given condition focuses on the number of data points, the variability of perfor-

mance, the level of performance, and the direction and de- gree of any trends in the data.

35. As a general rule, the more data in a condition and the greater the stability of those data, the more confidence one can place in the data path’s estimate of behavior during that time. The more variability in the behavioral measures during a condition, the greater the need for additional data.

36. Variability refers to the frequency and degree to which multiple measures of behavior yield different outcomes. A high degree of variability within a given condition usually indicates that little or no control has been achieved over the factors influencing the behavior.

37. Level refers to the value on the vertical axis around which a series of data points converges. When the data in a given condition all fall at or near a specific level, the behavior is considered stable with respect to level; to the extent that the behavioral measures vary considerably from one to an- other, the data are described as showing variability with re- spect to level. In cases of extreme variability, no particular level of performance is evidenced.

38. Mean or median level lines are sometimes added to graphic displays to represent the overall average or typical per- formance during a condition. Mean and median level lines should be used and interpreted with care because they can obscure important variability and trends in the data.

39. Trend refers to the overall direction taken by a data path; trends are described in terms of their direction (increas- ing, decreasing, or zero trend), degree (gradual or steep), and the extent of variability of data points around the trend.

40. Trend direction and degree can be visually represented by drawing a trend line, or line of progress, through a series of data points. Trend lines can be drawn freehand, using the least-squares regression equation, or using a method called the split-middle line of progress. Split-middle lines of progress can be drawn quickly and reliably and have proven useful in analyzing behavior change.

41. Visual analysis of data across conditions determines whether change in level, variability, and/or trend occurred and to what extent any changes were significant.

Constructing and Interpreting Graphic Displays of Behavioral Data

177

A-B design affirmation of the consequent ascending baseline baseline baseline logic confounding variable dependent variable descending baseline experimental control

experimental design experimental question external validity extraneous variable independent variable internal validity parametric analysis practice effects

prediction replication single-subject designs stable baseline steady state responding steady state strategy variable baseline verification

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 3: Principles, Processes, and Concepts

3–10 Define and provide examples of functional relations.

Content Area 5: Experimental Evaluation of Interventions

5–1 Systematically manipulate independent variables to analyze their effects on treatment.

5–2 Identify and address practical and ethical considerations in using various ex- perimental designs.

5–4 Conduct a parametric analysis (i.e., determining effective parametric values of consequences, such as duration or magnitude).

Analyzing Behavior Change: Basic Assumptions and Strategies

Key Terms

178

Measurement can show whether, when, and how much behavior has changed, but measurement alone cannot reveal why, or more accurately

how, behavior change occurred. A useful technology of behavior change requires an understanding of the spe- cific arrangements of environmental variables that will produce desired behavior change. Without this knowl- edge, efforts to change behavior could only be consid- ered one-shot affairs consisting of procedures selected randomly from a bag of tricks with little or no general- ity from one situation to the next.

The search for and demonstration of functional and reliable relations between socially important behavior and its controlling variables is a defining characteristic of applied behavior analysis. A major strength of applied behavior analysis is its insistence on experimentation as its method of proof, which enables and demands an on- going, self-correcting search for effectiveness.

Our technology of behavior change is also a technol- ogy of behavior measurement and of experimental design; it developed as that package, and as long as it stays in that package, it is a self-evaluating enterprise. Its successes are successes of known magnitude; its failures are almost immediately detected as failures; and whatever its outcomes, they are attributable to known inputs and procedures rather than to chance events or coincidences. (D. M. Baer, personal com- munication, October 21, 1982)

An experimental analysis must be accomplished to determine if and how a given behavior functions in rela- tion to specific changes in the environment. This chapter introduces the basic concepts and strategies that underlie the analysis in applied behavior analysis.1 The chapter begins with a brief review of some general conceptions of science, followed by a discussion of two defining fea- tures and two assumptions about the nature of behavior that dictate the experimental methods most conducive to the subject matter. The chapter then describes the neces- sary components of any experiment in applied behavior analysis and concludes by explaining the basic logic that guides the experimental methods used by applied be- havior analysts.

1The analysis of behavior has benefited immensely from two particu- larly noteworthy contributions to the literature on experimental method: Sidman’s Tactics of Scientific Research (1960/1988) and Johnston and Pennypacker’s Strategies and Tactics of Human Behavioral Research (1980, 1993a). Both books are essential reading and working references for any serious student or practitioner of behavior analysis, and we ac- knowledge the significant part each has played in the preparation of this chapter.

Concepts and Assumptions Underlying the Analysis of Behavior Scientists share a set of common perspectives that in- clude assumptions about the nature of the phenomena they study (determinism), the kind of information that should be gathered on the phenomena of interest (em- piricism), the way questions about the workings of na- ture are most effectively examined (experimentation), and how the results of experiments should be judged (with parsimony and philosophic doubt). These attitudes apply to all scientific disciplines, including the scientific study of behavior. “The basic characteristics of science are not restricted to any particular subject matter” (Skin- ner, 1953, p. 11).

The overall goal of science is to achieve an under- standing of the phenomena under study—socially sig- nificant behavior, in the case of applied behavior analysis. Science enables various degrees of understanding at three levels: description, prediction, and control. First, sys- tematic observation enhances the understanding of natural phenomena by enabling scientists to describe them ac- curately. Descriptive knowledge of this type yields a col- lection of facts about the observed events—facts that can be quantified and classified, a necessary and important element of any scientific discipline.

A second level of scientific understanding occurs when repeated observation discovers that two events con- sistently covary. That is, the occurrence of one event (e.g., marriage) is associated with the occurrence of another event at some reliable degree of probability (e.g., longer life expectancy). The systematic covariation between two events—termed a correlation—can be used to predict the probability that one event will occur based on the pres- ence of the other event.

The ability to predict successfully is a useful result of science; prediction allows preparation. However, the greatest potential benefits of science are derived from the third, and highest, level of scientific understanding, which comes from establishing experimental control. “The experimental method is a method for isolating the relevant variables within a pattern of events. Methods that depend merely on observed correlations, without ex- perimental intervention, are inherently ambiguous” (Dinsmoor, 2003, p. 152).

Experimental Control: The Path to and Goal of Behavior Analysis

Behavior is the interaction between an organism and its environment and is best analyzed by measuring changes in behavior that result from imposed variations on the

Analyzing Change: Basic Assumptions and Strategies

179

Analyzing Change: Basic Assumptions and Strategies

environment. This statement embodies the general strat- egy and the goal of behavioral research: to demonstrate that measured changes in the target behavior occur be- cause of experimentally manipulated changes in the en- vironment.

Experimental control is achieved when a pre- dictable change in behavior (the dependent variable) can be reliably produced by the systematic manipulation of some aspect of the person’s environment (the indepen- dent variable). Experimentally determining the effects of environmental manipulation on behavior and demon- strating that those effects can be reliably produced con- stitute the analysis in applied behavior analysis. An analysis of a behavior has been achieved when a reliable functional relation between the behavior and some spec- ified aspect of the environment has been demonstrated convincingly. Knowledge of functional relations enables the behavior analyst to reliably alter behavior in mean- ingful ways.

An analysis of behavior “requires a believable demonstration of the events that can be responsible for the occurrence or nonoccurrence of that behavior. An exper- imenter has achieved an analysis of a behavior when he can exercise control over it” (Baer, Wolf, & Risley, 1968, p. 94).2 Baer and colleague’s original definition of analy- sis highlights an important point. Behavior analysis’ seeking and valuing of the experimental isolation of a given environmental variable of which a behavior is shown to be a function has often been misinterpreted as support for a simplistic conception of the causes of be- havior. The fact that a behavior varies as a function of a given variable does not preclude its varying as a function of other variables. Thus, Baer and colleagues described an experimental analysis as a convincing demonstration that a variable can be responsible for the observed be- havior change. Even though a complete analysis (i.e., un- derstanding) of a behavior has not been achieved until all of its multiple causes have been accounted for, an app- lied (i.e., technologically useful) analysis has been ac- complished when the investigator has isolated an environmental variable (or group of variables that oper- ate together as a treatment package) that reliably pro- duces socially significant behavior change. An applied analysis of behavior also requires that the target behav- ior be a function of an environmental event that can be practically and ethically manipulated.

Experiments that show convincingly that changes in behavior are a function of the independent variable and are not the result of uncontrolled or unknown variables are said to have a high degree of internal validity. A

2The researcher’s audience ultimately determines whether a claimed func- tional relation is believable, or convincing. We will explore the believabil- ity of research findings further in chapter entitled “Planning and Evaluating Applied Behavior Analysis Research.”

study without internal validity can yield no meaningful statements regarding functional relations between the variables examined in the experiment, nor can it be used as the basis for any statements regarding the generality of the findings to other persons, settings, and/or behaviors.3

When initially planning an experiment and later when examining the actual data from an ongoing study, the investigator must always be on the lookout for threats to internal validity. Uncontrolled variables known or sus- pected to exert an influence on the dependent variable are called confounding variables. For example, suppose a researcher wants to analyze the effects of guided lecture notes on high school biology students’ learning as mea- sured by their scores on next-day quizzes. One potential confounding variable that the researcher would need to take into account would be each student’s changing level of interest in and background knowledge about the spe- cific curriculum content (e.g., a student’s high score on a quiz following a lecture on sea life may be due to his prior knowledge about fishing, not the guided notes pro- vided during that lecture).

A primary factor in evaluating the internal validity of an experiment is the extent to which it eliminates or controls the effects of confounding variables while still investigating the research questions of interest. It is impossible to eliminate all sources of uncontrolled vari- ability in an experiment, although the researcher always strives for that ideal. In reality, the goal of experimental design is to eliminate as many uncontrolled variables as possible and to hold constant the influence of all other variables except the independent variable, which is pur- posefully manipulated to determine its effects.

Behavior: Defining Features and Assumptions that Guide Its Analysis

Behavior is a difficult subject matter, not because it is inaccessible, but because it is extremely complex. Since it is a process, rather than a thing, it cannot easily be held still for observation. It is changing, fluid, evanes- cent, and for this reason it makes great technical de- mands upon the ingenuity and energy of the scientist.

—B. F. Skinner (1953, p. 15)

How a science defines its subject matter exerts profound influence and imposes certain constraints on the experi- mental strategies that will be most effective in an under- standing of it. “In order for the scientific study of behavior to be as effective as possible, it is necessary for the meth- ods of the science to accommodate the characteristics of its subject matter” (Johnston & Pennypacker, 1993a,

3External validity commonly refers to the degree to which a study’s results are generalizable to other subjects, settings, and/or behaviors. Strategies for assessing and extending the external validity of experimentally demon- strated functional relations are discussed in Chapter entitled “Planning and Evaluating Applied Behavior Analysis Research.”

180

Analyzing Change: Basic Assumptions and Strategies

p. 117). The experimental methods of behavior analysis are guided by two defining features of behavior: (a) the fact that behavior is an individual phenomenon and (b) the fact that behavior is a continuous phenomenon; and by two assumptions about its nature: (a) that behav- ior is determined and (b) that behavioral variability is ex- trinsic to the organism.

Behavior Is an Individual Phenomenon

If behavior is defined as a person’s interaction with the environment, it follows that a science seeking to discover general principles or laws that govern behavior must study the behavior of individuals. Groups of people do not behave; individual people do. Thus, the experimental strategy of behavior analysis is based on within-subject (or single-subject) methods of analysis.

The average performance of a group of individuals is often interesting and useful information and may, de- pending on the methods by which individuals were se- lected to be in the group, enable probability statements about the average performance within the larger popula- tion represented by the group. However, “group data” provide no information about the behavior of any indi- vidual or how any individual might perform in the fu- ture. For example, although administrators and taxpayers may be justifiably interested in the average increase in students’ reading comprehension from grade level to grade level, such information is of little use to the class- room teacher who must decide how to improve a given student’s comprehension skills.

Nonetheless, learning how behavior–environment relations work with many individuals is vital. A science of behavior contributes to a useful technology of be- havior change only to the extent that it discovers func- tional relations with generality across individuals. The issue is how to achieve that generality. Behavior ana- lysts have found that discovery of behavioral principles with generality across persons is best accomplished by replicating the already demonstrated functional relations with additional subjects.

Behavior Is a Dynamic, Continuous Phenomenon

Just as behavior cannot take place in an environmental void (it must happen somewhere), so must behavior occur at particular points in time. Behavior is not a static event; it takes place in and changes over time. Therefore, single measures, or even multiple measures sporadically dis- persed over time, cannot provide an adequate description of behavior. Only continuous measurement over time yields a complete record of behavior as it occurs in con-

tinuous measurement is seldom feasible in applied set- tings, the systematic repeated measurement of behavior has become the hallmark of applied behavior analysis.

Behavior Is Determined

All scientists hold the assumption that the universe is a lawful and orderly place and that natural phenomena occur in relation to other natural events.

The touchstone of all scientific research is order. In the experimental analysis of behavior, the orderliness of rela- tions between environmental variables and the subject’s behavior is at once the operating assumption upon which the experimenter proceeds, the observed fact that permits doing so, and the goal that continuously focuses experi- mental decisions. That is, the experimenter begins with the assumption that the subject’s behavior is the result of variables in the environment (as opposed to having no causes at all). (Johnston & Pennypacker, 1993a, p. 238)

In other words, the occurrence of any event is deter- mined by the functional relations it holds with other events. Behavior analysts consider behavior to be a nat- ural phenomenon that, like all natural phenomena, is de- termined. Although determinism must always remain an assumption—it cannot be proven—it is an assumption with strong empirical support.

Data gathered from all scientific fields indicate that de- terminism holds throughout nature. It has become clear that the law of determinism, that is, that all things are determined, holds for the behavioral area also. . . . When looking at actual behavior we’ve found that in situation 1, behavior is caused; in situation 2, behavior is caused; in situation 3, behavior is caused; . . . and in situation 1001, behavior is caused. Every time an experimenter introduces an independent variable that produces some behavior or some change in behavior, we have further empirical evidence that behavior is caused or determin- istic. (Malott, General, & Snapper, 1973, pp. 170, 175)

Behavioral Variability Is Extrinsic to the Organism

When all conditions during a given phase of an experi- ment are held constant and repeated measures of the be- havior result in a great deal of “bounce” in the data (i.e., the subject is not responding in a consistent fashion), the behavior is said to display variability.

The experimental approach most commonly used in psychology and other social and behavioral sciences (e.g., education, sociology, political science) makes two as- sumptions about such variability: (a) Behavioral vari- ability is an intrinsic characteristic of the organism, and

text with its environmental influences. Because true con-

181

Analyzing Change: Basic Assumptions and Strategies

(b) behavioral variability is distributed randomly among individuals in any given population. These two assump- tions have critical methodological implications: (a) At- tempting to experimentally control or investigate variability is a waste of time—it simply exists, it’s a given; and (b) by averaging the performance of individ- ual subjects within large groups, the random nature of variability can be statistically controlled or canceled out. Both of these assumptions about variability are likely false (empirical evidence points in the opposite direc- tion), and the methods they encourage are detrimental to a science of behavior. “Variables are not canceled statis- tically. They are simply buried so their effects are not seen” (Sidman, 1960/1988, p. 162).4

Behavior analysts approach variability in their data quite differently. A fundamental assumption underlying the design and guiding the conduct of experiments in behavior analysis is that, rather than being an intrinsic characteristic of the organism, behavioral variability is the result of environmental influence: the independent variable with which the investigator seeks to produce change, some uncontrolled aspect of the experiment it- self, and/or an uncontrolled or unknown factor outside of the experiment.

The assumption of extrinsic variability yields the fol- lowing methodological implication: Instead of averaging the performance of many subjects in an attempt to mask variability (and as a result forfeiting the opportunity to un- derstand and control it), the behavior analyst experi- mentally manipulates factors suspected of causing the variability. Searching for the causal factors contributes to the understanding of behavior, because experimental demonstration of a source of variability implies experi- mental control and thus another functional relation. In fact, “tracking down these answers may even turn out to be more rewarding than answering the original experi- mental question” (Johnston & Pennypacker, 1980 p. 226).

From a purely scientific viewpoint, experimentally tracking down sources of variability is always the pre- ferred approach. However, the applied behavior analyst, with a problem to solve, must often take variability as it presents itself (Sidman, 1960/1988). Sometimes the ap- plied researcher has neither the time nor the resources to experimentally manipulate even suspected and likely sources of variability (e.g., a teacher who interacts with a student for only part of the day has no hope of control- ling the many variables outside the classroom). In most settings the applied behavior analyst seeks a treatment variable robust enough to overcome the variability in-

4Some investigators use group comparison designs not just to cancel ran- domly distributed variability but also to produce results that they believe will have more external validity. Group comparison and within subject ex- perimental methods are compared in Chapter entitled “Planning and Eval-

duced by uncontrolled variables and produce the desired effects on the target behavior (Baer, 1977b).

Components of Experiments in Applied Behavior Analysis

Nature to be commanded must be obeyed. . . . But, that coin has another face. Once obeyed, nature can be commanded.

—B. F. Skinner (1956, p. 232)

Experimentation is the scientist’s way of discovering na- ture’s rules. Discoveries that prove valid and reliable can contribute to a technology of effective behavior change. All experiments in applied behavior analysis include these essential components:

• At least one participant (subject)

• At least one behavior (dependent variable)

• At least one setting

• A system for measuring the behavior and ongoing visual analysis of the data

• At least one treatment or intervention condition (independent variable)

• Manipulations of the independent variable so that its effects on the dependent variable, if any, can be detected (experimental design)

Because the reason for conducting any experiment is to learn something from nature, a well-planned exper- iment begins with a specific question for nature.

Experimental Question We conduct experiments to find out something we do not know.

—Murray Sidman (1960/1988, p. 214)

For the applied behavior analyst, Sidman’s “something we do not know” is cast in the form of a question about the existence and/or specific nature of a functional rela- tion between meaningful improvement in socially sig- nificant behavior and one or more of its controlling variables. An experimental question is “a brief but spe- cific statement of what the researcher wants to learn from conducting the experiment” (Johnston & Pennypacker (1993b, p. 366). In published reports of applied behavior analysis studies, the experimental (or research) question is sometimes stated explicitly in the form of a question, as in these examples:

• Which method of self-correction, after attempting each of 10 words or after attempting a list of 10uating Applied Behavior Analysis Research.”

182

Analyzing Change: Basic Assumptions and Strategies

words, will produce better effects on (a) the acqui- sition of new spelling words as measured by end- of-the-week tests, and (b) the maintenance of practiced spelling words as measured by 1-week maintenance tests by elementary school students with learning disabilities? (Morton, Heward, & Alber, 1998)

• What are the effects of training middle school students with learning disabilities to recruit teacher attention in the special education class- room on (a) the number of recruiting responses they emit in the general education classroom, (b) the number of teacher praise statements re- ceived by the students in the general education classroom, (c) the number of instructional feed- back statements received by the students in the general education classroom, and (d) the stu- dents’ academic productivity and accuracy in the general education classroom? (Alber, Heward, & Hippler, 1999, p. 255)

More often, however, the research question exam- ined by the experiment is implicit within a statement of the study’s purpose. For example:

• The purpose of the present study was to compare the relative effectiveness of nonremoval of the spoon and physical guidance as treatments for food refusal and to assess the occurrence of cor- ollary behaviors produced by each procedure. (Ahearn, Kerwin, Eicher, Shantz, & Swearingin, 1996, p. 322)

• The present study was conducted to determine if habit reversal is effective in treating verbal tics in children with Tourette syndrome. (Woods, Twohig, Flessner, & Roloff, 2003, p. 109)

• The purpose of this study was to determine if ob- served SIB during the tangible condition was con- founded by the simultaneous delivery of therapist attention. (Moore, Mueller, Dubard, Roberts, & Sterling-Turner, 2002, p. 283)

• The purpose of this study was to determine whether naturally occurring meals would affect performance adversely during postmeal sessions in which highly preferred food was used as re- inforcement. (Zhou, Iwata, & Shore, 2002, pp. 411–412)

Whether an experimental question is stated explicitly in the form of a question or implicit within a statement of purpose, all aspects of an experiment’s design and con- duct should follow from it.

5It has become commonplace in the behavior analysis literature to refer to the person(s) whose behavior is the dependent variable in an experiment as a participant, instead of the more traditional term, subject. We use both terms in this text and urge readers to consider Sidman’s (2002) perspective on the issue: “[W]e are no longer permitted to call our subjects ‘subjects.’ The term is supposed to be dehumanizing, and so we are supposed to call them ‘participants.’ I think this is completely misguided. Experimenters, too, are participants in their experiments. What does making them non- participants do to our perception of science and of scientists? Are experi- menters merely robots who follow prescribed and unbreakable scientific rules? Are they supposed just to manipulate variables and coldly record the results of their manipulations? Separating them as nonparticipating ma- nipulators and recorders of the behavior of participants really dehumanizes not only experimenters but, along with them, the whole scientific process.” (p. 9) 6An experiment by Rindfuss, Al-Attrash, Morrison, and Heward (1998) provides a good example of the extent to which the term single-subject re- search can be a misnomer. A within-subject reversal design was used to evaluate the effects of response cards on the quiz and exam scores of 85 stu- dents in five eighth-grade American history classes. Although a large group of subjects participated in this study, it actually consisted of 85 individual experiments; or 1 experiment and 84 replications!

A good design is one that answers the question convinc- ingly, and as such needs to be constructed in reaction to the question and then tested through arguments in that context (sometimes called, “thinking through”), rather than imitated from a textbook. (Baer, Wolf, & Risley, 1987, p. 319)

Subject

Experiments in applied behavior analysis are most often referred to as single-subject (or single-case) designs. This is not because behavior analysis studies are neces- sarily conducted with only one subject (though some are), but because the experimental logic or reasoning for ana- lyzing behavior changes often employs the subject as her own control.5 In other words, repeated measures of each subject’s behavior are obtained as she is exposed to each condition of the study (e.g., the presence and absence of the independent variable). A subject is often exposed to each condition several times over the course of an ex- periment. Measures of the subject’s behavior during each phase of the study provide the basis for comparing the effects of experimental variables as they are presented or withdrawn in subsequent conditions.

Although most applied behavior analysis studies in- volve more than one subject (four to eight is common), each subject’s data are graphed and analyzed separately.6

Instead of using single-subject design to refer to the ex- periments in which each subject serves as his or her own control, some authors use more aptly descriptive terms such as within-subject design or intra-subject design.

Sometimes the behavior analyst is interested in as- sessing the total effect of a treatment variable within a group of subjects—for example, the number of home- work assignments completed by members of a class of

183

Analyzing Change: Basic Assumptions and Strategies

fifth-grade students. In such cases the total number of as- signments completed may be measured, graphed, and an- alyzed as a dependent variable within a “single-subject” design. However, it must be remembered that unless each student’s data are individually graphed and interpreted, no individual student’s behavior has been analyzed, and the data for the group may not be representative of any indi- vidual subject.

Use of a single participant, or a small number of par- ticipants, each of whom is considered an intact experi- ment, stands in sharp contrast to the group comparison designs traditionally used in psychology and the other social sciences that employ large numbers of subjects.7

Proponents of group comparison designs believe that large numbers of subjects control for the variability dis- cussed earlier and increase the generality (or external va- lidity) of any findings to the population from which the subjects were selected. For now, we will leave this issue with Johnston and Pennypacker’s (1993b) astute observation:

When well done, the procedures of within-subject de- signs preserve the pure characteristics of behavior, un- contaminated by intersubject variability. In contrast, the best between groups design practices obfuscate the rep- resentation of behavior in various ways, particularly by mixing intersubject variability with treatment-induced variability. (p. 188)

Behavior: Dependent Variable

The target behavior in an applied behavior analysis ex- periment, or more precisely a measurable dimensional quantity of that behavior (e.g., rate, duration), is called the dependent variable. It is so labeled because the ex- periment is designed precisely to determine whether the behavior is, in fact, dependent on (i.e., a function of ) the independent variable(s) manipulated by the investigator.

In some studies more than one behavior is measured. One reason for measuring multiple behaviors is to provide data patterns that can serve as controls for evaluating and replicating the effects of an independent variable as it is

7For a history of single-case research, see Kennedy (2005).

on behaviors other than the response class to which it was applied directly. This strategy is used to determine whether the independent variable had any collateral ef- fects—either desired or undesired—on other behaviors of interest. Such behaviors are referred to as secondary dependent variables. The experimenter obtains regular measures of their rate of occurrence, though perhaps not with the same frequency with which measures of the pri- mary dependent variable are recorded.

Still another reason for measuring multiple behav- iors is to determine whether changes in the behavior of a person other than the subject occur during the course of an experiment and whether such changes might in turn explain observed changes in the subject’s behav- ior. This strategy is implemented primarily as a control strategy in assessing the effects of a suspected con- founding variable: The extra behavior(s) measured are not true dependent variables in the sense of undergoing analysis. For example, in a classic study analyzing the effects of the self-recording by a junior high school girl on her classroom study behavior, Broden, Hall, and Mitts (1971) observed and recorded the number of times the girl’s teacher paid attention to her throughout the experiment. If teacher attention had been found to co- vary with changes in study behavior, a functional rela- tion between self-recording and study behavior would not have been demonstrated. In that case, teacher at- tention would likely have been identified as a potential confounding variable, and the focus of the investigation would likely have shifted to include efforts to experi- mentally control it (i.e., to hold teacher attention con- stant) or to systematically manipulate and analyze its effects. However, the data revealed no functional rela- tion between teacher attention and study behavior dur- ing the first four phases of the experiment, when concern was highest that teacher attention may have been a confounding variable.

Setting Control the environment and you will see order in behavior.

—B. F. Skinner (1967, p. 399)

Functional relations are demonstrated when observed vari- ations in behavior can be attributed to specific operations imposed on the environment. Experimental control is

8This is the distinguishing feature of the multiple baseline across behav- iors design, an experimental tactic used widely in applied behavior analy- sis. Multiple baseline designs are presented in Chapter entitled “Multiple Baseline and Changing Criterion Designs ”.

sequentially applied to each of the behaviors.8 A second reason for multiple dependent measures is to assess the presence and extent of the independent variable’s effects

184

Analyzing Change: Basic Assumptions and Strategies

achieved when a predictable change in behavior (the de- pendent variable) can be reliably and repeatedly produced by the systematic manipulation of some aspect of the sub- ject’s environment (the independent variable). To make such attributions properly, the investigator must, among other things, control two sets of environmental variables. First, the investigator must control the independent variable by presenting it, withdrawing it, and/or varying its value. Second, the investigator must control, by holding constant, all other aspects of the experimental setting—extraneous variables—to prevent unplanned environmental variation. These two operations—precisely manipulating the inde- pendent variable and maintaining the constancy of every other relevant aspect of the experimental setting—define the second meaning of experimental control.

In basic laboratory research, experimental space is designed and furnished to maximize experimental con- trol. Lighting, temperature, and sound, for example, are all held constant, and programmed apparatus virtually guarantee the presentation of antecedent stimuli and the delivery of consequences as planned. Applied behavior analysts, however, conduct their studies in the settings where socially important behaviors naturally occur—the classroom, home, and workplace. It is impossible to con- trol every feature of an applied environment; and to add to the difficulty, subjects are typically in the experimen- tal setting for only part of each day, bringing with them the influences of events and contingencies operating in other settings.

In spite of the complexity and ever-changing nature of applied settings, the behavior analyst must make every effort to hold constant all seemingly relevant aspects of the environment. When unplanned variations take place, the investigator must either wait out their effects or try to incorporate them into the design of the experiment. In any event, repeated measures of the subject’s behavior are the barometer for assessing whether unplanned envi- ronmental changes are of concern.

Applied studies are often conducted in more than one setting. Researchers sometimes use concurrent mea- sures of the same behavior obtained in multiple settings as controls for analyzing the effects of an independent variable that is sequentially applied to the behavior in each setting.9 In addition, data are often collected in mul- tiple settings to assess the extent to which behavior changes observed in the primary setting have also oc- curred in the other setting(s).

9This analytic tactic is known as a multiple-baseline across settings de- sign. Multiple-baseline designs are presented inChapter entitled “Multiple Baseline and Changing Criterion Designs ”

Measurement System and Ongoing Visual Analysis

Beginning students of behavior analysis sometimes be- lieve that the discipline is preoccupied with issues and procedures related to the observation and measurement of behavior. They want to get on with the analysis. How- ever, the results of any experiment can be presented and interpreted only in terms of what was measured, and the observation and recording procedures used in the study determine not only what was measured, but also how well it was measured (i.e., how representative of the subject’s actual behavior is the estimate provided by the experi- mental data—all measurements of behavior, no matter how frequent and technically precise, are estimates of true values). It is critical that observation and recording procedures be conducted in a completely standardized manner throughout each session of an experiment. Stan- dardization involves every aspect of the measurement system, from the definition of the target behavior (de- pendent variable) to the scheduling of observations to the manner in which the raw data are transposed from record- ing sheets to session summary sheets to the way the data are graphed. An adventitious change in measurement tac- tics can result in unwanted variability or confounded treatment effects.

Advantages accrue to the behavioral researcher who maintains direct contact with the experimental data by ongoing visual inspection of graphic displays. The be- havior analyst must become skilled at recognizing changes in level, trend, and degree of variability as these changes develop in the data. Because behavior is a con- tinuous, dynamic phenomenon, experiments designed to discover its controlling variables must enable the inves- tigator to inspect and respond to the data continuously as the study progresses. Only in this way can the behavior analyst be ready to manipulate features of the environ- ment at the time and in the manner that will best reveal functional relations and minimize the effects of con- founding variables.

Intervention or Treatment: Independent Variable

Behavior analysts seek reliable relations between behav- ior and the environmental variables of which it is a func- tion. The particular aspect of the environment that the experimenter manipulates to find out whether it affects the subject’s behavior is called the independent vari- able. Sometimes called the intervention, treatment, or experimental variable, this component of an experiment is called the independent variable because the researcher.

185

Analyzing Change: Basic Assumptions and Strategies

can control or manipulate it independent of the subject’s behavior or any other event. (Though, as we will soon see, manipulating the independent variable without re- gard to what is happening with the dependent variable is unwise.) Whereas any changes that must be made in the experimental setting to conduct the study (e.g., the addi- tion of observers to measure behavior) are made with the goal of minimizing their effects on the dependent vari- able, “changes in the independent variable are arranged by the experimenter in order to maximize . . . its influence on responding” (Johnston & Pennypacker, 1980, p. 260).

Manipulations of the Independent Variable: Experimental Design

Experimental design refers to the particular arrange- ment of conditions in a study so that meaningful com- parisons of the effects of the presence, absence, or different values of the independent variable can be made. Independent variables can be introduced, withdrawn, in- creased or decreased in value, or combined across be- haviors, settings, and/or subjects in an infinite number of ways.10 However, there are only two basic kinds of inde- pendent variable changes that can be made with respect to the behavior of a given subject in a given setting.

A new condition can be introduced or an old condition can be reintroduced. . . . Experimental designs are merely temporal arrangements of various new and old conditions across behaviors and settings in ways that produce data that are convincing to the investigator and the audience. (Johnston & Pennypacker, 1980, p. 270)

In the simplest case—from an analytic perspective, but not necessarily a practical point of view—an inde- pendent variable can be manipulated so that it is either present or absent during each time period or phase of the study. When the independent variable is in either of these conditions during a study, the experiment is termed a non- parametric study. In contrast, a parametric analysis seeks to discover the differential effects of a range of val- ues of the independent variable. For example, Lerman, Kelley, Vorndran, Kuhn, and LaRue (2002) conducted a parametric study when they assessed the effects of dif- ferent reinforcer magnitudes (i.e., 20 seconds, 60 sec- onds, or 300 seconds of access to toys or escape from demands) on the duration of postreinforcement pause and resistance to extinction. Parametric experiments are sometimes used because a functional relation may have

10How many different experimental designs are there? Because an exper- iment’s design includes careful selection and consideration of all of the components discussed here (i.e., subject, setting, behavior, etc.), not count- ing the direct replication of experiments, one could say that there are as many experimental designs as there are experiments.

more generality if it is based on several values of the in- dependent variable.

Sometimes the investigator is interested in compar- ing the effects of several treatment alternatives. In this case multiple independent variables become part of the experiment. For example, perhaps two separate treat- ments are evaluated as well as the effects of a third treat- ment, which represents a combination of both variables. However, even in experiments with multiple independent variables, the researcher must heed a simple but funda- mental rule of experimentation: Change only one vari- able at a time. Only in this manner can the behavior analyst attribute any measured changes in behavior to a specific independent variable. If two or more variables are altered simultaneously and changes in the dependent variable are noted, no conclusions can be made with re- gard to the contribution of any one of the altered vari- ables to the behavior change. If two variables changed together, both could have contributed equally to the re- sultant behavior change; one variable could have been solely, or mostly, responsible for the change; or one vari- able may have had a negative or counterproductive ef- fect, but the other independent variable was sufficiently strong enough to overcome this effect, resulting in a net gain. Any of these explanations or combinations may have accounted for the change.

As stated previously, applied behavior analysts often conduct their experiments in “noisy” environments where effective treatments are required for reasons related to personal safety or exigent circumstances. In such cases, applied behavior analysts sometimes “package” multiple and well-documented and effective treatments, knowing that multiple independent variables are being introduced. As implied earlier, a package intervention is one in which multiple independent variables are being combined or bundled into one program (e.g., token reinforcement + praise, + self-recording + time out). However, from the perspective of experimental analysis, the rule still holds. When manipulating a treatment package, the experi- menter must ensure that the entire package is presented or withdrawn each time a manipulation occurs. In this situation, it is important to understand that the entire package is being evaluated, not the discrete components that make up the package. If at a later time, the analyst wishes to determine the relative contributions of each part of the package, a component analysis would need to be carried out.

There are no off-the-shelf experimental designs avail- able for a given research problem (Baer et al., 1987; John- ston & Pennypacker, 1980, 1993a; Sidman, 1960/1988; Skinner, 1956, 1966). The investigator must not get locked into textbook “designs” that (a) require a priori assumptions about the nature of the functional relations

186

Analyzing Change: Basic Assumptions and Strategies

one seeks to investigate and (b) may be insensitive to unanticipated changes in behavior. Instead, the behavior analyst should select and combine experimental tactics that best fit the research question, while standing ever ready to “explore relevant variables by manipulating them in an improvised and rapidly changing design” (Skinner, 1966, p. 21).

Simultaneous with, and to a large degree responsible for, the growth and success of applied behavior analysis have been the development and refinement of a powerful group of extremely flexible experimental tactics for ana- lyzing behavior-environment relations. However, to most effectively select, modify, and combine these tactics into convincing experiments, the behavior analyst must first fully understand the experimental reasoning, or logic, that provides the foundation for within-subject experimental comparisons.

Steady State Strategy and Baseline Logic Steady or stable state responding—which “may be de- fined as a pattern of responding that exhibits relatively little variation in its measured dimensional quantities over a period of time” (Johnston & Pennypacker, 1993a, p. 199)—provides the basis for a powerful form of ex- perimental reasoning commonly used in behavior analy- sis, called baseline logic. Baseline logic entails three elements—prediction, verification, and replication—each of which depends on an overall experimental approach called steady state strategy. Steady state strategy entails repeatedly exposing a subject to a given condition while trying to eliminate or control any extraneous influences on the behavior and obtaining a stable pattern of re- sponding before introducing the next condition.

Nature and Function of Baseline Data

Behavior analysts discover behavior–environment rela- tions by comparing data generated by repeated measures of a subject’s behavior under the different environmental conditions of the experiment. The most common method of evaluating the effects of a given variable is to impose it on an ongoing measure of behavior obtained in its ab- sence. These original data serve as the baseline against which any observed changes in behavior when the inde- pendent variable is applied can be compared. A baseline serves as a control condition and does not necessarily mean the absence of instruction or treatment as such, only the absence of a specific independent variable of exper- imental interest.

Why Establish a Baseline?

From a purely scientific or analytic perspective, the pri- mary purpose for establishing a baseline level of re- sponding is to use the subject’s performance in the absence of the independent variable as an objective basis for detecting the effects of the independent variable when it is introduced in the future. However, obtaining baseline data can yield a number of applied benefits. For one, sys- tematic observation of the target behavior before a treat- ment variable is introduced provides the opportunity to look for and note environmental events that occur just before and just after the behavior. Such empirically ob- tained descriptions of antecedent-behavior-consequent correlations are often invaluable in planning an effective intervention. For example, baseline observations reveal- ing that a child’s disruptive outbursts are consistently fol- lowed by parent or teacher attention can be used in designing an intervention of ignoring outbursts and con- tingent attention following desired behavior.

Second, baseline data can provide valuable guidance in setting initial criteria for reinforcement, a particularly important step when a contingency is first put into effect. If the criteria are too high, the subject never comes into contact with the contingency; if they are too low, little or no improvement can be expected.

From a practical perspective, a third reason for col- lecting baseline data concerns the merits of objective measurement versus subjective opinion. Sometimes the results of systematic baseline measurement convince the behavior analyst or significant others to alter their per- spectives on the necessity and value of attempting to change the behavior. For example, a behavior being con- sidered for intervention because of several recent and ex- treme instances is no longer targeted because baseline data show it is decreasing. Or, perhaps a behavior’s topog- raphy attracted undue attention from teachers or parents, but objective baseline measurement over several days re- veals that the behavior is not occurring at a frequency that warrants an intervention.

Types of Baseline Data Patterns

Examples of four data patterns sometimes generated by baseline measurement are shown in Figure 1. It must be stressed that these hypothetical baselines represent only four examples of the wide variety of baseline data pat- terns an experimenter or practitioner will encounter. The potential combinations of different levels, trends, and de- grees of variability are, of course, infinite. Nevertheless, in an effort to provide guidance to the beginning behav- ior analyst, some general statements will be given about the experimental decisions that might be warranted by the data patterns shown in Figure 1.

187

Analyzing Change: Basic Assumptions and Strategies

Time

(A)

R es

p o

n se

M ea

su re

Time

(B)

R es

p o

n se

M ea

su re

Time

(C)

R es

p o

n se

M ea

su re

Time

(D)

R es

p o

n se

M ea

su re

Figure 1 Data patterns illustrating stable (A), ascending (B), descending (C), and variable (D) baselines.

Graph A shows a relatively stable baseline. The data show no evidence of an upward or downward trend, and all of the measures fall within a small range of values. A stable baseline provides the most desirable basis, or con- text, against which to look for effects of an independent variable. If changes in level, trend, and/or variability co- incide with the introduction of an independent variable on a baseline as stable as that shown in Graph A, one can reasonably suspect that those changes may be related to the independent variable.

The data in Graphs B and C represent an ascending baseline and a descending baseline, respectively. The data path in Graph B shows an increasing trend in the be- havior over time, whereas the data path in Graph C shows a decreasing trend. The applied behavior analyst must treat ascending and descending baseline data cautiously. By definition, dependent variables in applied behavior analysis are selected because they represent target be- haviors that need to be changed. But ascending and de- scending baselines reveal behaviors currently in the process of changing. The effects of an independent vari- able introduced at this point are likely to be obscured or confounded by the variables responsible for the already- occurring change. But what if the applied investigator needs to change the behavior immediately? The applied perspective can help solve the dilemma.

Whether a treatment variable should be introduced depends on whether the trending baseline data represent

improving or deteriorating performance. When an as- cending or descending baseline represents behavior change in the therapeutically desired direction, the in- vestigator should withhold treatment and continue to monitor the dependent variable under baseline conditions. When the behavior ceases to improve (as evidenced by stable responding) or begins to deteriorate, the indepen- dent variable can be applied. If the trend does not level off and the behavior continues to improve, the original prob- lem may no longer be present, leaving no reason for in- troducing the treatment as planned (although the investigator might be motivated to isolate and analyze the variables responsible for the “spontaneous” im- provement). Introducing an independent variable to an already-improving behavior makes it difficult, and often impossible, to claim any continued improvement as a function of the independent variable.

An ascending or descending baseline that repre- sents significantly deteriorating performance signals an immediate application of the independent variable. From an applied perspective the decision to intervene is obvious: The subject’s behavior is deteriorating, and a treatment designed to improve it should be intro- duced. An independent variable capable of affecting desired behavior change in spite of other variables “pushing” the behavior in the opposite direction is most likely a robust variable, one that will be a welcome addition to the behavior analyst’s list of effective treat-

188

Analyzing Change: Basic Assumptions and Strategies

B eh

av io

Time

Figure 2 Solid data points represent actual measures of behavior that might be generated in a stable baseline; open data points within the box represent the level of responding that would be predicted on the basis of the obtained measures, should the environment remain constant.

ments. The decision to introduce a treatment variable on a deteriorating baseline is also a sound one from an analytic perspective, which will be discussed in the next section.

Graph D in Figure 1 shows a highly unstable or variable baseline. The data in Graph D show just one of many possible patterns of unstable responding. The data points do not consistently fall within a narrow range of values, nor do they suggest any clear trend. Introducing the independent variable in the presence of such vari- ability is unwise from an experimental standpoint. Vari- ability is assumed to be the result of environmental variables, which in the case shown by Graph D, seem to be operating in an uncontrolled fashion. Before the re- searcher can analyze the effects of an independent vari- able effectively, these uncontrolled sources of variability must be isolated and controlled.

Stable baseline responding provides an index of the degree of experimental control the researcher has estab- lished. Johnston and Pennypacker stressed this point in both editions of Strategies and Tactics of Human Behav- ioral Research:

If unacceptably variable responding occurs under base- line conditions, this is a statement that the researcher is probably not ready to introduce the treatment condi- tions, which involves adding an independent variable whose effects are in question. (1993a, p. 201)

These authors were more direct and blunt in the first edi- tion of their text:

If sufficiently stable responding cannot be obtained, the experimenter is in no position to add an independent variable of suspected but unknown influence. To do so would be to compound confusion and lead to further ig- norance. (1980, p. 229)

Again, however, applied considerations must be bal- anced against purely scientific pursuits. The applied problem may be one that cannot wait to be solved (e.g., severe self-injurious behavior). Or, confounding vari- ables in the subject’s environment and the setting(s) of the investigation may simply be beyond the experi- menter’s control.11 In such situations the independent variable is introduced with the hope of producing stable responding in its presence. Sidman (1960/1988) agreed that “the behavioral engineer must ordinarily take vari- ability as he finds it, and deal with it as an unavoidable fact of life” (p. 192).

11The applied researcher must guard very carefully against assuming au- tomatically that unwanted variability is a function of variables beyond his capability or resources to isolate and control, and thus fail to pursue the in- vestigation of potentially important functional relations.

12“The above reference to ‘relatively constant environmental conditions’ means only that the experimenter is not knowingly producing uncontrolled variations in functionally related environmental events” (Johnston & Pennypacker, 1980, p. 228).

Prediction

Prediction can be defined as “the anticipated outcome of a presently unknown or future measurement. It is the most elegant use of quantification upon which validation of all scientific and technological activity rests” (Johnston & Pennypacker, 1980, p. 120). Figure 2 shows a series of hypothetical measures representing a stable pattern of baseline responding. The consistency of the first five data points in the series encourage the prediction that—if no changes occur in the subject’s environment—subsequent measures will fall within the range of values obtained thus far. Indeed, a sixth measure is taken that gives cre- dence to this prediction. The same prediction is made again, this time with more confidence, and another mea- sure of behavior shows it to be correct. Throughout a baseline (or any other experimental condition), an ongo- ing prediction is made and confirmed until the investi- gator has every reason to believe that the response measure will not change appreciably under the present conditions. The data within the shaded portion of Figure 2 represent unobtained but predicted measures of fu- ture responding under “relatively constant environmen- tal conditions.”12 Given the stability of the obtained measures, few experienced scientists would quarrel with the prediction.

189

Analyzing Change: Basic Assumptions and Strategies

How many measures must be taken before an ex- perimenter can use a series of data points to predict fu- ture behavior with confidence? Baer and colleagues (1968) recommended continuing baseline measurement until “its stability is clear.” Even though there are no set answers, some general statements can be made about the predictive power of steady states. All things being equal, many measurements are better than a few; and the longer the period of time in which stable responding is obtained, the better the predictive power of those measures. Also, if the experimenter is not sure whether measurement has produced stable responding, in all likelihood it has not, and more data should be collected before the indepen- dent variable is introduced. Finally, the investigator’s knowledge of the characteristics of the behavior being studied under constant conditions is invaluable in decid- ing when to terminate baseline measurement and intro- duce the independent variable. That knowledge can be drawn from personal experience in obtaining stable base- lines on similar response classes and from familiarity with patterns of baseline responding found in the pub- lished literature.

It should be clear that guidelines such as “collect baseline data for at least five sessions” or “obtain base- line measures over two consecutive weeks” are misguided or naive. Depending on the situation, five data points ob- tained over one or two weeks of baseline conditions may or may not provide a convincing picture of steady state responding. The question that must be addressed is: Are the data sufficiently stable to serve as the basis for ex- perimental comparison? This question can be answered only by ongoing prediction and confirmation using re- peated measures in an environment in which all relevant conditions are held constant.

Behavior analysts are often interested in analyzing functional relations between an instructional variable and the acquisition of new skills. In such situations it is sometimes assumed that baseline measures are zero. For example, one would expect repeated observations of a child who has never tied her shoes to yield a per- fectly stable baseline of zero correct responses. How- ever, casual observations that have never shown a child to use a particular skill do not constitute a scientifi- cally valid baseline and should not be used to justify any claims about the effects of instruction. It could be that if given repeated opportunities to respond, the child would begin to emit the target behavior at a nonzero rate. The term practice effects refers to im- provements in performance resulting from repeated opportunities to emit the behavior so that baseline measurements can be obtained. For example, attempt- ing to obtain stable baseline data for students per- forming arithmetic problems can result in improved levels of responding simply because of the repeated

practice inherent in the measurement process. Practice effects confound a study, making it impossible to sep- arate and account for the effects of practice and in- struction on the student’s final performance. Repeated baseline measures should be used either to reveal the existence or to demonstrate the nonexistence of prac- tice effects. When practice effects are suspected or found, baseline data collection should be continued until steady state responding is attained.

The necessity to demonstrate a stable baseline and to control for practice effects empirically does not re- quire applied behavior analysts to withhold needed treat- ment or intervention. Nothing is gained by collecting unduly long baselines of behaviors that cannot reason- ably be expected to be in the subject’s repertoire. For ex- ample, many behaviors cannot be emitted unless the subject is competent in certain prerequisite behaviors; there is no legitimate possibility of a child’s tying his shoes if he currently does not pick up the laces, or of a student’s solving division problems if she cannot sub- tract and multiply. Obtaining extended baseline data in such cases is unnecessary pro forma measurement. Such measures would “not so much represent zero behavior as zero opportunity for behavior to occur, and there is no need to document at the level of well-measured data that behavior does not occur when it cannot” (Horner & Baer, 1978, p. 190).

Fortunately, applied behavior analysts need neither abandon the use of steady state strategy nor repeatedly measure nonexistent behavior at the expense of begin- ning treatment. The multiple probe design is an experi- mental tactic that enables the use of steady state logic to analyze functional relations between instruction and the acquisition of behaviors shown to be nonexistent in the subject’s repertoire prior to the introduction of the inde- pendent variable.

Affirmation of the Consequent

The predictive power of steady state responding enables the behavior analyst to employ a kind of inductive logic known as affirmation of the consequent (Johnston & Pennypacker, 1980). When an experimenter introduces an independent variable on a stable baseline, an explicit assumption has been made: If the independent variable were not applied, the behavior, as indicated by the base- line data path, would not change. The experimenter is also predicting that (or more precisely, questioning whether) the independent variable will result in a change in the behavior.

The logical reasoning behind affirmation of the con- sequent begins with a true antecedent–consequent (if-A- then-B) statement and proceeds as follows:

190

Analyzing Change: Basic Assumptions and Strategies

1. If A is true, then B is true.

2. B is found to be true.

3. Therefore, A is true.

The behavior analyst’s version goes like this:

1. If the independent variable is a controlling factor for the behavior (A), then the data obtained in the presence of the independent variable will show that the behavior has changed (B).

2. When the independent variable is present, the data show that the behavior has changed (B is true).

3. Therefore, the independent variable is a controlling variable for the behavior (therefore, A is true).

The logic, of course, is flawed; other factors could be responsible for the truthfulness of A. But, as will be shown, a successful (i.e., convincing) experiment affirms several if-A-then-B possibilities, each one reducing the likelihood of factors other than the independent variable being responsible for the observed changes in behavior.

Data shown in Figures 3 to 5 illustrate how pre- diction, verification, and replication are employed in a hypothetical experiment using the reversal design, one of the most common and powerful analytic tactics used by behavior analysts. Figure 3 shows a successful affir- mation of the consequent. Steady state responding during baseline enabled the prediction that, if no changes were made in the environment, continued measurement would yield data similar to those in the shaded portion of the graph. The independent variable was then introduced, and repeated measures of the dependent variable during this treatment condition showed that the behavior did in- deed change. This enables two comparisons, one real and one hypothetical. First, the real difference between the obtained measures in the presence of the independent variable and the baseline level of responding represents the extent of a possible effect of the independent variable and supports the prediction that treatment would change the behavior.

The second, hypothetical, comparison is between the data obtained in the treatment condition with the pre- dicted measures had the treatment variable not been introduced (i.e., the open data points within the boxed area of Figure 3). This comparison represents the be- havior analyst’s hypothetical approximation of the ideal but impossible-to-achieve experimental design: the si- multaneous measurement and comparison of the behav- ior of an individual subject in both the presence and absence of the treatment variable (Risley, 1969).

Although the data in Figure 3 affirm the initial antecedent–consequent statement—a change in the be- havior was observed in the presence of the independent

13Although two-phase experiments consisting of a pretreatment baseline condition followed by a treatment condition (called A–B design) enable neither verification of the prediction of continued responding at baseline levels nor replication of the effects of the independent variable, studies using A–B designs can nevertheless contribute important and useful find- ings (e.g., Azrin & Wesolowski, 1974; Reid, Parsons, Phillips, & Green, 1993).

variable—asserting a functional relation between the in- dependent and dependent variables at this point is un- warranted. The experiment has not yet ruled out the possibility of other variables being responsible for the change in behavior. For example, perhaps some other event that is responsible for the change in behavior oc- curred at the same time that the independent variable was introduced.13

A firmer statement about the relation between the treatment and the behavior can be made at this point, however, if changes in the dependent variable are not ob- served in the presence of the independent variable. As- suming accurate measures of the behavior and a measurement system sensitive to changes in the behavior, then no behavior change in the presence of the indepen- dent variable constitutes a disconfirmation of the conse- quent (B was shown not to be true), and the independent variable is eliminated as a controlling variable. However, eliminating a treatment from the ranks of controlling vari- ables on the basis of no observed effects presupposes ex- perimental control of the highest order (Johnston & Pennypacker, 1993a).

B eh

av io

Time

Baseline Treatment

Figure 3 Affirmation of the consequent supporting the possibility of a functional relation between the behavior and treatment variable. The measures obtained in the presence of the treatment variable differ from the predicted level of responding in the absence of the treatment variable (open data points within boxed area).

191

Analyzing Change: Basic Assumptions and Strategies

However, in the situation illustrated in Figure 3, a change in behavior was observed in the presence of the independent variable, revealing a correlation between the independent variable and the behavior change. To what extent was the observed behavior change a function of the independent variable? To pursue this question, the behavior analyst employs the next component of base- line logic: verification.

Verification

The experimenter can increase the probability that an ob- served change in behavior was functionally related to the introduction of the independent variable by verifying the original prediction of unchanging baseline measures. Verification can be accomplished by demonstrating that the prior level of baseline responding would have re- mained unchanged had the independent variable not been introduced (Risley, 1969). If that can be demonstrated, this operation verifies the accuracy of the original pre- diction of continued stable baseline responding and reduces the probability that some uncontrolled (con- founding) variable was responsible for the observed change in behavior. Again, the reasoning behind affir- mation of the consequent is the logic that underlies the

Figure 4 illustrates the verification of effect in our hypothetical experiment. When steady state responding has been established in the presence of the independent variable, the investigator removes the treatment variable, thereby returning to the previous baseline conditions. This tactic allows the possibility of affirming two differ- ent antecedent–consequent statements. The first state- ment and its affirmation follows this pattern:

1. If the independent variable is a controlling factor for the behavior (A), then its removal will coincide with changes in the response measure (B).

2. Removal of the independent variable is accompa- nied by changes in the behavior (B is true).

3. Therefore, the independent variable controls re- sponding (therefore, A is true).

The second statement and affirmation follows this pattern:

1. If the original baseline condition controlled the be- havior (A), then a return to baseline conditions will result in similar levels of responding (B).

2. The baseline condition is reinstated and levels of responding similar to those obtained during the original baseline phase are observed (B is true).

3. Therefore, the baseline condition controlled the be- havior both then and now (therefore, A is true).

The six measures within the shaded area obtained during Baseline 2 of our hypothetical experiment in Figure 4 verify the prediction made for Baseline 1. The open data points in the shaded area in Baseline 2 repre- sent the predicted level of responding if the independent variable had not been removed. (The prediction compo- nent of baseline logic applies to steady state responding obtained during any phase of an experiment, baseline and treatment conditions alike.) The difference between the data actually obtained during Treatment (solid data points) and the data obtained during Base-line 2 (solid data points) affirms the first if-A-then-B statement: If the treatment is a controlling variable, then its removal will result in changes in behavior. The similarity between measures obtained during Baseline 2 and those obtained during Baseline 1 confirms the second if-A-then-B state- ment: If baseline conditions controlled the behavior be- fore, reinstating baseline conditions will result in similar levels of responding.

Again, of course, the observed changes in behavior associated with the application and withdrawal of the in- dependent variable are subject to interpretations other

B eh

av io

Time

Baseline Treatment Baseline 2Figure 4 Verification of a previously predicted level of baseline responding by termination or withdrawal of the treatment variable. The measures obtained during Baseline 2 (solid data points within shaded area) show a successful verification and a second affirmation of the consequent based on a comparison with the predicted level of responding (open dots in Base- line 2) in the continued presence of the treatment variable.

experimental strategy.

192

Analyzing Change: Basic Assumptions and Strategies

than a claim of a functional relation between the two events. However, the case for the existence of a func- tional relation is becoming stronger. When the indepen- dent variable was applied, behavior change was observed; when the independent variable was withdrawn, behavior again changed and responding returned to baseline lev- els. To the extent that the experimenter effectively con- trols the presence and absence of the independent variable and holds constant all other variables in the experimen- tal setting that might influence the behavior, a functional relation appears likely: An important behavior change has been produced and reversed by the introduction and withdrawal of the independent variable. The process of verification reduces the likelihood that a variable other than the independent variable was responsible for the ob- served behavior changes.

Does this two-step strategy of prediction and verifi- cation constitute sufficient demonstration of a functional relation? What if some uncontrolled variable covaried with the independent variable as it was presented and withdrawn and this uncontrolled variable was actually responsible for the observed changes in behavior? If such was the case, claiming a functional relation between the target behavior and the independent variable would at best be inaccurate and at the worst perhaps end a search for the actual controlling variables whose identification and control would contribute to an effective and reliable technology of behavior change.

The appropriately skeptical investigator (and research consumer) will also question the reliability of the ob- tained effect. How reliable is this verified behavior change? Was the apparent functional relation a fleeting, one-time-only phenomenon, or will repeated application of the independent variable reliably (i.e., consistently) produce a similar pattern of behavior change? An effec- tive (i.e., convincing) experimental design yields data that are responsive to these important questions. To investigate uncertain reliability, the behavior analyst employs the final, and perhaps the most important, component of base- line logic and experimental design: replication.

14Replication also refers to the repeating of experiments to determine the reliability of functional relations found in previous experiments and the extent to which those findings can be extended to other subjects, settings, and/or behaviors (i.e., generality or external validity). The replication of ex- periments is examined in Chapter entitled “Planning and Evaluating Ap- plied Behavior Analysis Research ”

Replication Replication is the essence of believability.

—Baer, Wolf, and Risley (1968, p. 95)

Within the context of any given experiment, replication means repeating independent variable manipulations con- ducted previously in the study and obtaining similar out- comes.14 Replication within an experiment has two important purposes. First, replicating a previously ob- served behavior change reduces the probability that a variable other than the independent variable was respon- sible for the now twice-observed behavior change. Sec- ond, replication demonstrates the reliability of the behavior change; it can be made to happen again.

Figure 5 adds the component of replication to our hypothetical experiment. After steady state responding was obtained during Baseline 2, the independent variable is reintroduced; this is the Treatment 2 phase. To the ex- tent that the data obtained during the second application of the treatment (data points within area shaded with cross-hatched lines) resemble the level of responding ob- served during Treatment 1, replication has occurred. Our hypothetical experiment has now produced powerful ev- idence of a functional relation exists between the inde- pendent and the dependent variable. The extent to which one has confidence in the assertion of a functional rela- tion rests on numerous factors, some of the most impor- tant of which are the accuracy and sensitivity of the measurement system, the degree of control the experi- menter maintained over all relevant variables, the duration of experimental phases, the stability of responding within each phase, and the speed, magnitude, and consistency of behavior change between conditions. If each of these

B eh

av io

Time

Baseline Treatment 1 Baseline 2 Treatment 2 Figure 5 Replication of experi- mental effect accomplished by re- introducing the treatment variable. The measures obtained during Treatment 2 (data points within area shaded with cross hatching) enhance the case for a functional relation between the treatment variable and the target behavior.

193

Summary Introduction

1. Measurement can show whether and when behavior changes, but measurement alone cannot reveal how the change has come about.

2. Knowledge of specific functional relations between be- havior and environment is necessary if a systematic and useful technology of behavior change is to develop.

3. An experimental analysis must be performed to determine how a given behavior functions in relation to specific en- vironmental events.

Concepts and Assumptions Underlying the Analysis of Behavior

4. The overall goal of science is to achieve an understand- ing of the phenomena under study—socially important be- haviors, in the case of applied behavior analysis.

5. Science produces understanding at three levels: descrip- tion, prediction, and control.

6. Descriptive research yields a collection of facts about the observed events—facts that can be quantified and classified.

7. A correlation exists when two events systematically co- vary with one another. Predictions can be made about the probability that one event will occur based on the occur- rence of the other event.

8. The greatest potential benefits of science are derived from the third, and highest, level of scientific understanding, which comes from establishing experimental control.

9. Experimental control is achieved when a predictable change in behavior (the dependent variable) can be reliably produced by the systematic manipulation of some aspect of the person’s environment (the independent variable).

10. A functional analysis does not eliminate the possibility that the behavior under investigation is also a function of other variables.

11. An experiment that shows convincingly that changes in behavior are a function of the independent variable and not the result of uncontrolled or unknown variables has internal validity.

12. External validity refers to the degree to which a study’s results are generalizable to other subjects, settings, and/or behaviors.

13. Confounding variables exert unknown or uncontrolled in- fluences on the dependent variable.

14. Because behavior is an individual phenomenon, the ex- perimental strategy of behavior analysis is based on within- subject (or single-subject) methods of analysis.

15. Because behavior is a continuous phenomenon that occurs in and changes through time, the repeated measurement of behavior is a hallmark of applied behavior analysis.

16. The assumption of determinism guides the methodology of behavior analysis.

17. Experimental methods in behavior analysis are based on the assumption that variability is extrinsic to the organ- ism; that is, variability is imposed by environmental vari- ables and is not an inherent trait of the organism.

18. Instead of masking variability by averaging the perfor- mance of many subjects, behavior analysts attempt to iso- late and experimentally manipulate the environmental factors responsible for the variability.

Components of Experiments in Applied Behavior Analysis

19. The experimental question is a statement of what the re- searcher seeks to learn by conducting the experiment and should guide and be reflected in all aspects of the experi- ment’s design.

20. Experiments in applied behavior analysis are most often referred to as single-subject (or single-case) research de- signs because the experimental logic or reasoning for an- alyzing behavior change often employs the subject as her own control.

21. The dependent variable in an applied behavior analysis experiment is a measurable dimensional quantity of the target behavior.

22. Three major reasons behavior analysts use multiple- response measures (dependent variables) in some studies are (a) to provide additional data paths that serve as con- trols for evaluating and replicating the effects of an inde- pendent variable that is sequentially applied to each behavior, (b) to assess the generality of treatment effects to behaviors other than the response class to which the in- dependent variable was applied, and (c) to determine whether changes in the behavior of a person other than the subject occur during the course of an experiment and

Analyzing Change: Basic Assumptions and Strategies

considerations is satisfied by the experimental design and is supported by the data as displayed within the design, then replication of effect becomes perhaps the most crit- ical factor in claiming a functional relation.

An independent variable can be manipulated in an effort to replicate an effect many times within an exper-

iment. The number of replications required to demon- strate a functional relation convincingly is related to many considerations, including all of those just enumerated, and to the existence of other similar experiments that have produced the same effects.

194

Analyzing Change: Basic Assumptions and Strategies

whether such changes might in turn explain observed changes in the subject’s behavior.

23. In addition to precise manipulation of the independent variable, the behavior analyst must hold constant all other aspects of the experimental setting—extraneous vari- ables—to prevent unplanned environmental variation.

24. When unplanned events or variations occur in the exper- imental setting, the behavior analyst must either wait out their effects or incorporate them into the design of the experiment.

25. Observation and measurement procedures must be con- ducted in a standardized manner throughout an experiment.

26. Because behavior is a continuous and dynamic phenome- non, ongoing visual inspection of the data during the course of an experiment is necessary to identify changes in level, trend, and/or variability as they develop.

27. Changes in the independent variable are made in an effort to maximize its effect on the target behavior.

28. The term experimental design refers to the way the inde- pendent variable is manipulated in a study.

29. Although an infinite number of experimental designs are possible as a result of the many ways independent vari- ables can be manipulated and combined, there are only two basic kinds of changes in independent variables: introduc- ing a new condition or reintroducing an old condition.

30. A parametric study compares the differential effects of a range of different values of the independent variable.

31. The fundamental rule of experimental design is to change only one variable at a time.

32. Rather than follow rigid, pro forma experimental designs, the behavior analyst should select experimental tactics suited to the original research questions, while standing ready to “explore relevant variables by manipulating them in an improvised and rapidly changing design” (Skinner, 1966, p. 21).

Steady State Strategy and Baseline Logic

33. Stable, or steady state, responding enables the behavior analyst to employ a powerful form of inductive reasoning, sometimes called baseline logic. Baseline logic entails three elements: prediction, verification, and replication.

34. The most common method for evaluating the effects of a given variable is to impose it on an ongoing measure of be- havior obtained in its absence. These preintervention data serve as the baseline by which to determine and evaluate any subsequent changes in behavior.

35. A baseline condition does not necessarily mean the absence of instruction or treatment per se, only the absence of the specific independent variable of experimental interest.

36. In addition to the primary purpose of establishing a base- line as an objective basis for evaluating the effects of the

independent variable, three other reasons for baseline data collection are as follows: (a) Systematic observation of the target behavior prior to intervention sometimes yields information about antecedent-behavior-consequent corre- lations that may be useful in planning an effective inter- vention; (b) baseline data can provide valuable guidance in setting initial criteria for reinforcement; and (c) some- times baseline data reveal that the behavior targeted for change does not warrant intervention.

37. Four types of baseline data patterns are stable, ascending, descending, and variable.

38. The independent variable should be introduced when sta- ble baseline responding has been achieved.

39. The independent variable should not be introduced if ei- ther an ascending or descending baseline indicates im- proving performance.

40. The independent variable should be introduced if either an ascending or descending baseline indicates deteriorat- ing performance.

41. The independent variable should not be imposed on a highly variable, unstable baseline.

42. Prediction of future behavior under relatively constant en- vironmental conditions can be made on the basis of re- peated measures of behavior showing little or no variation.

43. In general, given stable responding, the more data points there are and the longer the time period in which they were obtained, the more accurate the prediction will likely be.

44. Practice effects refer to improvements in performance re- sulting from opportunities to emit the behavior that must be provided to obtain repeated measures.

45. Extended baseline measurement is not necessary for be- haviors that have no logical opportunity to occur.

46. The inductive reasoning called affirmation of the conse- quent lies at the heart of baseline logic.

47. Although the logic of affirming the consequent is not com- pletely sound (some other event may have caused the change in behavior), an effective experimental design con- firms several if-A-then-B possibilities, thereby eliminating certain other factors as responsible for the observed changes in behavior.

48. Verification of prediction is accomplished by demonstrat- ing that the prior level of baseline responding would have remained unchanged if the independent variable had not been introduced.

49. Replication within an experiment means reproducing a previously observed behavior change by reintroducing the independent variable. Replication within an experiment reduces the probability that a variable other than the inde- pendent variable was responsible for the behavior change and demonstrates the reliability of the behavior change.

195

A-B-A design A-B-A-B design alternating treatments design B-A-B design DRI/DRA reversal technique

DRO reversal technique irreversibility multielement design multiple treatment interference multiple treatment reversal design

(NCR) reversal technique reversal design sequence effects withdrawal design

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 5: Experimental Evaluation of Interventions

5-1 Systematically manipulate independent variables to analyze their effects on treatment.

(a) Use withdrawal designs.

(b) Use reversal designs.

5-2 Identify and address practical and ethical considerations in using various experimental designs.

Reversal and Alternating Treatments Designs

Key Terms

196

This chapter describes the reversal and alter- nating treatments designs, two types of experi- mental analysis tactics widely used by applied

behavior analysts. In a reversal design, the effects of in- troducing, withdrawing (or “reversing” the focus of), and reintroducing an independent variable are observed on the target behavior. In an alternating treatments analysis, two or more experimental conditions are rapidly alter- nated, and the differential effects on behavior are noted. We explain how each design incorporates the three ele- ments of steady state strategy—prediction, verification, and replication—and present representative examples il- lustrating the major variations of each. Considerations for selecting and using reversal and alternating treatments designs are also presented.

Reversal Design An experiment using a reversal design entails repeated measures of behavior in a given setting that requires at least three consecutive phases: (a) an initial baseline phase in which the independent variable is absent, (b) an inter- vention phase during which the independent variable is introduced and remains in contact with the behavior, and (c) a return to baseline conditions accomplished by with- drawal of the independent variable. In the widely used no- tation system for describing experimental designs in applied behavior analysis, the capital letters A and B de- note the first and second conditions, respectively, that are introduced in a study. Typically baseline (A) data are col- lected until steady state responding is achieved. Next, an intervention (B) condition is applied that signifies the pres- ence of a treatment—the independent variable. An ex- periment entailing one reversal is described as an A-B-A design. Although studies using an A-B-A design are reported in the literature (e.g., Christle & Schuster, 2003; Geller, Paterson, & Talbott, 1982; Jacobson, Bushell, & Risley, 1969; Stitzer, Bigelow, Liebson, & Hawthorne, 1982), an A-B-A-B design is preferred be- cause reintroducing the B condition enables the replication of treatment effects, which strengthens the demonstration of experimental control (see Figure 1).1

1Some authors use the term withdrawal design to describe experiments based on an A-B-A-B analysis and reserve the term reversal design for studies in which the behavioral focus of the treatment variable is reversed (or switched to another behavior), as in the DRO and DRI/DRA reversal techniques described later in this chapter (e.g., Leitenberg, 1973; Poling, Method, & LeSage, 1995). However, reversal design, as the term is used most often in the behavior analysis literature, encompasses both with- drawals and reversals of the independent variable, signifying the re- searcher’s attempt to demonstrate “behavioral reversibility” (Baer, Wolf, & Risley, 1968; Thompson & Iwata, 2005). Also, withdrawal design is some- times used to describe an experiment in which the treatment variable(s) are sequentially or partially withdrawn after their effects have been analyzed in an effort to promote maintenance of the target behavior (Rusch & Kazdin, 1981).

The A-B-A-B reversal is the most straightforward and generally most powerful within-subject design for demonstrating a functional relation between an environ- mental manipulation and a behavior. When a functional relation is revealed with a reversal design, the data show how the behavior works.

As explanations go, the one offered by the reversal de- sign was not at all a bad one. In answer to the question, “How does this response work?” we could point out demonstrably that it worked like so [e.g., see Figure 1]. Of course, it might also work in other ways; but, we would wait until we had seen the appropriate graphs before agreeing to any other way. (Baer, 1975, p. 19)

Baer’s point must not be overlooked: Showing that a behavior works in a predictable and reliable way in the presence and absence of a given variable provides only one answer to the question, How does this behavior work? There may be (and quite likely are) other control- ling variables for the targeted response class. Whether additional experimentation is needed to explore those other possibilities depends on the social and scientific importance of obtaining a more complete analysis.

Operation and Logic of the Reversal Design

Risley (2005) described the rationale and operation of the reversal design as follows:

The reversal or ABAB design that Wolf reinvented from Claude Bernard’s early examples in experimental medi- cine entailed establishing a baseline of repeated quanti- fied observations sufficient to see a trend and forecast that trend into the near future (A); to then alter conditions and see if the repeated observations become different than they were forecast to be (B); to then change back and see if the repeated observations return to confirm the original forecast (A); and finally, to reintroduce the al- tered conditions and see if the repeated observations again become different than forecast (B). (pp. 280–281)2

A brief review here of the roles of prediction, ver- ification, and replication in the reversal design will suf- fice. Figure 2 shows the same data from Figure 1 with the addition of the open data points representing pre- dicted measures of behavior if conditions in the previ- ous phase had remained unchanged. After a stable pattern of responding, or a countertherapeutic trend, is obtained during Baseline 1, the independent variable is

2Risley (1997, 2005) credits Montrose Wolf with designing the first ex- periments using the reversal and multiple baseline designs. “The research methods that Wolf pioneered in these studies were groundbreaking. That methodology came to define applied behavior analysis” (pp. 280–281).

Reversal and Alternating Treatments Designs

197

introduced. In our hypothetical experiment the measures obtained during Treatment 1, when compared with those from Baseline 1 and with the measures predicted by Baseline 1, show that behavior change occurred and that the change in behavior coincided with the intervention. After steady state responding is attained in Treatment 1, the independent variable is withdrawn and baseline con- ditions are reestablished. If the level of responding in Baseline 2 is the same as or closely approximates the measures obtained during Baseline 1, verification of the prediction made for Baseline 1 data is obtained. Stated otherwise, had the intervention not been introduced and had the initial baseline condition remained in effect, the predicted data path would have appeared as shown in Baseline 2. When withdrawal of the independent vari- able results in a reversal of the behavior change associ- ated with its introduction, a strong case builds that the intervention is responsible for the observed behavior change. If reintroduction of the independent variable in Treatment 2 reproduces the behavior change observed during Treatment 1, replication of effect has been achieved, and a functional relation has been demon- strated. Again stated in other terms, had the intervention continued and had the second baseline condition not been

introduced, the predicted data path of the treatment would have appeared as shown in Treatment 2.

Romaniuk and colleagues (2002) provided an ex- cellent example of the A-B-A-B design. Three students with developmental disabilities who frequently displayed problem behaviors (e.g., hitting, biting, whining, crying, getting out of seat, inappropriate gestures, noises, and comments) when given academic tasks participated in the study. Prior to the experiment a functional analysis had shown that each student’s problem behaviors were maintained by escape from working on the task (i.e., problem behavior occurred most often when followed by being allowed to take a break from the task). The re- searchers wanted to determine whether providing stu- dents with a choice of which task to work on would reduce the frequency of their problem behavior, even though problem behavior, when it occurred, would still result in a break. The experiment consisted of two con- ditions: no choice (A) and choice (B). The same set of teacher-nominated tasks was used in both conditions.

Each session during the no-choice condition began with the experimenter providing the student with a task and saying, “This is the assignment you will be working on today” or “It’s time to work on _____” (p. 353). Dur-

B eh

av io

Time

Baseline 1 Treatment 1 Baseline 2 Treatment 2Figure 2 Illustration of A-B-A-B reversal design. Open data points represent data predicted if conditions from previous phase remained in effect. Data collected during Baseline 2 (within shaded box) verify the prediction from Baseline 1. Treatment 2 data (cross- hatched shading) replicate the experi- mental effect.

B eh

av io

Baseline 1 Treatment 1 Baseline 2 Treatment 2

Time

Figure 1 Graphic prototype of the A-B-A-B reversal design.

Reversal and Alternating Treatments Designs

198

ing the choice condition (B), the experimenter placed the materials for four to six tasks on the table before the stu- dent and said, “Which assignment would you like to work on today?” (p. 353). The student was also told that he or she could switch tasks at any time during the session by requesting to do so. Occurrences of problem behavior in both conditions resulted in the experimenter stating, “You can take a break now” and giving a 10-second break.

Figure 3 shows the results of the experiment. The data reveal a clear functional relation between the op- portunity to choose which tasks to work on and reduced occurrence of problem behavior by all three students. The percentage of session time in which each student exhib- ited problem behavior (reported as a total duration mea- sure obtained by recording the second of onset and offset as shown by the VCR timer) decreased sharply from the no-choice (baseline) levels when the choice condition was implemented, returned (reversed) to baseline levels when choice was withdrawn, and decreased again when choice was reinstated. The A-B-A-B design enabled Ro- maniuk and colleagues to conduct a straightforward, un-

ambiguous demonstration that significant reductions in problem behavior exhibited by each student were a func- tion of being given a choice of tasks.

In the 1960s and early 1970s applied behavior ana- lysts relied almost exclusively on the A-B-A-B reversal design. The straightforward A-B-A-B design played such a dominant role in the early years of applied behavior analysis that it came to symbolize the field (Baer, 1975). This was no doubt due, at least in part, to the rever- sal design’s ability to expose variables for what they are—strong and reliable or weak and unstable. An- other reason for the reversal design’s dominance may have been that few alternative analytic tactics were available at that time that effectively combined the intrasubject experimental elements of prediction, verification, and replication. Although the reversal design is just one of many experimental designs available to applied behavior analysts today, the simple, unadorned A-B-A-B design continues to play a major role in the behavior analysis literature (e.g., Anderson & Long, 2002; Ash- baugh & Peck, 1998; Cowdery, Iwata, & Pace, 1990;

100

P ro

b le

m B

eh av

io r

(% o

f Se

ss io

n T

im e)

1 5 9 13 17 21 25

No Choice (No Ch)

Choice (Ch)

No Ch Ch

Brooke

100

P ro

b le

m B

eh av

io r

(% o

f Se

ss io

n T

im e)

No Ch Ch No Ch

1 5 9 13 17 21

Ch No Ch Ch

Maggie

100

P ro

b le

m B

eh av

io r

(% o

f Se

ss io

n T

im e)

No Ch Ch No Ch

Gary

5 9 13 17 21 25 29

Sessions

Figure 3 An A-B-A-B reversal design. From “The Influence of Activity Choice on Problems Behaviors Maintained by Escape versus Attention” by C. Romaniuk, R. Miltenberger, C. Conyers, N. Jenner, M. Jurgens, and C. Ringenberg, 2002, Journal of Applied Behavior Analysis, 35, p. 357. Copyright 2002 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Reversal and Alternating Treatments Designs

199

Reversal and Alternating Treatments Designs

Deaver, Miltenberger, & Stricker, 2001; Gardner, Heward, & Grossi, 1994; Levondoski & Cartledge, 2000; Lindberg, Iwata, Kahng, & DeLeon, 1999; Mazaleski, Iwata, Rodgers, Vollmer, & Zarcone, 1994; Taylor & Alber, 2003; Umbreit, Lane, & Dejud, 2004).

Variations of the A-B-A-B Design

Many applied behavior analysis studies use variations or extensions of the A-B-A-B design.

Repeated Reversals

Perhaps the most obvious variation of the A-B-A-B re- versal design is a simple extension in which the inde- pendent variable is withdrawn and reintroduced a second time; A-B-A-B-A-B (see the graph for Maggie in Figure 3). Each additional presentation and withdrawal that reproduces the previously observed effects on behavior increases the likelihood that the behavior changes are the result of manipulating the independent variable. All other things being equal, an experiment that incorporates mul- tiple reversals presents a more convincing and compelling demonstration of a functional relation than does an ex- periment with one reversal (e.g., Fisher, Lindauer, Alter- son, & Thompson, 1998; Steege et al., 1990). That said, it is also possible to reach a point of redundancy beyond which the findings of a given analysis are no longer en- hanced significantly by additional reversals.

B-A-B Design

The B-A-B design begins with the application of the in- dependent variable: the treatment. After stable responding has been achieved during the initial treatment phase (B), the independent variable is withdrawn. If the behavior worsens in the absence of the independent variable (the A condition), the treatment variable is reintroduced in an attempt to recapture the level of responding obtained dur- ing the first treatment phase, which would verify the pre- diction based on the data path obtained during the initial treatment phase.

Compared to the A-B-A design, the B-A-B design is preferable from an applied sense in that the study ends with the treatment variable in effect. However, in terms of demonstrating a functional relation between the inde- pendent variable and dependent variable, the B-A-B de- sign is the weaker of the two because it does not enable an assessment of the effects of the independent variable on the preintervention level of responding. The noninter- vention (A) condition in a B-A-B design cannot verify a prediction of a previous nonexistent baseline. This weak-

ness can be remedied by withdrawing and then reintro- ducing the independent variable, as in a B-A-B-A-B design (e.g., Dixon, Benedict, & Larson, 2001.

Because the B-A-B design provides no data to de- termine whether the measures of behavior taken during the A condition represent preintervention performance, sequence effects cannot be ruled out: The level of be- havior observed during the A condition may have been in- fluenced by the fact that the treatment condition preceded it. Nevertheless, there are exigent situations in which ini- tial baseline data cannot be collected. For instance, the B-A-B design may be appropriate with target behaviors that result in physical harm or danger to the participant or to others. In such instances, withholding a possibly effective treatment until a stable pattern of baseline re- sponding can be obtained may present ethical problems. For example, Murphy, Ruprecht, Baggio, and Nunes (1979) used a B-A-B design to evaluate the effectiveness of mild punishment combined with reinforcement on the number of self-choking responses by a 24-year-old man with profound mental retardation. After the treatment was in effect for 24 sessions, it was withdrawn for three sessions, during which an immediate and large increase in self-choking was recorded (see Figure 4). Reintro- duction of the treatment package reproduced behavior levels noted during the first treatment phase. The average number of self-chokes during each phase of the B-A-B study was 22, 265, and 24, respectively.

Despite the impressive reduction of behavior, the re- sults of Murphy and colleagues’ study using a B-A-B de- sign may have been enhanced by gathering and reporting objectively measured data on the level of behavior prior to the first intervention. Presumably, Murphy and colleagues chose not to collect an initial baseline for ethical and prac- tical reasons. They reported anecdotally that self-chokes averaged 434 per day immediately prior to their interven- tion when school staff had used a different procedure to re- duce the self-injurious behavior. This anecdotal information increased the believability of the functional relation sug- gested by the experimental data from the B-A-B design.

At least two other situations exist in which a B-A-B design might be warranted instead of the more conven- tional A-B-A-B design. These include (a) when a treat- ment is already in place (e.g., Marholin, Touchette, & Stuart, 1979; Pace & Troyer, 2000) and (b) when the be- havior analyst has limited time in which to demonstrate practical and socially significant results. For instance, Robinson, Newby, and Ganzell (1981) were asked to de- velop a behavior management system for a class of 18 hyperactive boys with the stipulation that the program’s effectiveness be demonstrated within 4 weeks. Given “the stipulation of success in 4 weeks, a B-A-B design was used” (pp. 310–311).

200

Reversal and Alternating Treatments Designs

300

275

250

225

200

175

150

125

100

N u

m b

er o

f Se

lf -C

h o

ki n

g R

es p

o n

se s

5 10 15 20 25 30 35 40 45 50 55 60

Treatment Rev Treatment

Sessions

Figure 4 A B-A-B reversal design. From “The Use of Mild Punishment in Combination with Reinforcement of Alternate Behaviors to Reduce the Self- Injurious Behavior of a Profoundly Retarded Individual” by R. J. Murphy, M. J. Ruprecht, P. Baggio, and D. L. Nunes, 1979, AAESPH Review, 4, p. 191. Copyright 1979 by the AAESPH Review. Reprinted by permission.

Multiple Treatment Reversal Designs

Experiments that use the reversal design to compare the effects of two or more experimental conditions to base- line and/or to one another are said to use a multiple treat- ment reversal design. The letters C, D, and so on, denote additional conditions, as in the A-B-C-A-C-B-C design used by Falcomata, Roane, Hovanetz, Kettering, and Keeney (2004); the A-B-A-B-C-B-C design used by Free- land and Noell (1999); the A-B-C-B-C-B-C design used by Lerman, Kelley, Vorndran, Kuhn, and LaRue (2002); the A-B-A-C-A-D-A-C-A-D design used by Weeks and Gaylord-Ross (1981); and the A-B-A-B-B+C-B-B+C de- sign of Jason and Liotta (1982). As a whole, these de- signs are considered variations of the reversal design because they embody the experimental method and logic of the reversal tactic: Responding in each phase provides baseline (or control condition) data for the subsequent phase (prediction), independent variables are withdrawn in an attempt to reproduce levels of behavior observed in a previous condition (verification), and each indepen- dent variable that contributes fully to the analysis is introduced at least twice (replication). Independent vari- ables can be introduced, withdrawn, changed in value, combined, and otherwise manipulated to produce an end- less variety of experimental designs.

For example, Kennedy and Souza (1995) used an A-B-C-B-C-A-C-A-C design to analyze and compare the

effects of two kinds of competing sources of stimulation on eye poking by a 19-year-old student with profound disabilities. Geoff had a 12-year history of poking his forefinger into his eyes during periods of inactivity, such as after lunch or while waiting for the bus. The two treat- ment conditions were music (B) and a video game (C). During the music condition, Geoff was given a Sony Walkman radio with headphones. The radio was tuned to a station that his teacher and family thought he pre- ferred. Geoff had continuous access to the music during this condition, and he could remove the headphones at any time. During the video game condition, Geoff was given a small handheld video game on which he could observe a variety of visual patterns and images on the screen with no sound. As with the music condition, Geoff had continuous access to the video game and could dis- continue using it at any time.

Figure 5 shows the results of the study. Following an initial baseline phase (A) in which Geoff averaged 4 eye pokes per hour, the music condition (B) was intro- duced and eye pokes decreased to a mean of 2.8 per hour. The video game (C) was implemented next, and eye pokes decreased further to 1.1 per hour. Measures ob- tained during the next two phases—a reintroduction of music (B) followed by a second phase of the video game (C)—replicated previous levels of responding under each condition. This B-C-B-C portion of the experiment

201

Reversal and Alternating Treatments Designs

5 10 15 20 25 30

Baseline (BL) Video Game Video Video Music Music BL BL

Video

School Days

N u

m b

er o

f Ey

e- Po

ke s

p er

H o

u r

Figure 5 Example of multiple-treatment reversal design (A-B-C-B-C-A-C-A-C). From “Functional Analysis and Treatment of Eye Poking” by C. H. Kennedy and G. Souza, 1995, Journal of Applied Behavior Analysis, 28, p. 33. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

revealed a functional relation between video game con- dition and lower frequency of eye pokes compared to music. The final five phases of the experiment (C-A-C- A-C) provided an experimental comparison of the video game and baseline (no-treatment) condition.

In most instances, extended designs involving mul- tiple independent variables are not preplanned. Instead of following a predetermined, rigid structure that dictates when and how experimental manipulations must be made, the applied behavior analyst makes design deci- sions based on ongoing assessments of the data.

In this sense, a single experiment may be viewed as a number of successive designs that are collectively neces- sary to clarify relations between independent and depen- dent variables. Thus, some design decisions might be made in response to the data unfolding as the investiga- tion progresses. This sense of design encourages the ex- perimenter to pursue in more dynamic fashion the solutions to problems of experimental control immedi- ately upon their emergence. (Johnston & Pennypacker, 1980, pp. 250–251)

Students of applied behavior analysis should not in- terpret this description of experimental design as a rec- ommendation for a completely free-form approach to the manipulation of independent variables. The researcher must always pay close attention to the rule of changing only one variable at a time and must understand the op- portunities for legitimate comparisons and the limitations that a given sequence of manipulations places on the con- clusions that can be drawn from the results.

Experiments that use the reversal design to compare two or more treatments are vulnerable to confounding by sequence effects. Sequence effects are the effects on a subject’s behavior in a given condition that are the result of the subject’s experience with a prior condition. For ex- ample, caution must be used in interpreting the results from the A-B-C-B-C design that results from the fol- lowing fairly common sequence of events in practice: After baseline (A), an initial treatment (B) is implemented

and little or no behavioral improvements are noted. A second treatment (C) is then tried, and the behavior im- proves. A reversal is then conducted by reintroducing the first treatment (B), followed by reinstatement of the sec- ond treatment (C) (e.g., Foxx & Shapiro, 1978). In this case, we can only speak knowingly about the effects of C when it follows B. Recapturing the original baseline levels of responding before introducing the second treat- ment condition (i.e., an A-B-A-C-A-C sequence) reduces the threat of sequence effects (or helps to expose them for what they are).

An A-B-A-B-C-B-C design, for instance, enables di- rect comparisons of B to A and C to B, but not of C to A. An experimental design consisting of A-B-A-B-B+C-B- B+C (e.g., Jason & Liotta, 1982) permits an evaluation of the additive or interactive effects of B+C, but does not reveal the independent contribution of C. And in both of these examples, it impossible to determine what effects, if any, C may have had on the behavior if it had been im- plemented prior to B. Manipulating each condition so that it precedes and follows every other condition in the experiment (e.g., A-B-A-B-C-B-C-A-C-A-C) is the only way to know for sure. However, manipulating multiple conditions requires a large amount of time and resources, and such extended designs become more susceptible to confounding by maturation and other historical variables not controlled by the experimenter.

NCR Reversal Technique

With interventions based on positive reinforcement, it can be hypothesized that observed changes in behavior are the result of the participant’s feeling better about him- self because of the improved environment created by the reinforcement, not because a specific response class has been immediately followed by contingent reinforcement. This hypothesis is most often advanced when interven- tions consisting of social reinforcement are involved. For example, a person may claim that it doesn’t matter how

202

Reversal and Alternating Treatments Designs

the teacher’s praise and attention were given; the stu- dent’s behavior improved because the praise and attention created a warm and supporting environment. If, however, the behavioral improvements observed during a contin- gent reinforcement condition are lost during a condition when equal amounts of the same consequence are deliv- ered independent of the occurrence of the target behav- ior, a functional relation between the reinforcement contingency and behavior change is demonstrated. In other words, such an experimental control technique can show that behavior change is the result of contingent re- inforcement, not simply the presentation of or contact with the stimulus event (Thompson & Iwata, 2005).

A study by Baer and Wolf (1970a) on the effects of teachers’ social reinforcement on the cooperative play of a preschool child provides an excellent example of the NCR reversal technique (Figure 6). The authors de- scribed the use and purpose of the design as follows:

[The teachers first collected] baselines of cooperative and other related behaviors of the child, and of their own interaction with the child. Ten days of observation indi- cated that the child spent about 50% of each day in prox- imity with other children (meaning within 3 feet of them indoors, or 6 feet outdoors). Despite this frequent prox- imity, however, the child spent only about 2% of her day in cooperative play with these children. The teachers, it was found, interacted with this girl about 20% of the

day, not all of it pleasant. The teachers, therefore, set up a period of intense social reinforcement, offered not for cooperative play but free of any response requirement at all: the teachers took turns standing near the girl, attend- ing closely to her activities, offering her materials, and smiling and laughing with her in a happy and admiring manner. The results of 7 days of this noncontingent ex- travagance of social reinforcement were straightforward: the child’s cooperative play changed not at all, despite the fact that the other children of the group were greatly attracted to the scene, offering the child nearly double the chance to interact with them cooperatively. These 7 days having produced no useful change, the teachers then began their planned reinforcement of cooperative behavior. . . . Contingent social reinforcement, used in amounts less than half that given during the noncontin- gent period, increased the child’s cooperative play from its usual 2% to a high of 40% in the course of 12 days of reinforcement. At that point, in the interests of certainty, the teachers discontinued contingent reinforcement in favor of noncontingent. In the course of 4 days, they lost virtually all of the cooperative behavior they had gained during the reinforcement period of the study, the child showing about a 5% average of cooperative play over that period of time. Naturally, the study concluded with a return to the contingent use of social reinforcement, a re- covery of desirable levels of cooperative play, and a gradual reduction of the teacher’s role in maintaining that behavior. (pp. 14–15)

P er

ce n

ta ge

o f

Se ss

io n

in C

o o

p er

at iv

e P

la y

1 5 10 15 20 25 30 35 40 Days

Baseline

N o

n co

n ti

n ge

n t

R ei

n fo

rc em

en t 1 Contingent

Reinforcement1

N o

n co

n ti

n ge

n t

R ei

n fo

rc em

en t 2 Contingent

Reinforcement2

Figure 6 Reversal design using noncontin- gent reinforcement (NCR) as a control technique. From “Recent Examples of Behavior Modification in Pre-School Settings” by D. M. Baer and M. M. Wolf in Behavior Modification in Clinical Psychology, pp. 14–15, edited by C. Neuringer and J. L. Michael, 1970, Upper Saddle River, NJ: Prentice Hall. Copyright 1970 by Prentice Hall. Adapted by permission.

203

Reversal and Alternating Treatments Designs

Using NCR as a control conditions to demonstrate a functional relation is advantageous when it is not possi- ble or appropriate to eliminate completely the event or activity used as a contingent reinforcement. For exam- ple, Lattal (1969) employed NCR as a control condition to “reverse” the effects of swimming as reinforcement for tooth brushing by children in a summer camp. In the contingent reinforcement condition, the campers could go swimming only if they had brushed their teeth; in the NCR condition, swimming was available whether or not tooth brushing occurred. The campers brushed their teeth more often in the contingent reinforcement condition.

The usual procedure is to deliver NCR on a fixed or variable time schedule independent of the subject’s be- havior. A potential weakness of the NCR control proce- dure becomes apparent when a high rate of the desired behavior has been produced during the preceding con- tingent reinforcement phase. It is probable in such situ- ations that at least some instances of NCR, delivered according to a predetermined time schedule, will follow occurrences of the target behavior closely in time, and thereby function as adventitious, or “accidental rein- forcement” (Thompson & Iwata, 2005). In fact, an inter- mittent schedule of reinforcement might be created inadvertently that results in even higher levels of perfor- mance than those obtained under contingent reinforce- ment. In such cases the investigator might consider using one of the two control techniques described next, both of which involve “reversing” the behavioral focus of the contingency.3

DRO Reversal Technique

One way to ensure that reinforcement will not immedi- ately follow the target behavior is to deliver reinforce- ment immediately following the subject’s performance of any behavior other than the target behavior. With a DRO reversal technique, the control condition consists of delivering the event suspected of functioning as rein- forcement following the emission of any behavior other than the target behavior (e.g., Baer, Peterson, & Sher- man, 1967; Osbourne, 1969; Poulson, 1983). For exam- ple, Reynolds and Risley (1968) used contingent teacher attention to increase the frequency of talking in a 4-year- old girl enrolled in a preschool program for disadvan-

3Strictly speaking, using NCR as an experimental control technique to demonstrate that the contingent application of reinforcement is requisite to its effectiveness is not a separate variation of the A-B-A reversal design. Technically, the NCR reversal technique, as well as the DRO and DRI/DRA reversal techniques described next, is a multiple treatment design. For ex- ample, the Baer and Wolf (1970a) study of social reinforcement shown in Figure 6 used an A-B-C-B-C design, with B representing the NCR con- ditions and C representing the contingent reinforcement conditions.

taged children. After a period of teacher attention contin- gent on verbalization, in which the girl’s talking increased from a baseline average of 11% of the intervals observed to 75%, a DRO condition was implemented during which the teachers attended to the girl for any behavior except talking. During the 6 days of DRO, the girl’s verbalization dropped to 6%. Teacher attention was then delivered con- tingent on talking, and the girl’s verbalization “immedi- ately increased to an average of 51%” (p. 259).

DRI/DRA Reversal Technique

During the control condition in a DRI/DRA reversal technique, occurrences of a specified behavior that is ei- ther incompatible with the target behavior (i.e., the two behaviors cannot possibly be emitted at the same time) or an alternative to the target behavior are immediately fol- lowed by the same consequence previously delivered as contingent reinforcement for the target behavior. Goetz and Baer’s (1973) investigation of the effects of teacher praise on preschool children’s creative play with building blocks illustrates the use of a DRI control condition. Figure 7 shows the number of different block forms (e.g., arch, tower, roof, ramp) constructed by the three children who participated in the study. During baseline (data points indicated by the letter N), “the teacher sat by the child as she built with the blocks, watching closely but quietly, displaying neither criticism nor enthusiasm about any particular use of the blocks” (p. 212). During the next phase (the D data points), “the teacher remarked with in- terest, enthusiasm, and delight every time that the child placed and/or rearranged the blocks so as to create a form that had not appeared previously in that session’s con- struction(s). . . . ‘Oh, that’s very nice—that’s different!’” (p. 212). Then, after increasing form diversity was clearly established, instead of merely withdrawing verbal praise and returning to the initial baseline condition, the teacher provided descriptive praise only when the children had constructed the same forms (the S data points). “Thus, for the next two to four sessions, the teacher continued to display interest, enthusiasm, and delight, but only at those times when the child placed and/or rearranged a block so as to create a repetition of a form already apparent in that session’s construction(s). . . . Thus, no first usage of a form in a session was reinforced, but every second usage of that form and every usage thereafter within the session was. . . . ‘How nice—another arch!’” (p. 212). The final phase of the experiment entailed a return to de- scriptive praise for different forms. Results show that the form diversity of children’s block building was a func- tion of teacher praise and comments. The DRI reversal tactic allowed Goetz and Baer to determine that it was not just the delivery of teacher praise and comment that

204

Reversal and Alternating Treatments Designs

Sally

N N

D S

D D

Kathy

N D

N N

DDD

D DD

D D

S SS

Mary

N D

NNN

D D D

S S

Constructions

N—no reinforcement D—reinforce only different forms S—reinforce only same forms

Fo rm

D iv

er si

ty (n

u m

b er

o f

d if

fe re

n t

fo rm

s p

er c

o n

st ru

ct io

n )

5 10 15

Figure 7 Reversal design using a DRI control technique. From “Social Control of Form Diversity and the Emergence of New Forms in Children’s Blockbuilding” by E. M. Goetz and D. M. Baer, 1973, Journal of Applied Behavior Analysis, 6, p. 213. Copyright 1973 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

resulted in more creative block building by the children; the praise and attention had to be contingent on different forms to produce increasing form diversity.4

Considering the Appropriateness of the Reversal Design

The primary advantage of the reversal design is its abil- ity to provide a clear demonstration of the existence (or absence) of a functional relation between the indepen- dent and dependent variables. An investigator who reli- ably turns the target behavior on and off by presenting and withdrawing a specific variable makes a clear and

4The extent to which the increased diversity of the children’s block build- ing can be attributed to the attention and praise (“That’s nice”) or the de- scriptive feedback (“. . . that’s different”) in the teacher’s comments cannot be determined from this study because social attention and descriptive feedback were delivered as a package.

5Additional manipulations in the form of the partial or sequential with- drawal of intervention components are made when it is necessary or de- sirable for the behavior to continue at its improved level in the absence of the complete intervention (cf., Rusch & Kazdin, 1981).

convincing demonstration of experimental control. In ad- dition, the reversal design enables quantification of the amount of behavior change over the preintervention level of responding. And the return to baseline provides infor- mation on the need to program for maintenance. Fur- thermore, a complete A-B-A-B design ends with the treatment condition in place.5

In spite of its strengths as a tool for analysis, the re- versal design entails some potential scientific and social disadvantages that should be considered prior to its use. The considerations are of two types: irreversibility, which affects the scientific utility of the design; and the social, educational, and ethical concerns related to withdrawing a seemingly effective intervention.

205

Reversal and Alternating Treatments Designs

Irreversibility: A Scientific Consideration

A reversal design is not appropriate in evaluating the ef- fects of a treatment variable that, by its very nature, can- not be withdrawn once it has been presented. Although independent variables involving reinforcement and pun- ishment contingencies can be manipulated with some certainty—the experimenter either presents or withholds the contingency—an independent variable such as pro- viding information or modeling, once presented, cannot simply be removed. For example, a reversal design would not be an effective element of an experiment in- vestigating the effects of attending an in-service training workshop for teachers during which participants ob- served a master teacher use contingent praise and atten- tion with students. After the participants have listened to the rationale for using contingent praise and attention and observed the master teacher model it, the exposure provided by that experience could not be withdrawn. Such interventions are said to be irreversible.

Irreversibility of the dependent variable must also be considered in determining whether a reversal would be an effective analytic tactic. Behavioral irreversibility means that a level of behavior observed in an earlier phase cannot be reproduced even though the experimental conditions are the same as they were during the earlier phase (Sidman, 1960). Once improved, many target behaviors of interest to the applied behavior analyst remain at their newly en- hanced level even when the intervention responsible for the behavior change is removed. From a clinical or educational standpoint, such a state of affairs is desirable: The behav- ior change is shown to be durable, persisting even in the absence of continued treatment. However, irreversibility is a problem if demonstration of the independent variable’s role in the behavior change depends on verification by re- capturing baseline levels of responding.

For example, baseline observations might reveal very low, almost nonexistent, rates of talking and social inter- action for a young child. An intervention consisting of teacher-delivered social reinforcement for talking and in- teracting could be implemented, and after some time the girl might talk to and interact with her peers at a frequency and in a manner similar to that of her classmates. The in- dependent variable, teacher-delivered reinforcement, could be terminated in an effort to recapture baseline rates of talking and interacting. But the girl might continue to talk to and interact with her classmates even though the intervention, which may have been responsible for the initial change in her behavior, is withdrawn. In this case a source of reinforcement uncontrolled by the experi- menter—the girl’s classmates talking to and playing with her as a consequence of her increased talking and inter- acting with them—could maintain high rates of behavior after the teacher-delivered reinforcement is no longer pro-

vided. In such instances of irreversibility, an A-B-A-B design would fail to reveal a functional relation between the independent variable and the target behavior.

Nonetheless, one of the major objectives of applied behavior analysis is establishing socially important behavior through experimental treatments so that the be- havior will contact natural “communities of reinforce- ment” to maintain behavioral improvements in the absence of treatment (Baer & Wolf, 1970b). When irreversibility is suspected or apparent, in addition to considering DRO or DRI/DRA conditions as control techniques, investiga- tors can consider other experimental tactics.

Withdrawing an Effective Intervention: A Social, Educational, and Ethical Consideration

Although it can yield an unambiguous demonstration of experimental control, withdrawing a seemingly effective intervention to evaluate its role in behavior change pre- sents a legitimate cause for concern. One must question the appropriateness of any procedure that allows (indeed, seeks) an improved behavior to deteriorate to baseline levels of responding. Various concerns have been voiced over this fundamental feature of the reversal design. Al- though there is considerable overlap among the concerns, they can be classified as having primarily a social, edu- cational, or ethical basis.

Social Concerns. Applied behavior analysis is, by definition, a social enterprise. Behaviors are selected, defined, observed, measured, and modified by and for people. Sometimes the people involved in an applied behavior analysis—administrators, teachers, parents, and participants—object to the withdrawal of an inter- vention they associate with desirable behavior change. Even though a reversal may provide the most unquali- fied picture of the behavior–environment relation under study, it may not be the analytic tactic of choice be- cause key participants do not want the intervention to be withdrawn. When a reversal design offers the best experimental approach scientifically and poses no ethi- cal problems, the behavior analyst may choose to ex- plain the operation and purpose of the tactic to those who do not favor it. But it is unwise to attempt a rever- sal without the full support of the people involved, es- pecially those who will be responsible for withdrawing the intervention (Tawney & Gast, 1984). Without their cooperation the procedural integrity of the experiment could easily be compromised. For example, people who are against the withdrawal of treatment might sabotage the return to baseline conditions by implementing the

206

Reversal and Alternating Treatments Designs

intervention, or at least those parts of it that they con- sider the most important.

Ethical Concerns. A serious ethical concern must be addressed when the use of a reversal design is con-

sidered for evaluating a treatment for self-injurious or dangerous behaviors. With mild self-injurious or ag- gressive behaviors, short reversal phases consisting of one or two baseline probes can sometimes provide the empirical evidence needed to reveal a functional rela- tion (e.g., Kelley, Jarvie, Middlebrook, McNeer, & Drabman, 1984; Luce, Delquadri, & Hall, 1980; Mur- phy et al., 1979 [Figure 4]). For example, in their study evaluating a treatment for elopement (i.e., run- ning away from supervision) by a child with attention- deficient/hyperactivity disorder, Kodak, Grow, and Northrup (2004) returned to baseline conditions for a single session (see Figure 8).

Nonetheless, with some behaviors it may be deter- mined that withdrawing an intervention associated with improvement for even a few one-session probes would be inappropriate for ethical reasons. In such cases ex- perimental designs that do not rely on the reversal tactic must be used.

Alternating Treatments Design An important and frequently asked question by teachers, therapists, and others who are responsible for changing behavior is, Which of these treatments will be most effective with this student or client? In many situations, the research literature, the analyst’s experience, and/or logical extensions of the principles of behavior point to

100

Pe rc

en ta

ge o

f T im

e El

o p

in g

Baseline BaselineTreatment Treatment

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Sessions

Figure 8 Reversal design with a single-session return-to-baseline probe to evaluate and verify effects of treatment for a potentially dangerous behavior. From “Functional Analysis and Treatment of Elopement for a Child with Attention Deficit Hyperactivity Disorder” by T. Kodak, L. Grow, and J. Northrup, 2004, Journal of Applied Behavior Analysis, 37, p. 231. Copyright 2004 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Educational and Clinical Issues. Educational or clinical issues concerning the reversal design are often raised in terms of instructional time lost during the re- versal phases, as well as the possibility that the behav- ioral improvements observed during intervention may not be recaptured when treatment is resumed after a re- turn to baseline conditions. We agree with Stolz (1978) that “extended reversals are indefensible.” If preinter- vention levels of responding are reached quickly, rever- sal phases can be quite short in duration. Sometimes only three or four sessions are needed to show that ini- tial baseline rates have been reproduced (e.g., Ash- baugh & Peck, 1998; Cowdery Iwata, & Pace, 1990). Two or three brief reversals can provide an extremely convincing demonstration of experimental control. Concern that the improved levels of behavior will not return when the treatment variable is reintroduced, while understandable, has not been supported by em- pirical evidence. Hundreds of published studies have shown that behavior acquired under a given set of envi- ronmental conditions can be reacquired rapidly during subsequent reapplication of those conditions.

207

Reversal and Alternating Treatments Designs

several possible interventions. Determining which of sev- eral possible treatments or combination of treatments will produce the greatest improvement in behavior is a pri- mary task for applied behavior analysts. As described earlier, although a multiple treatment reversal design (e.g., A-B-C-B-C) can be used to compare the effects of two or more treatments, such designs have some inherent limi- tations. Because the different treatments in a multiple treatment reversal design are implemented during sepa- rate phases that occur in a particular order, the design is particularly vulnerable to confounding because of se- quence effects (e.g., Treatment C may have produced its effect only because it followed Treatment B, not because it was more robust in its own right). A second disadvan- tage of comparing multiple treatments with the reversal tactic is the extended time required to demonstrate dif- ferential effects. Most behaviors targeted for change by teachers and therapists are selected because they need immediate improvement. An experimental design that will quickly reveal the most effective treatment among several possible approaches is important for the applied behavior analyst.

The alternating treatments design provides an exper- imentally sound and efficient method for comparing the effects of two or more treatments. The term alternating treatments design, proposed by Barlow and Hayes (1979), accurately communicates the operation of the design. Other terms used in the applied behavior analysis litera- ture to refer to this analytic tactic include multielement design (Ulman & Sulzer-Azaroff, 1975), multiple sched- ule design (Hersen & Barlow, 1976), concurrent sched- ule design (Hersen & Barlow, 1976), and simultaneous treatment design (Kazdin & Hartmann, 1978).6

Operation and Logic of the Alternating Treatments Design

The alternating treatments design is characterized by the rapid alternation of two or more distinct treatments (i.e., independent variables) while their effects on the tar- get behavior (i.e., dependent variable) are measured. In contrast to the reversal design in which experimental ma- nipulations are made after steady state responding is achieved in a given phase of an experiment, the different

Treatment A

Treatment B

Sessions

R es

p o

n se

M ea

su re

Figure 9 Graphic prototype of an alternating treat- ments design comparing the differential effects of two treatments (A and B).

6A design in which two or more treatments are concurrently or simultane- ously presented, and in which the subject chooses between treatments, is correctly termed a concurrent schedule or simultaneous treatment design. Some published studies described by their authors as using a simultaneous treatment design have, in fact, employed an alternating treatments design. Barlow and Hayes (1979) could find only one true example of a simulta- neous treatment design in the applied literature: a study by Browning (1967) in which three techniques for reducing the bragging of a 10-year-old boy were compared.

interventions in an alternating treatments design are ma- nipulated independent of the level of responding. The de- sign is predicated on the behavioral principle of stimulus discrimination. To aid the subject’s discrimination of which treatment condition is in effect during a given ses- sion, a distinct stimulus (e.g., a sign, verbal instructions, different colored worksheets) is often associated with each treatment.

The data are plotted separately for each intervention to provide a ready visual representation of the effects of each treatment. Because confounding factors such as time of administration have been neutralized (presum- ably) by counterbalancing, and because the two treat- ments are readily discriminable by subjects through instructions or other discriminative stimuli, differences in the individual plots of behavior change corresponding with each treatment should be attributable to the treat- ment itself, allowing a direct comparison between two (or more) treatments. (Barlow & Hayes, 1979, p. 200)

Figure 9 shows a graphic prototype of an alternat- ing treatments design comparing the effects of two treat- ments, A and B, on some response measure. In an alternating treatments design, the different treatments can be alternated in a variety of ways. For example, the treat- ments might be (a) alternated across daily sessions, one treatment in effect each day; (b) administered in separate sessions occurring within the same day; or (c) imple- mented each during a portion of the same session. Coun- terbalancing the days of the week, times of day, sequence in which the different treatments occur (e.g., first or sec- ond each day), persons delivering the different treatments, and so forth, reduces the probability that any observed differences in behavior are the result of variables other than the treatments themselves. For example, assume that Treatments A and B in Figure 9 were each administered

208

Reversal and Alternating Treatments Designs

for a single 30-minute session each day, with the daily sequence of the two treatments determined by a coin flip.

The data points in Figure 9 are plotted on the hor- izontal axis to reflect the actual sequence of treatments each day. Thus, the horizontal axis is labeled Sessions, and each consecutive pair of sessions occurred on a sin- gle day. Some published reports of experiments using an alternating treatments design in which two or more treat- ments were presented each day (or session) plot the mea- sures obtained during each treatment above the same point on the horizontal axis, thus implying that the treat- ments were administered simultaneously. This practice masks the temporal order of events and has the unfortu- nate consequence of making it difficult for the researcher or reader to discover potential sequence effects.

The three components of steady state strategy—pre- diction, verification, and replication—are found in the alternating treatments design. However, each component is not readily identified with a separate phase of the de- sign. In an alternating treatments design, each successive data point for a specific treatment plays all three roles: It provides (a) a basis for the prediction of future levels of responding under that treatment, (b) potential verification of the previous prediction of performance under that treat- ment, and (c) the opportunity for replication of previous effects produced by that treatment.

To see this logic unfold, the reader should place a piece of paper over all the data points in Figure 9 ex- cept those for the first five sessions of each treatment. The visible portions of the data paths provide the basis for predicting future performance under each respective treatment. Moving the paper to the right reveals the two data points for the next day, each of which provides a de- gree of verification of the previous predictions. As more data are recorded, the predictions of given levels of re- sponding within each treatment are further strengthened by continued verification (if those additional data con- form to the same level and/or trend as their predecessors). Replication occurs each time Treatment A is reinstated and measurement reveals responding similar to previous Treatment A measures and different from those obtained when Treatment B is in effect. Likewise, another mini- replication is achieved each time a reintroduction of Treatment B results in measures similar to previous Treat- ment B measures and different from Treatment A levels of responding. A consistent sequence of verification and replication is evidence of experimental control and strengthens the investigator’s confidence of a functional relation between the two treatments and different levels of responding.

The presence and degree of experimental control in an alternating treatments design is determined by visual inspection of the differences between (or among) the data paths representing the different treatments. Experimental

control is defined in this instance as objective, believable evidence that different levels of responding are pre- dictably and reliably produced by the presence of the dif- ferent treatments. When the data paths for two treatments show no overlap with each other and either stable levels or opposing trends, a clear demonstration of experimen- tal control has been made. Such is the case in Figure 9, in which there is no overlap of data paths and the picture of differential effects is clear. When some overlap of data paths occurs, a degree of experimental control over the target behavior can still be demonstrated if the majority of data points for a given treatment fall outside the range of values of the majority of data points for the contrast- ing treatment.

The extent of any differential effects produced by two treatments is determined by the vertical distance— or fractionation—between their respective data paths and quantified by the vertical axis scale. The greater the ver- tical distance, the greater the differential effect of the two treatments on the response measure. It is possible for ex- perimental control to be shown between two treatments but for the amount of behavior change to be socially in- significant. For instance, experimental control may be demonstrated for a treatment that reduces a person’s se- vere self-injurious behavior from 10 occurrences per hour to 2 per hour, but the participant is still engaged in self- mutilation. However, if the vertical axis is scaled mean- ingfully, the greater the separation of data paths on the vertical axis, the higher the likelihood that the difference represents a socially significant effect.

Data from an experiment that compared the effects of two types of group-contingent rewards on the spelling accuracy of fourth-grade underachievers (Morgan, 1978) illustrate how the alternating treatments design reveals experimental control and the quantification of differential effects. The six children in the study were divided into two equally skilled teams of three on the basis of pretest scores. Each day during the study the students took a five- word spelling test. The students received a list of the words the day before, and a 5-minute study period was provided just prior to the test. Three different conditions were used in the alternating treatments design: (a) no game, in which the spelling tests were graded immedi- ately and returned to the students, and the next scheduled activity in the school day was begun; (b) game, in which test papers were graded immediately, and each member of the team who had attained the highest total score re- ceived a mimeographed Certificate of Achievement and was allowed to stand up and cheer; and (c) game plus, consisting of the same procedure as the game condition, plus each student on the winning team also received a small trinket (e.g., a sticker or pencil).

The results for Student 3 (see Figure 10) show that experimental control over spelling accuracy was obtained

209

Reversal and Alternating Treatments Designs

N u

m b

er o

f W

o rd

s Sp

el le

d C

o rr

ec tl

5 10 15 20

game (4.16) game plus (4.14)

no game (2.16)

no game

game

game plus

Days

Student 3Figure 10 Alternating treatments design comparing the effects of three different treat- ments on the spelling accuracy of a fourth- grade student. From Comparison of Two “Good Behavior Game” Group Contin- gencies on the Spelling Accuracy of Fourth-Grade Students by Q. E. Morgan, 1978, unpublished master’s thesis, The Ohio State University. Reprinted by permission.

between the no-game condition and both the game and the game-plus conditions. Only the first two no-game data points overlap the lower range of scores obtained during the game or the game-plus conditions. However, the data paths for the game and game-plus conditions overlap completely and continuously throughout the study, revealing no difference in spelling accuracy be- tween the two treatments. The vertical distance between the data paths represents the amount of improvement in spelling accuracy between the no-game condition and the game and the game-plus conditions. The mean difference between the two game conditions and the no-game con- dition was two words per test. Whether such a difference represents a significant improvement is an educational question, not a mathematical or statistical one, but most educators and parents would agree that an increase of two words spelled correctly out of five is socially signif- icant, especially if that gain can be sustained from week to week. The cumulative effect over a 180-day school year would be impressive. There was virtually no differ- ence in Student 3’s spelling performance between the game and game-plus conditions. However, even a larger mean difference would not have contributed to the con- clusions of the study because of the lack of experimen-

tal control between the game and the game-plus treat- ments.

Student 6 earned consistently higher spelling scores in the game-plus condition than he did in the game or no- game conditions (see Figure 11). Experimental control was demonstrated between the game-plus and the other two treatments for Student 6, but not between the no- game and game conditions. Again, the difference in re- sponding between treatments is quantified by the vertical distance between the data paths. In this case there was a mean difference of 1.55 correctly spelled words per test between the game-plus and no-game conditions.

Figures 10 and 11 illustrate two other important points about the alternating treatments design. First, the two graphs show how an alternating treatments design enables a quick comparison of interventions. Although the study would have been strengthened by the collection of additional data, after 20 sessions the teacher had suf- ficient empirical evidence for selecting the most effective consequences for each student. If only two conditions had been compared, even fewer sessions may have been re- quired to identify the most effective intervention. Second, these data underscore the importance of evaluating treat- ment effects at the level of the individual subject. All six

N u

m b

er o

f W

o rd

s Sp

el le

d C

o rr

ec tl

5 10 15 20

game (3.40) no game (3.30)

game plus (4.85)

no game

game

game plus

Days

Student 6Figure 11 Alternating treatments design comparing the effects of three different treat- ments on the spelling accuracy of a fourth- grade student. From Comparison of Two “Good Behavior Game” Group Contin- gencies on the Spelling Accuracy of Fourth-Grade Students by Q. E. Morgan, 1978, unpublished master’s thesis, The Ohio State University. Reprinted by permission.

210

Reversal and Alternating Treatments Designs

children spelled more words correctly under one or both of the game conditions than they did under the no-game condition. However, Student 3’s spelling accuracy was equally enhanced by either the game or the game-plus contingency, whereas Student 6’s spelling scores im- proved only when a tangible reward was available.

Variations of the Alternating Treatments Design

The alternating treatments design can be used to com- pare one or more treatments to a no-treatment or baseline condition, assess the relative contributions of individual components of a package intervention, and perform parametric investigations in which different values of an independent variable are alternated to determine differ- ential effects on behavior change. Among the most com- mon variations of the alternating treatments design are the following:

• Single-phase alternating treatments design without a no-treatment control condition

• Single-phase design in which two or more condi- tions, one of which is a no-treatment control condi- tion, are alternated

• Two-phase design consisting of an initial baseline phase followed by a phase in which two or more conditions (one of which may be a no-treatment control condition) are alternated

• Three-phase design consisting of an initial base- line, a second phase in which two or more condi- tions (one of which may be a no-treatment control condition) are alternated, and a final phase in which only the treatment that proved most effec- tive is implemented

Alternating Treatments Design without a No-Treatment Control Condition

One application of the alternating treatments design con- sists of a single-phase experiment in which the effects of two or more treatment conditions are compared (e.g., Barbetta, Heron, & Heward, 1993; McNeish, Heron, & Okyere, 1992; Morton, Heward, & Alber, 1998). A study by Belfiore, Skinner, and Ferkis (1995) provides an ex- cellent example of this design. They compared the ef- fects of two instructional procedures—trial-repetition and response-repetition—on the acquisition of sight words by three elementary students with learning disabilities in reading. An initial training list of five words for each con- dition was created by random selection from a pool of unknown words (determined by pretesting each student). Each session began with a noninstructional assessment of

unknown and training words, followed by both condi- tions. The order of instructional conditions was counter- balanced across sessions. Words spoken correctly on three consecutive noninstructional assessments were consid- ered mastered and replaced as training words with un- known words.

The trial-repetition condition consisted of one re- sponse opportunity within each of five interspersed prac- tice trials per word. The experimenter placed a word card on the table and said, “Look at the word, and say the word.” If the student made a correct response within 3 seconds, the experimenter said, “Yes, the word is _____.” (p. 347). If the student’s initial response was incorrect, or the student made no response within 3 seconds, the ex- perimenter said, “No, the word is _____,” and the student repeated the word. The experimenter then presented the next word card and the procedure repeated until five prac- tice trials (antecedent-response-feedback) were provided with each word.

The response-repetition condition also consisted of five response opportunities per word, but all five re- sponses occurred within a single practice trial for each word. The experimenter placed a word card on the table and said, “Look at the word, and say the word.” If the student made a correct response within 3 seconds, the experimenter said, “Yes, the word is _____, please repeat the word four more times” (p. 347). If the student made an incorrect response or no response within 3 seconds, the experimenter said, “No, the word is _____.” The student then repeated the word and was instructed to repeat it four more times.

Figure 12 shows the cumulative number of words mastered by each student under both conditions. Even though the number of correct responses per word during instruction was identical in both conditions, all three stu- dents had higher rates of learning new words in the trial- repetition condition than in the response-repetition condition. These results obtained within the simple al- ternating treatments design enabled Belfiore and col- leagues (1995) to conclude that, “Response repetition outside the context of the learning trial (i.e., of the three- term contingency) was not as effective as repetition that included antecedent and consequent stimuli in relation to the accurate response” (p. 348).

Alternating Treatments Design with No-Treatment Control Condition

Although not a requirement of the design, a no-treatment condition is often incorporated into the alternating treat- ments design as one of the treatments to be compared. For example, the no-game condition in the Morgan (1978) study served as a no-treatment control condition against which the students’ spelling scores in the game

211

Reversal and Alternating Treatments Designs

26 24 22 20 18 16 14 12 10

8 6 4 2

C u

m u

la ti

ve N

u m

b er

o f

W o

rd s

M as

te re

Intervention

3 6 9 12 15 18 21 24

Maint

Trial Repetition

Response Repetition

Lou

26 24 22 20 18 16 14 12 10

8 6 4 2

Intervention

3 6 9 12 15 18 21 24

Maint

Trial Repetition

Response Repetition

Carl

Session

Intervention

3 6 9 12 15 18 21 24

Maint

Trial Repetition

Response Repetition

Hal

Figure 12 Single-phase alternating treatments design without a no-treatment control condition. From “Effects of Response and Trial Repetition on Sight-Word Training for Students with Learning Disabilities” by P. J. Belfiore, C. H. Skinner, and M. A. Ferkis, 1995, Journal of Applied Behavior Analysis, 28, p. 348. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

and game-plus conditions were compared (see Figures 10 and 11).

Including a no-treatment control condition as one of the experimental conditions in an alternating treatments design provides valuable information on any differences in responding under the intervention treatment(s) and no treatment. However, the measures obtained during the no-treatment control condition should not be considered representative of an unknown preintervention level of re- sponding. It may be that the measures obtained in the no- treatment condition represent only the level of behavior under a no-treatment condition when it is interspersed within an ongoing series of treatment condition(s), and do not represent the level of behavior that existed before the alternating treatments design was begun.

Alternating Treatments Design with Initial Baseline

Investigators using the alternating treatments tactic often use a two-phase experimental design in which baseline measures are collected until a stable level of responding or countertherapeutic trend is obtained prior to the alter- nating treatments phase (e.g., Martens, Lochner, & Kelly, 1992 [see Figure 6 of Chapter entitled “Schedules of Reinforcement”]). Sometimes the baseline condition is continued during the alternating treatments phase as a no-treatment control condition.

A study by J. Singh and N. Singh (1985) provides an excellent example of an alternating treatments design incorporating an initial baseline phase. The experiment evaluated the relative effectiveness of two procedures for reducing the number of oral reading errors by students with mental retardation. The first phase of the study con- sisted of a 10-day baseline condition in which each stu- dent was given a new 100-word passage three times each day and told, “Here is the story for this session. I want you to read it. Try your best not to make any errors”

(p. 66). The experimenter sat nearby but did not assist the student, correct any errors, or attend to self-corrections. If a student requested help with new or difficult words, he was prompted to continue reading.

During the alternating treatments phase of the study, three different conditions were presented each day in sep- arate sessions of about 5 minutes each: control (the same procedures as during baseline), word supply, and word analysis. To minimize any sequence or carryover effects from one condition to another, the three conditions were presented in random order each day, each condition was preceded with specific instructions identifying the pro- cedure to be implemented, and an interval of at least 5 minutes separated consecutive sessions. During the word- supply condition, each student was instructed, “Here is the story for this session. I want you to read it. I will help you if you make a mistake. I will tell you the correct word while you listen and point to the word in the book. After that, I want you to repeat the word. Try your best not to make any errors” (p. 67). The experimenter supplied the correct word when an oral reading error was made, had the child repeat the correct word once, and instructed the child to continue reading. During the word-analysis con- dition, each student was instructed, “Here is the story for this session. I want you to read it. I will help you if you make a mistake. I will help you sound out the word and then you can read the word correctly before you carry on reading the rest of the story. Try your best not to make any errors” (p. 67). When errors were made in this condition, the experimenter directed the child’s attention to the pho- netic elements of the word and coaxed the child to sound out correctly each part of the word. Then the experi- menter had the student read the entire word at the nor- mal speed and instructed him or her to continue reading the passage.

The results for the four students who participated in the study are shown in Figure 13. Each baseline data

212

Reversal and Alternating Treatments Designs

Michael

Jane

N u

m b

er o

f O

ra l R

ea d

in g

Er ro

Leanne

0 5 10 15 20 25 30

Days

Rex

Baseline Alternating Treatments Control Word Supply Word Analysis

Figure 13 Alternating treatments design with an initial baseline. From “Comparison of Word-Supply and Word-Analysis Error-Correction Procedures on Oral Reading by Mentally Retarded Children” by J. Singh and N. Singh, 1985, American Journal of Mental Deficiency, 90, p. 67. Copyright 1985 by the American Journal of Mental Deficiency. Reprinted by permission.

point is the mean number of errors for the three daily ses- sions. Although the data in each condition are highly vari- able (perhaps because of the varied difficulty of the different passages used), experimental control is evident. All four students committed fewer errors during the word- supply and the word-analysis conditions than they did during the control condition. Experimental control of oral reading errors, although not complete because of some overlap of the data paths, is also demonstrated between the word-supply and word-analysis conditions, with all four students making fewer errors during the word- analysis condition.

By beginning the study with a baseline phase, J. Singh and N. Singh (1985) were able to compare the level of responding obtained during each of the treat- ments to the natural level of performance uncontami- nated by the introduction of either error-correction

7The mand is one of six types of elementary verbal operants identified by Skinner (1957). Chapter entitled “Verbal Behavior” describes Skinner’s analysis of verbal behavior and its importance to applied behavior analysis. 8Stimulus preference assessment procedures are described in chapter en- titled “Positive Reinforcement.”

intervention. In addition, the initial baseline served as the basis for predicting and assessing the measures ob- tained during the control sessions of the alternating treat- ments phase of the study. The measures obtained in the alternating control condition matched the relatively high frequency of errors observed during the initial baseline phase, providing evidence that (a) the vertical distance between the data paths for the word-supply and word- analysis conditions and the data path for the control con- dition represents the true amount of improvement produced by each treatment and (b) the frequency of er- rors during the control condition was not influenced by reduced errors during the other two treatments (i.e., no generalized reduction in oral reading errors from the treated passages to untreated passages occurred).

Alternating Treatments Design with Initial Baseline and Final Best Treatment Phase

A widely used variation of the alternating treatments de- sign consists of three sequential phases: an initial base- line phase, a second phase comparing alternating treatments, and a final phase in which only the most ef- fective treatment is administered (e.g., Heckaman, Alber, Hooper, & Heward, 1998; Kennedy & Souza, 1995, Study 4; Ollendick, Matson, Esvelt-Dawson, & Shapiro, 1980; N. Singh, 1990; N. Singh & J. Singh, 1984; N. Singh & Winton, 1985). Tincani (2004) used an alter- nating treatments design with an initial baseline and final best treatment phase to investigate the relative effective- ness of sign language and picture exchange training on the acquisition of mands (requests for preferred items) by two children with autism.7 A related research ques- tion was whether a relation existed between students’ pre- existing motor imitation skills and their abilities to learn mands through sign language or by picture exchange. Two assessments were conducted for each student prior to baseline. A stimulus preference assessment (Pace, Ivancic, Edwards, Iwata, & Page, 1985) was conducted to identify a list of 10 to 12 preferred items (e.g., drinks, edibles, toys), and each student’s ability to imitate 27 hand, arm, and finger movements similar to those re- quired for sign language was assessed.8

The purpose of baseline was to ensure that the par- ticipants were not able to request preferred items with picture exchange, sign language, or speech prior to train- ing. Baseline trials consisted of giving the student 10 to 20 seconds of noncontingent access to a preferred item, removing the item briefly, and then placing it out of the

213

Reversal and Alternating Treatments Designs

student’s reach. A laminated 2-inch-by-2-inch picture of the item was placed in front of the student. If the student placed the picture symbol in the experimenter’s hand, signed the name of the item, or spoke the name of the item within 10 seconds, the experimenter provided ac- cess to the item. If not, the item was removed and the next item on the list was presented. Following a three- session baseline, during which neither participant emit- ted an independent mand in any modality, the alternating treatments phase was begun.

The sign language training procedures were adapted from Sundberg and Partington’s (1998) Teaching Lan- guage to Children with Autism or Other Developmental Disabilities. The simplest sign from American Sign Lan- guage for each item was taught. Procedures used in the PECS training condition were adapted from Bondy and Frost’s (2002) The Picture Exchange Communication

System Training Manual. In both conditions, training on each preferred item continued for five to seven trials per session, or until the participant showed no interest in the item. At that time training then began on the next item and continued until all 10 or 12 items on the participant’s list of preferred items had been presented. During the study’s final phase, each participant received either sign language or PECS training only, depending on which method had been most successful during the alternating treatments phase.

The percentage of independent mands by the two students throughout the study is shown in Figures 14 (Jennifer) and 15 (Carl). Picture exchange training was clearly more effective than sign language for Jennifer. Jennifer demonstrated weak motor imitation skills in the prebaseline assessment, correctly imitating 20% of the motor movements attempted in the prebaseline imitation

100

P er

ce n

ta ge

o f

In d

ep en

d en

t M

an d

5 10 15 20 25 30

Jennifer

Sessions

Baseline PECS IIPECS I and Sign Language PECS IIIbPECS IIIa

PECS

Sign Language

Figure 14 Alternating treatments design with an initial baseline and a final best-treatment-only condition. From “Comparing the Picture Exchange Communication System and Sign Language Training for Children with Autism” by M. Tincani, 2004, Focus on Autism and Other Developmental Disabilities, 19, p. 160. Copyright 2004 by Pro-Ed. Used by permission.

100

P er

ce n

ta ge

o f

In d

ep en

d en

t M

an d

5 10 15 20 25 30 Sessions

Baseline

Modified Sign Training

PECS I and Sign Language Sign Language “Best Treatment”

PECS

Sign Language

Carl

Figure 15 Alternating treat- ments design with an initial baseline and a final best-treatment-only condition. From “Comparing the Picture Exchange Communication System and Sign Language Training for Children with Autism” by M. Tincani, 2004, Focus on Autism and Other Developmental Disabilities, 19, p. 159. Copyright 2004 by Pro-Ed. Used by permission.

214

Reversal and Alternating Treatments Designs

assessment. After a slight modification in sign language training procedures was implemented to eliminate Carl’s prompt dependency, he emitted independent mands more often during sign language training than with picture ex- change training. Carl’s preexisting motor imitation skills were better than Jennifer’s. He imitated correctly 43% of the attempted motor movements in the prebaseline imi- tation assessment.

This study highlights the importance of individual analyses and exploring the possible influence of variables not manipulated during the study. In discussing the study’s results, Tincani (2004) noted that

For learners without hand-motor imitation skills, includ- ing many children with autism, PECS training may be more appropriate, at least in terms of initial mand acqui- sition. Jennifer had weak hand-motor imitation skills prior to intervention and learned picture exchange more rapidly than sign language. For learners who have mod- erate hand-motor imitation skills, sign language training may be equally, if not more, appropriate. Carl had mod- erate hand-motor imitation skills prior to intervention and learned sign language more rapidly than picture ex- change. (p. 160)

Advantages of the Alternating Treatments Design

The alternating treatments design offers numerous ad- vantages for evaluating and comparing two or more in- dependent variables. Most of the benefits cited here were described by Ulman and Sulzer-Azaroff (1975), who are credited with first bringing the rationale and possibilities of the alternating treatments design to the attention of the applied behavior analysis community.

Does Not Require Treatment Withdrawal

A major advantage of the alternating treatments design is that it does not require the investigator to withdraw a seemingly effective treatment to demonstrate a functional relation. Reversing behavioral improvements raises eth- ical issues that can be avoided with the alternating treat- ments design. Regardless of ethical concerns, however, administrators and teachers may be more likely to accept an alternating treatments design over a reversal design even when one of the alternating treatments is a no-treatment control condition. “It would appear that a return to base- line conditions every other day or every third day is not as disagreeable to a teacher as is first establishing a high level of desirable behavior for a prolonged period, and then reinstating the baseline behaviors” (Ulman & Sulzer- Azaroff, 1975, p. 385).

Speed of Comparison

The experimental comparison of two or more treatments can often be made quickly with the alternating treatments design. In one study an alternating treatments design en- abled the superiority of one treatment over another in in- creasing the cooperative behavior of a 6-year-old boy to be determined after only 4 days (McCullough, Cornell, McDaniel, & Mueller, 1974). The alternating treatments design’s ability to produce useful results quickly is a major reason that it is the basic experimental tactic used in functional behavior analysis.

When the effects of different treatments become ap- parent early in an alternating treatments design, the in- vestigator can then switch to programming only the most effective treatment. The efficiency of the alternating treat- ments design can leave a researcher with meaningful data even when an experiment must be terminated early (Ulman & Sulzer-Azaroff, 1975). A reversal or multiple baseline design, on the other hand, must be carried through to completion to show a functional relation.

Minimizes Irreversibility Problem

Some behaviors, even though they have been brought about or modified by the application of the intervention, do not return to baseline levels when the intervention is withdrawn and thereby resist analysis with an A-B-A-B design. However, rapidly alternating treatment and no- treatment (baseline) conditions may reveal differences in responding between the two conditions, especially early in an experiment before responding in the no-treatment condition begins to approximate the level of responding in the treatment condition.

Minimizes Sequence Effects

An alternating treatments design, when properly con- ducted, minimizes the extent to which an experiment’s results are confounded by sequence effects. Sequence ef- fects pose a threat to the internal validity of any experi- ment, but especially to those involving multiple treatments. The concern over sequence effects can be summed up by this simple question: Would the results have been the same if the sequence of treatments had been different? Sequence effects can be extremely diffi- cult to control in experiments using reversal or multiple tactics to compare two or more independent variables because each experimental condition must remain in effect for a fairly long period of time, thereby produc- ing a specific sequence of events. However, in an al- ternating treatments design, the independent variables

215

Reversal and Alternating Treatments Designs

are rapidly alternated with one another in a random fash- ion that produces no particular sequence. Also, each treat- ment is in effect for short periods of time, reducing the likelihood of carryover effects (O’Brien, 1968). The abil- ity to minimize sequence effects makes the alternating treatments design a powerful tool for achieving complex behavior analyses.

Can Be Used with Unstable Data

Determining functional behavior–environment relations in the presence of unstable data presents a serious prob- lem for the applied behavior analyst. Using steady state responding to predict, verify, and replicate behavioral changes is the foundation of experimental reasoning in behavior analysis (Sidman, 1960). Obtaining stable base- line responding, however, is extremely difficult with many socially important behaviors of interest to applied behavior analysts. Merely providing a subject with re- peated opportunities to emit a target response can result in gradually improved performance. Although practice effects are worthy of empirical investigation because of their applied and scientific importance (Greenwood, Delquadri, & Hall, 1984; Johnston & Pennypacker, 1993a), the unstable baselines they create pose problems for the analysis of intervention variables. The changing levels of task difficulty inherent in moving through a cur- riculum of progressively more complex material also make obtaining steady state responding for many acade- mic behaviors difficult.

Because the different treatment conditions are alter- nated rapidly in an alternating treatments design, because each treatment is presented many times throughout each time period encompassed by the study, and because no single condition is present for any considerable length of time, it can be presumed that any effects of practice, change in task difficulty, maturation, or other historical variables will be equally represented in each treatment condition and therefore will not differentially affect any one condition more or less than the others. For example, even though each of two data paths representing a stu- dent’s reading performance under two different teaching procedures shows variable and ascending trends that might be due to practice effects and uneven curriculum materials, any consistent separation and vertical distance between the data paths can be attributed to differences in the teaching procedures.

Can Be Used to Assess Generalization of Effects

By alternating various conditions of interest, an experi- menter can continually assess the degree of generalization of behavior change from an effective treatment to other

conditions of interest. For example, by alternating dif- ferent therapists in the final phase of their study of pica behavior, N. Singh and Winton (1985) were able to de- termine the extent to which the overcorrection treatment was effective when presented by different persons.

Intervention Can Begin Immediately

Although determining the preintervention level of re- sponding is generally preferred, the clinical necessity of immediately attempting to change some behaviors pre- cludes repeated measurement in the absence of interven- tion. When necessary, an alternating treatments design can be used without an initial baseline phase.

Considering the Appropriateness of the Alternating Treatments Design

The advantages of the alternating treatments design are significant. As with any experimental tactic, however, the alternating treatments design presents certain disadvan- tages and leaves unanswered certain questions that can be addressed only by additional experimentation.

Multiple Treatment Interference

The fundamental feature of the alternating treatments de- sign is the rapid alternation of two or more independent variables irrespective of the behavioral measures obtained under each treatment. Although the rapid alternation min- imizes sequence effects and reduces the time required to compare treatments, it raises the important question of whether the effects observed under any of the alternated treatments would be the same if each treatment were im- plemented alone. Multiple treatment interference refers to the confounding effects of one treatment on a subject’s behavior being influenced by the effects of another treat- ment administered in the same study.

Multiple treatment interference must always be sus- pected in the alternating treatments design (Barlow & Hayes, 1979; McGonigle, Rojahn, Dixon, & Strain, 1987). However, by following the alternating treatments phase with a phase in which only the most effective treat- ment condition is in effect, the experimenter can assess the effects of that treatment when administered in isolation.

Unnatural Nature of Rapidly Alternating Treatments

The rapid back-and-forth switching of treatments does not reflect the typical manner in which clinical and edu- cational interventions are applied. From an instructional

216

Reversal and Alternating Treatments Designs

perspective, rapid switching of treatments can be viewed as artificial and undesirable. In most instances, however, the quick comparison of treatments offered by the alter- nating treatments design compensates for concerns about its contrived nature. The concern of whether participants might suffer detrimental effects from the rapid alternation of conditions is an empirical question that can be deter- mined only by experimentation. Also, it is helpful for practitioners to remember that one purpose of the alter- nating treatments design is to identify an effective inter- vention as quickly as possible so that the participant does not have to endure ineffective instructional approaches or treatments that would delay progress toward educa- tional goals. On balance, the advantages of rapidly switching treatments to identify an efficacious interven- tion outweigh any undesirable effects that such manipu- lation may cause.

Limited Capacity

Although the alternating treatments design enables an el- egant, scientifically sound method for comparing the dif- ferential effects of two or more treatments, it is not an open-ended design in which an unlimited number of treat- ments can be compared. Although alternating treatments designs with up to five conditions have been reported (e.g., Didden, Prinson, & Sigafoos, 2000), in most situ- ations a maximum of four different conditions (one of which may be a no-treatment control condition) can be compared effectively within a single phase of an alter- nating treatments design, and in many instances only two different treatments can be accommodated. To separate the effects of each treatment condition from any effects that may be caused by aspects of the alternating treat- ments design, each treatment must be carefully counter- balanced across all potentially relevant aspects of its administration (e.g., time of day, order of presentation, settings, therapists). In many applied settings the logistics of counterbalancing and delivering more than two or three treatments would be cumbersome and would cause the experiment to require too many sessions to complete. Also, too many competing treatments can decrease the subject’s ability to discriminate between treatments, thereby reducing the design’s effectiveness.

Selection of Treatments

Theoretically, although an alternating treatments design can be used to compare the effects of any two discrete treatments, in reality the design is more limited. To en- hance the probability of discrimination between condi- tions (i.e., obtaining reliable, measurable differences in

behavior), the treatments should embody significant dif- ferences from one to the other. For example, an investi- gator using an alternating treatments design to study the effects of group size on students’ academic performance during instruction might include conditions of 4, 10, and 20 students. Alternating conditions of 6, 7, and 8 stu- dents, however, is less likely to reveal a functional rela- tion between group size and performance. However, a treatment condition should not be selected for inclusion in an alternating treatments design only because it might yield a data path that is easily differentiated from that of another condition. The applied in applied behavior analy- sis encompasses the nature of treatment conditions as well as the nature of behaviors investigated (Wolf, 1978). An important consideration in selecting treatment con- ditions should be the extent to which they are represen- tative of current practices or practices that could conceivably be implemented. For example, although an experiment comparing the effects of 5 minutes, 10 min- utes, and 30 minutes of math homework per school night on math achievement might be useful, a study comparing the effects of 5 minutes, 10 minutes, and 3 hours of math homework per night probably would not be. Even if such a study found 3 hours of nightly math homework ex- tremely effective in raising students’ achievement in math, few teachers, parents, administrators, or students would carry out a program of 3 hours of nightly homework for a single content area.

Another consideration is that some interventions may not produce important behavior change unless and until they have been implemented consistently over a contin- uous period of time.

When a multielement baseline design is employed, over- lapping data do not necessarily rule out the possible effi- cacy of an experimental procedure. The session-by- session alternation of conditions might obscure effects that could be observed if the same condition was pre- sented during several consecutive sessions. It is there- fore possible that a given treatment may prove to be effective with a reversal or multiple baseline design, but not with a multielement baseline design. (Ulman & Sulzer-Azaroff, 1975, p. 382)

The suspicion that a given treatment may be effective if it is presented in isolation for an extended period is an empirical question that can be explored properly only through experimentation. At one level, if extended ap- plication of a single treatment results in behavioral im- provement, the practitioner might be satisfied, and no further action would be needed. However, the prac- titioner-researcher who is interested in determining experimental control might return to an alternating treatments design and compare the performance of the single treatment with that of another intervention.

217

Reversal and Alternating Treatments Designs

Summary Reversal Design

1. The reversal tactic (A-B-A) entails repeated measurement of behavior in a given setting during three consecutive phases: (a) a baseline phase (absence of the independent variable), (b) a treatment phase (introduction of the inde- pendent variable), and (c) a return to baseline conditions (withdrawal of the independent variable).

2. The reversal design is strengthened tremendously by rein- troducing the independent variable in the form of an A-B-A-B design. The A-B-A-B design is the most straight- forward and generally most powerful intrasubject design for demonstrating functional relations.

Variations of the A-B-A-B Design

3. Extending the A-B-A-B design with repeated reversals may provide a more convincing demonstration of a func- tional relation than a design with one reversal.

4. The B-A-B reversal design can be used with target be- haviors for which an initial baseline phase is inappropri- ate or not possible for ethical or practical reasons.

5. Multiple treatment reversal designs use the reversal tactic to compare the effects of two or more experimental con- ditions to baseline and/or to one another.

6. Multiple treatment reversal designs are particularly sus- ceptible to confounding by sequence effects.

7. The NCR reversal technique enables the isolation and analysis of the contingent aspect of reinforcement.

8. Reversal techniques incorporating DRO and DRI/DRA control conditions can also be used to demonstrate the ef- fects of contingent reinforcement.

Considering the Appropriateness of the Reversal Design

9. An experimental design based on the reversal tactic is in- effective in evaluating the effects of a treatment variable that, by its very nature, cannot be withdrawn once it has been presented (e.g., instruction, modeling).

10. Once improved, some behaviors will not reverse to base- line levels even though the independent variable has been withdrawn. Such behavioral irreversibility precludes ef- fective use of the reversal design.

11. Legitimate social, educational, and ethical concerns are often raised over withdrawing a seemingly effective treat- ment variable to provide scientific verification of its func- tion in changing behavior.

12. Sometimes very brief reversal phases, or even one-session baseline probes, can demonstrate believable experimental control.

Alternating Treatments Design

13. The alternating treatments design compares two or more distinct treatments (i.e., independent variables) while their effects on the target behavior (i.e., dependent variable) are measured.

14. In an alternating treatments design, each successive data point for a specific treatment plays three roles: it provides (a) a basis for the prediction of future levels of responding under that treatment, (b) potential verification of the pre- vious prediction of performance under that treatment, and (c) the opportunity for replication of previous effects pro- duced by that treatment.

15. Experimental control is demonstrated in the alternating treatments design when the data paths for two different treatments show little or no overlap.

16. The extent of any differential effects produced by two treat- ments is determined by the vertical distance between their re- spective data paths and quantified by the vertical axis scale.

Variations of the Alternating Treatments Design

17. Common variations of the alternating treatments design include the following:

• Single-phase alternating treatments design without a no- treatment control condition

• Single-phase design with a no-treatment control condition

• Two-phase design: initial baseline phase followed by the alternating treatments phase

• Three-phase design: initial baseline phase followed by the alternating treatments phase and a final best treatment phase

Advantages of the Alternating Treatments Design

18. Advantages of the alternating treatments design include the following:

• Does not require treatment withdrawal.

• Quickly compares the relative effectiveness of treatments.

• Minimizes the problem of irreversibility.

• Minimizes sequence effects.

• Can be used with unstable data patterns.

• Can be used to assess generalization of effects.

• Intervention can begin immediately.

Considering the Appropriateness of the Alternating Treatments Design

19. The alternating treatments design is susceptible to multi- ple treatment interference. However, by following the al- ternating treatments phase with a phase in which only one

218

22. The alternating treatments design is most effective in re- vealing the differential effects of treatment conditions that differ significantly from one another.

23. The alternating treatments design is not effective for as- sessing the effects of an independent variable that pro- duces important changes in behavior only when it is consistently administered over a continuous period of time.

Reversal and Alternating Treatments Designs

treatment is administered, the experimenter can assess the effects of that treatment in isolation.

20. The rapid back-and-forth switching of treatments does not reflect the typical manner in which interventions are ap- plied and may be viewed as artificial and undesirable.

21. An alternating treatments phase is usually limited to a maximum of four different treatment conditions.

219

changing criterion design delayed multiple baseline design multiple baseline across

behaviors design

multiple baseline across settings design

multiple baseline across subjects design

multiple baseline design multiple probe design

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 5: Experimental Evaluation of Interventions

5-1 Systematically manipulate independent variables to analyze their effects on treatment.

(d) Use changing criterion design.

(e) Use multiple baseline designs.

5-2 Identify and address practical and ethical considerations in using various experimental designs.

Multiple Baseline and Changing Criterion Designs

Key Terms

220

This chapter describes two additional experi- mental tactics for analyzing behavior–environ- ment relations—the multiple baseline design and

the changing criterion design. In a multiple baseline de- sign, after collecting initial baseline data simultaneously across two or more behaviors, settings, or people, the be- havior analyst then applies the treatment variable sequen- tially across these behaviors, settings, or people and notes the effects. The changing criterion design is used to ana- lyze improvements in behavior as a function of stepwise, incremental criterion changes in the level of responding required for reinforcement. In both designs, experimental control and a functional relation are demonstrated when the behaviors change from a steady state baseline to a new steady state after the introduction of the independent vari- able is applied, or a new criterion established.

Multiple Baseline Design The multiple baseline design is the most widely used ex- perimental design for evaluating treatment effects in ap- plied behavior analysis. It is a highly flexible tactic that enables researchers and practitioners to analyze the ef- fects of an independent variable across multiple behav- iors, settings, and/or subjects without having to withdraw the treatment variable to verify that the improvements in behavior were a direct result of the application of the treatment. As you recall, the reversal design by its very nature requires that the independent variable be with- drawn to verify the prediction established in baseline. This is not so with the multiple baseline design.

Operation and Logic of the Multiple Baseline Design

Baer, Wolf, and Risley (1968) first described the multiple baseline design in the applied behavior analysis litera- ture. They presented the multiple baseline design as an al- ternative to the reversal design for two situations: (a) when the target behavior is likely to be irreversible or (b) when it is undesirable, impractical, or unethical to re- verse conditions. Figure 1 illustrates Baer and col- leagues’ explanation of the basic operation of the multiple baseline design.

In the multiple baseline technique, a number of re- sponses are identified and measured over time to pro- vide baselines against which changes can be evaluated. With these baselines established, the experimenter then applies an experimental variable to one of the behaviors, produces a change in it, and perhaps notes little or no change in the other baselines. If so, rather than reversing the just-produced change, he instead applies the experi- mental variable to one of the other, as yet unchanged, re- sponses. If it changes at that point, evidence is accruing

1Although most of the graphic displays created or selected for this text as examples of experimental design tactics show data plotted on noncumula- tive vertical axes, the reader is reminded that repeated measurement data collected within any type of experimental design can be plotted on both noncumulative and cumulative graphs. For example, Lalli, Zanolli, and Wohn (1994) and Mueller, Moore, Doggett, and Tingstrom (2000) used cumulative graphs to display the data they collected in multiple baseline de- sign experiments; and Kennedy and Souza (1995) and Sundberg, Endicott, and Eigenheer (2000) displayed the data they obtained in reversal designs on cumulative graphs. Students of applied behavior analysis should be careful not to confuse the different techniques for graphically displaying data with tactics for experimental analysis.

that the experimental variable is indeed effective, and that the prior change was not simply a matter of coinci- dence. The variable then may be applied to still another response, and so on. The experimenter is attempting to show that he has a reliable experimental variable, in that each behavior changes maximally only when the experi- mental variable is applied to it. (p. 94)

The multiple baseline design takes three basic forms:

• The multiple baseline across behaviors design, consisting of two or more different behaviors of the same subject

• The multiple baseline across settings design, con- sisting of the same behavior of the same subject in two or more different settings, situations, or time periods

• The multiple baseline across subjects design, con- sisting of the same behavior of two or more differ- ent participants (or groups)

Although only one of the multiple baseline design’s basic forms is called an “across behaviors” design, all multiple baseline designs involve the time-lagged appli- cation of a treatment variable across technically different (meaning independent) behaviors. That is, in the multi- ple baseline across settings design, even though the sub- ject’s performance of the same target behavior is measured in two or more settings, each behavior–setting combination is conceptualized and treated as a different behavior for analysis. Similarly, in a multiple baseline across subjects design, each subject–behavior combina- tion functions as a different behavior in the operation of the design.

Figure 2 shows the same data set displayed in Figure 1 with the addition of data points representing predicted measures if baseline conditions were not changed and shaded areas illustrating how the three ele- ments of baseline logic—prediction, verification, and replication—are operationalized in the multiple baseline design.1 When stable baseline responding has been achieved for Behavior 1, a prediction is made that if the environment were held constant, continued measurement

Multiple Baseline and Changing Criterion Designs

221

B eh

av io

r 1

Baseline Treatment

B eh

av io

r 2

B eh

av io

r 3

R es

p o

n se

M ea

su re

Sessions 5 10 15 20 25 30

Figure 1 Graphic prototype of a multiple baseline design.

would reveal similar levels of responding. When the re- searcher’s confidence in such a prediction is justifiably high, the independent variable is applied to Behavior 1. The open data points in the treatment phase for Behavior 1 represent the predicted level of responding. The solid data points show the actual measures obtained for Be- havior 1 during the treatment condition. These data show a discrepancy with the predicted level of responding if no changes had been made in the environment, thereby suggesting that the treatment may be responsible for the change in behavior. The data collected for Behavior 1 in a multiple baseline design serve the same functions as the data collected during the first two phases of an A-B-A-B reversal design.

Continued baseline measures of the other behaviors in the experiment offer the possibility of verifying the prediction made for Behavior 1. In a multiple baseline design, verification of a predicted level of responding for one behavior (or tier) is obtained if little or no change is observed in the data paths of the behaviors (tiers) that are

still exposed to the conditions under which the predic- tion was made. In Figure 2 those portions of the base- line condition data paths for Behaviors 2 and 3 within the shaded boxes verify the prediction for Behavior 1. At this point in the experiment, two inferences can be made: (a) The prediction that Behavior 1 would not change in a constant environment is valid because the environment was held constant for Behaviors 2 and 3 and their levels of responding remained unchanged; and (b) the observed changes in Behavior 1 were brought about by the inde- pendent variable because only Behavior 1 was exposed to the independent variable and only Behavior 1 changed.

In a multiple baseline design, the independent vari- able’s function in changing a given behavior is inferred by the lack of change in untreated behaviors. However, verification of function is not demonstrated directly as it is with the reversal design, thereby making the multiple baseline design an inherently weaker tactic (i.e., less con- vincing from the perspective of experimental control) for revealing a functional relation between the independent

Multiple Baseline and Changing Criterion Designs

222

Multiple Baseline and Changing Criterion Designs

B eh

av io

r 1

Baseline Treatment

B eh

av io

r 2

B eh

av io

r 3

R es

p o

n se

M ea

su re

Sessions 5 10 15 20 25 30

Figure 2 Graphic prototype of a multiple baseline design with shading added to show elements of baseline logic. Open data points represent predicted measures if baseline conditions were unchanged. Baseline data points for Behaviors 2 and 3 within shaded areas verify of the prediction made for Behavior 1. Behavior 3 baseline data within Bracket A verify the prediction made for Behavior 2. Data obtained during the treatment condition for Behaviors 2 and 3 (cross-hatched shading) provide replications of the experi- mental effect.

variable and a target behavior. However, the multiple baseline design compensates somewhat for this weak- ness by providing the opportunity to verify or refute a se- ries of similar predictions. Not only is the prediction for Behavior 1 in Figure 2 verified by continued stable baselines for Behavior 2 and 3, but the bracketed portion of the baseline data for Behavior 3 also serves as verifi- cation of the prediction made for Behavior 2.

When the level of responding for Behavior 1 under the treatment condition has stabilized or reached a prede- termined performance criterion, the independent variable is then applied to Behavior 2. If Behavior 2 changes in a manner similar to the changes observed for Behavior 1, replication of the independent variable’s effect has been achieved (shown by the data path shaded with cross-hatch- ing). After Behavior 2 has stabilized or reached a prede- termined performance criterion, the independent variable is applied to Behavior 3 to see whether the effect will be replicated. The independent variable may be applied to additional behaviors in a similar manner until a convinc-

ing demonstration of the functional relation has been es- tablished (or rejected) and all of the behaviors targeted for improvement have received treatment.

As with verification, replication of the independent variable’s specific effect on each behavior in a multiple baseline design is not manipulated directly. Instead, the generality of the independent variable’s effect across the behaviors comprising the experiment is demonstrated by applying it to a series of behaviors. Assuming accurate measurement and proper experimental control of rele- vant variables (i.e., the only environmental factor that changes during the course of the experiment should be the presence—or absence—of the independent variable), each time a behavior changes when, and only when, the independent variable is introduced, confidence in the ex- istence of a functional relation increases.

How many different behaviors, settings, or subjects must a multiple baseline design include to provide a be- lievable demonstration of a functional relation? Baer, Wolf, and Risley (1968) suggested that the number of

223

Multiple Baseline and Changing Criterion Designs

replications needed in any design is ultimately a matter to be decided by the consumers of the research. In this sense, an experiment using a multiple baseline design must contain the minimum number of replications nec- essary to convince those who will be asked to respond to the experiment and to the researcher’s claims (e.g., teach- ers, administrators, parents, funding sources, journal ed- itors). A two-tier multiple baseline design is a complete experiment and can provide strong support for the effec- tiveness of the independent variable (e.g., Lindberg, Iwata, Roscoe, Worsdell, & Hanley, 2003; McCord, Iwata, Galensky, Ellingson, & Thomson, 2001; New- strom, McLaughlin, & Sweeney, 1999; Test, Spooner, Keul, & Grossi, 1990). McClannahan, McGee, MacDuff, and Krantz (1990) conducted a multiple baseline design study in which the independent variable was sequentially implemented in an eight-tier design across 12 partici- pants. Multiple baseline designs of three to five tiers are most common. When the effects of the independent vari- able are substantial and reliably replicated, a three- or four-tier multiple baseline design provides a convincing demonstration of experimental effect. Suffice it to say that the more replications one conducts, the more con- vincing the demonstration will be.

Some of the earliest examples of the multiple base- line design in the applied behavior analysis literature were studies by Risley and Hart (1968); Barrish, Saunders, and Wolf (1969); Barton, Guess, Garcia, and Baer (1970); Panyan, Boozer, and Morris (1970); and Schwarz and Hawkins (1970). Some of the pioneering applications of the multiple baseline technique are not readily apparent with casual examination: The authors may not have iden- tified the experimental design as a multiple baseline de- sign (e.g., Schwarz & Hawkins, 1970), and/or the now-common practice of stacking the tiers of a multiple baseline design one on the other so that all of the data can be displayed graphically in the same figure was not always used (e.g., Maloney & Hopkins, 1973; McAllis- ter, Stachowiak, Baer, & Conderman, 1969; Schwarz & Hawkins, 1970).

In 1970, Vance Hall, Connie Cristler, Sharon Cranston, and Bonnie Tucker published a paper that de- scribed three experiments, each an example of one of the three basic forms of the multiple baseline design: across behaviors, across settings, and across subjects. Hall and colleagues’ paper was important not only because it pro- vided excellent illustrations that today still serve as mod- els of the multiple baseline design, but also because the studies were carried out by teachers and parents, indi- cating that practitioners “can carry out important and sig- nificant studies in natural settings using resources available to them” (p. 255).

Multiple Baseline across Behaviors Design

The multiple baseline across behaviors design begins with the concurrent measurement of two or more behav- iors of a single participant. After steady state responding has been obtained under baseline conditions, the investi- gator applies the independent variable to one of the be- haviors while maintaining baseline conditions for the other behavior(s). When steady state or criterion-level performance has been reached for the first behavior, the independent variable is applied to the next behavior, and so on (e.g., Bell, Young, Salzberg, & West, 1991; Gena, Krantz, McClannahan, & Poulson, 1996; Higgins, Williams, & McLaughlin, 2001 [see Figure 26.8]).

Ward and Carnes (2002) used a multiple baseline across behaviors design to evaluate the effects of self- set goals and public posting on the execution of three skills by five linebackers on a college football team: (a) reads, in which the linebacker positions himself to cover a specified area on the field on a pass play or from the line of scrimmage on a run; (b) drops, in which the line- backer moves to the correct position depending on the of- fensive team’s alignment; and (c) tackles. A video camera recorded the players’ movements during all prac- tice sessions and games. Data were collected for the first 10 opportunities each player had with each skill. Reads and drops were recorded as correct if the player moved to the zone identified in the coaches’ playbook; tackles were scored as correct if the offensive ball carrier was stopped.

Following baseline, each player met with one of the researchers, who described the player’s mean baseline performance for a given skill. Players were asked to set a goal for their performances during practice sessions; no goals were set for games. The correct performances dur- ing baseline for all five players ranged from 60 to 80%, and all players set goals of 90% correct performance. The players were informed that their performance in each day’s practice would be posted on a chart prior to the next practice session. A Y (yes) or an N (no) was placed next to each player’s name to indicate whether he had met his goal. A player’s performance was posted on the chart only for the skill(s) in intervention. The chart was mounted on a wall in the locker room where all players on the team could see it. The head coach explained the purpose of the chart to other players on the team. Players’ performances during games were not posted on the chart.

The results for one of the players, John, are shown in Figure 3. John met or exceeded his goal of 90% correct performance during all practices for each of the three skills. Additionally, his improved performance general- ized to games. The same pattern of results was obtained for each of the other four players in the study, illustrating

224

Multiple Baseline and Changing Criterion Designs

100

John

Baseline Public Posting

Game

Practice

Reads

100

0 Drops

100

0 Tackles

5 10 15 20 25 30 35 40 Sessions

P er

ce n

ta ge

o f

C o

rr ec

t P

er fo

rm an

ce s

Figure 3 A multiple baseline across behaviors design showing percentage of correct reads, drops, and tackles by a college football player during practices and games. From “Effects of Posting Self-Set Goals on Collegiate Football Players’ Skill Execution During Practice and Games” by P. Ward and M. Carnes, 2002, Journal of Applied Behavior Analysis, 35, p. 5. Copyright 2002 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

that the multiple baseline across behaviors design is a single-subject experimental strategy in which each sub- ject serves as his own control. Each player constituted a complete experiment, replicated in this case with four other participants.

Multiple Baseline across Settings Design

In the multiple baseline across settings design, a single behavior of a person (or group) is targeted in two or more different settings or conditions (e.g., locations, times of day). After stable responding has been demonstrated under baseline conditions, the independent variable is in- troduced in one of the settings while baseline conditions

remain in effect in the other settings. When maximum behavior change or criterion-level performance has been achieved in the first setting, the independent variable is applied in the second setting, and so on.

Roane, Kelly, and Fisher (2003) employed a multi- ple baseline across settings design to evaluate the effects of a treatment designed to reduce the rate at which an 8- year-old boy put inedible objects in his mouth. Jason, who had been diagnosed with autism, cerebral palsy, and moderate mental retardation, had a history of putting ob- jects such as toys, cloth, paper, tree bark, plants, and dirt into his mouth.

Data on Jason’s mouthing were obtained concur- rently in a classroom, a playroom, and outdoors—three settings that contained a variety of inedible objects and

225

Multiple Baseline and Changing Criterion Designs

where caretakers had reported Jason’s mouthing to be problematic. Observers in each setting unobtrusively tal- lied the number of times Jason inserted an inedible object past the plane of his lips during 10-minute sessions. The researchers reported that Jason’s object mouthing usu- ally consisted of a series of discrete episodes, rather than an extended, continuous event, and that he often placed multiple objects (inedible objects and food) in his mouth simultaneously.

Roane and colleagues (2003) described the baseline and treatment conditions for Jason as follows:

The baseline condition was developed based on the functional analysis results, which showed that mouthing was maintained by automatic reinforcement and oc- curred independent of social consequences. During baseline, a therapist was present (approximately 1.5 to 3 m from Jason), but all occurrences of mouthing were ignored (i.e., no social consequences were arranged for mouthing, and Jason was allowed to place items in his mouth). No food items were available during baseline. The treatment condition was identical to baseline except that Jason had continuous access to foods that had been

2Functional analysis and automatic reinforcement are described in chapters entitled “Functional Behavior Assessment” and “Positive Reinforcement”, respectively.

previously identified to compete with the occurrence of object mouthing: chewing gum, marshmallows, and hard candy. Jason wore a fanny pack containing these items around his waist. (pp. 580–581)2

The staggered sequence in which the treatment was implemented in each setting and the results are shown in Figure 4. During baseline, Jason’s mouthed objects at mean rates of 0.9, 1.1, and 1.2 responses per minute in the classroom, a playroom, and outdoor settings, re- spectively. Introduction of the fanny pack with food in each setting produced an immediate drop to a zero or near zero rate of mouthing. During treatment, Jason put items of food from the fanny pack into his mouth at mean rates of 0.01, 0.01, and 0.07 responses per minute in the classroom, a playroom, and outdoor settings, respec- tively. The multiple baseline across settings design re- vealed a clear functional relation between the treatment and the frequency of Jason’s object mouthing. No mea- sures obtained during the treatment condition were as

1.5

0.5

Baseline Fanny Pack (Food)

Classroom

5 10 15 202

1.5

0.5

Playroom

5 10 15 20 2

1.5

0.5

Outdoor

5 10 15 20 Sessions

R es

p o

n se

s p

er M

in u

te (

M o

u th

in g)

Figure 4 A multiple baseline across settings design showing the number of object mouthing responses per minute during baseline and treatment conditions. From “The Effects of Noncontingent Access to Food on the Rate of Object Mouthing across Three Settings” by H. S. Roane, M. L. Kelly, and W. W. Fisher, 2003, Journal of Applied Behavior Analysis, 36, p. 581. Copyright 2003 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

226

Multiple Baseline and Changing Criterion Designs

high as the lowest measures in baseline. During 22 of 27 treatment sessions across the three settings, Jason put no inedible objects in his mouth.

As was done in the study by Roane and colleagues (2003), the data paths that comprise the different tiers in a multiple baseline across settings design are typically obtained in different physical environments (e.g., Cush- ing & Kennedy, 1997; Dalton, Martella, & Marchand- Martella, 1999). However, the different “settings” in a multiple baseline across settings design may exist in the same physical location and be differentiated from one another by different contingencies in effect, the presence or absence of certain people, and/or the different times of the day. For example, in a study by Parker and colleagues (1984), the presence or absence of other people in the training room constituted the different settings (environ- ments) in which the effects of the independent variable were evaluated. The attention, demand, and no-attention conditions (i.e., contingencies in effect) defined the dif- ferent settings in a multiple baseline design study by Kennedy, Meyer, Knowles, and Shukla (2000). The afternoon and the morning portions of the school day functioned as different settings in the multiple baseline across settings design used by Dunlap, Kern-Dunlap, Clarke, and Robbins (1991) to analyze the effects of cur- ricular revisions on a student’s disruptive and off-task behaviors.

In some studies using a multiple baseline across set- tings design, the participants are varied, changing, and perhaps even unknown to the researchers. For example, Van Houten and Malenfant (2004) used a multiple base- line design across two crosswalks on busy streets to eval- uate the effects of an intensive driver enforcement program on the percentage of drivers yielding to pedestrians and the number of motor vehicle–pedestrian conflicts. Watson (1996) used a multiple baseline design across men’s rest rooms on a college campus to assess the effectiveness of posting signs in reducing bathroom graffiti.

Multiple Baseline across Subjects Design

In the multiple baseline across subjects design, one tar- get behavior is selected for two or more subjects (or groups) in the same setting. After steady state responding has been achieved under baseline conditions, the inde- pendent variable is applied to one of the subjects while baseline conditions remain in effect for the other sub- jects. When criterion-level or stable responding has been attained for the first subject, the independent variable is applied to another subject, and so on. The multiple base- line across subjects design is the most widely used of all three forms of the design, in part because teachers, clin- icians, and other practitioners are commonly confronted

by more than one student or client needing to learn the same skill or eliminate the same problem behavior (e.g., Craft, Alber, & Heward, 1998; Kahng, Iwata, DeLeon, & Wallace, 2000; Killu, Sainato, Davis, Ospelt, & Paul, 1998; Kladopoulos & McComas, 2001). Sometimes a multiple baseline design is conducted across “groups” of participants (e.g., Dixon & Holcomb, 2000; Lewis, Pow- ers, Kelk, & Newcomer, 2002; White & Bailey, 1990).

Krantz and McClannahan (1993) used a multiple baseline across subjects design to investigate the effects of introducing and fading scripts to teach children with autism to interact with their peers. The four participants, ages 9 to 12, had severe communication deficits and min- imal or absent academic, social, leisure skills. Prior to the study each of the children had learned to follow first photographic activity schedules (Wacker & Berg, 1983) and later written activity schedules that prompted them through chains of academic, self-care, and leisure activ- ities. Although their teachers modeled social interactions, verbally prompted the children to interact, and provided contingent praise and preferred snacks and activities for doing so, the children consistently failed to initiate in- teractions without adult prompts.

Each session consisted of a continuous 10-minute interval in which observers recorded the number of times each child initiated and responded to peers while engaged in three art activities—drawing, coloring, and painting— that were rotated across sessions throughout the study. Krantz and McClannahan (1993) described the depen- dent variables as follows:

Initiation to peers was defined as understandable state- ments or questions that were unprompted by an adult, that were directed to another child by using his or her name or by facing him or her, and that were separated from the speaker’s previous vocalizations by a change in topic or a change in recipient of interaction. . . . Scripted interactions were those that matched the written script, . . . e.g., “Ross, I like your picture.” Unscripted interactions differed from the script by more than changes in con- junctions, articles, prepositions, pronouns, or changes in verb tense; the question, “Would you like some more paper?” was scored as an unscripted initiation because the noun “paper” did not occur in the script. A response was defined as any contextual utterance (word, phrase, or sentence) that was not prompted by the teacher and that occurred within 5 s of a statement or question di- rected to the target child. . . . Examples of responses were “what?” “okay,” and “yes, I do.” (p. 124)

During baseline, each child found art materials at his or her place and a sheet of paper with the written in- structions, “Do your art” and “Talk a lot.” The teacher prompted each child to read the written instructions, then moved away. During the script condition, the two written

227

Multiple Baseline and Changing Criterion Designs

instructions in baseline were supplemented by scripts con- sisting of 10 statements and questions such as, “{Name}, did you like to {swing/rollerskate/ride the bike} outside today?” “{Name}, do you want to use one of my pen- cils/crayons/brushes}?” (p. 124). Immediately before each session, the teacher completed blank portions of the scripts so that they reflected activities the children had completed or were planning and objects in the classroom environment. Each child’s script included the three other children’s names, and the order of the questions or state- ments varied across sessions and children.

The script condition was implemented with one child at a time, in staggered fashion (see Figure 5). Initially the teacher manually guided the child through the script, prompting him or her to read the statement to another child and to pencil a check mark next to it after doing so.

Krantz and McClannahan (1993) described the prompt- ing and script-fading procedures as follows:

Standing behind a participant, the teacher manually guided him or her to pick up a pencil, point to an in- struction or a scripted statement or question, and move the pencil along below the text. If necessary, the teacher also manually guided the child’s head to face another child to whom a statement or question was addressed. If the child did not verbalize the statement or questions within 5 s, the manual guidance procedure was repeated. If the child read or said a statement or read or asked a question, the teacher used the same type of manual guid- ance to ensure that the child placed a check mark to the left of that portion of the script.

Manual prompts were faded as quickly as possible; no prompts were delivered to Kate, Mike, Walt, and

Baseline Script

Kate Scripted

Unscripted

1 2 3 4 5

0 Mike

1 2 3 4 5

0 Walt

1 2 3 4 5

0 Ross

1 2 3 4 5

Sessions

0 5 10 15 20 25 30 35 40

2 Month Follow-Up

N u

m b

er o

f Sc

ri p

te d

a n

d U

n sc

ri p

te d

I n

it ia

ti o

n s

Figure 5 A multiple baseline across subjects design showing the number of scripted and unscripted initiations to peers and responses by four children with autism during baseline, script, and follow-up sessions. Arrows indicate when fading steps occurred. From “Teaching Children with Autism to Initiate to Peers: Effects of a Script-Fading Procedure” by P. J. Krantz and L. E. McClannahan, 1993, Journal of Applied Behavior Analysis, 26, p. 129. Copyright 1993 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

228

Multiple Baseline and Changing Criterion Designs

Ross after Sessions 15, 18, 23, and 27, respectively, and the teacher remained at the periphery of the classroom throughout subsequent sessions. After manual guidance had been faded for a target child, fading of the script began. Scripts were faded from end to beginning in five phases. For example, the fading steps for the question “Mike, what do you like to do best on Fun Friday?” were (a) “Mike, what do you like to do best,” (b) “Mike, what do you,” (c) “Mike, what,” (d) “M,” and (e) “.” (p. 125)

Kate and Mike, who never initiated during baseline, had mean initiations per session of 15 and 13, respectively, during the script condition. Walt’s initiations increased from a baseline mean of 0.1 to 17 during the script condi- tion, and Ross averaged 14 initiations per session during script compared to 2 during baseline. As the scripts were faded, each child’s frequency of unscripted initiations in- creased. After the scripts were faded, the four participants’ frequency of initiations were within the same range as that of a sample of three typically developing children. The re- searchers implemented the script-fading steps with each participant in response to his or her performance, not ac- cording to a predetermined schedule, thereby retaining the flexibility needed to pursue the behavior–environment re- lations that are the focus of the science of behavior.

However, because each subject did not serve as his or her own control, this study illustrates that the multiple baseline across subjects design is not a true single-subject design. Instead, verification of predictions based on the baseline data for each subject must be inferred from the relatively unchanging measures of the behavior of other subjects who are still in baseline, and replication of ef- fects must be inferred from changes in the behavior of other subjects when they come into contact with the in- dependent variable. This is both a weakness and a po- tential advantage of the multiple baseline across subjects design (Johnston & Pennypacker, 1993a), discussed later in the chapter.

Variations of the Multiple Baseline Design

Two variations of the multiple baseline design are the mul- tiple probe design and the delayed multiple baseline de- sign. The multiple probe design enables the behavior analyst to extend the operation and logic of the multiple baseline tactic to behaviors or situations in which con- current measurement of all behaviors comprising the de- sign is unnecessary, potentially reactive, impractical, or too costly. The delayed multiple baseline technique can be used when a planned reversal design is no longer possible or proves ineffective; it can also add additional tiers to an already operational multiple baseline design, as would be the case if new subjects were added to an ongoing study.

Multiple Probe Design

The multiple probe design, first described by Horner and Baer (1978), is a method of analyzing the relation between the independent variable and the acquisition of a succes- sive approximation or task sequence. In contrast to the multiple baseline design—in which data are collected si- multaneously throughout the baseline phase for each be- havior, setting, or subject in the experiment—in the multiple probe design intermittent measures, or probes, provide the basis for determining whether behavior change has occurred prior to intervention. According to Horner and Baer, when applied to a chain or sequence of related behaviors to be learned, the multiple probe design pro- vides answers to four questions: (a) What is the initial level of performance on each step (behavior) in the se- quence? (b) What happens when sequential opportunities to perform each step in the sequence are provided prior to training on that step? (c) What happens to each step as training is applied? and (d) What happens to the perfor- mance of untrained steps in the sequence as criterion-level performance is reached on the preceding steps?

Figure 6 shows a graphic prototype of the multiple probe design. Although researchers have developed many variations of the multiple probe technique, the basic de- sign has three key features: (a) An initial probe is taken to determine the subject’s level of performance on each behavior in the sequence; (b) a series of baseline mea- sures is obtained on each step prior to training on that step; and (c) after criterion-level performance is reached on any training step, a probe of each step in the sequence is obtained to determine whether performance changes have occurred in any other steps.

Thompson, Braam, and Fuqua (1982) used a multi- ple probe design to analyze the effects of an instructional procedure composed of prompts and token reinforcement on the acquisition of a complex chain of laundry skills by three students with developmental disabilities. Observa- tions of people doing laundry resulted in a detailed task analysis of 74 discrete responses that were organized into seven major components (e.g., sorting, loading washer). Each student’s performance was assessed via probe and baseline sessions that preceded training on each compo- nent. Probe and baseline sessions began with instructions to the student to do the laundry. When an incorrect re- sponse was emitted or when no response occurred within 5 seconds of a prompt to continue, the student was seated away from the laundry area. The trainer then performed the correct response and called the student back to the area so that assessment of the rest of the laundry sequence could continue.

Probe sessions differed from baseline sessions in two ways. First, a probe measured each response in the entire

229

Multiple Baseline and Changing Criterion Designs

100

B eh

av io

r 1

Baseline Instruction Follow-Up

100

B eh

av io

r 2

100

B eh

av io

r 3

100

B eh

av io

r 4

1 10 20 30 40 50 9174 Sessions

P er

ce n

t A

cc u

ra cy

Figure 6 Graphic prototype of a multiple probe design. Square data points represent results of probe sessions in which the entire sequence or set of behaviors (1–4) are tested.

chain and occurred immediately prior to baseline and training for every component. Baseline sessions occurred following the probe and measured only previously trained components plus the component about to be trained. Baseline data were gathered on a variable num- ber of consecutive sessions immediately prior to training sessions. Second, no tokens or descriptive praise were delivered during probes. During baseline, tokens were delivered for previously trained responses only. . . . Fol- lowing baseline, each component was trained using a graduated 3-prompt procedure (Horner & Keilitz, 1975), consisting of verbal instruction, modeling, and graduated guidance. If one prompt level failed to produce a correct response within 5 sec, the next level was introduced. . . . When the student performed a component at 100% accu- racy for two consecutive trials, he was required to per- form the entire laundry chain from the beginning through the component most recently mastered. The en- tire chain of previously mastered components was

trained (chain training condition) until it was performed without errors or prompts for two consecutive trials. (Thompson, Braam, & Fuqua, 1982, p. 179)

Figure 7 shows the results for Chester, one of the students. Chester performed a low percentage of correct responses during the probe and baseline sessions, but per- formed with 100% accuracy after training was applied to each component. During a generalization probe conducted at a community laundromat after training, Chester per- formed correctly 82% of the 74 total responses in the chain. Five additional training sessions were needed to retrain responses performed incorrectly during the gen- eralization probe and to train “additional responses ne- cessitated by the presence of coin slots and minor differences between the training and laundromat equip- ment” (p. 179). On two follow-up sessions conducted 10 months after training, Chester performed at 90% accu-

230

Multiple Baseline and Changing Criterion Designs

100

0 100

So rt

in g

Lo ad

W as

h er

Chester

100

0Se t

W as

h er

D ia

100

Em p

ty W

as h

100

Li n

t Sc

re en

100

0Se t

D ry

er D

ia ls

100

Em p

D ry

P er

ce n

t C

o rr

ec t

R es

p o

n se

B as

el in

e T

ra in

in g

G en

er al

iz e

C h

ai n

T ra

in in

C h

ai n

T ra

in in

C h

ai n

T ra

in in

C h

ai n

T ra

in in

C h

ai n

T ra

in in

C h

ai n

T ra

in in

C h

ai n

T ra

in in

Fo llo

w U

5 10 15 20 25 30 3540 45 50 55 60 65 70 75

Sessions

Li gh

D ar

B o

Figure 7 A multiple probe design showing the percentage of correct responses for each trial on each component of a laundry task by a young adult male with mental retardation. Heavy vertical lines on the horizontal axis represent successive training sessions; lighter and shorter vertical lines indicate trials within a session. From “Training and Generalization of Laundry Skills: A Multiple-Probe Evaluation with Handicapped Persons” by T. J. Thompson, S. J. Braam, and R. W. Fuqua, 1982, Journal of Applied Behavior Analysis, 15, p. 180. Copyright 1982 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

racy even though he had not performed the laundry task for the past 2 months. Similar results were obtained for the other two students who participated in the study.

Thompson and colleagues (1982) added the chain training condition to their study because they believed that components trained as independent skills were un- likely to be emitted in correct sequence without such practice. It should be noted that the experimenters did not begin training a new component until stable re- sponding had been achieved during baseline observations (see the baseline data for the bottom four tiers in Figure 7). Delaying the training in this manner enabled a clear demonstration of a functional relation between training and skill acquisition.

The multiple probe design is particularly appropri- ate for evaluating the effects of instruction on skill se- quences in which it is highly unlikely that the subject can improve performance on later steps in the sequence with-

out acquiring the prior steps. For example, the repeated measurement of the accuracy in solving division prob- lems of a student who possesses no skills in addition, sub- traction, and multiplication would add little to an analysis. Horner and Baer (1978) made this point exceedingly well:

The inevitable zero scores on the division baseline have no real meaning: division could be nothing else than zero (or chance, depending on the test format), and there is no real point in measuring it. Such measures are pro forma: they fill out the picture of a multiple baseline, true, but in an illusory way. They do not so much repre- sent zero behavior as zero opportunity for the behavior to occur, and there is no need to document at the level of well-measured data that behavior does not occur when it cannot. (p. 190)

Thus, the multiple probe design avoids the necessity of collecting ritualistic baseline data when the perfor- mance of any component of a chain or sequence is

231

Multiple Baseline and Changing Criterion Designs

impossible or unlikely before acquisition of its preceding components. In addition to the two uses already men- tioned—analysis of the effects of instruction on complex skill sequences and reduction in the amount of baseline measurement for behaviors that have no plausible op- portunity to occur—the multiple probe technique is also an effective experimental strategy for situations in which extended baseline measurement may prove reactive, im- practical, or costly. The repeated measurement of a skill under nontreatment conditions can prove aversive to some students; and extinction, boredom, or other undesirable responses can occur. In his discussion of multiple base- line designs, Cuvo (1979) suggested that researchers should recognize that “there is a trade-off between re- peatedly administering the dependent measure to estab- lish a stable baseline on one hand and risking impaired performance by subjecting participants to a potentially punishing experience on the other hand” (pp. 222–223). Furthermore, complete assessment of all skills in a se- quence may require too much time that could otherwise be spent on instruction.

Other examples of the multiple probe design can be found in Arntzen, Halstadtr, and Halstadtr (2003); Cole- man-Martin & Wolff Heller (2004); O’Reilly, Green, and Braunling-McMorrow, (1990); and Werts, Caldwell and

Delayed Multiple Baseline Design

The delayed multiple baseline design is an experimen- tal tactic in which an initial baseline and intervention are begun, and subsequent baselines are added in a staggered or delayed fashion (Heward, 1978). Figure 8 shows a graphic prototype of the delayed multiple baseline de- sign. The design employs the same experimental reason- ing as a full-scale multiple baseline design with the exception that data from baselines begun after the inde- pendent variable has been applied to previous behaviors, settings, or subjects cannot be used to verify predictions based on earlier tiers of the design. In Figure 8 baseline measurement of Behaviors 2 and 3 was begun early

Baseline Treatment

B eh

av io

r 1

B eh

av io

r 2

B eh

av io

r 3

B eh

av io

r 4

R es

p o

n se

M ea

su re

5 10 15 20 25 Sessions

Figure 8 Graphic prototype of a delayed multiple baseline design.

Wolery.

232

Multiple Baseline and Changing Criterion Designs

enough for those data to be used to verify the prediction made for Behavior 1. The final four baseline data points for Behavior 3 also verify the prediction for Behavior 2. However, baseline measurement of Behavior 4 began after the independent variable had been applied to each of the previous behaviors, thus limiting its role in the de- sign to an additional demonstration of replication.

A delayed multiple baseline design may allow the behavior analyst to conduct research in certain environ- ments in which other experimental tactics cannot be implemented. Heward (1978) suggested three such situations.

• A reversal design is no longer desirable or possible. In applied settings the research environment may shift, negating the use of a previously planned reversal de- sign. Such shifts may involve changes in the subject’s environment that make the target behavior no longer likely to reverse to baseline levels, or changes in the behavior of parents, teachers, administrators, the sub- ject/client, or the behavior analyst that, for any number of reasons, make a previously planned reversal design no longer desirable or possible. . . . If there are other behaviors, settings, or subjects appropriate for applica- tion of the independent variable, the behavior analyst could use a delayed multiple baseline technique and still pursue evidence of a functional relation.

• Limited resources, ethical concerns, or practical diffi- culties preclude a full-scale multiple baseline design. This situation occurs when the behavior analyst only controls resources sufficient to initially record and in- tervene with one behavior, setting, or subject, and an- other research strategy is inappropriate. It may be that as a result of the first intervention, more resources be- come available for gathering additional baselines. This might occur following the improvement of certain be- haviors whose pretreatment topography and/or rate re- quired an inordinate expenditure of staff resources. Or, it could be that a reluctant administrator, after seeing the successful results of the first intervention, provides the resources necessary for additional analysis. Ethical concerns may preclude extended baseline measure- ment of some behaviors (e.g., Linscheid, Iwata, Rick- etts, Williams, & Griffin, 1990). Also under this heading would fall the “practical difficulties” cited by Hobbs and Holt (1976) as a reason for delaying base- line measurement in one of three settings.

• A “new” behavior, setting, or subject becomes avail- able. A delayed multiple baseline technique might be employed when another research design was originally planned but a multiple baseline analysis becomes the preferred approach due to changes in the environment (e.g., the subject begins to emit another behavior ap- propriate for intervention with the experimental vari- able, the subject begins to emit the original target behavior in another setting, or additional subjects dis- playing the same target behavior become available.) (adapted from pp. 5–6)

Researchers have used the delayed multiple baseline technique to evaluate the effects of a wide variety of in- terventions (e.g., Baer, Williams, Osnes, & Stokes, 1984; Copeland, Brown, & Hall, 1974; Hobbs & Holt, 1976; Jones, Fremouw, & Carples, 1977; Linscheid et al., 1990; Risley & Hart, 1968; Schepis, Reid, Behrmann, & Sutton, 1998; White & Bailey, 1990. Poche, Brouwer, and Swearingen (1981) used a delayed multiple baseline de- sign to evaluate the effects of a training program designed to prevent children from being abducted by adults. Three typically developing preschool children were selected as subjects because, during a screening test, each readily agreed to leave with an adult stranger. The dependent variable was the level of appropriateness of self-protec- tive responses emitted by each child when an adult sus- pect approached the child and attempted to lure her away with a simple lure (“Would you like to go for a walk?”), an authoritative lure (“Your teacher said it was alright for you to come with me.”), or an incentive lure (“I’ve got a nice surprise in my car. Would you like to come with me and see it?”).

Each session began with the child’s teacher bringing the child outdoors, then pretending to have to return to the building for some reason. The adult suspect (a confeder- ate of the experimenters but unknown to the child) then approached the child and offered one of the lures. The confederate also served as observer, scoring the child’s re- sponse on a 0 to 6 scale, with a score of 6 representing the desired response (saying, “No, I have to go ask my teacher” and moving at least 20 feet away from the sus- pect within 3 seconds) and a score of 0 indicating that the child moved some distance away from the school building with the suspect. Training consisted of model- ing, behavioral rehearsal, and social reinforcement for correct responses.

Figure 9 shows the results of the training program. During baseline, all three children responded to the lures with safety ratings of 0 or 1. All three children mastered correct responses to the incentive lure in one to three training sessions, with one or two more sessions required for each child to master correct responses to the other two lures. Overall, training took approximately 90 min- utes per child distributed over five or six sessions. All three children responded correctly when the lures were administered in generalization probes on sidewalk loca- tions 150 to 400 feet from the school.

Although each baseline in this study was of equal length (i.e., had an equal number of data points), contra- dicting the general rule that the baselines in a multiple base- line design should vary significantly in length, there are two good reasons that Poche and colleagues began train- ing when they did with each subject. First, the nearly total stability of the baseline performance of each child pro- vided an ample basis for evaluating the training program

233

Multiple Baseline and Changing Criterion Designs

Baseline Training Generality Follow-Up

Patty

School Grounds

Community

Stan

John

5 10 15 20 2 5

Le ve

l o

f A

p p

ro p

ri at

en es

s o

f Se

lf -P

ro te

ct iv

e R

es p

o n

se s

Days Type of Lure

Simple

Authority

Incentive

Figure 9 A delayed multiple baseline design showing the level of appropriateness of self-protective responses during baseline, training, and generality probes in school and community settings. Closed symbols indicate data gathered near the school; open symbols, in a location away from the school. From “Teaching Self-Protection to Young Children” by C. Poche, R. Brouwer, and M. Swearingen, 1981, Journal of Applied Be- havior Analysis, 14, p. 174. Copyright 1981 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

(the only exception to complete susceptibility to the adult suspect’s lures occurred when Stan stayed near the sus- pect instead of actually going away with him on his fourth baseline observation). Second, and more important, the nature of the target behavior required that it be taught to each child as soon as possible. Although continuing base- line measurement for varying lengths across the different tiers of any multiple baseline design is good practice from a purely experimental viewpoint, the ethics of such a practice in this instance would be highly questionable, given the potential danger of exposing the children to adult lures repeatedly while withholding training.

The delayed multiple baseline design presents several limitations (Heward, 1978). First, from an applied stand- point the design is not a good one if it requires the be- havior analyst to wait too long to modify important behaviors, although this problem is inherent in all multi- ple baseline designs. Second, in a delayed multiple base- line design there is a tendency for the delayed baseline phases to contain fewer data points than are found in a standard multiple baseline design, in which all baselines are begun simultaneously, resulting in baseline phases of considerable and varying length. Long baselines, if sta- ble, provide the predictive power that permits convincing demonstrations of experimental control. Behavior ana- lysts using any type of multiple baseline design must be

sure that all baselines, regardless of when they are begun, are of sufficient and varied length to provide a believable basis for comparing experimental effects. A third limita- tion of the delayed multiple baseline design is that it can mask the interdependence of dependent variables.

The strength of any multiple baseline design is that little or no change is noticed in the other, as yet untreated, be- haviors until, and only until, the experimenter applies the independent variable. In a delayed multiple baseline design, the “delayed baseline” data gathered for subse- quent behaviors may represent changed performance due to the experimental manipulation of other behaviors in the design and, therefore, may not be representative of the true, preexperimental operant level. . . . In such instances, the delayed multiple baseline might result in a “false negative,” and the researcher may erroneously conclude that the intervention was not effective on the subsequent behavior(s), when in reality the lack of si- multaneous baseline data did not permit the discovery that the behaviors covaried. This is a major weakness of the delayed multiple baseline design and makes it a re- search tactic of second choice whenever a full-scale multiple baseline can be employed. However, this limi- tation can and should be combated whenever possible by beginning subsequent baselines at least several sessions prior to intervention on previous baselines. (Heward, 1978, pp. 8–9)

234

Multiple Baseline and Changing Criterion Designs

Both the multiple probe design and the delayed mul- tiple baseline design offer the applied behavior analyst alternative tactics for pursuing a multiple baseline analy- sis when extended baseline measurement is unnecessary, impractical, too costly, or unavailable. Perhaps the most useful application of the delayed multiple baseline tech- nique is in adding tiers to an already operational multi- ple baseline design. Whenever a delayed baseline can be supplemented by probes taken earlier in the course of the study, experimental control is strengthened. As a general rule, the more baseline data, the better.

Assumptions and Guidelines for Using Multiple Baseline Designs

Like all experimental tactics, the multiple baseline de- sign requires the researcher to make certain assumptions about how the behavior–environment relations under in- vestigation function, even though discovering the exis- tence and operation of those relations is the very reason for conducting the research. In this sense, the design of behavioral experiments resembles an empirical guessing game—the experimenter guesses; the data answer. The investigator makes assumptions, hypotheses in the in- formal sense, about behavior and its relation to control- ling variables and then constructs experiments designed to produce data capable of verifying or refuting those conjectures.3

Because verification and replication in the multiple baseline design depends on what happens, or does not happen, to other behaviors as a result of the sequential application of the independent variable, the experimenter must be particularly careful to plan and carry out the de- sign in a manner that will afford the greatest degree of confidence in any relations suggested by the data. Al- though the multiple baseline design appears deceptively simple, its successful application entails much more than selecting two or more behaviors, settings, or subjects, collecting some baseline data, and then introducing a treatment condition to one behavior after the other. We

3Hypothesis, as we are using the term here, should not be confused with the formal hypothesis testing models that use inferential statistics to confirm or reject a hypothesis deduced from a theory. As Johnston and Pennypacker (1993a) pointed out, “Researchers do not need to state hypotheses if they are asking a question about nature. When the experimental question sim- ply asks about the relation between independent and dependent variables, there is no scientific reason to make a prediction about what will be learned from the data” (p. 48). However, Johnston and Pennypacker (1980) also rec- ognized that “more modest hypotheses are constantly being subjected to ex- perimental tests, if only to establish greater confidence in the details of the suspected controlling relations. Whenever an experimenter arranges to af- firm the consequent of a particular proposition, he or she is testing a hy- pothesis, although it is rare to encounter the actual use of such language [in behavior analysis]. Hypothesis testing in this relatively informal sense guides the construction of experiments without blinding the researcher to the importance of unexpected results” (pp. 38–39).

suggest the following guidelines for designing and con- ducting experiments using multiple baseline designs.

Select Independent, yet Functionally Similar, Baselines

Demonstration of a functional relation in a multiple base- line design depends on two occurrences: (a) the behav- ior(s) still in baseline showing no change in level, variability, or trend while the behavior(s) in contact with the independent variable changes; and (b) each behavior changes when, and only when, the independent variable has been applied to it. Thus, the experimenter must make two, at times seemingly contradictory, assumptions about the behaviors targeted for analysis in a multiple baseline design. The assumptions are that the behaviors are func- tionally independent of one another (the behaviors will not covary with one another), and yet the behaviors share enough similarity that each will change when the same in- dependent variable is applied to it (Tawney & Gast, 1984). An error in either assumption can result in a fail- ure to demonstrate a functional relation.

For example, let us suppose that the independent variable is introduced with the first behavior, and changes in level and/or trend are noted, but the other behaviors still in baseline also change. Do the changes in the still- in-baseline behaviors mean that an uncontrolled variable is responsible for the changes in all of the behaviors and that the independent variable is an effective treatment? Or do the simultaneous changes in the untreated behav- iors mean that the changes in the first behavior were af- fected by the independent variable and have generalized to the other behaviors? Or, let us suppose instead that the first behavior changes when the independent variable is introduced, but subsequent behaviors do not change when the independent variable is applied. Does this failure to replicate mean that a factor other than the independent variable was responsible for the change observed in the first behavior? Or does it mean only that the subsequent behaviors do not operate as a function of the experimen- tal variable, leaving open the possibility that the change noted in the first behavior was affected by the indepen- dent variable?

Answers to these questions can be pursued only by further experimental manipulations. In both kinds of fail- ure to demonstrate experimental control, the multiple base- line design does not rule out the possibility of a functional relation between the independent variable and the behav- ior(s) that did change when the variable was applied. In the first instance, the failure to demonstrate experimental control with the originally planned design is offset by the opportunity to investigate and possibly isolate the variable robust enough to change multiple behaviors simul- taneously. Discovery of variables that reliably produce

235

Multiple Baseline and Changing Criterion Designs

generalized changes across behaviors, settings, and/or sub- jects is a major goal of applied behavior analysis; and if the experimenter is confident that all other relevant variables were held constant before, during, and after the observed behavior changes, the original independent variable is the first candidate for further investigation.

In the second situation, with its failure to replicate changes from one behavior to another, the experimenter can pursue the possibility of a functional relation between the independent variable and the first behavior, perhaps using a reversal technique, and seek to discover later an ef- fective intervention for the behavior(s) that did not change. Another possibility is to drop the original independent variable altogether and search for another treatment that might be effective with all of the targeted behaviors.

Select Concurrent and Plausibly Related Multiple Baselines

In an effort to ensure the functional independence of be- haviors in a multiple baseline design, experimenters should not select response classes or settings so unrelated to one another as to offer no plausible means of compar- ison. For the ongoing baseline measurement of one be- havior to provide the strongest basis for verifying the prediction of another behavior that has been exposed to an independent variable, two conditions must be met: (a) The two behaviors must be measured concurrently, and (b) all of the relevant variables that influence one behav- ior must have an opportunity to influence the other be- havior. Studies that employ a multiple baseline approach across subjects and settings often stretch the logic of the design beyond its capabilities. For example, using the stable baseline measures of one child’s compliance with parental requests as the basis for verifying the effect of in- tervention on the compliance behavior of another child living with another family is questionable practice. The sets of variables influencing the two children are surely differentiated by more than the presence or absence of the experimental variable.

There are some important limits to designating multiple behavior/setting combinations that are intended to func- tion as part of the same experiment. In order for the use of multiple behaviors and settings to be part of the same design and thus augment experimental reasoning, the general experimental conditions under which the two re- sponses (whether two from one subject or one from each of two subjects) are emitted and measured must be on- going concurrently. . . . Exposure [to the independent variable] does not have to be simultaneous for the differ- ent behavior/setting combinations, [but] it must be the identical treatment conditions along with the associated extraneous variables that impinge on the two responses and/or settings. This is because the conditions imposed

4A related series of A-B designs across different behaviors, settings, and/or participants in which each A-B sequence is conducted at a different point in time is sometimes called a nonconcurrent multiple baseline design (Wat- son & Workman, 1981). The absence of concurrent measurement, how- ever, violates and effectively neuters the experimental logic of the multiple baseline design. Putting the graphs of three A-B designs on the same page and tying them together with a dogleg dashed line might produce something that “looks like” a multiple baseline design, but doing so is of questionable value and is likely to mislead readers by suggesting a greater degree of ex- perimental control than is warranted. We recommend describing such a study as a series or collection of A-B designs and graphing the results in a manner that clearly depicts the actual time frame in which each A-B se- quence occurred with respect to the others (e.g., Harvey, May, & Kennedy, 2004, Figure 2).

on one behavior/setting combination must have the op- portunity of influencing the other behavior/setting com- bination at the same time, regardless of the condition that actually prevails for the second. . . . It follows that using responses of two subjects each responding in different settings would not meet the requirement that there be a coincident opportunity for detecting the treatment effect. A treatment condition [as well as the myriad other vari- ables possibly responsible for changes in the behavior of one subject] could not then come into contact with the responding of the other subject, because the second sub- ject’s responding would be occurring in an entirely dif- ferent location. . . . Generally, the greater the plausibility that the two responses would be affected by the single treatment [and all other relevant variables], the more powerful is the demonstration of experimental control evidenced by data showing a change in only one behav- ior. (Johnston and Pennypacker, 1980, pp. 276–278)

The requirements of concurrency and plausible in- fluence must be met for the verification element of base- line logic to operate in a multiple baseline design. However, replication of effect is demonstrated each time a baseline steady state is changed by the introduction of the independent variable, more or less regardless of where or when the variable is applied. Such nonconcurrent and/or unrelated baselines can provide valuable data on the generality of a treatment’s effectiveness.4

This discussion should not be interpreted to mean that a valid (i.e., logically complete) multiple baseline design cannot be conducted across different subjects, each responding in different settings. Numerous studies using mixed multiple baselines across subjects, responses classes, and/or settings have contributed to the develop- ment of an effective technology of behavior change (e.g., Dixon et al., 1998; Durand, 1999; Ryan, Ormond, Im- wold, & Rotunda, 2002).

Let us consider an experiment designed to analyze the effects of a particular teacher training intervention, perhaps a workshop on using tactics to increase each stu- dent’s opportunity to respond during group instruction. Concurrent measurement is begun on the frequency of student response opportunities in the classrooms of the teachers who are participating in the study. After stable

236

Multiple Baseline and Changing Criterion Designs

baselines have been established, the workshop is pre- sented first to one teacher (or group of teachers) and eventually, in staggered multiple baseline fashion, to all of the teachers.

In this example, even though the different subjects (teachers) are all behaving in different environments (dif- ferent classrooms), comparison of their baseline condi- tions is experimentally sound because the variables likely to influence their teaching styles operate in the larger, shared environment in which they all behave (the school and teaching community). Nevertheless, whenever ex- periments are proposed or published that involve differ- ent subjects responding in different settings, researchers and consumers should view the baseline comparisons with a critical eye toward their logical relation to one other.

Do Not Apply the Independent Variable to the Next Behavior Too Soon

To reiterate, for verification to occur in a multiple base- line design, it must be established clearly that as the in- dependent variable is applied to one behavior and change is noted, little or no change is observed in the other, as- yet-untreated behaviors. The potential for a powerful demonstration of experimental control has been destroyed in many studies because the independent variable was applied to subsequent behaviors too soon. Although the operational requirement of sequential application in the multiple baseline tactic is met by introduction of the in- dependent variable even in adjacent time intervals, the experimental reasoning afforded by such closely spaced manipulations is minimal.

The influence of unknown, concomitant, extraneous variables that might be present could still be substantial, even a day or two later. This problem can be avoided by demonstrating continued stability in responding for the second behavior/setting combination during and after the introduction of the treatment for the first combina- tion until a sufficient period of time has elapsed to detect any effect on the second combination that might appear. (Johnston & Pennypacker, 1980, p. 283)

Vary Significantly the Lengths of Multiple Baselines

Generally, the more the baseline phases in a multiple base- line design differ in length from one another, the stronger the design will be. Baselines of significantly different lengths allow the unambiguous conclusion (assuming an effective treatment variable) that each behavior not only changes when the independent variable is applied, but also that each behavior does not change until the inde- pendent variable has been applied. If the different base- lines are of the same or similar length, the possibility

exists that changes noted when the independent variable is introduced are the result of a confounding variable, such as practice or reactivity to observation and mea- surement, and not a function of the experimental variable.

Those effects . . . called practice, adaptation, warm-up, self-analysis, etc.; whatever they may be and whatever they may be called, the multiple baseline design controls for them by systematically varying the length of time (sessions, days, weeks) in which they occur prior to the introduction of the training package. . . . Such control is essential, and when the design consists of only two baselines, then the number of data points in each prior to experimental intervention should differ as radically as possible, at least by a factor of 2. I cannot see not systematically varying lengths of baselines prior to in- tervention, and varying them as much as possible/practi- cal. Failure to do that . . . weakens the design too much for credibility. (D. M. Baer, personal communication, June 2, 1978)

Intervene on the Most Stable Baseline First

In the ideal multiple baseline design, the independent vari- able is not applied to any of the behaviors until steady state responding has been achieved for each. However, the applied behavior analyst is sometimes denied the op- tion of delaying treatment just to increase the strength of an experimental analysis. When intervention must begin before stability is evident across each tier of the design, the independent variable should be applied to the behavior, setting, or subject that shows the most stable level of base- line responding. For example, if a study is designed to evaluate the effects of a teaching procedure on the rate of math computation of four students and there is no a pri- ori reason to teach the students in any particular sequence, instruction should begin with the student showing the most stable baseline. However, this recommendation should be followed only when the majority of the baselines in the de- sign show reasonable stability.

Sequential application of the independent variable should be made in the order of greatest stability at the time of each subsequent application. Again, however, the realities of the applied world must be heeded. The social significance of changing a particular behavior must some- times take precedence over the desire to meet the re- quirements of experimental design.

Considering the Appropriateness of Multiple Baseline Designs

The multiple baseline design offers significant advan- tages, which no doubt have accounted for its widespread use by researchers and practitioners. Those advantages,

237

Multiple Baseline and Changing Criterion Designs

however, must be weighed against the limitations and weaknesses of the design to determine its appropriate- ness in any given situation.

Advantages of the Multiple Baseline Design

Probably the most important advantage of the multiple baseline design is that it does not require withdrawing a seemingly effective treatment to demonstrate experi- mental control. This is a critical consideration for target behaviors that are self-injurious or dangerous to others. This feature of the multiple baseline design also makes it an appropriate method for evaluating the effects of inde- pendent variables that cannot, by their nature, be with- drawn and for investigating target behaviors that are likely or that prove to be irreversible (e.g., Duker & van Lent, 1991). Additionally, because the multiple baseline design does not necessitate a reversal of treatment gains to base- line levels, parents, teachers, or administrators may accept it more readily as a method of demonstrating the effects of an intervention.

The requirement of the multiple baseline design to sequentially apply the independent variable across mul- tiple behaviors, settings, or subjects complements the usual practice of many practitioners whose goal is to de- velop multiple behavior changes. Teachers are charged with helping multiple students learn multiple skills to be used in multiple settings. Likewise, clinicians typically need to help their clients improve more than one response class and emit more adaptive behavior in several settings. The multiple baseline design is ideally suited to the eval- uation of the progressive, multiple behavior changes sought by many practitioners in applied settings.

Because the multiple baseline design entails con- current measurement of two or more behaviors, settings, or subjects, it is useful in assessing the occurrence of gen- eralization of behavior change. The simultaneous moni- toring of several behaviors gives the behavior analyst the opportunity to determine their covariation as a result of manipulations of the independent variable (Hersen & Barlow, 1976). Although changes in behaviors still under baseline conditions eliminate the ability of the multiple baseline design to demonstrate experimental control, such changes reveal the possibility that the independent vari- able is capable of producing behavioral improvements with desirable generality, thereby suggesting an addi- tional set of research questions and analytic tactics (e.g., Odom, Hoyson, Jamieson, & Strain, 1985).

Finally, the multiple baseline design has the advan- tage of being relatively easy to conceptualize, thereby of- fering an effective experimental tactic for teachers and parents who are not trained formally in research method- ology (Hall et al., 1970).

Limitations of the Multiple Baseline Design

The multiple baseline design presents at least three sci- entific limitations or considerations. First, a multiple base- line design may not allow a demonstration of experimental control even though a functional relation exists between the independent variable and the behaviors to which it is applied. Changes in behaviors still under baseline condi- tions and similar to concurrent changes in a behavior in the treatment condition preclude the demonstration of a func- tional relation within the original design. Second, from one perspective, the multiple baseline design is a weaker method for showing experimental control than the rever- sal design. This is because verification of the baseline pre- diction made for each behavior within a multiple baseline design is not directly demonstrated with that behavior, but must be inferred from the lack of change in other behav- iors. This weakness of the multiple baseline design, how- ever, should be weighed against the design’s advantage of providing multiple replications across different behav- iors, settings, or subjects. Third, the multiple baseline de- sign provides more information about the effectiveness of the treatment variable than it does about the function of any particular target behavior.

Consistently [the] multiple baseline is less an experi- mental analysis of the response than of the technique used to alter the response. In the reversal design, the re- sponse is made to work again and again; in the multiple- baseline designs, it is primarily the technique that works again and again, and the responses either work once each [if different responses are used] or else a single re- sponse works once each per setting or once each per subject. Repetitive working of the same response in the same subject or the same setting is not displayed. But, while repetitive working of the response is foregone, repetitive and diverse working of the experimental tech- nique is maximized, as it would not be in the reversal design. (Baer, 1975, p. 22)

Two important applied considerations that must be evaluated in determining the appropriateness of the mul- tiple baseline design are the time and resources required for its implementation. Because the treatment variable cannot be applied to subsequent behaviors, settings, or subjects until its effects have been observed on previous behaviors, settings, or subjects, the multiple baseline de- sign requires that intervention be withheld for some be- haviors, settings, or subjects, perhaps for a long time. This delay raises practical and ethical concerns. Treatment can- not be delayed for some behaviors; their importance makes delaying treatment impractical. And as Stolz (1978) pointed out, “If the intervention is generally acknowl- edged to be effective, denying it simply to achieve a multiple-baseline design might be unethical” (p. 33). Sec- ond, the resources needed for the concurrent measure-

238

Multiple Baseline and Changing Criterion Designs

ment of multiple behaviors must be considered. Use of a multiple baseline design can be particularly costly when behavior must be observed and measured in several set- tings. However, when the use of intermittent probes dur- ing baseline can be justified in lieu of continuous measurement (Horner & Baer, 1978), the cost of concur- rently measuring multiple behaviors can be reduced.

Changing Criterion Design The changing criterion design can be used to evaluate the effects of a treatment that is applied in a graduated or stepwise fashion to a single target behavior. The chang- ing criterion design was first described in the applied be- havior analysis literature in two papers coauthored by Vance Hall (Hall & Fox, 1977; Hartmann & Hall, 1976).

Operation and Logic of the Changing Criterion Design

The reader can refer to Figure 10 before and after read- ing Hartmann and Hall’s (1976) description of the changing criterion design.

The design requires initial baseline observations on a single target behavior. This baseline phase is followed by implementation of a treatment program in each of a series of treatment phases. Each treatment phase is asso- ciated with a step-wise change in criterion rate for the target behavior. Thus, each phase of the design provides a baseline for the following phase. When the rate of the target behavior changes with each stepwise change in the criterion, therapeutic change is replicated and exper- imental control is demonstrated. (p. 527)

The operation of two elements of baseline logic— prediction and replication—is clear in the changing cri-

terion design. When stable responding is attained within each phase of the design, a prediction of future respond- ing is made. Replication occurs each time the level of be- havior changes in a systematic way when the criterion is changed. Verification of the predictions based on each phase is not so obvious in this design but can be ap- proached in two ways. First, varying the lengths of phases systematically enables a form of self-evident verification. The prediction is made that the level of responding will not change if the criterion is not changed. When the cri- terion is not changed and stable responding continues, the prediction is verified. When it can be shown within the design that levels of responding do not change unless the criterion is changed, regardless of the varied lengths of phases, experimental control is evident. Hall and Fox (1977) suggested another possibility for verification: “The experimenter may return to a former criterion and if the behavior conforms to this criterion level there is also a co- gent argument for a high degree of behavioral control” (p. 154). Such a reversed criterion is shown in the next- to-last phase of Figure 10. Although returning to an earlier criterion level requires a brief interruption of the steady improvement in behavior, the reversal tactic strengthens the analysis considerably and should be in- cluded in the changing criterion design unless other fac- tors indicate its inappropriateness.

One way to conceptualize the changing criterion de- sign is as a variation of the multiple baseline design. Both Hartmann and Hall (1976, p. 530) and Hall and Fox (1977, p. 164) replotted data from changing criterion de- sign experiments in a multiple baseline format with each tier of the multiple baseline showing the occurrence or nonoccurrence of the target behavior at one of the criterion levels used in the experiment. A vertical condi- tion change line doglegs through the tiers indicating when the criterion for reinforcement was raised to the level

N u

m b

er o

f R

es p

o n

se s

Baseline Treatment

5 10 15 20 25 30 35 Sessions

Figure 10 Graphic prototype of a changing criterion design.

239

Multiple Baseline and Changing Criterion Designs

represented by each tier. By graphing whether the target behavior was emitted during each session at or above the level represented on each tier both before and after the change in criterion to that level, a kind of multiple base- line analysis is revealed. However, the strength of the multiple baseline argument is not quite so convincing be- cause the “different” behaviors represented by each tier are not independent of one another. For example, if a tar- get behavior is emitted 10 times in a given session, all of the tiers representing criteria below 10 responses would have to show that the behavior occurred, and all of the tiers representing criteria of 11 or more would have to show no occurrence of the behavior, or zero responding. The majority of the tiers that would appear to show ver- ification and replication of effect, in fact, could only show these results because of the events plotted on another tier. A multiple baseline design provides its convincing demonstration of experimental control because the mea- sures obtained for each behavior in the design are a func- tion of the controlling variables for that behavior, not artifacts of the measurement of another behavior. Thus, recasting the data from a changing criterion design into a many-tiered multiple baseline format will often result in a biased picture in favor of experimental control.

Even though the multiple baseline design is not com- pletely analogous, the changing criterion design can be conceptualized as a method of analyzing the develop- ment of new behaviors. As Sidman (1960) pointed out, “It is possible to make reinforcement contingent upon a spec- ified value of some aspect of behavior, and to treat that value as a response class in its own right” (p. 391). The changing criterion design can be an effective tactic for showing the repeated production of new rates of behav- ior as a function of manipulations of the independent vari- able (i.e., criterion changes).

Other than the experiments included in the Hartmann and Hall (1976) and Hall and Fox (1977) papers, there have been relatively few examples of pure changing cri- terion designs published in the applied behavior analysis literature (e.g., DeLuca & Holborn, 1992; Foxx & Rubi- noff, 1979; Johnston & McLaughlin, 1982). Some re- searchers have employed a changing criterion tactic as an analytic element within a larger design (e.g., Martella, Leonard, Marchand-Martella, & Agran, 1993; Schleien, Wehman, & Kiernan, 1981).

Allen and Evans (2001) used a changing criterion design to evaluate the effects of an intervention to reduce the excessive checking of blood sugar levels by Amy, a 15-year-old girl diagnosed with insulin-dependent dia- betes about 2 years prior to the study. Persons with this form of diabetes must guard against hypoglycemia (i.e., low blood sugar), a condition that produces a cluster of symptoms such as headaches, dizziness, shaking, im- paired vision, and increased heart rate, and can lead to

seizures and loss of consciousness. Because hypo- glycemic episodes are physically unpleasant and can be a source of social embarrassment, some patients become hypervigilent in avoiding them, checking for low blood sugar more often than is necessary and deliberately main- taining high blood glucose levels. This leads to poor metabolic control and increased risk of complications such as blindness, renal failure, and heart disease.

At home Amy’s parents helped her monitor her blood sugar levels and insulin injections; at school Amy checked her blood glucose levels independently. Her physician rec- ommended that Amy keep her blood sugar levels between 75 and 150 mg/dl, which required her to check her blood sugar 6 to 12 times per day. Soon after she had been di- agnosed with diabetes, Amy experienced a single hypo- glycemic episode in which her blood sugar fell to 40 mg/dl, and she experienced physical symptoms but no loss of consciousness. After that episode Amy began checking her glucose levels more and more often, until at the time of her referral she was conducting 80 to 90 checks per day, which cost her parent approximately $600 per week in reagent test strips. Amy was also maintaining her blood sugar level between 275 to 300 mg/dl, far above the recommended levels for good metabolic control.

Following a 5-day baseline condition, a treatment was begun in which Amy and her parents were exposed to a gradually decreasing amount of information about her blood glucose level. Over a 9-month period Amy’s parents gradually reduced the number of test strips she was given each day, beginning with 60 strips during the first phase of the treatment. Allen and Evans (2001) ex- plained the treatment condition and method for chang- ing criteria as follows:

The parents expressed fears, however, that regardless of the criterion level, Amy might encounter a situation in which additional checking would be necessary. Con- cerns about adherence to the exposure protocol by the parents resulted in a graduated protocol in which Amy could earn a small number of additional test strips above and beyond the limit set by the parents. One additional test strip could be earned for each half hour of engage- ment in household chores. Amy was allowed to earn a maximum of five additional tests above the criterion when the criterion was set at 20 test strips or higher. Amy was allowed two additional test strips when the criterion was set below 20. Access to test strips was re- duced in graduated increments, with the parents setting criteria to levels at which they were willing to adhere. Criteria changes were contingent upon Amy success- fully reducing total test strip use to below the criterion on 3 successive days. (p. 498)

Figure 11 shows the criterion changes and the num- ber of times Amy monitored her blood glucose level dur- ing the last 10 days of each criterion level. The results

240

Multiple Baseline and Changing Criterion Designs

N u

m b

er o

f C

h ec

k s

20 18

12 14

Baseline

5/16 5/29 7/22 8/29 10/16 11/15 12/19 3/3

Days

Exposure-Based Treatment

Figure 11 A changing criterion design showing the number of blood glucose monitoring checks conducted during the last 10 days of each criterion level. Dashed lines and corresponding numbers indicate the maximum number of test strips allotted at each level. Checks above the criterion levels were conducted with additional test strips earned by Amy. From “Exposure-Based Treatment to Control Excessive Blood Glucose Monitoring” by K. D. Allen and J. H. Evans, 2001, Journal of Applied Behavior Analysis, 12, p. 499. Copyright 2001 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

clearly show that Amy responded well to the treatment and rarely exceeded the criterion. Over the course of the 9-month treatment program, Amy reduced the number of times she monitored her blood sugar from 80 to 95 times per day during baseline to fewer than 12 tests per day, a level that she maintained at a 3-month follow-up. Amy’s parents indicated that they did not plan to decrease the cri- terion any further. A concern was that Amy might main- tain high blood sugar levels during treatment. The authors reported that her blood sugar levels increased initially during treatment, but gradually decreased over the treat- ment program to a range of 125 to 175 mg/dl, within or near the recommended level.

Although the figure shows data only for the final 10 days of each criterion level, it is likely that the phases varied in length.5 The study consisted of seven criterion changes of two magnitudes, 20 and 2. Although greater variation in the magnitude of criterion changes and a re- turn to a previously attained higher criterion level may

5Data on the number of checks by Amy throughout the intervention are available from Allen and Evans (2001).

have provided a more convincing demonstration of ex- perimental control, the practical and ethical considera- tions of doing so would be questionable. As always, the applied behavior analyst must balance experimental con- cerns with the need to improve behavior in the most ef- fective, efficient, ethical manner.

This study illustrates very well the changing crite- rion design’s flexibility and is a good example of behav- ior analysts and clients working together. “Because the parents were permitted to regulate the extent of each cri- terion change, the intervention was quite lengthy. How- ever, by allowing the parents to adjust their own exposure to acceptable levels, adherence to the overall procedure may have been improved.” (Allen & Evans, 2001, p. 500)

Guidelines for Using the Changing Criterion Design

Proper implementation of the changing criterion design requires the careful manipulation of three design factors: length of phases, magnitude of criterion changes, and number of criterion changes.

241

Length of Phases

Because each phase in the changing criterion design serves as a baseline for comparing changes in responding measured in the next phase, each phase must be long enough to achieve stable responding. “Each treatment phase must be long enough to allow the rate of the target behavior to restabilize at a new and changed rate; it is stability after change has been achieved, and before in- troduction of the next change in criterion, that is crucial to producing a convincing demonstration of control” (Hartmann & Hall, 1976, p. 531). Target behaviors that are slower to change therefore require longer phases.

The length of phases in a changing criterion design should vary considerably to increase the design’s valid- ity. For experimental control to be evident in a changing criterion design, the target behavior not only must change to the level required by each new criterion in a predictable (preferably immediate) fashion, but also must conform to the new criterion for as long as it is in effect. When the target behavior closely follows successively more de- manding criteria that are held in place for varied periods of time, the likelihood is reduced that the observed changes in behavior are a function of factors other than the independent variable (e.g., maturation, practice ef- fects). In most situations, the investigator should not set a predetermined number of sessions for which each cri- terion level will remain in effect. It is best to let the data guide ongoing decisions whether to extend the length of a current criterion phase or introduce a new criterion.

Magnitude of Criterion Changes

Varying the size of the criterion changes enables a more convincing demonstration of experimental control. When changes in the target behavior occur not only at the time a new criterion is implemented but also to the level spec- ified by the new criterion, the probability of a functional relation is strengthened. In general, a target behavior’s immediate change to meet a large criterion change is more impressive than a behavior change in response to a small criterion change. However, two problems arise if criterion changes are too large. First, setting aside prac- tical considerations, and speaking from a design stand- point only, large criterion changes may not permit inclusion of a sufficient number of changes in the design (the third design factor) because the terminal level of per- formance is reached sooner. The second problem is from an applied view: Criterion changes cannot be so large that they conflict with good instructional practice. Crite- rion changes must be large enough to be detectable, but not so large as to be unachievable. Therefore, the vari- ability of the data in each phase must be considered in determining the size of criterion changes. Smaller crite-

rion changes can be employed with very stable levels of responding, whereas larger criterion changes are required to demonstrate behavior change in the presence of vari- ability (Hartmann & Hall, 1976).

When using a changing criterion design, behavior analysts must guard against imposing artificial ceilings (or floors) on the levels of responding that are possible in each phase. An obvious mistake of this sort would be to give a student only five math problems to complete when the criterion for reinforcement is five. Although the stu- dent could complete fewer than five problems, the pos- sibility of exceeding the criterion has been eliminated, resulting perhaps in an impressive-looking graph, but one that is badly affected by poor experimental procedure.

Number of Criterion Changes

In general, the more times the target behavior changes to meet new criteria, the more convincing the demonstra- tion of experimental control is. For example, eight crite- rion changes, one of which was a reversal to a previous level, were implemented in the changing design illus- trated in Figure 10, and Allen and Evans (2001) con- ducted seven criterion changes (Figure 11). In both of these cases, a sufficient number of criterion changes oc- curred to demonstrate experimental control. The experi- menter cannot, however, simply add any desired number of criterion changes to the design. The number of crite- rion changes that are possible within a changing crite- rion design is interrelated with the length of phases and the magnitude of criterion changes. Longer phases mean that the time necessary to complete the analysis increases; with a limited time to complete the study, the greater the number of phases, the shorter each phase can be.

Considering the Appropriateness of the Changing Criterion Design

The changing criterion design is a useful addition to the behavior analyst’s set of tactics for evaluating system- atic behavior change. Like the multiple baseline design, the changing criterion design does not require that im- provement in behavior be reversed. However, partial re- versals to earlier levels of performance enhance the design’s capability to demonstrate experimental control. Unlike the multiple baseline design, only one target be- havior is required.

Several characteristics of the changing criterion de- sign limit its effective range of applications. The design can be used only with target behaviors that are already in the subject’s repertoire and that lend themselves to step- wise modification. However, this is not as severe a limi- tation as it might seem. For example, students perform

Multiple Baseline and Changing Criterion Designs

242

many academic skills to some degree, but not at a useful rate. Many of these skills (e.g., solving math problems, reading) are appropriate for analysis with a changing cri- terion design. Allowing students to progress as efficiently as possible while meeting the design requirements of changing criterion analysis can be especially difficult. Tawney and Gast (1984) noted that “the challenge of identifying criterion levels that will permit the demon- stration of experimental control without impeding opti- mal learning rates” is problematic with all changing criterion designs (p. 298).

Although the changing criterion design is sometimes suggested as an experimental tactic for analyzing the ef- fects of shaping programs, it is not appropriate for this purpose. In shaping, a new behavior that initially is not in the person’s repertoire is developed by reinforcing re-

sponses that meet a gradually changing criterion, called successive approximations, toward the terminal behav- ior. However, the changing response criteria employed in shaping are topographical in nature, requiring differ- ent forms of behavior at each new level. The multiple probe design (Horner & Baer, 1978), however, is an ap- propriate design for analyzing a shaping program because each new response criterion (successive approximation) represents a different response class whose frequency of occurrence is not wholly dependent on the frequency of behaviors meeting other criteria in the shaping program. Conversely, the changing criterion design is best suited for evaluating the effects of instructional techniques on stepwise changes in the rate, frequency, accuracy, dura- tion, or latency of a single target behavior.

Summary Multiple Baseline Design

1. In a multiple baseline design, simultaneous baseline mea- surement is begun on two or more behaviors. After stable baseline responding has been achieved, the independent variable is applied to one of the behaviors while baseline conditions remain in effect for the other behavior(s). After maximum change has been noted in the first behavior, the independent variable is then applied in sequential fashion to the other behaviors in the design.

2. Experimental control is demonstrated in a multiple base- line design by each behavior changing when, and only when, the independent variable is applied.

3. The multiple baseline design takes three basic forms: (a) a multiple baseline across behaviors design consisting of two or more different behaviors of the same subject; (b) a multiple baseline across settings design consisting of the same behavior of the same subject in two or more differ- ent settings; and (c) a multiple baseline across subjects de- sign consisting of the same behavior of two or more different participants.

Variations of the Multiple Baseline Design

4. The multiple probe design is effective for evaluating the effects of instruction on skill sequences in which it is highly unlikely that the subject’s performance on later steps in the sequence can improve without instruction or mastery of the earlier steps in the chain. The multiple probe design is also appropriate for situations in which prolonged baseline measurement may prove reactive, im- practical, or too costly.

5. In a multiple probe design, intermittent measurements, or probes, are taken on all of the behaviors in the design at the

outset of the experiment. Thereafter, probes are taken each time the subject has achieved mastery of one of the be- haviors or skills in the sequence. Just prior to instruction on each behavior, a series of true baseline measures are taken until stability is achieved.

6. The delayed multiple baseline design provides an analytic tactic in situations in which (a) a planned reversal design is no longer desirable or possible; (b) limited resources preclude a full-scale multiple baseline design; or (c) a new behavior, setting, or subject appropriate for a multiple baseline analysis becomes available.

7. In a delayed multiple baseline design, baseline mea- surement of subsequent behaviors is begun sometime after baseline measurement was begun on earlier be- haviors in the design. Only baselines begun while ear- lier behaviors in the design are still under baseline conditions can be used to verify predictions made for the earlier behaviors.

8. Limitations of the delayed multiple baseline design in- clude (a) having to wait too long to modify certain be- haviors, (b) a tendency for baseline phases to contain too few data points, and (c) the fact that baselines begun after the independent variable has been applied to earlier be- haviors in the design can mask the interdependence (co- variation) of behaviors.

Assumptions and Guidelines for Using Multiple Baseline Designs

9. Behaviors comprising multiple baseline designs should be functionally independent of one another (i.e., they do not covary) and should share a reasonable likelihood that each will change when the independent variable is applied to it.

Multiple Baseline and Changing Criterion Designs

243

Multiple Baseline and Changing Criterion Designs

10. Behaviors selected for a multiple baseline design must be measured concurrently and must have an equal opportunity of being influenced by the same set of relevant variables.

11. In a multiple baseline design, the independent variable should not be applied to the next behavior until the previ- ous behavior has changed maximally and a sufficient pe- riod of time has elapsed to detect any effects on behaviors still in baseline conditions.

12. The length of the baseline phases for the different behav- iors comprising a multiple baseline design should vary significantly.

13. All other things being equal, the independent variable should be applied first to the behavior showing the most stable level of baseline responding.

14. Conducting a reversal phase in one or more tiers of a mul- tiple baseline design can strengthen the demonstration of a functional relation.

Considering the Appropriateness of Multiple Baseline Designs

15. Advantages of the multiple baseline design include the fact that (a) it does not require withdrawing a seemingly ef- fective treatment, (b) sequential implementation of the in- dependent variable parallels the practice of many teachers and clinicians whose task is to change multiple behaviors in different settings and/or subjects, (c) the concurrent mea- surement of multiple behaviors allows direct monitoring of generalization of behavior change, and (d) the design is relatively easy to conceptualize and implement.

16. Limitations of the multiple baseline design include the fact that (a) if two or more behaviors in the design covary, the multiple baseline design may not demonstrate a functional relation even though one exists; (b) because verification must be inferred from the lack of change in other behav- iors, the multiple baseline design is inherently weaker than the reversal design in showing experimental control be- tween the independent variable and a given behavior; (c) the multiple baseline design is more an evaluation of the

independent variable’s general effectiveness than an analy- sis of the behaviors involved in the design; and (d) con- ducting a multiple baseline design experiment requires considerable time and resources.

Changing Criterion Design

17. The changing criterion design can be used to evaluate the effects of a treatment on the gradual or stepwise improve- ment of a behavior already in the subject’s repertoire.

18. After stable baseline responding has been achieved, the first treatment phase is begun, in which reinforcement (or punishment) is usually contingent on the subject’s per- forming at a specified level (criterion). The design entails a series of treatment phases, each requiring an improved level of performance over the previous phase. Experi- mental control is demonstrated in the changing criterion design when the subject’s behavior closely conforms to the gradually changing criteria.

19. Three features combine to determine the potential of a changing criterion design to demonstrate experimental control: (a) the length of phases, (b) the magnitude of cri- terion changes, and (c) the number of criterion changes. The believability of the changing criterion design is en- hanced if a previous criterion is reinstated and the sub- ject’s behavior reverses to the level previously observed under that criterion.

Considering the Appropriateness of the Changing Criterion Design

20. The primary advantages of the changing criterion design are that (a) it does not require a withdrawal or reversal of a seemingly effective treatment, and (b) it enables an ex- perimental analysis within the context of a gradually im- proving behavior, thus complementing the practice of many teachers.

21. Limitations of the changing criterion design are that the target behavior must already be in the subject’s repertoire, and that incorporating the necessary features of the design may impede optimal learning rates.

244

component analysis direct replication double-blind control placebo control

procedural fidelity replication systematic replication treatment drift

treatment integrity Type I error Type II error

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 1: Ethical Considerations

1-12 Give preference to assessment and intervention methods that have been scientifically validated, and use scientific methods to evaluate those that have not yet been scientifically validated.

Content Area 5: Experimental Evaluation of Interventions

5-1 Systematically manipulate independent variables to analyze their effects on treatment.

5-3 Conduct a component analysis (i.e., determining effective component(s) of an intervention package).

Content Area 10: Systems Support

10-3 Design and use systems for monitoring procedural integrity.

Planning and Evaluating Applied Behavior Analysis Research

Key Terms

245

Planning and Evaluating Applied Behavior Analysis Research

Previously outlined were considerations and procedures for selecting target behaviors, de- tailed strategies for designing measurement sys-

tems, presented guidelines for displaying and interpreting behavioral data, and described experimental tactics for revealing whether observed changes in a target behavior can be attributed to an intervention. This chapter supple- ments the information described thus far by examining questions and considerations that should be addressed when designing, replicating, and evaluating behavioral research. We begin by reviewing the central role of the in- dividual subject in behavioral research, and follow with a discussion of the value of flexibility in experimental design.

Importance of the Individual Subject in Behavioral Research To achieve maximum effectiveness, the research meth- ods of any science must respect the defining characteris- tics of that science’s subject matter. Behavior analysis—a science devoted to discovering and understanding the controlling variables of behavior—defines its subject matter as the activity of living organisms, a dynamic phe- nomenon that occurs at the level of the individual organ- ism. It follows that the research methods most often used by behavior analysts feature repeated measures of be- havior of individual organisms (the only place, by defi- nition, where behavior can be found). This focus on the behavior of individual subjects has enabled applied be- havior analysts to discover and refine effective interven- tions for a wide range of socially significant behavior.

To further explain the importance that this focus on the individual subject or client holds for applied behavior analy- sis, we will now contrast it with a research model that re- volves around comparisons of data representing the aggregate measures of different groups of subjects. This groups-comparison approach to designing and evaluating experiments has predominated “behavioral research” in psy- chology, education, and other social sciences for decades.

Brief Outline of a Groups-Comparison Experiment

The basic format for a groups-comparison experiment can be described as follows.1 A pool of subjects (e.g., 60 first-grade nonreaders) is selected randomly from the pop-

1This brief sketch of the simplest form of a groups-comparison design study omits many important details and controls. Readers interested in a thorough explication and examples of group research methods should con- sult an authoritative text such as Campbell and Stanley (1966).

2Perhaps in part because the researcher must attend to “the purely logisti- cal demands forced by the management of a battalion of subjects” (John- ston & Pennypacker, 1980, p. 256), group designs are characterized by few measures of the dependent variable (often just two measures: pretest and posttest).

ulation (e.g., all first-grade nonreaders in a school dis- trict) relevant to the research question (e.g., Will the XYZ intensive phonics program improve first-grade nonread- ers’ ability to decode unpredictable text?). The subjects are divided randomly into two groups: the experimental group and the control group. An initial measure (pretest) of the dependent variable (e.g., score on a test of decod- ing skills) is obtained for all subjects in the study, the in- dividual pretest scores for the subjects in each group are combined, and the mean and standard deviation are cal- culated for each group’s performance on the pretest. Sub- jects in the experimental group are then exposed to the independent variable (e.g., 6 weeks of the XYZ program), which is not provided to subjects in the control group. After the treatment program has been completed, a posttest measure of the dependent variable is obtained for all subjects, and the mean and standard deviation posttest scores for each group are computed.2 The researcher then compares any changes in each group’s scores from pretest to posttest, applying various statistical tests to the data that enable inferences regarding the likelihood that any differences between the two groups’ performances can be attributed to the independent variable. For example, as- suming that the mean pretest scores for the experimental and control groups were similar, and the posttest measure revealed an improved mean score for the experimental group but not for the control group, statistical analyses would indicate the mathematical probability that the dif- ference was due to chance. When a statistical test rules out chance as a likely factor, the researcher infers that that independent variable was responsible for effects on the dependent variable (e.g., the experimental group’s im- provement from pretest to posttest).

Researchers who combine and compare measures of groups of subjects in this way do so for two primary rea- sons. First, advocates of group designs assume that av- eraging the measures of many subjects’ performance controls for intersubject variability; thus, they assume that any changes in performance are the work of the independent variable. The second rationale for using large groups of subjects is the assumption that increas- ing the number of subjects in a study increases the ex- ternal validity of the findings. That is, a treatment variable found effective with the subjects in the exper- imental group will also be effective with other subjects in the population from which the sample subjects were selected. The assumption of increased generality of findings is discussed later in this chapter in the section on replication. In the next section we comment on

246

Planning and Evaluating Applied Behavior Analysis Research

3A complete discussion of the many problems posed by commingling data from multiple subjects is beyond the scope of this text. Students wishing to obtain a more complete understanding of this important issue are en- couraged to read Johnston and Pennypacker (1980, 1993b) and Sidman (1960).

the first reason for the use of groups of subjects—that doing so controls for intersubject variability. Our dis- cussion identifies three fundamental concerns with typi- cal groups-comparison designs that bear heavily on experimental reasoning.3

Group Data May Not Represent the Performance of Individual Subjects

By definition, applied behavior analysis is concerned with improving the behavior of the individual. Knowing that the average performance of a group of subjects changed may not reveal anything about the performance of indi- vidual subjects. It is quite possible for the average per- formance of subjects in the experimental group to have improved, while the performance of some subjects stayed the same, and the performance of others even deterio- rated. It is even possible for the majority of subjects to show no improvement, for some subjects to get worse, and for a few subjects to improve sufficiently to yield an overall average improvement of statistical significance.

In defense of the groups-comparison approach, it might be said that it can show that a treatment is gener- ally effective, that no treatment works with everyone, that people respond differently, and so on. But the fact that a group’s average performance improves with a treatment is insufficient reason to adopt it, particularly for people in dire need of help with academic, social, or other be- havioral challenges. General effectiveness is insufficient; the factors responsible for one subject’s improvement with the treatment and another’s lack of improvement must be discovered. To be most useful, a treatment must be understood at the level at which people come into con- tact with it and are affected by it: the individual level.

The two graphs in Figure 10.1 suggest some of the many faulty conclusions that are possible when an in-

of the man whose bare feet were in a bucket of ice while his head was on fire. When asked how he was feeling, the man replied, “On the average, I feel fine.”

vestigator’s interpretation of a study is based on group mean scores. Each graph presents hypothetical data for the individual and average performances of two groups, each group consisting of two subjects. The data show no change in the mean response measure from pretest to posttest for either group. The pre- and posttest group data in both graphs in Figure1 would suggest that the in- dependent variable had no effect on the subjects’ behav- ior. However, the left-hand graph in Figure 1 shows that Subject A’s performance improved from pretest to posttest, and Subject B’s behavior deteriorated over the same period of time.4 The right-hand graph shows that although the pre- and posttest measures for Subjects C and D were identical, if repeated measures of Subject C’s and Subject D’s behavior between the pretest and posttest had been conducted, significant variability within and be- tween the two subjects would have been revealed.

Group Data Masks Variability in the Data

A second problem associated with the mean performance of a group of subjects is that it hides variability in the data. Even if repeated measures of Subject C’s and Sub- ject D’s behavior between the pre- and posttest had been conducted as shown in Figure 1, a researcher who re- lied on the group’s mean performance as the primary in- dicator of behavior change would be ignorant of the variability that occurred within and between subjects.

When repeated measurement reveals significant lev- els of variability, an experimental search with the goal of identifying and controlling the factors responsible for the variability is in order. The widespread belief that the ef- fects of uncontrolled variables in a study can be somehow controlled by statistical manipulations of the dependent variable is faulty.

Subject A Subject B Group Mean

R es

p o

n se

M ea

su re

Subject C Subject D Group Mean

R es

p o

n se

M ea

su re

Time Time

Figure 1 Hypothetical data showing that the mean performance of a group of subjects may not represent an individual subject’s behavior.

The posttest data point in the left-hand graph of Figure 1 is reminiscent

247

Planning and Evaluating Applied Behavior Analysis Research

Statistical control is never a substitute for experimental control. . . . The only way to determine whether or not uncontrolled variables are influencing the data is to in- spect the data at the finest available level of decomposi- tion, usually point-by-point for each individual subject. No purpose is served by combining the data statistically to obscure such effects. (Johnston & Pennypacker, 1980, p. 371)

Instead of controlling its sources before the fact, the be- tween groups approach emphasizes controlling variabil- ity statistically after the fact. These two tactics do not have the same effects on the database. Whereas efforts to control actual variability lead to improved control over responding and, thus, a clearer picture of the effects of each condition, statistical manipulation of variable data cannot remove the influences already represented in the data. (Johnston & Pennypacker, 1993b, p. 184)

Attempting to “cancel out” variability through sta- tistical manipulation neither eliminates it from the data nor controls the variables responsible for it. And the re- searcher who attributes the effects of unknown or un- controlled variables to chance removes himself or herself even further from the identification and analysis of im- portant variables. In his monumental work, Tactics of Sci- entific Research, Sidman (1960) dealt repeatedly and forcefully with this critical issue.

To some experimenters, chance is simply a name for the combined effects of uncontrolled variables. If such vari- ables are, in fact, controllable, then chance in this sense is simply an excuse for sloppy experimentation, and no further comment is required. If the uncontrolled variables are actually unknown, then chance is, as Boring (1941) has pointed out, a synonym for ignorance. . . . One of the most discouraging and at the same time challenging as- pects of behavioral science is the sensitivity of behavior to a tremendous array of variables. . . . But variables are not canceled statistically. They are simply buried so that their effects cannot be seen. The rationale for statistical immobilization of unwanted variables is based on the as- sumed random nature of such variables. . . . Not only is the assumption of randomness with respect to the uncon- trolled variables an untested one but it is also highly im- probable. There are few, if any, random phenomena in the behavioral world. (pp. 45, 162–163)

Sidman (1960) also commented on an experimenter’s use of statistics in an attempt to deal with troublesome se- quence effects.

He has a neat trick up his sleeve. By averaging together the data for both subjects under Condition A, and again under Condition B, he “cancels out” the order effect, and completely bypasses the problem of irreversibility. By a simple arithmetical operation, two subjects have become one, and a variable has been eliminated.

It has not, in fact, gone anywhere. Numbers may be made to disappear by adding and subtracting them from

each other. Five apples minus three apples are two ap- ples. The numbers are easily changed by a few strokes of the pen, but some eating has to be done before the ap- ples themselves will vanish. (p. 250)

The “eating” that must be done to control the effects of any variable can be accomplished in only two ways: (a) holding the variable constant throughout the experiment, or (b) isolating the suspected factor as an independent variable and manipulating its presence, absence, and/or value during the experiment.

Intrasubject Replication Is Absent from Group Designs

A third weakness of the groups-comparison statistical in- ference research model is that the power of replicating effects with individual subjects is lost. One of the great strengths of within-subject experimental designs is the convincing demonstration of a functional relation made possible by replication within the design itself. Even though multiple subjects are typically involved in applied behavior analysis research, each subject is always treated as a separate experiment. Although behavior analysts often display and describe the data for all subjects as a group, data from individual subjects are used as the basis for determining and interpreting experimental effects. Applied behavior analysts are wise to heed Johnston and Pennypacker’s (1980) admonition, “An effect that emerges only after individual data have been combined is probably artifactual and not representative of any real be- havioral processes” (p. 257).

This discussion should not be interpreted to mean that the overall performance of groups of subjects cannot, or should not, be studied with the strategies and tactics of applied behavior analysis. There are many applied situ- ations in which the overall performance of a group is so- cially significant. For example, Brothers, Krantz, and McClannahan (1994) evaluated an intervention to in- crease the number of pounds of recyclable office paper re- cycled by 25 staff members at a school. Still, it is important to remember that group data may not repre- sent the performance of individual participants, and vice versa. For example, Lloyd, Eberhardt, and Drake (1996) compared the effects of group versus individual rein- forcement contingencies within the context of collabo- rative group study conditions on quiz scores by students in a Spanish language class. The results showed that the group contingencies resulted in higher mean quiz scores for the class as a whole compared to the individual con- tingencies condition. However, overall benefits at the class level were mitigated by differential results for in- dividual students. When group results do not represent individual performances, researchers should supplement group data with individual results, ideally in the form of

248

Planning and Evaluating Applied Behavior Analysis Research

graphic displays (e.g., Lloyd et al., 1996; Ryan & Hemmes, 2005).

In some instances, however, the behavior analyst may not be able to control the access of subjects to the exper- imental setting and contingencies or even be able to iden- tify who the subjects are (e.g., Van Houten & Malenfant, 2004; Watson, 1996). The dependent variable must then consist of all of the responses made by individuals who enter the experimental setting. This approach is used fre- quently in community-based behavior analysis research. For example, group data have been collected and ana- lyzed on such dependent variables as litter control on a university campus (Bacon-Prue, Blount, Pickering, & Drabman, 1980), car pooling by university students (Ja- cobs, Fairbanks, Poche, & Bailey, 1982), drivers’ com- pliance and caution at stop signs (Van Houten & Malenfant, 2004), the use of child safety belts in shopping carts (Barker, Bailey, & Lee, 2004), and reducing graffiti on restroom walls (Watson, 1996).

Importance of Flexibility in Experimental Design On one level, an effective experimental design is any arrangement of type and sequence of independent vari- able manipulations that produces data that are interesting and convincing to the researcher and the audience. In this context the word design is particularly appropriate as a verb as well as a noun; the effective behavioral researcher must actively design each experiment so that each achieves its own unique design. There are no ready-made experimental designs awaiting selection. The prototype designs are examples of analytic tactics that afford a form of experimental reasoning and control that has proven effective in advancing our understanding of a wide range of phenomena of interest to applied behavior analysts. Johnston and Pennypacker (1980, 1993a) have been clear and consistent in stating that the “suspicion some may hold that generic categories of design types exist and should be botanized” (1980, p. 293) is counterproductive to the practice of the science of behavior.

In order to explain how to design and interpret within subject comparisons, it is tempting to develop cate- gories of similar arrangements or designs. This almost requires giving each category a label, and the labeled categories then imply that there is something impor- tantly different that distinguishes each from the others. (1993a, p. 267)

The requirements for creating useful comparisons cannot be reduced to a cookbook of simple rules or for- mulas. . . . It misleads students by suggesting that partic- ular types of arrangements have specific functions and

5The analytic tactics presented in Chapters entitled “Reversal and Alter- nating Treatments Designs” and “Multiple Baseline and Changing Crite- rion Designs” should not be considered soley as experimental designs for another reason: All experiments incorporate design elements in addition to the type and sequence of independent variable manipulations (e.g., sub- jects, setting, dependent variable, measurement system).

by failing to encourage them to understand the underly- ing considerations that open up unlimited experimental options. (1993a, p. 285)

Sidman (1960) was even more adamant in his warn- ing regarding the undesirable effects of researchers’ be- lieving in the existence of a given set of rules for experimental design.

The examples may be accepted as constituting a set of rules that must be followed in the design of experiments. I cannot emphasize too strongly that this would be disas- trous. I could make the trite statement that every rule has its exception, but this is not strong enough. Nor is the more relaxed statement that the rules of experimental design are flexible, to be employed only where appropri- ate. The fact is that there are no rules of experimental design. (p. 214)

We agree with Sidman. The student of applied be- havior analysis should not be led to believe that any of the analytic tactics constitute experimental designs per se.5

Still, we believe that it is useful to present the most com- monly used analytic tactics in design form for two rea- sons. First, the vast majority of studies that have advanced the field of applied behavior analysis have used experi- mental designs that incorporated one or more of the an- alytic tactics. Second, we believe that the beginning student of behavior analysis benefits from an examination of specific examples of isolated experimental tactics and their application; it is one step in learning the assump- tions and strategic principles that guide the selection and arrangement of analytic tactics into an experimental de- sign that effectively and convincingly addresses the re- search question(s) at hand.

Experimental Designs That Combine Analytic Tactics

Combining multiple baseline and reversal tactics may allow a more convincing demonstration of experimental control than either tactic alone. For example, by with- drawing the treatment variable (a return to baseline) and then reapplying it within one or more tiers in a multiple baseline design, researchers are able to determine the existence of a functional relation between the indepen- dent variable and each behavior, setting, or subject of the multiple baseline element and also to analyze the ef- fectiveness of the independent variable across the tiers

249

Planning and Evaluating Applied Behavior Analysis Research

(e.g., Alexander, 1985; Ahearn, 2003; Barker, Bailey, & Lee, 2004; Blew, Schwartz, & Luce, 1985; Bowers, Woods, Carlyon, & Friman, 2000; Heward & Eachus,

To investigate the research questions of interest, in- vestigators often build experimental designs that entail a combination of analytic tactics. For example, it is not un- common for experimenters to evaluate multiple treat- ments by sequentially applying each in a multiple baseline fashion (e.g., Bay-Hinitz, Peterson, & Quilitch, 1994; Iwata, Pace, Cowdery, & Miltenberger, 1994; Van Houten, Malenfant, & Rolider, 1985; Wahler & Fox, 1980; Yeaton & Bailey, 1983). Experimental designs that combine multiple baseline, reversal, and/or alternating treatments tactics can also provide the basis for compar- ing the effects of two or more independent variables or conducting a component analysis of elements of a treat- ment package. For example, the experimental designs used by L. J. Cooper and colleagues’ (1995) used alter- nating treatments comparisons within a sequence of mul- tiple treatment reversals to identify the active variables in treatment packages for children with feeding disorders.

Haring and Kennedy (1990) used multiple baseline across settings and reversal tactics in their experimental design that compared the effectiveness of time-out and differential reinforcement of other behavior (DRO) on the frequency of problem behaviors by two secondary students with severe disabilities (see Figure 2).6 San- dra and Raff each frequently engaged in repetitive, stereo- typic problem behaviors (e.g., body rocking, loud vocalizations, hand flapping, spitting) that interfered with classroom and community activities. In addition to as- sessing the effects of the time-out and DRO interventions against a no-treatment baseline condition, the design also enabled the researchers to conduct two comparisons of the relative effects of each treatment during an instruc- tional task and leisure context. The design enabled Har- ing and Kennedy to discover that the time-out and DRO interventions produced different outcomes depending on the activity context in which they were applied. For both students, DRO was more effective than time-out in sup- pressing problem behavior in the task context; the oppo- site results were obtained in the leisure context, where time-out suppressed problem behavior and DRO proved ineffective.

Experimenters have also incorporated alternating treatments into experimental designs containing multi- ple baseline elements. For example, Ahearn, Kerwin,

6Time-out and differential reinforcement of other behavior (DRO) are ex- plained in Chapters entitled “Punishment by Removal of a Stimulus” and “Differential Reinforcement,” respectively.

Eicher, Shantz, and Swearingin (1996) evaluated the rel- ative effects of two treatments for food refusal in an al- ternating treatments design implemented in a multiple baseline across subjects format. Likewise, McGee, Krantz, and McClannahan (1985) evaluated the effects of several procedures for teaching language to autistic children with an experimental design that incorporated alternating treatments within a multiple baseline across behaviors component that was, in turn, nested within an overall multiple baseline across subjects format. Zanolli and Daggett (1998) investigated the effects of reinforce- ment rate on the spontaneous social initiations of socially withdrawn preschoolers with an experimental design con- sisting of multiple baseline, alternating treatments, and re- versal tactics.

Figure 3 shows how Sisson and Barrett (1984) in- corporated a multiple probe across behaviors component, an alternating treatments analysis, and a multiple base- line across behaviors element in a design comparing the effects of two language-training procedures. The design enabled the investigators to discover the superiority of the total communication method for these two children, as well as the fact that direct application of the treatment was required for learning to occur on specific sentences. Re- sults for a third subject revealed a functional relation of the same form and direction as that found for the two chil- dren whose results are shown in Figure 3, but one not so strongly in favor of the total communication procedure.

Our intent in describing several experiments that com- bined analytic tactics is not to offer any of these examples as model designs. They are presented instead as illustra- tions of the infinite number of experimental designs that are possible by arranging different combinations and se- quences of independent variable manipulations. In every instance the most effective (i.e., convincing) experimental designs are those that use an ongoing evaluation of data from individual subjects as the basis for employing the three elements of baseline logic—prediction, verification, and replication.

Internal Validity: Controlling Potential Sources of Confounding in Experimental Design An experiment is interesting and convincing, and yields the most useful information for application, when it pro- vides an unambiguous demonstration that the indepen- dent variable was solely responsible for the observed behavior change. Experiments that demonstrate a clear functional relation are said to have a high degree of in- ternal validity. The strength of an experimental design is

1979; Miller & Kelley, 1994; Zhou, Goff, & Iwata, 2000).

250

Planning and Evaluating Applied Behavior Analysis Research

100

T as

Baseline DRO T.O. DRO T.O. DRO

100

Le is

u re

Baseline T.O.

Sandra

10 20 30

100

T as

BL. T.O. DRO T.O. DRO T.O. DRO

100

Le is

u re

Baseline T.O.

Raff

P er

ce n

ta ge

o f

In te

rv al

s w

it h

S te

re o

ty p

10 20 Sessions

Figure 2 Experimental design employing multiple baselines across settings and re- versal tactics counterbalanced across two subjects to analyze the effects of time-out (TO) and differential reinforcement of other behavior (DRO) treatment conditions. From “Contextual Control of Problem Behavior” by T. G. Haring and C. H. Kennedy, 1990, Journal of Applied Behavior Analysis, 23, pp. 239–240. Copyright 1990 by the Society for the Experi- mental Analysis of Behavior, Inc. Reprinted by permission.

determined by the extent to which it (a) demonstrates a re- liable effect (i.e., repeated manipulation of the indepen- dent variable produces a consistent pattern of behavior change) and (b) eliminates or reduces the possibility that factors other than the independent variable produced the behavior change (i.e., controls for confounding variables).

Implicit in the term experimental control, which is often used to signify a researcher’s ability to reliably pro- duce a specified behavior change by manipulating an in- dependent variable, is the idea that the researcher controls the subject’s behavior. However, “control of behavior” is inaccurate because the experimenter can control only some aspect of the subject’s environment. Therefore, the level of experimental control obtained by a researcher

refers to the extent to which she controls all relevant vari- ables in a given experiment. The researcher exerts this control within the context of an experimental design that, even though carefully planned at the outset, takes its ul- timate form from the researcher’s ongoing examination and response to the data.

An effective experimental design simultaneously reveals a reliable functional relation between indepen- dent and dependent variables (if one exists) and mini- mizes the likelihood that the observed behavior changes are the result of unknown or uncontrolled variables. An experiment has high internal validity when changes in the dependent variable are demonstrated to be a function only of the independent variable. When

251

Planning and Evaluating Applied Behavior Analysis Research

0Se n

te n

ce 1

& 1

Eli Total Communication

Oral

Baseline Alternating Treatments Total Communication

0Se n

te n

ce 2

& 2

0Se n

te n

ce 3

& 3

1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 M

ea n

N u

m b

er o

f Se

n te

n ce

P ar

ts Sessions

4 3 2 1 0

Se n

te n

ce 1

& 1

Mick

Baseline Alternating Treatments Total Communication

4 3 2 1 0

Se n

te n

ce 2

& 2

4 3 2 1 0

Se n

te n

ce 3

& 3

1 5 10 15 20 25 30 35 40 45 50 55 60 65 95

M ea

n N

u m

b er

o f

Se n

te n

ce P

ar ts

Sessions 70 75 80 85 90

Figure 3 Experimental design employing an alternating treatments tactic, a multiple probe, and a multiple baseline across behaviors analysis. From “Alternating Treatments Comparison of Oral and Total Communication Training with Minimally Verbal Retarded Children” by L. A. Sisson and R. P. Barrett, 1984, Journal of Applied Behavior Analysis, 17, p. 562. Copyright 1984 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

planning an experiment and later when examining the ac- tual data from an ongoing study, the investigator must al- ways be on the lookout for threats to internal validity. Uncontrolled factors known or suspected to have exerted influence on the dependent variable are called confound- ing variables. Much of a researcher’s efforts during the course of a study are aimed at eliminating or controlling confounding variables.

The attainment of steady state responding is the pri- mary means by which applied behavior analysts assess the degree of experimental control. Separating the effects of the independent variable from the effects of a poten- tially confounding variable requires clear, empirical ev- idence that the potentially confounding variable is no longer present, has been held constant across experi- mental conditions, or has been isolated for manipulation as an independent variable. Any experiment can be af-

fected by a virtually unlimited number of potential con- founds; and as with every aspect of experimental design, there are no set rules for identifying and controlling con- founding variables to which researchers can turn. How- ever, some common and likely sources of confounding can be identified as well as tactics that can be considered to control them. Confounding variables can be viewed as related primarily to one of four elements of an experi- ment: subject(s), setting, measurement of the dependent variable, and independent variable.

Subject Confounds

A variety of subject variables can confound the results of a study. Maturation, which refers to changes that take place in a subject over the course of an experiment, is a potential confounding variable. For example, a subject’s

252

Planning and Evaluating Applied Behavior Analysis Research

improved performance during the later phases of a study may be the result of physical growth or the acquisition of academic, social, or other behaviors and be unrelated to manipulations of the independent variable. Experimen- tal designs that incorporate rapidly changing conditions or multiple introductions and withdrawals of the inde- pendent variable over time usually control for matura- tion effectively.

In most applied behavior analysis research, a subject is in the experimental setting and contacting the contin- gencies implemented by the investigator for only a por- tion of the day. As it is in any study, the assumption is made that each subject’s behavior during each session will be primarily a function of the experimental condi- tions in effect. In reality, however, each subject’s behav- ior may also be influenced by events that have occurred outside of the experiment. For example, suppose that the frequency of contributions to a class discussion is the de- pendent variable in a study. Now suppose that just prior to a session a student who has been contributing to dis- cussions at a high rate was involved in a fight in the lunch- room and emits substantially fewer contributions compared to his level of responding in previous sessions. This change in the student’s behavior may, or may not, be a result of the lunchroom fight. If the lunchroom fight coincided with a change in the independent variable, it would be especially difficult to detect or separate any ef- fects of the experimental conditions from those of the extra-experimental event.

Although the researcher may be aware of some events that are likely causes of variability during a study, many other potential confounds go undetected. Repeated mea- surement is both the control for and the means to detect the presence and effects of such variables. The uncon- trolled variables responsible for a subject’s having a “bad day” or an unusually “good day” are particularly trouble- some in research designs with few and/or widely spaced measurements of the dependent variable. This is one of the major weaknesses of using pretest-posttest compar- isons to evaluate the effects of a treatment program.

Because groups-comparison experiments are predi- cated on subjects’ similarity in relevant characteristics (e.g., gender, age, ethnicity, cultural and linguistic back- ground, current skills), they are vulnerable to confound- ing by differences among subjects. Concern that the characteristics of one or more subjects may confound an experiment’s results is generally not an issue in the single- subject experiments of applied behavior analysis. First, a person should participate in a study because she will ben- efit if the target behavior is changed successfully. Sec- ond, a subject’s idiosyncratic characteristics cannot confound a study using a true within-single experimental design. With the exception of a multiple baseline across

subjects analysis, each participant in a behavioral study serves as her own control, which guarantees identically matched subjects in all experimental conditions because those subjects are the same person. Third, the external validity of results from a single-subject analysis is not de- pendent on the extent to which the subject(s) shares cer- tain characteristics with others. The extent to which a functional relation applies to other subjects is established by replicating the experiment with different subjects.

Setting Confounds

Most applied behavior analysts conduct studies in nat- ural settings where a host of variables are beyond their control. Studies in natural settings are more prone to con- founding by uncontrolled events than are studies con- ducted in laboratories where extraneous variables can be more tightly managed. Even so, the applied experimenter is not without resources to mitigate the detrimental ef- fects of setting confounds. For instance, when the applied researcher observes that an uncontrolled event has coin- cided with changes in the data, he should hold all possi- ble aspects of the experiment constant until repeated measurement again reveals stable responding. If the un- planned event appears to have a robust effect on the tar- get behavior, or is otherwise of interest to the investigator, and is amenable to experimental manipulation, the in- vestigator should treat it as an independent variable and explore its possible effects experimentally.

Applied researchers concerned about setting con- founds must also be on the lookout for the availability of “bootleg” reinforcement within and outside the experi- mental situation. A good example of how a setting con- found operates occurs when, unbeknownst to the experimenter, subjects have ready access to potential re- inforcers. In such a case, the effectiveness of those con- sequences as reinforcers diminishes.

Measurement Confounds

Many factors should be considered in designing an accurate and nonreactive measurement system. Still, numerous sources of confounding may exist within a well-planned measurement system. For instance, data might be confounded by observer drift, the influence of the experimenter’s behavior on observers, and/or observer bias. Although admittedly difficult to accom- plish in applied settings where observers often see the independent variable being implemented, keeping observers naive to the conditions and expected outcomes of an experiment reduces the potential of confounding by observer bias. On a related note,

253

Planning and Evaluating Applied Behavior Analysis Research

when observers score permanent products, the products should not contain identifying marks that indicate who produced each product and under what experimental con- ditions it was produced. Having observers score papers from baseline and treatment conditions in randomized order reduces the likelihood that observer drift or bias will confound the data within one treatment condition phase. (This procedure is more suitable to controlling for drift or bias by observers conducting postexperiment accuracy or IOA assessments.)

Unless a completely unobtrusive measurement sys- tem is devised (e.g., a covert system using one-way mir- rors, or observations conducted at some distance from the subject), reactivity to the measurement procedure must always be considered as a possible confound. To offset this possible confound, the experimenter must maintain baseline conditions long enough for any reactive effects to run their course and for stable responding to be obtained. If reactivity to measurement produces un- desirable effects (e.g., aggressive behavior, cessation of productivity) and a more unobtrusive measurement pro- cedure cannot be devised, intermittent probes should be considered. Measures can also be confounded by practice, adaptation, and warm-up effects, especially during the initial stages of baseline. Again, the proper procedure is to continue baseline conditions until stable responding is obtained or variability is reduced to minimal levels. In- termittent probes should not be used for baseline mea- surement of behaviors for which practice effects would be expected. This is because, if the target behavior is sus- ceptible to practice effects, those effects will occur dur- ing the intervention condition when more frequent measures are conducted, thereby confounding any effects of the independent variable.

Independent Variable Confounds

Most independent variables are multifaceted; that is, there is usually more to a treatment condition than the specific variable of interest to the investigator. For ex- ample, the effects of a token economy on students’ aca- demic productivity may be confounded by variables such as the personal relationship between the students and the teacher who delivers the tokens, social interactions as- sociated with delivering and exchanging tokens, the ex- pectation of teacher and students that performance will improve when the token system is implemented, and so on. If the intent is to analyze the effects of token rein- forcement per se, these potentially confounding variables must be controlled.

Schwarz and Hawkins (1970) provided a good ex- ample of a control procedure for ruling out an aspect as- sociated with a treatment as responsible for behavior

change. The researchers evaluated the effects of token reinforcement on three maladaptive behaviors of an ele- mentary student who was described as severely with- drawn. During treatment, the therapist and the girl met each day after school and viewed a videotape that had been made earlier that day of the student’s classroom be- havior. The therapist also administered tokens contingent on the girl’s videotaped behavior displaying progressively fewer occurrences of maladaptive behaviors.

Schwarz and Hawkins recognized an independent confound potentially lurking in the design of their study. They reasoned that if improvement occurred as a function of the treatment, the question would remain whether the student’s behavior had improved because the therapist-delivered positive attention and rewards improved her self-concept, which in turn changed her maladaptive behaviors in the classroom, which were symptomatic of her poor self-concept. In that case, Schwarz and Hawkins could not be certain that the con- tingent tokens played an important role in changing the behavior. Schwarz and Hawkins, anticipating this pos- sible confound, controlled for it in a simple and direct way. Following baseline, they implemented a condition in which the therapist met with the girl each day after school and provided her with social attention and token reinforcement contingent on improvements in hand- writing. During this control phase, the three target behaviors—face touching, slouching, and low voice volume—showed no change, thereby increasing their confidence in a conclusion that the girl’s ultimate be- havioral improvements during the subsequent interven- tion phases were due to the token reinforcement.

When medical researchers design experiments to test the effects of a drug, they use a technique called a placebo control to separate effects that may be produced by a subject’s perceived expectations of improvement be- cause of taking the drug apart from the effects actually produced by the drug. In the typical groups-comparison design, the subjects in the experimental group receive the real drug, and subjects in the control group receive placebo pills. Placebo pills contain an inert substance, but they look, feel, and taste exactly like the pills con- taining the real drug being tested.

Applied behavior analysts have also employed placebo controls in single-subject experiments. For example, in their study evaluating a pharmacological treatment of the impulsivity by students with attention- deficit/hyperactivity disorder (ADHD), Neef, Bicard, Endo, Coury, and Aman (2005) had a pharmacist prepare placebos and medications in identical gelatin capsules in 1-week supplies for each child. Neither the students nor the observers knew if a child had taken the medication or had taken the placebos. When neither the subject(s) nor the observers know whether the independent variable is

254

Planning and Evaluating Applied Behavior Analysis Research

present or absent from session to session, this type of control procedure is called a double-blind control. A double-blind control procedure eliminates confounding by subject expectations, parent and teacher expectations, differential treatment by others, and observer bias.

Treatment Integrity

The results of many experiments have been confounded by the inconsistent application of the independent vari- able. The researcher must make a concerted effort to en- sure that the independent variable is applied exactly as planned and that no other unplanned variables are ad- ministered inadvertently along with the planned treat- ment. The terms treatment integrity and procedural fidelity refer to the extent to which the independent vari- able is implemented or carried out as planned.

Low treatment integrity invites a major source of confounding into an experiment, making it difficult, if not impossible, to interpret the results with confidence. Data from an experiment in which the independent vari- able was administered improperly, applied inconsistently, conducted piecemeal, and/or delivered in overdose or un- derdose form often lead to conclusions that—depending on the results obtained—represent either a false positive (claiming a functional relation when no such relation ex- ists) or a false negative (failing to detect a functional re- lation when one actually does exist). If a functional relation is apparent from the analysis of the data, one can- not be sure whether the treatment variable as described by the experimenter was responsible or whether the effects were a function of extraneous, uncontrolled elements of the intervention as it was actually applied. On the other hand, it may be equally erroneous to interpret the failure to produce significant behavior change as evidence that the independent variable is ineffective. In other words, had the independent variable been implemented as planned, it might have been effective.

Numerous threats to treatment integrity exist in ap- plied settings (Billingsley, White, & Munson, 1980; Gre- sham, Gansle, & Noell, 1993; Peterson, Homer, & Wonderlich, 1982). Experimenter bias can cause the re- searcher to administer the independent variable in such a way that it enjoys an unfair advantage over the baseline or comparative conditions. Treatment drift occurs when the application of the independent variable during later phases of an experiment differs from the way it was ap- plied at the outset of the study. Treatment drift can result from the complexity of the independent variable, which can make it difficult for practitioners to implement all of the elements consistently over the course of an experi- ment. Contingencies influencing the behavior of those responsible for implementing the independent variable can also result in treatment drift. For example, a teacher

might implement only those aspects of a procedure that she favors and might implement the full intervention only when the experimenter is present.

Precise Operational Definition. Achieving a high level of treatment integrity begins with developing a complete and precise operational definition of the treat- ment procedures. Besides providing the basis for train- ing the persons who will implement an intervention and judging the level of treatment integrity attained, opera- tional definitions of treatment conditions are a requisite for meeting the technological dimension of applied be- havior analysis (Baer et al., 1968). An investigator’s failure to provide explicit operational definitions of the treatment variable hampers the dissemination and proper use of the intervention by practitioners, and makes it difficult for other researchers to replicate and ultimately validate the findings.

Gresham and colleagues (1993) recommended that descriptions of the independent variable be judged by the same standards of explicitness that are used in assessing the quality of definitions of dependent variable. That is, they should be clear, concise, unambiguous, and objec- tive. More specifically, Gresham and colleagues also sug- gested that treatments be operationally defined in each of four dimensions: verbal, physical, spatial, and tempo- ral. They used Mace, Page, Ivancic, and O’Brien’s (1986) definition of a time-out procedure as an example of an operational definition of an independent variable.

(a) Immediately following the occurrence of a target behavior (temporal dimension), (b) the therapist said “No, go to time-out” (verbal dimension), (c) led the child by the arm to a prepositioned time-out chair (physical dimension), (d) seated the child facing the corner (spatial dimension). If the child’s buttocks were raised from the time-out chair or if the child’s head was turned more than 45° (spatial dimension), the therapist used the least amount of force necessary to guide com- pliance with the time-out procedure (physical dimen- sion). (f ) At the end of 2 min (temporal dimension), the therapist turned the time-out chair 45° from the corner (physical and spatial dimensions) and walked away (physical dimension). (pp. 261–262)

Simplify, Standardize, and Automate. When plan- ning an experiment, placing a high priority on simplify- ing and standardizing the independent variable and providing criterion-based training and practice for the people who will be responsible for implementing it en- hances treatment integrity. Treatments that are simple, precise, and brief, and that require relatively little ef- fort, are more likely to be delivered with consistency than those that are not. Simple, easy-to-implement techniques also have a higher probability of being ac- cepted and used by practitioners than those that are not,

255

Planning and Evaluating Applied Behavior Analysis Research

and thus possess a certain degree of self-evident social validity. Simplicity is, of course, a relative concern, not a mandate; effecting change in some socially important behaviors may require the application of intense, com- plex interventions over a long period of time and may involve many people. Baer (1987) made this point suc- cinctly when he stated:

Long problems are simply those in which the task analy- sis requires a series of many behavior changes, perhaps in many people, and although each of them is relatively easy and quick, the series of them requires not so much effort as time, and so it is not arduous but merely tedious. (pp. 336–337)

Practitioners need not be thwarted or dismayed by complex interventions; they simply need to realize the treatment integrity implications. All things being equal, however, a simple and brief treatment will probably be applied more accurately and consistently than will a com- plex and extended one.

To ensure the consistent implementation of the in- dependent variable, experimenters should standardize as many of its aspects as cost and practicality allow. Stan- dardization of treatment can be accomplished in a vari- ety of ways. When a treatment requires a complex and/or extended sequence of behaviors, a script for the person administering it may improve the accuracy and consis- tency with which the independent variable is applied. For example, Heron, Heward, Cooke, and Hill (1983) used a scripted lesson plan with overhead transparencies to en- sure that a classwide peer tutor training program was im- plemented consistently across groups of children.

If automating the intervention will not compromise it in any way, researchers might consider “canning” the independent variable so that an automated device can be used for its delivery. Although a videotaped tutor training presentation in Heron and colleagues’ (1983) study would have eliminated any potential confounding caused by the teacher’s slightly different presentations of the lesson from group to group and across sets of tutoring skills, using a canned presentation would also have eliminated the desired interactive and personal aspects of the train- ing program. Some treatment variables are well suited to automated presentation in that automation neither limits the desirability of the treatment nor seriously reduces its social validity in terms of acceptability or practicability (e.g., use of videotaped programs to model residential energy conservation).

Training and Practice. Training and practice in properly implementing the independent variable provides the person(s) who will be responsible for conducting the treatment or experimental sessions with the necessary

7Different levels or degrees of treatment integrity can be manipulated as an independent variable to analyze the effects of full versus partial imple- mentations of an intervention, or various kinds of treatment “mistakes” (e.g., Holcombe, Wolery, & Snyder, 1994; Vollmer, Roane, Ringdahl, & Marcus, 1999).

skills and knowledge to carry out the treatment. It would be a mistake for the researcher to assume that a person’s general competence and experience in the experimental setting (e.g., a classroom) guarantees correct and consis- tent application of an independent variable in that setting (e.g., implementing a peer-mediated tutoring program).

As stated earlier, scripts detailing treatment proce- dures and cue cards or other devices that remind and prompt people through steps of an intervention can be helpful. Researchers should not, however, assume that merely providing the intervention agent with a detailed script will ensure a high degree of treatment integrity. Mueller and colleagues (2003) found that a combination of verbal instructions, modeling, and/or rehearsal was re- quired for parents to implement pediatric feeding proto- cols with a high level of treatment integrity. Performance feedback has also been shown to improve the integrity with which parents and practitioners implement behavior support plans and explicit teaching techniques (e.g., Cod- ding, Feinberg, Dunn, & Pace, 2005; Sarakoff & Strumey, 2004; Witt, Noell, LaFleur, & Mortenson, 1997).

Assessing Treatment Integrity. Although simpli- fication, standardization, and training help increase the degree of treatment integrity, they do not guarantee it. If there is any doubt about the correct and consistent application of the independent variable, investigators should provide data on the accuracy and reliability of the independent variable (Peterson et al., 1982; Wolery, 1994). Treatment integrity (or procedural fidelity) data reveal the extent to which the actual implementation of all of the experimental conditions over the course of a study matches their descriptions in the method section of a research report.7

Even though the effective control of the presence and absence of the independent variable is a requisite to an internally valid experiment, applied behavior analysts have not always made sufficient efforts to assure the in- tegrity of the independent variable. Two reviews of arti- cles published in the Journal of Applied Behavior Analysis from 1968 to 1990 found that the majority of authors did not report data assessing the degree to which the inde- pendent variable was properly and consistently applied (Gresham et al., 1993; Peterson et al., 1982). Peterson and colleagues noted that a “curious double standard” had developed in applied behavior analysis in which data on the interobserver agreement of dependent variable mea-

256

Planning and Evaluating Applied Behavior Analysis Research

sures were required for publication, but such data were seldom provided or required for the independent variable.

Peterson and colleagues (1982) suggested that the technology developed for assessing and increasing the accuracy and believability of measures of the dependent variable is fully applicable to the collection of procedural fidelity data. Importantly, observation and recording of the independent variable provides the experimenter with data indicating whether calibration of the treatment agent is necessary (i.e., bringing the intervention agent’s be- havior into agreement with the true value of the inde- pendent variable). Observation and calibration give the researcher an ongoing ability to use retraining and prac- tice to ensure a high level of treatment integrity over the course of an experiment.

Figure 4 shows the data collection form used by trained observers to collect treatment integrity data in a study evaluating the effects of different qualities and du- rations of reinforcement for problem behavior, compli- ance, and communication within a treatment package for escape-maintained problem behavior (Van Norman, 2005). The observers viewed videotapes of randomly se- lected sessions representing approximately one third to one half of all sessions in each condition and phase of the study. The percentage of treatment integrity for each condition was calculated by dividing the number of steps the experimenter completed correctly during a session by the total number of steps completed.

This overview of sources of potential confounding variables is, of necessity, incomplete. A complete inven-

tory of all possible threats to the internal validity of ex- perimental research would be well beyond the scope of this text. And presenting such a list might suggest that a researcher need only control for the variables listed and not worry about anything else. In truth, the list of poten- tial confounds is unique to every experiment. The effec- tive researcher is the one who questions and probes the influence of as many relevant variables as possible. No experimental design can control all potential confounds; the challenge is to reduce, eliminate, or identify the in- fluence of as many potentially confounding variables as possible.

Social Validity: Assessing the Applied Value of Behavior Changes and the Treatments That Accomplish Them In his landmark article, “Social Validity: The Case for Subjective Measurement or How Applied Behavior Analy- sis Is Finding Its Heart,” Montrose Wolf (1978) proposed the then “radical concept that clients (including parents and guardians of dependent people, and even those whose taxes support social programs) must understand and ad- mire the goals, outcomes, and methods of an interven- tion” (Risley, 2005, p. 284). Wolf recommended that the social validity of a study in applied behavior analysis

1. The instructor delivers a task prompt at the beginning of the session, e.g., "it's time to work" or similar.

2. If the participant does not respond, the instructor represents the choice by replacing materials or restating the contingencies.

3. The break card (or similar) is presented simultaneously (within 3 seconds) with work task materials.

4. Following work choice (touching materials associated with work) a. Removes the task materials b. Presents a timer with green colored cue card c. Provides access to high preference items d. Engages in play with the participant for 1 minute

Procedural Steps Phase 1/A (SD)

5. Following break requests a. Removes the task materials b. Presents timer with yellow cue card c. Provides access to moderately preferred tangible items and neutral commenting for 30 seconds

6. Following problem behavior within 10 s the instructor a. Removes the task/play materials b. Presents a timer with red cue card c. Provides no attentiion or tangible items for 10 seconds

7. Which side was the break card (or similar) presented (the participant's R = right or L = left) for each choice presentation.

Opportunities Correct % Correct Yes

R R L L L R L R R L R L R L L R

N/A

Video Clip # Rater Initials: Date:1-1AL 7/6/05

16/16

7/7

8/8

E. B.

Figure 4 Example of a form used to record treatment integrity data. Adapted from “The Effects of Functional Communication Training, Choice Making, and an Adjusting Work Schedule on Problem Behavior Maintained by Negative Reinforcement” by R. K. Van Norman, 2005, p. 204. Unpublished doctoral dissertation. Columbus, OH: The Ohio State University. Used by permission.

257

Planning and Evaluating Applied Behavior Analysis Research

should be assessed in three ways: the social significance of the target behavior, the appropriateness of the proce- dures, and the social importance of the results.

Although social validity assessments may increase the likelihood that a study will be published and can be helpful in the marketing and public relations of behavioral programs (Hawkins, 1991; Winett, Moore, & Anderson, 1991), the ultimate purpose of social validity assessments is “to help choose and guide [behavior change] program developments and applications” (Baer & Schwartz, 1991, p. 231). Social validity assessments are most often ac- complished by asking the direct consumers of a behavior change program (the learners, clients, research subjects) and/or a group of indirect consumers (e.g., family mem- bers, teachers, therapists, community people) questions about how satisfied they are with the relevance and im- portance of the goals of the program, the acceptability of the procedures, and the value of the behavior change out- comes achieved.8

Verbal statements by practitioners and consumers that they find a treatment or program acceptable and ef- fective should not be viewed as either evidence that the program was effective, or if it was, that consumers will continue to use the methods. Noting Baer, Wolf, and Ris- ley’s (1968) admonition that “a subject’s verbal descrip- tion of his own nonverbal behavior usually would not be accepted as a measure of his actual behavior” (p. 93), Hawkins (1991) recommended that the term consumer satisfaction be used instead of social validity because it acknowledges that what is typically obtained in social validity assessments “is essentially a collection of con- sumer opinions” (p. 205), the validity of which has not been determined.

In measuring consumers’ verbal judgments, we are only hoping that these verbal behaviors are substantially con- trolled by variables directly relevant to the habilitation task at hand, and thus that they predict habilitative out- comes to some degree. The validity of such consumer judgments has yet to be established; they should not be viewed as a validity criterion but rather as a second opin- ion from a lay person who may or may not be better in- formed and less biased than the professional is. (p. 212)

Validating the Social Importance of Behavior Change Goals

The social validity of behavior change goals begins with a clear description of those goals.

8Detailed discussions of social validity and procedures for assessing it can be found in Fuqua and Schwade (1986), Van Houten (1979), Wolf (1978), and the special section on social validity in the summer 1991 issue of the Journal of Applied Behavior Analysis.

To assess the social importance of goals, the researcher must be precise about the goals of the behavior change effort at the levels of (a) the broad social goal (e.g., im- proved parenting, enhanced social skills, improved car- diovascular health, increased independence), (b) the categories of behavior hypothesized to be related to the broad goal (e.g., parenting—providing instructional feedback, using time-out, etc.), and/or (c) the responses that comprise the behavioral category of interest (e.g., using time-out—directing the child to a location away from other people, instructing the child to “sit out” for a specified duration, etc.). Social validation may be con- ducted for any of these levels of goals. (Fawcett, 1991, pp. 235–236)

Van Houten (1979) suggested two basic approaches to determining socially valid goals: (a) Assess the per- formance of persons considered competent, and (b) ex- perimentally manipulate different levels of performance to determine empirically which produces optimal results. Observations of the performance of typical performers can be used to identify and validate behavior change goals and target levels of performance. To arrive at a socially valid performance criterion for a social skills training program for two adults with disabilities who worked in a restaurant, Grossi, Kimball, and Heward (1994) ob- served four restaurant employees without disabilities over a period of 2 weeks to determine the frequency with which they acknowledged verbal initiations from cowork- ers. Results from these observations revealed that the em- ployees without disabilities acknowledged an average of 90% of initiations directed toward them. This level of performance was selected as the goal for the two target employees in the study.

A study by Warren, Rogers-Warren, and Baer (1976) provided a good example of testing the effects of differ- ent levels of performance to determine socially valid out- comes. The researchers assessed the effect of different frequencies of children’s offers to share play materials with their peers on the peers’ reactions to those offers. They found that peers accepted offers to share most con- sistently when those offers were made at a middle fre- quency; that is, not too frequently, not too seldom.

Validating the Social Acceptance of Interventions

Several scales and questionnaires for obtaining consumers’ opinions of the acceptability of behavioral interventions have been developed. For example, the Intervention Rat- ing Profile is a 15-item Likert-type scale for assessing the acceptability of classroom interventions (Martens, Witt, Elliott, & Darveaux, 1985). The Treatment Accept- ability Rating Form (TARF) consists of 20 questions with which parents rate the acceptability of behavioral treat-

258

Planning and Evaluating Applied Behavior Analysis Research

ments used in outpatient clinic (Reimers & Wacker, 1988).

the TARF used by Van Norman (2005) to obtain treat- ment acceptability information from each participant’s parents, teachers, therapists, and behavior support staff. Although some of the people whose opinions were being sought had witnessed, or had watched a video of, the in- tervention being used with the student, the following description of the intervention was read to each con- sumer before he or she was asked to answer each of the questions:

First we conducted an assessment to find out what moti- vated Zachary to engage in challenging behavior(s) such as throwing materials, hitting people, and dropping to the floor. We found that Zachary engaged in challenging behavior(s), at least in part, in order to escape or avoid task demands.

Next, we taught Zachary to ask for a break as a re- placement behavior for challenging behavior by using physical prompting and attaching the response of asking for a break to access to a highly preferred item, lots of attention, and a long duration break (3 min).

Then we gave Zachary the choice to ask for work by simply touching the work materials (essentially en- gaging in the first step of the task) and getting access to highly preferred items, attention, and long duration break (1 min) or asking for a break and getting access to moderately preferred items for a shorter duration break (30 sec). At any time during this procedure if Zachary engaged in problem behavior he was given a 10 s break with no attention and no activities/items.

Finally, we continued to give Zachary the choice to ask for work, a break or engage in problem behavior, however now we required Zachary to comply with a greater number of task-related instructions before he was given access to the highly preferred activities, atten- tion, and a 1 min break. Each session we increased the number of task-related instructions that were given and needed to be complied with before access to the highly preferred break.

Physical prompting was only used during the initial phase to teach Zachary new responses, specifically how to ask for a break and how to ask for work. Otherwise Zachary was making all independent choices as they were presented. (p. 247)

Validating the Social Importance of Behavior Changes

Methods for assessing the social validity of outcomes in- clude (a) comparing participants’ performance to the per- formance of a normative sample, (b) asking consumers to rate the social validity of participants’ performance, (c) asking experts to evaluate participants’ performance, (d) using a standardized assessment instrument, and (e) test-

ing participants’ newly learned level of performance in the natural environment.

Normative Sample

Van den Pol and colleagues (1981) used the performance of a normative sample of typical fast-food restaurant cus- tomers to assess the social validity of the posttraining performance of the young adults with disabilities whom they had taught to order or pay for a meal without assis- tance. The researchers simply observed 10 randomly se- lected, typical customers who ordered and ate a meal in fast-food restaurants. They recorded the accuracy with which these customers performed each step of a 22-step task analysis. The students’ performance at the follow- up probe equaled or exceeded that of the customers in the normative sample in all but 4 of 22 specific skills.

Using normative samples to assess the social valid- ity of behavior change is not limited to posttreatment comparisons. Comparing subjects’ behavior to ongoing probes of the behavior of a normative sample provides a formative assessment measure of how much improve- ment has been made and how much is still needed. An ex- cellent sample of ongoing social validity assessment is a study by Rhode, Morgan, and Young (1983), in which token reinforcement and self-evaluation procedures were used to improve the classroom behavior of six students with behavior disorders. The overall goal of the study was to help the six students improve their appropriate classroom behavior (e.g., following classroom rules, com- pleting teacher-assigned tasks, volunteering relevant re- sponses) and decrease inappropriate behavior (e.g., talking out, noncompliance, aggression) so that they would be accepted and successful in regular (general ed- ucation) classrooms. At least once per day throughout the course of the 17-week study, the researchers randomly selected classmates in the regular classrooms for obser- vation. The same observation codes and procedures that were used to measure the six target students’ behavior were used to obtain the normative sample data.

Figure 6 shows the mean and range of the six stu- dents’ appropriate behavior during each condition and phase of the study compared to the normative sample. (Individual graphs showing the percentage of appropriate behavior in the resource and regular classroom of all six subjects in each of nearly 90 sessions were also included in Rhode and colleagues’ article.) During baseline, the six boys’ levels of appropriate behavior were well below those of their nondisabled peers. During Phase I of the study, in which the subjects learned to self-evaluate, their behavior in the resource room improved to a level match- ing that of their regular classroom peers. However, when the subjects were in the regular classroom during Phase I, their behavior compared poorly with that of the other

Figure 5 shows the experimenter-modified version of

259

Planning and Evaluating Applied Behavior Analysis Research

Figure 5 Examples of questions adapted from the Treatment Acceptability Rating Form—Revised (Reimers and Wacker, 1988) to obtain consumers’ opinions of the acceptability of intervention procedures used to treat challenging behaviors of secondary students with severe disabilities.

Treatment Acceptability Rating Form—Revised (TARF-R)

1. How clear is your understanding of the suggested procedures? _____ _____ _____ _____ _____ _____ _____ Not at Neutral Very

all clear clear

2. How acceptable do you find the strategies to be regarding your concerns about the identified learner? _____ _____ _____ _____ _____ _____ _____ Not at all Neutral Very

acceptable acceptable

3. How willing are you to implement the suggested procedures as you heard them described? _____ _____ _____ _____ _____ _____ _____ Not at Neutral Very

all willing willing

4. Given the learner’s behavior issues, how reasonable do you find the suggested procedures? _____ _____ _____ _____ _____ _____ _____ Not at all Neutral Very

reasonable reasonable

5. How costly will it be to implement these strategies? _____ _____ _____ _____ _____ _____ _____ Not at Neutral Very

all costly costly

11. How disruptive will it be to your classroom to implement the suggested procedures? _____ _____ _____ _____ _____ _____ _____ Not at all Neutral Very disruptive disruptive

13. How affordable are these procedures? _____ _____ _____ _____ _____ _____ _____ Not at all Neutral Very affordable affordable

14. How much do you like the proposed procedures? _____ _____ _____ _____ _____ _____ _____

Do not like Neutral Like them them at all very much

17. How much discomfort is your learner likely to experience as a result of these procedures? _____ _____ _____ _____ _____ _____ _____

No discomfort Neutral Very much at all discomfort

19. How willing would you be to change your classroom routine to implement these procedures? _____ _____ _____ _____ _____ _____ _____ Not at Neutral Very

all willing willing

20. How well will carrying out these procedures fit into your classroom routine? _____ _____ _____ _____ _____ _____ _____ Not at Neutral Very all well well

From “The Effects of Functional Communication Training, Choice Making, and an Adjusting Work Schedule on Problem Behavior Maintained by Negative Reinforcement” by R. K. Van Norman, 2005, pp. 248–256. Unpublished doctoral dissertation. Columbus, OH: The Ohio State University. Used by permission.

260

R ie

n ./

Fe ed

. 100

M ea

n %

o f

A p

p ro

p ri

at e

B eh

av io

B as

el in

M at

ch /1

M at

ch /5

M at

ch /3

3 1/

M at

ch /1

6 2/

S- Ev

al ./

S- Ev

al ./

S- Ev

al ./

S- Ev

al ./

V R

2 D

ay s

S- Ev

al ./

N o

P o

in ts

S- Ev

al ./

V er

b . 6

0 m

in .

S- Ev

al ./

V er

b . V

R 2

D ay

Baseline Phase I Phase II

= Resource room (subjects) = Regular classroom (subjects) = Regular classroom (peers) = Range/regular classroom (Subjects)

N o

S -E

va l.

Experimental Conditions

Figure 6 Example of using measures of the behavior of a normative sample standard for assessing the social validity of outcomes of a behavior change program. From “Generalization and Maintenance of Treatment Gains of Behaviorally Handicapped Students from Resource Rooms to Regular Classrooms Using Self- Evaluation Procedures” by G. Rhode, D. P. Morgan, and K. R. Young, 1983, Journal of Applied Behavior Analysis, 16, p. 184. Copyright 1984 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

students in the normative sample. As Phase II progressed, which involved various strategies for generalization and maintenance of the treatment gains, the mean level of ap- propriate behavior by the six students matched that of their nondisabled peers, and variability among the six students decreased (except for one subject who exhibited no appropriate behavior on one session in the next-to- last condition).

Consumer Opinion

The most frequently used method for assessing social va- lidity is to ask consumers, including subjects or clients whenever possible, if they thought behavior changes oc- curred during the study or program, and if so, if they thought those behavior changes were important and valu- able. Figure 7 shows the questionnaire Van Norman (2005) used to obtain opinions of consumers (i.e., the subjects’ parents, teachers, and instructional aides; school administrators; behavior support staff; an occupational therapist; a school psychologist; and a psychology aide) on the social validity of results of an intervention de- signed to reduce challenging behavior maintained by es- cape. Van Norman created a series of 5-minute video clips

from randomly selected before-intervention and after- intervention sessions and placed the clips in random order on a CD. The social validity evaluators did not know whether each clip represented a before- or after- intervention session. After viewing each clip, the con-

Expert Evaluation

Experts can be called on to judge the social validity of some behavior changes. For example, as one measure of the social validity of changes in the unaided notetaking skills of high school students with learning disabilities in social studies lectures as a result of having been ex- posed to teacher-prepared guided notes, White (1991) asked 16 secondary social studies teachers to rate the stu- dents’ baseline and post-intervention lecture notes on three dimensions: (1) accuracy and completeness com- pared to lecture content; (2) usefulness for study for tests over the lecture content; and (3) how the notes compared to those taken by typical general education students. (The teachers did not know whether each set of notes they were rating was from baseline or post-intervention condition.)

Planning and Evaluating Applied Behavior Analysis Research

sumer completed the questionnaire shown in Figure 7.

261

Planning and Evaluating Applied Behavior Analysis Research

intervention used to treat challenging behavior of secondary students with severe disabilities.

Social Validity of Results Questionnaire

Directions: Video Clip # _____

Please view the video and then circle one of the five choices that best describes the extent to which you agree or disagree with each of the three statements below.

1. The student is engaged in academic or vocational work tasks, sitting appropriately (bottom in seat), and attending to the teacher or materials.

1 2 3 4 5 Strongly Disagree Disagree Undecided Agree Strongly Agree

2. The student is engaged in challenging behavior and not attending to the teacher or materials.

1 2 3 4 5 Strongly Disagree Disagree Undecided Agree Strongly Agree

3. The student appears to have a positive affect (e.g., smiling, laughing).

1 2 3 4 5 Strongly Disagree Disagree Undecided Agree Strongly Agree

Comments about the student’s behavior in the video clip: ___________________________

___________________________________________________________________________

General comments about this student’s behavior:__________________________________

___________________________________________________________________________

Name (optional): _____________________________________________________________

Relation to the learner in video (optional): ________________________________________

Adapted from “The Effects of Functional Communication Training, Choice Making, and an Adjusting Work Schedule on Problem Behavior Maintained by Negative Reinforcement” by R. K. Van Norman, 2005, p. 252. Unpublished doctoral dissertation. Columbus, OH: The Ohio State University. Used by permission.

Fawcett (1991) observed that, “If expert ratings are not sufficiently high, [the researcher should] consider what else might be done to program for social validity. Despite the best efforts, assessments may show that re- search goals are regarded by consumer judges as in- significant, interventions procedures as unacceptable, or results as unimportant” (p. 238).

Standardized Tests

Standardized tests can be used to assess the social valid- ity of some behavior change program outcomes. Iwata, Pace, Kissel, Nau, and Farber (1990) developed the Self- Injury Trauma Scale (SITS) to enable researchers and therapists to measure the number, type, severity, and lo- cation of injuries produced by self-injurious behavior.

The SITS yields a Number Index and Severity Index with scores of 0 to 5, and an estimate of current risk. Although the data collected in a treatment program may show sig- nificant decreases in the behaviors that produce self-injury (e.g., eye poking, face slapping, head banging), the social significance of the treatment must be validated by evi- dence of reduced injury. Iwata and colleagues wrote:

. . . the social relevance of the behavior lies in its trau- matic outcome. The measurement of physical injuries prior to treatment can establish the fact that a client or subject actually displays behavior warranting serious at- tention. . . . Conversely, injury measurement following treatment can corroborate observed changes in behavior because reduction of an injury-producing response below a certain level should be reflected in the eventual disappearance of observable trauma. In both of these in-

Figure 7 Form for obtaining consumer opinions of social validity of results of an

262

Planning and Evaluating Applied Behavior Analysis Research

stances, data on injuries provide a means of assessing social validity. (Wolf, 1978, pp. 99–100)

Twohig and Woods (2001) used the SITS to validate the outcomes of a habit-reversal treatment for the chronic skin picking of two typically developing adult males. Both men reported that they had engaged in skin picking since childhood, digging the fingernails into the ends of a finger and pulling or scrapping the skin, which some- times caused bleeding, scarring, and infections. Two ob- servers independently rated pretreatment, posttreatment, and follow-up photographs of the two men’s hands with the SITS. The Number Index (NI) and Severity Index (SI) SITS scores on the pretreatment photographs for both men were 1 and 2, respectively, indicating one to four in- juries on either hand and distinct but superficial breaks in the skin. NI and SI scores of 0 on the posttreatment pho- tos for both men indicated no apparent injuries. On the follow-up photos taken 4 months after treatment had ended, both men had SITS NI and SI scores of 1, indi- cating red or irritated skin.

Real-World Test

Perhaps the most socially valid way to assess the social validity of a learner’s newly acquired behavior is to put it to an authentic test in the natural environment. For in- stance, the validity of what three adolescents with learn- ing disabilities had learned about road signs and traffic laws was validated when they passed the Ohio Depart- ment of Motor Vehicles test and earned their temporary driver’s permits (Test & Heward, 1983).

In a similar way, the social validity of the cooking skills being learned by three secondary students with de- velopmental disabilities and visual impairments was tested frequently when their friends arrived at the end of probe sessions to share the food they had just prepared (Trask-Tyler, Grossi, & Heward, 1994). In addition to providing a direct and authentic assessment of social validity, real-world tests put the learner’s repertoire in contact with naturally occurring contingencies of rein- forcement, which may promote maintenance and gener- alization of the newly acquired behaviors.

External Validity: Replicating Experiments to Determine the Generality of Research Findings External validity refers to the degree to which a func- tional relation found reliable and socially valid in a given experiment holds under different conditions. An inter- vention that works only within a circumscribed set of

conditions and proves ineffective when any aspect in the original experiment is altered makes a limited contribu- tion to the development of a reliable and useful technol- ogy of behavior change. When a carefully controlled experiment has shown that a particular treatment produces consistent and socially significant improvements in the target behavior of a given subject, a series of important questions should then be asked: Will this same treatment be as effective if it is applied to other behaviors? Will the procedure continue to work if it is changed in some way (e.g., if implemented at a different time of the day, by an- other person, on a different schedule)? Will it work in a setting different from the original experiment? Will it work with participants of different ages, backgrounds, and repertoires? Questions about external validity are not abstract or rhetorical; they are empirical questions that can be addressed only by empirical methods.

A functional relation with external validity, or gen- erality, will continue to operate under a variety of con- ditions. External validity is a matter of degree, not an all-or-nothing property. A functional relation that cannot be reproduced under any conditions other than the exact set of original variables (including the original subject) possesses no external validity. At the other end of the continuum, a procedure that is effective at any time, under any conditions, in any setting, with any behavior, and for any subject has complete generality (an improbable sit- uation). Most functional relations fall somewhere be- tween the two ends of this continuum, and those found to have higher degrees of generality make the greater con- tribution to applied behavior analysis. Investigators who use groups-comparison research methods approach the issue of external validity quite differently than do inves- tigators who use within-subjects research methods.

External Validity and Groups Design Research

As stated previously, practitioners of group-comparison experimental designs claim two advantages for the use of large groups of subjects. In addition to the assumption that aggregating the data of a group of subjects will con- trol for intersubject variability, researchers who employ group designs assume that including many subjects in an experiment increases the external validity of a study’s re- sults. On the surface this assumption is perfectly logical, and when viewed at the proper level of extrapolation, it is also true. The more subjects with which a functional re- lation has been demonstrated, the more likely it is that that functional relation will also be effective with other subjects who share similar characteristics. And in fact, demonstrating a functional relation with various subjects in different settings is exactly how applied behavior an- alysts document external validity.

263

Planning and Evaluating Applied Behavior Analysis Research

However, the investigator who claims that the find- ings of a groups-comparison study possess generality to other individuals in the population from which the ex- perimental subjects were chosen violates a fundamental premise of the groups-comparison method and ignores a defining characteristic of behavior. The proper inferences about the results of a groups-design study are from the sample to the population, not from the sample to the in- dividual (Fisher, 1956). The careful methods of random sampling used in groups-design research are followed to ensure that the participants in the study represent a het- erogeneous sample of all the relevant characteristics found in the population from which they were selected. Indeed, the better the sample represents the population from which it is drawn, the less meaningful are the results for any individual subject. “The only statement that can be made concerns the average response of a group with that particular makeup which, unfortunately, is unlikely to be duplicated” (Hersen & Barlow, 1976, p. 56).

A second problem inherent in attempting to extend the results of a groups-comparison study to other people (and unless care is taken, sometimes even to a subject who participated in the study, as was illustrated in Figure 1) is that the groups-design experiment does not demonstrate a functional relation between the behavior of any subject and some aspect of his or her environment. In other words, from the perspective of behavior analysis, there is nothing in the results of a groups-design experi- ment that can have external validity; there is nothing to generalize. Johnston and Pennypacker (1993a) made this point repeatedly and well.

Between groups designs tend to put the cart before the horse. The tactic of exposing different levels of the inde- pendent variable to different groups composed of a large number of subjects and treating their responses collec- tively by asking if they represent the untested portion of the population provides comparisons that describe no member of any group. By failing to focus on individuals with careful attention to experiment control, these tradi- tional methods greatly decrease the chances of discover- ing orderly relationships in the first place, thereby making the questions of subject generality moot. (1993a, p. 352)

The researcher’s first objective is to obtain data that truly represent the relationship between the experimen- tal conditions and the dependent variable for each sub- ject. If this is not accomplished, nothing else matters. Only when the findings are “true” does the question of the meaningfulness or universality of these results be- come relevant. (1993a, p. 250)

Groups-comparison designs and statistical inference have long dominated research in psychology, education, and the other social sciences. Despite its long-standing dominance, the extent to which this research tradition has

contributed to an effective technology of behavior change is highly questionable (Baer, 1977b; Birnbrauer, 1981; Michael, 1974). The field of education is perhaps the most telling example of the inability of groups-design re- search to provide data that lead to improved practice (Greer, 1983; Heward & Cooper, 1992). Instructional methods in the classroom are often influenced more by fad, the personal style of individual teachers, and ideol- ogy than by the cumulative knowledge and understand- ing provided by rigorous and sustained experimental analysis of the variables of which learning is a function (Heron, Tincani, Peterson, & Miller, 2005; Kozloff, 2005; Zane, 2005).

The methods of groups-comparison experimentation are simply inappropriate for answering the questions of primary interest to the applied behavior analyst—em- pirical questions that can be pursued only by analysis of repeated measures of individual behavior under all rele- vant conditions. We agree with Johnston and Penny- packer (1993b):

We find the reasoning underlying all such procedures alien to both the subject matter and the goals of a natural science of behavior and regard the utility of group com- parisons as extremely limited, no matter how elegant the mathematical treatment of data they afford. . . . [group comparison experimentation constitutes] a process of scientific inquiry that is almost totally inverted; instead of using questions about natural phenomena to guide de- cisions about experimental design, models of design are allowed to dictate both the form and content of the ques- tions asked. Not only is this antithetical to the estab- lished role of experimentation in science, the types of questions allowed by groups comparison designs are largely inappropriate or irrelevant to gaining an under- standing of the determinants of behavior. (pp. 94–95)

Our discussion of the inherent limitations of groups- comparison designs for behavior analysis should not be confused with a position that group designs and statisti- cal inference have no value as research methods for seek- ing empirical knowledge about the world. On the contrary, group-comparison designs and statistical infer- ence are highly effective tools for seeking answers to the kinds of questions for which they were devised. Properly designed and well-executed groups-comparison experi- ments can provide answers with a specific degree of con- fidence (i.e., probability) to questions involved in many large-scale evaluations. For example, a government body is less interested in the effects of a new regulation on any individual person (and even less interested in whether a functional relation exists between the regulation and that person’s behavior) than it is in the probability that the behavior of a predictable percentage of the population will be affected by the regulation. The former concern is

264

Planning and Evaluating Applied Behavior Analysis Research

a behavioral one, and the experimental methods of be- havior analysis provide the means to address it. The lat- ter concern is an actuarial one, and it is best pursued with the methods of random sampling, groups comparison, and statistical inference.

External Validity and Applied Behavior Analysis

The external validity (generality) of research findings in applied behavior analysis is assessed, established, and specified through the replication of experiments.

In order to know whether a particular result will be ob- tained for a subject in another study or under applied circumstances, what we really need to know is what variables are necessary to make the effect occur, what variables will prevent it from occurring, and what vari- ables will modulate it. . . . This information cannot be learned by increasing the size of the control and experi- mental groups. It requires conducting a series of experi- ments that identify and study variables that might fall into one of these three categories. (Johnson & Penny- packer, 1993a, p. 251)

Replication in this context means repeating a pre- vious experiment.9 Sidman (1960) described two major types of scientific replication—direct and systematic.

Direct Replication

In a direct replication, the researcher makes every ef- fort to duplicate exactly the conditions of an earlier ex- periment. If the same subject is used in a direct replication, the study is an intrasubject direct replication. Intrasubject replication within experiments is a defining characteristic of applied behavior analysis research and the primary tactic for establishing the existence and reli- ability of a functional relation. An intersubject direct replication maintains every aspect of the earlier experi- ment except that different, although similar, subjects are involved (i.e., same age, similar repertoires). Intersub- ject replication is the primary method for determining the extent to which research findings have generality across subjects.

The many uncontrolled variables in natural settings make the direct replication of experiments outside of the

9Johnston and Pennypacker (1980) pointed out a distinction between repli- cating an experiment and reproducing its results. They stated that the qual- ity of the replication should be judged only by “the extent to which equivalent environmental manipulations associated with [the original ex- periment] are duplicated. . . . Thus, one replicates procedures in an effort to reproduce effects” (pp. 303–304). However, when most researchers re- port a “failure to replicate,” they mean that the results of the replication did not match those obtained in the earlier research (e.g., Ecott, Foate, Tay- lor, & Critchfield, 1999; Friedling & O’Leary, 1979).

laboratory extremely difficult. Nevertheless, intersubject replication is the rule rather than the exception in applied behavior analysis. Although numerous single-subject studies involve just one subject (e.g., Ahearn, 2003; Dixon & Falcomata, 2004; Kodak, Grow, & Northrup, 2004; Tarbox, Williams, & Friman, 2004), the vast ma- jority of published studies in applied behavior analysis include direct intersubject replications. This is because each subject is usually considered an intact experiment. For example, a behavior analysis study in which the in- dependent variable is manipulated in exactly the same way for six subjects in the same setting yields five inter- subject replications.

Systematic Replication

The direct replication of experiments demonstrates the reliability of a functional relation, but the generality of that finding to other conditions can be established only through repeated experimentation in which the conditions of interest are purposefully and systematically varied. In a systematic replication the researcher purposefully varies one or more aspects of an earlier experiment. When a systematic replication successfully reproduces the re- sults of previous research, it not only demonstrates the reliability of the earlier findings but also adds to the ex- ternal validity of the earlier findings by showing that the same effect can be obtained under different conditions. In a systematic replication, any aspect of a previous ex- periment can be altered: subjects, setting, administration of the independent variable, target behaviors.

Although systematic replication offers greater po- tential rewards than direct replication does because it can provide new knowledge about the variables under investigation, it entails some risk. Sidman (1960) de- scribed systematic replication as a gamble, but one well worth taking.

If systematic replication fails, the original experiment will still have to be redone, else there is no way of deter- mining whether the failure to replicate stemmed from the introduction of new variables in the second experi- ment, or whether the control of relevant factors was in- adequate in the first experiment.

On the other hand, if systematic replication suc- ceeds, the pay-off is handsome. Not only is the relia- bility of the original finding increased, but also its generality with respect to other organisms and to other experimental procedures is greatly enhanced. Further- more, additional data are now available which could not have been obtained by a simple repetition of the first ex- periment. (pp. 111–112)

Sidman went on to explain that economic husbandry of limited resources must also play an important role in

265

Planning and Evaluating Applied Behavior Analysis Research

the scientist’s determination of how a research program should proceed. Direct replication of a long and costly experiment can provide data on only the reliability of a functional relation, whereas systematic replication can provide information on the reliability and generality of the phenomena under investigation, as well as new in- formation for additional experimentation.

The external validity of results of groups-comparison research is viewed as an inherent characteristic of a given experiment, as something that can be directly assessed by examining the methods used to conduct the study (e.g., sampling procedures). If that logic is extended to single- subject experiments, then the findings of single-subject experiments cannot be said to have any external validity. But as Birnbrauer (1981) pointed out, external validity is not something a single study has, but rather the prod- uct of many studies. External validity can be pursued only through the active process of systematic replication.

Generality is established, or more likely limited, by ac- cumulating studies which are internally valid and by placing the results into a systematic context, i.e., seek- ing out the principles and parameters that particular procedures appear to be enunciating. The most informa- tive studies ask how can an earlier positive result be re- peated in the present circumstances, with the present problem? (p. 122)

Much of the applied behavior analysis literature con- sists of systematic replications. Indeed, one could argue quite persuasively that almost any applied behavior analy- sis study is a systematic replication of at least some as- pect of an earlier experiment. Even when the authors have not pointed it out, virtually every published experiment reveals significant procedural similarity with previous experiments. However, as we are using the term here, systematic replication refers to concerted and directed efforts to establish and specify the generality of a func- tional relation. For example, Hamlet, Axelrod, and Kuer- schner (1984) found a functional relation between demanded eye contact (e.g., “[Name], turn around”) and compliance with adult instructions in two 11-year-old school children. Included in the same published report were the results of six replications conducted by the same researchers over a period of one year with nine students aged 2 to 21 years. Similar results were reproduced in eight of the nine replication subjects. Although some might consider this an example of direct intersubject replication, Hamlet and colleagues’ replications were con- ducted in various settings (i.e., classrooms, homes, insti- tutions) and therefore should be viewed as a series of systematic replications that demonstrated not only the re- liability of the results but also considerable generality across subjects of different ages in different settings.

10Functional communication training (FCT) is described in Chapter enti- tled “Functional Behavior Assessment.”

Systematic replications across subjects sometimes reveal different patterns of effects, which the researcher might then study as a function of specific subject char- acteristics or contextual variables. For example, Hago- pian, Fisher, Sullivan, Acquisto, and LeBlanc (1998) reported the results of a series of systematic replications with 21 inpatient cases of functional communication training with and without extinction and punishment.10

Lerman, Iwata, Shore, and DeLeon (1997) found that thinning FR 1 schedules of punishment to intermittent punishment produced different effects on the self- injurious behavior of five adults with profound mental retardation.

Some systematic replications are attempts to repro- duce the results reported by another researcher in a slightly different situation or context. For example, Saigh and Umar (1983) successfully reproduced in a Sudanese classroom the positive results originally reported with

Researchers sometimes report multiple experiments, with each experiment serving as a systematic replication investigating the variables influencing a given functional relation. For example, Fisher, and colleagues (1993) con- ducted four studies designed to explore the effectiveness of functional communication training (FCT) with and without extinction and punishment.

Systematic replication is evident when a research team pursues a consistent line of related studies over time. Examples of this approach to replication can be found in Van Houten and colleagues’ studies investigating vari- ables affecting driver behavior and pedestrian safety (e.g., Huybers, Van Houten, & Malenfant, 2004; Van Houten & Nau, 1981, 1983; Van Houten, Nau, & Marini, 1980; Van Houten & Malenfant, 2004; Van Houten, Malenfant, & Rolider, 1985; Van Houten & Retting, 2001); Neef, Markel, and colleagues’experiments on the impulsivity of students with attention-deficit/hyperactivity disorder (ADHD) (e.g., Bicard & Neef, 2002; Ferreri, Neef, & Wait, 2006; Neef, Bicard, & Endo, 2001; Neef, Bicard, Endo, Coury, & Aman, 2005; Neef, Markel et al., 2005); and Miltenberger and colleagues’ line of research on teaching safety skills to children (e.g., Himle, Mil- tenberger, Flessner, & Gatheridge, 2004; Himle, Mil- tenberger, Gatheridge, & Flessner, & 2004; Johnson, Miltenberger et al., 2005; Johnson, Miltenberger et al., 2006; Miltenberger et al., 2004; Miltenberger et al., 2005).

the Good Behavior Game in an American classroom

ported that a “considerable degree of support for the cross-cultural utility of the game was established” (p. 343).

(Barrish, Saunders & Wolf, 1969). Saigh and Umar re-

266

Planning and Evaluating Applied Behavior Analysis Research

In many instances, the systematic replications nec- essary to explore and extend a significant line of research require the independent efforts of investigators at differ- ent sites who are aware of, and build on, one another’s work. When independent teams of researchers at differ- ent geographical locations report similar findings, the net result is a body of knowledge with significant scientific integrity and technological value. This collective effort speeds and enhances the refinement and rigorous testing of interventions that is necessary to the development and refinement of evidence-based practices (Horner et al., 2005; Peters & Heron, 1993). One such example of in- dependent research teams at various sites reporting sys- tematic replications is a growing body of studies exploring the effects of response cards on students’ aca- demic engagement, learning, and deportment during group instruction. Investigators have reported a similar pattern of results—increased participation during in- struction, improved retention of lesson content, and/or reductions in off-task and disruptive behaviors—with re- sponse cards in a wide range of students (general educa- tion students, special education students, and ESL learners), curriculum content (e.g., math, science, social studies, spelling), and instructional settings (e.g., ele- mentary, middle, secondary, and college classrooms) (e.g., Armendariz & Umbreit, 1999; Cavanaugh, Heward, & Donelson, 1996; Christle & Schuster, 2003; Davis & O’Neill, 2004; Gardner, Heward, & Grossi, 1994; Kel- lum, Carr, & Dozier, 2001; Lambert, Cartledge, Lo, & Heward, 2006; Marmolejo, Wilder, & Bradley, 2004).

Evaluating Applied Behavior Analysis Research A list of all the expectations and characteristics of ex- emplary research in applied behavior analysis would be very long. Thus far we have identified a considerable number of requirements for good applied behavior analy- sis. Our purpose now is to summarize those requirements in a sequence of questions one might ask in evaluating the quality of research in applied behavior analysis. Those questions can be organized under four major headings: in- ternal validity, social validity, external validity, and sci- entific and theoretical significance.

Internal Validity

To determine whether an analysis of behavior has been made, the reader of an applied behavior analysis study must decide whether a functional relation has been demonstrated. This decision requires a close examina- tion of the measurement system, the experimental design,

and the degree to which the researcher controlled poten- tial confounds, as well as a careful visual analysis and interpretation of the data.

Definition and Measurement of the Dependent Variable

The initial step in evaluating internal validity is to decide whether to accept the data as valid and accurate measures of the target behavior over the course of the experiment. Some of the important issues to be considered in this de-

Graphic Display

If the data are accepted as a valid and accurate repre- sentation of the dependent variable over the course of the experiment, the reader should next assess the extent of stability of the target behavior during each phase of the study. Before evaluating the stability of the data paths, however, the reader should examine the graphic display for any sources of distortion (e.g., scaling of axes, dis- tortions of time on the horizontal axis). The researcher or consumer who suspects that any element of the graph may encourage interpretations unwarranted by the data should replot the data using a new set of appropriately scaled axes. In an assessment of the stability of the de- pendent variable within the different phases of an ex- periment, the length of the phase or condition must be considered as well as the presence of trends in the data path. The reader should ask whether the conditions in effect during each phase were conducive to practice ef- fects. If so, were these effects allowed to play themselves out before experimental variables were manipulated?

Meaningfulness of Baseline Conditions

The representativeness or fairness of the baseline condi- tions as the basis for evaluating subsequent performance in the presence of the independent variable should be as- sessed. In other words, were the baseline conditions meaningful in relation to the target behavior, setting, and research questions addressed by the experiment? For ex- ample, consider two experiments by Miller, Hall, and Heward (1995) that evaluated the effects of two proce- dures for conducting 1-minute time trials during a daily 10-minute practice session on the rate and accuracy with which students answered math problems. Throughout all conditions and phases of both experiments, the students were instructed to answer as many problems as they could

cision are captured by the questions shown in Figure 8.

267

Planning and Evaluating Applied Behavior Analysis Research

Figure 8 Questions that should be asked when evaluating the definition and measurement of the dependent variable in an applied behavior analysis study.

• Was the dependent variable precisely, completely, and unambiguously defined? • Were examples and nonexamples of the target behavior provided, if doing so would

enhance clarity? • Were the most relevant, measurable dimensions of the target behavior specified (e.g.,

rate, duration)? • Were important concomitant behaviors also measured? • Were the observation and recording procedures appropriate for the target behavior? • Did the measurement provide valid (i.e., meaningful) data for the problem or research

question addressed? • Was the measurement scale broad and sensitive enough to capture the socially

significant changes in the behavior? • Have the authors provided sufficient information on the training and calibration of

observers? • What procedures were used to assess and ensure the accuracy of measurement? • Were interobserver agreement (IOA) assessments reported at the levels at which the

study’s results are presented (e.g., by subject and experimental condition)? • Were observation sessions scheduled during the times, activities, and places most

relevant to the problem or research question? • Did observations occur often enough and closely enough in time to provide a

convincing estimate of behavior change over time? • Were any contingencies operating in the study that may have influenced the observers’

behavior? • Was there any expectation or indication that the dependent variable may have been

reactive to the measurement system? If so, were procedures taken to assess and/or control reactivity?

• Were appropriate accuracy and/or reliability assessments of the data reported?

and they received feedback on their performance. Throughout both experiments, students’ worksheets were marked and scored as follows:

Experimenters marked each student’s worksheets by putting an ‘X’ next to incorrect answers. The number of correct answers over the total number of problems at- tempted was marked at the top of the first worksheet along with a positive comment to encourage the stu- dents to keep trying. If a student’s score was lower than his or her highest previous score, comments such as “Keep trying, Sally!” “Work faster!” or “Keep working on it!” were written. Whenever a student achieved his or her highest score to date, comments such as “Great job, Jimmy! This is your best ever!” were written on the packet. In the event a student equaled her highest previ- ous score, “You tied your best score!” was written on the packet.

At the beginning of each session, scored and marked worksheets from the prior day’s session were returned to the students. Each session during the 10-minute contin- uous work period condition that functioned as the base- line condition began with the classroom teachers saying to the students: “I want you to work hard and try to do your best. Answer as many problems as you can. Don’t

worry if you do not answer all of the problems. There are more problems in the packet than anyone can do. Just try your best” (p. 326).

The initial baseline (A) phase was followed by the two time-trial conditions (B and C) in an A-B-A-B-C-B- C design. Results for students in both classrooms showed a clear functional relation between both of the time-trial conditions and increased correct rate and accuracy over the baseline condition. However, if the classroom teach- ers had not instructed and reminded the students to do their best and to answer as many problems as they could prior to each baseline session, and if the students had not received feedback on their worksheets, the improved per- formance during the time-trial conditions would have been suspect. Even if a clear functional relation had been demonstrated against such baseline conditions, applied researchers and consumers could, and should, question the importance of such results. Maybe the children sim- ply did not know they were expected to work fast. Per- haps the students would have solved problems in the baseline condition at the same high rates that they did in the time-trial conditions if they had been told to “go fast” and received feedback on their performance, praise for their improvements, and encouragement to answer more

268

Planning and Evaluating Applied Behavior Analysis Research

Correct Conclusion Type I Error

(false positive)

Type II Error (false negative)

Correct Conclusion

Y es

N o

R es

ea rc

h er

C o

n cl

u d

es F

u n

ct io

n al

R el

at io

n s

E x

is ts

Yes No

Functional Relations Exist in Nature Figure 9 Ideally, an experimental design and methods of data analysis help a researcher conclude correctly that a functional relation between the indepen- dent and dependent variables exists (or does not exist) when in fact such a relation does (or does not) exist in nature. Concluding that the results of an experiment reveal a functional relation when no such relation exists in nature is a Type I error. Conversely, concluding that that independent variable did not have an effect on the dependent variable when such a relation did occur is a Type II error.

problems. By including the daily instruction to work hard and answer as many problems as they could and return- ing the worksheets to the students as components of the baseline condition, Miller and colleagues obtained mean- ingful data paths during baseline against which to test and compare the effects of the two time-trial conditions.

Experimental Design

The experimental design should be examined to deter- mine the type of experimental reasoning it affords. What elements of the design enable prediction, verification, and replication? Is the design appropriate for the research questions addressed by the study? Does the design ef- fectively control for confounding variables? Does the de- sign provide the basis for component and/or parametric analyses if such questions are warranted?

Visual Analysis and Interpretation

Although various statistical methods for evaluating be- havioral data and determining the existence of functional relations in single-subject designs have been recom- mended (e.g., Gentile, Rhoden, & Klein, 1972; Hart- mann, 1974; Hartmann et al., 1980; Jones, Vaught, & Weinrott, 1977; Pfadt & Wheeler, 1995; Sideridis & Greenwood, 1996), visual inspection remains the most commonly used, and we believe the most appropriate, method for interpreting data in applied behavior analysis. We will briefly present four factors that favor visual analysis over tests of statistical significance in applied behavior analysis.

First, applied behavior analysts have little interest in knowing that a behavior change is a statistically signifi- cant outcome of an intervention. Applied behavior ana- lysts are concerned with producing socially significant behavior changes: “If a problem has been solved, you

can see that; if you must test for statistical significance, you do not have a solution” (Baer, 1977a, p. 171).

Second, visual analysis is well suited for identifying variables that produce strong, large, and reliable effects, which contribute to an effective, robust technology of behavior change. On the other hand, powerful tests of statistical analysis can detect the slightest possible correlation between the independent and dependent vari- ables, which may lead to the inclusion of weak, unreliable variables in the technology.

Two types of errors are possible when determining experimental effect (see Figure 9). A Type I error (also called a false positive) is made when the researcher concludes that the independent variable had an effect on the dependent variable, when in truth no such relation exists in nature. A Type II error (also called a false neg- ative) is the opposite of a Type I error. In this case, the re- searcher concludes that an independent variable did not have an effect on the dependent variable, when in truth it did. Ideally, a researcher using well-reasoned experi- mental tactics coupled with a sound experimental design and buttressed by appropriate methods of data analysis will conclude correctly that a functional relation between the independent and dependent variables exists (or does not exist).

Baer (1977b) pointed out that the behavior analyst’s reliance on visual inspection to determine experimental effects results in a low incidence of Type I errors but in- creases the commission of Type II errors. The researcher who relies on tests of statistical significance to determine experimental effects makes many more Type I errors than the behavior analyst, but misses few, if any, variables that might produce some effect.

Scientists who commit relatively many Type 1 errors are bound to memorize very long lists of variables that are supposed to affect diverse behaviors, some predictable portion of which are not variables at all. By contrast,

269

Planning and Evaluating Applied Behavior Analysis Research

scientists who commit very few Type 1 errors have rela- tively short lists of variables to remember. Furthermore, and much more important, it is usually only the very ro- bust, uniformly effective variables that will make their list. Those who will risk Type 1 errors more often will uncover a host of weak variables. Unquestionably, they will know more, although some of that more is wrong, and much of it is tricky. . . . Those who keep their proba- bility of Type 2 errors low do not often reject an actually functional variable, relative to those whose Type 2 error probability is higher. Again, unquestionably, the practi- tioner with the lower probability of Type 2 errors will know more; but again, the nature of that more is seen often in its weakness, inconsistency of function, or its tight specialization. . . . Individual-subject-design practi- tioners . . . necessarily fall into very low probabilities of Type 1 errors and very high probabilities of Type 2 er- rors, relative to their group-paradigm colleagues. As a result, they learn about fewer variables, but these vari- ables are typically more powerful, general, dependable, and—very important—sometimes actionable. These are exactly the variables on which a technology of behavior might be built. (Baer, 1977b, pp. 170–171)

A third problem with using statistical methods to determine the existence of functional relations in be- havioral data occurs with borderline data sets containing significant amounts of variability. Such data sets should motivate a researcher to engage in additional experi- mentation in an effort to achieve more consistent exper- imental control and to discover the factors causing the variability. The researcher who forgoes the additional experimentation in favor of accepting the results of a test of statistical significance as evidence of a functional re- lation risks leaving important findings in the realm of the unknown.

The situation where a significance test might seem help- ful is typically one involving sufficient uncontrolled variability in the dependent variable that neither the ex- perimenter nor his readers can be sure that there is an interpretable relationship. This is evidence that the relevant behavior is not under good experimental con- trol, a situation calling for more effective experimenta- tion, not a more complex judgmental aid. (Michael, 1974, p. 650)

Fourth, statistical tests of significance can be ap- plied only to data sets that conform to predetermined criteria. If statistical methods for determining experi- mental effects were to become highly valued in applied behavior analysis, researchers might begin to design ex- periments so that such tests could be computed. The re- sultant loss of flexibility in experimental design would be counterproductive to the continued development of behavior analysis (Johnston & Pennypacker, 1993b; Michael, 1974).

Social Validity

The reader of a published study in applied behavior analy- sis should judge the social significance of the target be- havior, the appropriateness of the procedures, and the social importance of the outcomes (Wolf, 1978).

Many considerations should guide the applied be- havior analyst’s selection of target behaviors. The social validity of the dependent variable should be assessed in light of those factors. Ultimately, all of the issues and considerations relative to target behavior selection point to one question: Will an increase (or decrease) in the mea- sured dimension of this behavior improve the person’s life directly or indirectly?

The independent variable should be evaluated not only in terms of its effects on the dependent variable, but also in terms of its social acceptability, complexity, prac- ticality, and cost. Regardless of their effectiveness, treat- ments that are perceived by practitioners, parents, and/or clients as unacceptable or undesirable for whatever rea- son are unlikely to be used. Consequently, such treat- ments will never have the chance to contribute to a technology of behavior change. The same can be said of independent variables that are extremely complex and thus difficult to learn, teach, and apply. Similarly, treat- ment procedures that require large amounts of time and/or money to implement have less social validity than do pro- cedures that can be applied quickly and/or inexpensively.

Even though behavior change is clearly visible on a graphic display, it may not represent a socially valid im- provement for the participant and/or significant others in his environment. In evaluating the results of an applied behavior analysis study, the reader should ask questions such as these: Is the participant (or significant others in the participant’s life) better off now that the behavior has changed? Will this new level of performance result in in- creased reinforcement (or decreased punishment) for the subject now or in the future? (Hawkins, 1984). In some instances it is relevant to ask whether the subject (or sig- nificant others) believes that her behavior has improved (Wolf, 1978).

Maintenance and Generalization of Behavior Change

Improvements in behavior are most beneficial when they are long-lasting, appear in other appropriate environ- ments, and spill over to other related behaviors. Produc- ing these kinds of effects is a major goal of applied behavior analysis. When evaluating applied behavior

270

Planning and Evaluating Applied Behavior Analysis Research

analysis research, consumers should consider the main- tenance and generalization of behavior change in their evaluation of a study. An impressive behavior change that does not last or is limited to a specialized training setting may not be socially significant. Did the researchers report the results of assessment of maintenance and general- ization through follow-up observations and measurement in nontraining environments? Better yet, if maintenance and/or generalization were not evident in such follow-up observations, did the experimenters modify their design and implement procedures in an attempt to produce and analyze the occurrence of maintenance and/or general- ization? Additionally, the reader should ask whether re- sponse generalization—changes in functionally similar but untreated behaviors concomitant with changes in the target behavior(s)—is an appropriate concern in a given study. If so, did the experimenters attempt to assess, an- alyze, or discuss this phenomenon?

External Validity

As discussed earlier in this chapter, the generality of the findings of a given experiment to other subjects, set- tings, and behaviors cannot be assessed solely on in- herent aspects of the study itself. The generality of a behavior–environment relation can be established only through the active process of systematic replication. Therefore, the reader of an applied behavior analysis study should compare the study’s results with those of other published research with which it shares relevant features. The authors of a published report identify in the paper’s introduction the experiments that they be- lieve are most relevant. To make an effective judgment of the external validity of the data from a given study, the reader must often locate previous studies in the lit- erature and compare the results of those studies with those of the current experiment.

Even though external validity should not be consid- ered a characteristic of a study per se (Birnbrauer, 1981), various features of an experiment suggest to the reader an expected, or likely, level of generality for the results. For example, an experiment that demonstrated a functional relation of similar form and degree in six subjects of dif- ferent ages, backgrounds, and current repertoires would indicate a higher probability of generality to other sub- jects than would an identical study demonstrating the same results across six subjects of the same age, back- ground, and current repertoire. Similarly, if the experi- ment was conducted in various settings and a number of different people administered the independent variable, additional confidence in the external validity of the re- sults may be warranted.

11It is important to remember that although some research in applied be- havior analysis can rightfully be criticized as superficial because it adds little to our conceptual understanding of behavior, studies in which mean- ingful target behaviors are improved to a socially valid level by the appli- cation of a socially valid treatment variable (whether a package or not) are never superficial to the participants and the significant others who share their environment.

Theoretical Significance and Conceptual Sense

A published experiment should also be evaluated in terms of its scientific merit. It is possible for a study to clearly demonstrate a functional relation between the in- dependent variable and a socially important target be- havior—and thus be judged significant from an applied perspective—yet contribute little to the advancement of the field.11 It is possible to reliably reproduce an impor- tant behavior change while at the same time not fully understand which variables are responsible for the ob- served functional relation. Sidman (1960) differentiated this kind of simple reliability from “knowledgeable re- producibility,” a more complete level of analysis in which all of the important factors have been identified and are controlled.

The Need for More Thorough Analyses of Socially Important Behavior

Even though no behavior analyst would argue the ne- cessity of systematic replication and the central role it plays in the development of an effective technology of behavior change, and even though the literature provides evidence that at least a loose form of systematic repli- cation is commonly practiced, a more critical examina- tion of the literature suggests the need for more thorough analyses of the functional relations under study. Nu- merous authors have discussed the importance of fo- cusing on the analytic side of applied behavior analysis as much as the applied side (e.g., Baer, 1991; Birnbrauer, 1979, 1981; Deitz, 1982; Hayes, 1991; Iwata, 1991; Michael, 1980; Morris, 1991; Johnston, 1991; Penny- packer, 1981). After examining the majority of the ex- perimental articles published in the first 10 volumes of the Journal of Applied Behavior Analysis (1968 to 1977), Hayes, Rincover, and Solnick (1980) concluded that a technical drift had occurred in the field away from con- ceptual analyses and toward an emphasis on client cure. They warned of a likely loss of scientific understanding as a result of focusing purely on the technical aspects of improving behavior in applied settings, and they rec- ommended an increased effort to perform more thorough analyses of behavior.

271

The importance of component analyses, parametric analyses, and other more sophisticated analytic at- tempts are often to be found less in “control” (in an im- mediately applied sense) and more in “understanding” (in a scientific sense). One may easily control, say, aggressive behavior through the use of punishment without having contributed significantly to an under- standing of aggression. . . . For example, if one has a package program that is effective, there may be little obvious value in doing a component analysis. But these more complicated analyses may increase our knowl- edge of the actual functional variables and subsequently increase our ability to generate more efficient and gen- eral behavioral programs. Perhaps, we have gone too far in our attempt to be immediately applied at the ex- pense of being ultimately more effective, in failing to encourage more analogue and analytical studies that have treatment implications. (Hayes, Rincover, & Sol- nick, 1980, pp. 282–283)

Baer, Wolf, and Risley (1987), writing in the 20th anniversary issue of the Journal of Applied Behavior Analysis, emphasized the need to shift from demonstra- tions of behavior changes—as convincing as they might be—to a more complete analysis and conceptual under- standing of the principles that underlie the successful demonstrations.

Twenty years ago, analytic meant a convincing experi- mental design, and conceptual meant relevance to a comprehensive theory about behavior. Now, applied be- havior analysis is considered an analytic discipline only when it demonstrated convincingly how to make speci- fied behavior changes and when its behavior-change methods make systematic, conceptual sense. In the past 20 years, we have sometimes demonstrated convincingly that we had changed behavior as specified, but by meth- ods that did not make systematic, conceptual sense—it was not clear why those methods had worked. Such cases let us see that we were sometimes convincingly applied and behavioral, yet even so, not sufficiently ana- lytic. (p. 318)

We agree with the need for more sophisticated, thor- ough analyses of the variables controlling socially im- portant behavior. Fortunately, examination of the recent literature reveals numerous examples of the component and parametric analyses that are necessary steps to a more complete understanding of behavior—an understanding that is prerequisite to the development of a thoroughly effective technology of behavior change. Several of the studies cited earlier in this chapter as examples of sys- tematic replication incorporated component and para- metric analyses.

The extent of a phenomenon’s generality is known only when all of the necessary and sufficient conditions

for its reproducibility have been specified. Only when all of the variables influencing a functional relation have been identified and accounted for can an analysis be con- sidered complete. Even then, the notion of a complete analysis is misleading: “Further dissection or elaboration of either variable in a functional relation inevitably re- veals fresh variability, and analysis proceeds anew. . . . the analysis of behavior can never be complete” (Penny- packer, 1981, p. 159).

Evaluation of scientific significance takes into con- sideration such things as the authors’ technological de- scription of the experiment as well as their interpretation and discussion of the results. Are the procedures de- scribed in sufficient detail so that at least the unique as- pects of the study can be replicated?12

Readers should consider the level of conceptual in- tegrity displayed in an experimental report. Does the literature review reveal a careful integration of the study with previous research? Does the literature review pro- vide sufficient justification for the study’s research questions? Are the authors’ conclusions based on the data obtained in the study? Have the authors respected the difference between basic principles of behavior and behavior change tactics? Do the authors speculate be- yond the data without making it clear that they are doing so? Do the authors suggest directions for addi- tional research to further analyze the problem studied? Is the study important for reasons other than the results actually obtained? For example, an experiment that demonstrates a new measurement technique, investi- gates a new dependent or independent variable, or in- corporates a novel tactic for controlling a confounding variable can contribute to the scientific advancement of behavior analysis, even though the study failed to achieve experimental control or produce socially sig- nificant behavior change.

Numerous criteria and considerations are involved in evaluating the “goodness” of a published study in ap- plied behavior analysis. Although each criterion is im- portant on one level or another, it is unlikely that any experiment will meet all of the criteria. And, in fact, it is unnecessary for an experiment to do so to be considered good. Nevertheless, incorporating as many of these con- siderations as possible into a study enhances its social significance and scientific value as an applied behavior analysis.

Planning and Evaluating Applied Behavior Analysis Research

12Ideally, published procedural descriptions should include sufficient detail to allow an experienced investigator to replicate the experiment. However, space limitations of most journals often prohibit such detail. The common and recommended practice in replicating published studies is to request complete experimental protocols from the original investigator(s).

272

Summary Importance of the Individual Subject in Behavioral Research

1. The focus on the behavior of individual subjects has en- abled applied behavior analysts to discover and refine ef- fective interventions for a wide range of socially significant behavior.

2. Knowing that the average performance of a group of sub- jects changed may not reveal anything about the perfor- mance of individual subjects.

3. To be most useful, a treatment must be understood at the level at which people come into contact with it and are af- fected by it: the individual level.

4. When repeated measurement reveals significant variabil- ity, the researcher should seek to identify and control the factors responsible for it.

5. Attempting to cancel out variability through statistical ma- nipulation neither eliminates its presence in the data nor controls the variables responsible for it.

6. The researcher who attributes the effects of unknown or uncontrolled variables to chance is unlikely to identify and analyze important variables.

7. To control the effects of any variable, a researcher must ei- ther hold it constant throughout the experiment or manip- ulate it as an independent variable.

8. A great strength of within-subject experimental designs is the convincing demonstration of a functional relation made possible by replication within the design itself.

9. The overall performance of a group is socially significant in many situations.

10. When group results do not represent individual perfor- mances, researchers should supplement group data with individual results.

11. When the behavior analyst cannot control access to the experimental setting or identify individual subjects, the dependent variable must consist of the responses made by individuals who enter the experimental setting.

Importance of Flexibility in Experimental Design

12. A good experimental design is any sequence and type of independent variable manipulations that produces data that effectively and convincingly address the research question(s).

13. To investigate the research question(s) of interest, an ex- perimenter must often build an experimental design that employs a combination of analytic tactics.

14. The most effective experimental designs use ongoing eval- uation of data from individual subjects as the basis for em-

ploying the three elements of baseline logic—prediction, verification, and replication.

Internal Validity: Controlling Potential Sources of Confounding in Experimental Design

15. Experiments that demonstrate a clear functional relation between the independent variable and the target behavior are said to have a high degree of internal validity.

16. The strength of an experimental design is determined by the extent to which it (a) demonstrates a reliable effect and (b) eliminates or reduces the possibility that factors other than the independent variable produced the behav- ior change.

17. The phrase control of behavior is technically inaccurate because the experimenter controls only some aspect of the subject’s environment.

18. A confounding variable is an uncontrolled factor known or suspected to have exerted influence on the depen- dent variable.

19. Steady state responding is the primary means by which applied behavior analysts assess the degree of experimen- tal control.

20. Confounding variables can be viewed as related primar- ily to one of four elements of an experiment: subject, set- ting, measurement of the dependent variable, and independent variable.

21. A placebo control is designed to separate any effects that may be produced by a subject’s expectations of improve- ment as a result of receiving treatment from the effects ac- tually produced by the treatment.

22. With a double-blind control procedure neither the sub- ject(s) nor the observers know when the independent vari- able is present or absent.

23. Treatment integrity and procedural fidelity refer to the ex- tent to which the independent variable is implemented or carried out as planned.

24. Low treatment integrity invites a major source of con- founding into an experiment, making it difficult, if not im- possible, to interpret the results with confidence.

25. One threat to treatment integrity, treatment drift, occurs when application of the independent variable during later phases of an experiment differs from the way the treat- ment was applied at the outset of the study.

26. Achieving a high level of treatment integrity begins with an operational definition of treatment procedures.

27. Treatments that are simple, precise, and brief, and that re- quire relatively little effort, are more likely to be delivered with consistency than those that are not.

Planning and Evaluating Applied Behavior Analysis Research

273

Planning and Evaluating Applied Behavior Analysis Research

28. Researchers should not assume that a person’s general competence or experience in the experimental setting, or that providing the intervention agent with detailed writ- ten instructions or a script, will ensure a high degree of treatment integrity.

29. Treatment integrity (or procedural fidelity) data measure the extent to which the actual implementation of experi- mental procedures matches their descriptions in the method section of a research report.

Social Validity: Assessing the Applied Value of Behavior Changes and the Treatments That Accomplish Them

30. The social validity of an applied behavior analysis can be assessed in three ways: the social significance of the tar- get behavior, the appropriateness of the procedures, and the social importance of the results.

31. Social validity assessments are most often accomplished by seeking consumer opinions.

32. Socially valid goals can be determined empirically by as- sessing the performance of individuals judged to be highly competent, and experimentally manipulating different lev- els of performance to determine socially valid outcomes.

33. Several scales and questionnaires for obtaining consumers’ opinions of the acceptability of behavioral interventions have been developed.

34. Methods for assessing the social validity of outcomes in- clude (a) comparing participants’ performance to the per- formance of a normative sample, (b) using a standardized assessment instrument, (c) asking consumers to rate the social validity of participants’ performance, (d) asking ex- perts to evaluate participants’ performance, and (e) test- ing participants’ newly learned level of performance in the natural environment.

External Validity: Replicating Experiments to Determine the Generality of Research Findings

35. External validity refers to the degree to which a functional relation found reliable and socially valid in a given ex- periment will hold under different conditions.

36. The proper inferences about the results of a groups-design study are from the sample to the population, not from the sample to the individual.

37. Because a groups-design experiment does not demonstrate a functional relation between the behavior of any subject and some aspect of his or her environment, the external validity of the results is moot.

38. Although groups-comparison designs and tests of statisti- cal significance are necessary and effective tools for cer-

tain types of research questions, they have contributed lit- tle to an effective technology of behavior change.

39. The generality of research findings in applied behavior analysis is assessed, established, and specified by the repli- cation of experiments.

40. In a direct replication the researcher makes every effort to duplicate exactly the conditions of an earlier experiment.

41. In a systematic replication the researcher purposefully varies one or more aspects of an earlier experiment.

42. When a systematic replication successfully reproduces the results of previous research, it not only demonstrates the reliability of the earlier findings but also adds to the ex- ternal validity of the earlier findings by showing that the same effect can be obtained under different conditions.

43. Systematic replications occur in both planned and un- planned ways through the work of many experimenters in a given area, and they result in a body of knowledge with significant scientific integrity and technological value.

Evaluating Applied Behavior Analysis Research

44. The quality and value of an applied behavior analysis study may be evaluated by seeking answers to a sequence of questions related to the internal validity, social validity, external validity, and the scientific and theoretical signif- icance of the study.

45. A Type I error occurs when a researcher concludes that the independent variable had an effect on the dependent variable when it did not. A Type II error occurs when a re- searcher concludes that the independent variable did not have an effect on the dependent variable when it did.

46. Visual analysis effectively identifies variables that pro- duce strong, large, and reliable effects, which contribute to an effective and robust technology of behavior change. Statistical analysis detects the slightest possible correla- tions between the independent and dependent variables, which may lead to the identification and inclusion of weak and unreliable variables in the technology.

47. A study can demonstrate a functional relation between the independent variable and a socially important target be- havior—and thus be significant from an applied perspec- tive—yet contribute little to the advancement of the field.

48. Only when all of the variables influencing a functional re- lation have been identified and accounted for can an analy- sis be considered complete.

49. When evaluating the scientific significance of a research report, readers should consider the technological descrip- tion of the experiment, the interpretation and discussion of the results, and the level of conceptual sense and integrity.

274

275

Positive Reinforcement

Key Terms

automatic reinforcement conditioned reinforcer generalized conditioned reinforcer positive reinforcement

positive reinforcer Premack principle reinforcer assessment

response-deprivation hypothesis stimulus preference assessment unconditioned reinforcer

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 3: Principles, Processes, and Concepts

3-3 Define and provide examples of positive (and negative) reinforcement.

3-4 Define and provide examples of conditioned and unconditioned reinforcement.

3-14 Describe and provide examples of the operant conditioning paradigm.

3-19 Define and provide examples of contingency-shaped and rule-governed behavior and distinguish between them by providing examples.

Content Area 9: Behavior Change Procedures

9-2 Use positive (and negative) reinforcement:

(a) Identify and use reinforcers.

(b) Use appropriate parameters and schedules of reinforcement.

276

In looking back, it seems to me that the most important thing I learned in graduate school was from another stu- dent, Burrhus Frederic Skinner (I called him Burrhus, others called him Fred). This man had a box, within which was a smaller box, within which he would place a hungry laboratory rat. When the animal, in its explo- rations, would depress a lever that projected from one wall, a pellet of food would be discharged into a tray be- neath the lever. Under such conditions, the rat would learn, in a matter of minutes, sometimes seconds, how to get its meal by depression of the lever. It would even keep on pressing, sometimes at a rapid rate, when pellets were delivered only now and then; and if the food sup- ply was cut off entirely, the animal would still keep working for awhile.

—Fred Keller (1982, p. 7)

Although some people still believe that findings from laboratory research on animal learning are not applicable to human behavior, by the mid-

1960s applied researchers had established the significance of positive reinforcement in education and treatment. “It is safe to say that without Skinner’s detailed laboratory analyses of reinforcement (Skinner, 1938), there would be no field of ‘applied behavior analysis’ today, least not as we know it” (Vollmer & Hackenberg, 2001, p. 241). Positive reinforcement is the most important and most widely applied principle of behavior analysis.

Fittingly, the lead article in the first issue of the Journal of Applied Behavior Analysis reported several experiments showing the effects of positive reinforce- ment on student behavior (Hall, Lund, & Jackson, 1968). Six elementary students who were disruptive or dawdled frequently participated in this classic study. The depen- dent variable, study behavior, was defined individually for each student depending on the subject matter being

taught, but it generally consisted of the student being seated and oriented toward the appropriate object or per- son (e.g., looking at course materials or the lecturing teacher) and class participation (e.g., writing the assign- ment, answering the teacher’s question). The indepen- dent variable was teacher attention, cued by an observer who held up a small square of colored paper not likely to be noticed by the target student. On this signal, the teacher attended to the child by moving to his desk, mak- ing a verbal comment, giving him a pat on the shoulder, or the like.

The effects of contingent teacher attention on the be- havior of all six students were striking. Figure 1 shows the results for Robbie, a third-grader chosen to participate because he was “a particularly disruptive student who studied very little” (p. 3). During baseline, Robbie en- gaged in study behavior for an average of 25% of the ob- served intervals. The remainder of the time he snapped rubber bands, played with objects in his pocket, talked and laughed with classmates, and played with an empty milk carton from this earlier-served drink. The majority of attention Robbie received during baseline followed these nonstudy behaviors. During baseline, Robbie’s teacher often urged him to work, told him to put his milk carton away, and told him to stop bothering his classmates.

Following baseline, the experimenters showed the teacher a graph of Robbie’s study behavior, presented the results of previous studies in which contingent adult at- tention had improved child behavior, and discussed the fundamentals of providing social reinforcement. Hall and colleagues (1968) described the procedure implemented during the two reinforcement phases as follows:

Whenever Robbie had engaged in 1 min of continuous study the observer signaled his teacher. On this cue, the

100

P er

ce n

t St

u d

y B

eh av

io r

5 10 15 20 25 30 35 Sessions

Baseline Reinforcement1 Reversal Reinforcement2 Postchecks

Robbie

Figure 1 Percentage of intervals of study behavior by a third-grade student during baseline and reinforcement conditions. The arrow to the first postcheck data point shows when cueing the teacher to provide attention was discontinued. From “Effects of Teacher Attention on Study Behavior” by R. V. Hall, D. Lund, and D. Jackson, 1968, Journal of Applied Behavior Analysis, 1, p. 3. Copyright 1968 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Positive Reinforcement

277

teacher approached Robbie, saying, “Very good work Robbie,” “I see you are studying,” or some similar re- mark. She discontinued giving attention for nonstudy behaviors including those which were disruptive to the class. (p. 4)

During Reinforcement 1, Robbie’s study behavior increased to a mean of 71%. When a reversal to baseline conditions was introduced, his study behavior decreased to a mean of 50%; but when Robbie’s teacher again pro- vided attention for study behavior (Reinforcement 2), his study behavior recovered and stabilized at a level rang- ing between 70 and 80% of the observed intervals. Re- sults of follow-up observations over a 14-week period after signaling the teacher had been discontinued showed that Robbie’s study behavior had maintained at 79%. The teacher reported positive behavior changes associated with Robbie’s increased study behavior. By the final week of Reinforcement 2, Robbie completed his spelling as- signments more consistently, his disruptive behavior had diminished, and he continued to study while drinking his milk and did not play with the carton afterwards.

The intervention used by Robbie’s teacher to help him be more successful in the classroom was based on the principle of positive reinforcement. In this chapter we examine the definition and nature of positive reinforce- ment, describe methods for identifying potential rein- forcers and assessing their effects, outline experimental control techniques for verifying whether a positive re- inforcement contingency is responsible for increased responding, and offer guidelines for using positive rein- forcement effectively.

Definition and Nature of Positive Reinforcement The principle of reinforcement is deceptively simple. “The basic operant functional relation for reinforcement is the following: When a type of behavior (R) is followed by reinforcement (SR) there will be an increased future frequency of that type of behavior” (Michael, 2004, p. 30).1 However, as Michael and other authors have pointed out, three qualifications must be considered regarding the conditions under which the effects reinforcement will occur. These qualifications are (a) the delay between the response and onset of the consequence, (b) the stimulus conditions in effect when the response was emitted, and

1Various terms, such as strengthening the behavior or increasing the like- lihood of future responding, are sometimes used by behavior analysts to de- scribe the basic effect of reinforcement. Although such terms appear occasionally in this book, recognizing Michael’s (1995) concern that use of such terms “encourages a language of intervening variables, or an im- plied reference to something other than an observable aspect of behavior” (p. 274), we most often use increased future frequency to refer to the pri- mary effect of reinforcement.

2When the consequence that produced the increase in responding is best de- scribed as the termination or withdrawal of an already present stimulus, negative reinforcement has occurred. The fundamental nature and quali- fying conditions for positive reinforcement and negative reinforcement are the same. Negative reinforcement is examined in detail in chapter entitled “Negative Reinforcement.”

(c) the strength of the current motivation with respect to the consequence. In this section we examine these qual- ifications and several other concepts requisite to acquir- ing a full understanding of how reinforcement “works.”

Operation and Defining Effect of Positive Reinforcement

Positive reinforcement has occurred when a response is followed immediately by the presentation of a stimulus and, as a result, similar responses occur more frequently in the future. Figure 2 illustrates the two-term contin- gency—a response followed closely in time by the pre- sentation of a stimulus—and the effect on future responding that define positive reinforcement. This two- term contingency is the fundamental building block for the selection of all operant behavior (Glenn, Ellis, & Greenspoon, 1992).

The stimulus presented as a consequence and re- sponsible for the subsequent increase in responding is called a positive reinforcer, or, more simply, a reinforcer. Teacher attention in the form of positive comments was the reinforcer that increased Robbie’s study behavior. Cold water flowing into a cup and the sight of a colorful bird are the reinforcers for the two behaviors shown in Figure 2.

It is important to remember that a reinforcer does not (and cannot) affect the response that it follows. Rein- forcement only increases the frequency with which sim- ilar responses are emitted in the future.

It is not correct to say the operant reinforcement “strengthens the response which precedes it.” The re- sponse has already occurred and cannot be changed. What is changed is the future probability of responses in the same class. It is the operant as a class of behavior, rather than the response as a particular instance, which is conditioned. (Skinner, 1953, p. 87)

Skinner (1966) used rate of responding as the funda- mental datum for his research on reinforcement. To strengthen an operant is to make it occur more frequently.2

However, rate (or frequency) is not the only dimension of behavior selected, shaped, and maintained by reinforce- ment. Reinforcement can also strengthen the duration, la- tency, magnitude, and/or topography of behavior. For example, if reinforcement only follows those responses that fall within a range of magnitude—that is, above a min- imum force but below a maximum force—and responses

Positive Reinforcement

278

R SR+

Hold cup under tap and push lever

Cold water flows into cup

Turn head and look left See colorful bird

Effect on Future Frequency of Similar Responses

Figure 2 Two-term contin- gency illustrating positive reinforce- ment: A response (R) is followed closely in time by a stimulus change (SR+) that results in an increased frequency of similar responses in the future.

outside of that range of magnitudes are not followed by re- inforcement, the effect will be a higher frequency of re- sponses within that range. Reinforcement contingent on responses meeting multiple criteria will strengthen a sub- set of responses meeting those criteria (e.g., responses by a golfer practicing 10-foot putts must fall within a narrow range of force and form to be successful).

Importance of Immediacy of Reinforcement

Emphasizing the importance of the immediacy of rein- forcement is essential. The direct effects of reinforce- ment involve “temporal relations between behavior and its consequences that are on the order of a few seconds” (Michael, 2004, p. 161). Research with nonhumans sug- gests that at one end of the continuum as much as 30 sec- onds can elapse without critical loss of effect (e.g., Byrne, LeSage, & Poling, 1997; Critchfield & Lattal, 1993; Wilkenfeld, Nickel, Blakely, & Poling, 1992). However, a response-to-reinforcement delay of 1 second will be less effective than a 0-second delay. This is because be- haviors other than the target behavior occur during the delay; the behavior temporally closest to the presenta- tion of the reinforcer will be strengthened by its presen- tation. As Sidman (1960) described, “If the reinforcer does not immediately follow the response that was re- quired for its production, then it will follow some other behavior. Its major effect will then be upon the behavior that bears, adventitiously to be sure, the closest prior tem- poral relationship to the reinforcement” (p. 371).

Malott and Trojan Suarez (2004) discussed the im- portance of immediacy as follows:

If the reinforcer is to reinforce a particular response, it must immediately follow that response. But how imme- diate is immediate? We don’t have any experimental

data on this one for human beings, but the research on nonverbal animals suggests that a minute or two pushes the limit (even 30 seconds is hard). And if you talk to most behavior analysts working with nonverbal children, they’d agree. They’d quit their jobs if they had to wait 60 seconds before delivering each reinforcer to their children. Such a delay is a good way to ensure that no learning would occur, even with people—at least no de- sirable learning.

So, if you’re trying to reinforce a response, don’t push that 60-second limit. Push the other end—the 0-second end. The direct effect of reinforcement drops off quickly as you increase the delay, even to 3 or 4 sec- onds. And even a 1-second delay may reinforce the wrong behavior. If you ask a young child to look at you and deliver the reinforcer 1 second after the response, you’re liable to reinforce looking in the wrong direction. So one problem with delayed reinforcement is that it re- inforces the wrong response—the one that occurred just before the delivery of the reinforcer. (p. 6)

A common misconception is that delayed conse- quences can reinforce behavior, even if the consequences occur days, weeks, or even years after the responses oc- curred. “When human behavior is apparently affected by long-delayed consequences, the change is accomplished by virtue of the human’s complex social and verbal his- tory, and should not be thought of as an instance of the simple strengthening of behavior by reinforcement” (Michael, 2004, p. 36).

For example, suppose that a piano student practiced dutifully every day for several months in preparation for a statewide competition, at which she received a first- place award for her solo piano performance. Although some might believe that the award reinforced her persis- tent daily practice, they would be mistaken. Delayed con- sequences do not reinforce behavior directly. Delayed consequences can, when combined with language, influence future behavior through instructional control

Positive Reinforcement

279

and rule following. A rule is a verbal description of a be- havioral contingency (e.g., “Turnip seeds planted by Au- gust 15 will yield a crop before a killing freeze”). Learning to follow rules is one way that a person’s be- havior can come under the control of consequences that are too delayed to influence behavior directly. A state- ment by the piano teacher such as “If you practice your assignments every day for one hour between now and the competition, you could win first place” could have func- tioned as a rule that influenced the piano student’s daily practice. The student’s daily practice was rule governed if daily practice occurred because of her teacher’s rule.3

The following conditions provide strong indicators that behavior is the result of instructional control or rule fol- lowing rather than a direct effect of reinforcement (Mal- ott, 1988, Michael, 2004).

• No immediate consequence for the behavior is apparent.

• The response–consequence delay is greater than 30 seconds.

• Behavior changes without reinforcement.

• A large increase in the frequency of the behavior occurs following one instance of reinforcement.

• No consequence for the behavior exists, including no automatic reinforcement, but the rule exists.

Reinforcement Is Not a Circular Concept

A commonly held misconception is that reinforcement is the product of circular reasoning and therefore con- tributes nothing to our understanding of behavior. Cir- cular reasoning is a form of faulty logic in which the name used to describe an observed effect is mistaken as the cause for the phenomenon. This confusion of cause and effect is circular because the observed effect is the sole basis for identifying the presumed cause. In circular reasoning, the suspected cause is not independent of its effect—they are one and the same.

Here is an example of circular reasoning that occurs often in education. A student’s persistent difficulties in learning to read (effect) leads to a diagnosis of a learning disability, which is then offered as an explanation for the reading problem: “Paul’s reading problem is due to his learning disability.” How do you know Paul has a learn- ing disability? Because he hasn’t learned to read. Why

3Excellent discussions of rule-governed behavior can be found in Baum (1994); Chase and Danforth (1991); Hayes (1989); Hayes, Zettle, and Rosenfarb (1989); Malott and Garcia (1991); Malott and Trojan Suarez (2004); Reitman and Gross (1996); and Vaughan (1989).

hasn’t Paul learned to read? Because his learning dis- ability has prevented him from learning to read. And around and around it goes.

Similarly, it would be circular reasoning if we said that teacher attention increased Robbie’s study behav- ior because it is a reinforcer. However, the correct use is to say that because Robbie’s study behavior increased when (and only when) it was followed immediately by teacher attention, teacher attention is a reinforcer. The difference is more than the direction of the relation, or a semantic sleight-of-hand. In circular reasoning the sus- pected cause is not manipulated as an independent vari- able to see whether it affects the behavior. In circular reasoning such experimental manipulation is impossi- ble because the cause and effect are the same. Paul’s learning disability cannot be manipulated as an inde- pendent variable because, as we used the concept in this example, it is nothing more than another name for the dependent variable (effect).

Reinforcement is not a circular concept because the two components of the response–consequence relation can be separated, allowing the delivery of a consequence to be manipulated to determine whether it increases the frequency of the behavior it follows. Epstein (1982) de- scribed it as follows:

If we can show that a response increases in frequency because (and only because) it is followed by a particular stimulus, we call that stimulus a reinforcer and its pre- sentation, reinforcement. Note the lack of circularity. Reinforcement is a term we invoke when we observe certain relations between events in the world. . . . [However,] If we say, for example, that a particular stimulus strengthens a response behavior because it is a reinforcer, we are using the term reinforcer in a circu- lar fashion. It is because it strengthens behavior that we call the stimulus a reinforcer. (p. 4)

Epstein (1982) went on to explain the difference be- tween using an empirically demonstrated principle such as reinforcement in a theoretical account of behavior and using a circular argument.

In some of his writings, Skinner speculates that certain behavior (for example, verbal behavior) has come about through reinforcement. He may suggest, for example, that certain behavior is strong because it was reinforced. This use of the concept is not circular, only speculative or interpretive. Using the language of reinforcement in this way is reasonable when you have accumulated a large data base. . . . When Skinner attributes some every- day behavior to past reinforcers, he is making a plausi- ble guess based on a large data base and principles of behavior established under controlled conditions. (p. 4)

Used properly, reinforcement describes an empiri- cally demonstrated (or speculative, in a theoretical or con- ceptual analysis) functional relation between a stimulus

Positive Reinforcement

280

Table 1 The Vocabulary of Reinforcement*

Term Restrictions Examples

reinforcer (noun) A stimulus Food pellets were used as reinforcers for the rat’s lever presses.

reinforcing (adjective) A property of a stimulus The reinforcing stimulus was produced more often than the other, nonreinforcing stimuli.

reinforcement (noun) As an operation, the delivery of consequences The fixed-ratio schedule of reinforcement delivered when a response occurs food after every tenth key peck.

As a process, the increase in responding that The experiment with monkeys demonstrated rein- results from the reinforcement forcement produced by social consequences.

to reinforce (verb) As on operation, to deliver consequences when a When a period of free play was used to reinforce response occurs; responses are reinforced and not the child’s completion of school work, the child’s organisms grades improved.

As a process, to increase responding through the The experiment was designed to find out whether reinforcement operation gold stars would reinforce cooperative play among

first-graders.

*This vocabulary is appropriate if and only if three conditions exist: (1) A response produces consequences; (2) that type of response occurs more often when it produces those consequences than when it does not produce them; and (3) the increased responding occurs because the response has those consequences. A parallel vocabulary is appropriate to punishment (including punisher as a stimulus and punish as a verb), with the difference being that a punishing consequence makes responding occur less rather than more often. From Learning, Interim (4th ed.) by A. C. Catania, 2007, p. 69. Cornwall-on-Hudson, NY: SLoan Publishing.

change (consequence) immediately following a response and an increase in the future frequency of similar re- sponses. Table 1 shows restrictions and examples of appropriate use of the terms reinforcer, reinforcing, re- inforcement, and to reinforce suggested by Catania (1998). Box 1 describes four mistakes commonly made when speaking and writing about reinforcement.

Reinforcement Makes Antecedent Stimulus Conditions Relevant

Reinforcement does more than increase the future fre- quency of behavior it follows; it also changes the func- tion of stimuli that immediately precede the reinforced behavior. By virtue of being temporally paired with the response-reinforcer contingency, certain antecedent events acquire the ability to evoke (make more likely) in- stances of the reinforced response class. A discriminative stimulus (SD, pronounced “ess-dee”) is an antecedent stimulus correlated with the availability of reinforcement for a particular response class. Responding in the pres- ence of the SD produces reinforcement, and responding in the SD’s absence (a condition called stimulus delta [S , pronounced “ess-delta”]) does not. As a result of this his- tory of reinforcement, a person learns to make more re- sponses in the presence of the SD than in its absence. The behavior is then considered to be under stimulus control.

With the addition of the SD, the two-term contin- gency for reinforcement is transformed to the three-term

contingency of the discriminated operant. Figure 3 shows examples of three-term contingencies for positive reinforcement. Assuming that cold water is currently re- inforcing and the person has a history of receiving cold water only under blue taps, he is more likely to hold his cup under the blue tap on the cooler (than, say, a red tap). Similarly, assuming that seeing a colorful bird is cur- rently reinforcing and a person has a history of seeing birds more often when looking toward chirping sounds (than, say, other sounds or silence), turning one’s head and looking to the left will occur at a higher frequency when the chirping sound is heard.

Reinforcement Depends on Motivation

The phrase assuming that cold water is currently rein- forcing in the previous paragraph holds another key to understanding reinforcement. Although reinforcement is commonly thought of as a way of motivating people— and it can be—the momentary effectiveness of any stim- ulus change as reinforcement depends on an existing level of motivation with respect to the stimulus change in ques- tion. Motivating operations alter the current value of stim- ulus changes as reinforcement (Michael 2004).

Motivating operations (MOs) are environmental vari- ables that have two effects on behavior: (1) They alter the operant reinforcing effectiveness of some specific stimuli, objects, or events (the value-altering effect);

Positive Reinforcement

∆

281

A standard set of technical terms is prerequisite to the meaningful description of any scientific activity. Effec- tively communicating the design, implementation, and outcomes of an applied behavior analysis depends on the accurate use of the discipline’s technical language. The language of reinforcement includes some of the most im- portant elements of the behavior analyst’s vocabulary.

In this box we identify four mistakes made fre- quently by students of applied behavior analysis when describing reinforcement-based interventions. Perhaps the most common mistake—confusing negative rein- forcement with punishment—is not discussed here.

Reinforcing the Person

Although it is proper to speak of presenting a reinforcer to a learner (e.g., “The teacher gave a token to Bobby each time he asked a question”), statements such as, “The teacher reinforced Bobby when he asked a question” and “Chloe was reinforced with praise each time she spelled a word correctly” are incorrect. Behaviors are reinforced, not people. Bobby’s teacher reinforced question asking, not Bobby. Of course, reinforcement acts on and affects the overall person, in that it strengthens behaviors within the person’s repertoire. However, the procedural focus and the primary effect of reinforcement are on the be- haviors that it follows.

Practice As Reinforcement for a Skill

Educators will sometimes say that students should prac- tice a skill because “practicing reinforces the skill.” The phrase poses no problem if the speaker is describing a common outcome of practice with the everyday language connotation of reinforce, as in “to make something stronger” (e.g., to reinforce concrete by embedding steel rods in it). Well-designed drill and practice on a skill usu- ally yields stronger performance in the form of better re- tention, reduced latency, higher response rates, and/or increased endurance (e.g., Johnson & Layng, 1994; Swanson & Sachse-Lee, 2000). Unfortunately, a phrase such as “practicing reinforces the skill” is often misused and misinterpreted as technical usage of the language of operant conditioning.

Although a skill that has been practiced is often stronger as a result of the practice, the practice itself could not be a reinforcer for the behavior practiced.

Practice refers to the form and manner in which the tar- get skill is emitted (e.g., answering as many math prob- lems as you can in 1 minute). Practicing is a behavior that could be reinforced with various consequences such as a preferred activity (e.g., “Practice solving these math problems; then you can have 10 minutes of free time”). Depending on a learner’s history and preferences, the opportunity to practice a certain skill may function as a reinforcer for practicing another skill (e.g., “Finish your math problems; then you’ll get to do 10 minutes of re- peated reading practice”).

Artificial Reinforcement

A distinction between natural and artificial reinforcers is made sometimes, as in this statement, “As the students’ success rates improved, we gradually stopped using ar- tificial reinforcers, such as stickers and trinkets, and in- creased the use of natural reinforcers.” Some authors have suggested that applications of the principles of behavior re- sult in “artificial control” (e.g., Smith, 1992). A behavior– consequence contingency may be effective or ineffective as reinforcement, but none of its elements (the behavior, the consequence, or the resultant behavior change) is, or can be, artificial.

The reinforcement contingencies and stimuli used as reinforcers in any behavior change program are al- ways contrived—otherwise there would be no need for the program—but they are never artificial (Skinner, 1982). The meaningful distinction when talking about reinforcement contingencies is not between the natural and the artificial, but between contingencies that already exist in a given setting prior to a behavior change pro- gram and contingencies that are contrived as part of the program (Kimball & Heward, 1993). Although the ulti- mate effectiveness of a behavior change program may depend on shifting control from contrived to naturally occurring contingencies, there is no such thing as artifi- cial reinforcement.

Reinforcement and Feedback As Synonyms

Some speakers and writers mistakenly use reinforcement and feedback interchangeably. The two terms refer to dif- ferent operations and outcomes, though some of each term encompasses parts of the other term’s meaning. Feedback is information a person receives about a particular aspect of his or her behavior following its completion (e.g., “Very good, Kathy. Two quarters equal 50 cents.”). Feedback is

Box 1 Common Mistakes in Talking and Writing

about Reinforcement

Positive Reinforcement

282

most often provided in the form of verbal descriptions of performance, but it can also be provided by other means such as vibration or lights (e.g., Greene, Bailey, & Barber, 1981). Because feedback is a consequence that often re- sults in the increased future frequency of behavior, it sometimes leads to the faulty assumption that reinforce- ment must involve feedback or that reinforcement is just a behaviorist’s term for feedback.

Reinforcement always increases the future frequency of responding. Feedback may result in (a) an increase in the future frequency of the student’s performance as a reinforcement effect and/or as a prompt or instruction on how to respond next time (e.g. “You’re handwriting is improving, Jason, but don’t forget to cross your Ts”), and/or (b) a reduction in the frequency of some aspect of the learner’s performance as a function of punishment or instruction (e.g., “You dropped your elbow on that pitch. Don’t do that.”). Feedback may have multiple ef- fects, increasing one aspect of performance and de- creasing another. Feedback may also have no effect on future responding whatsoever.

Reinforcement is defined functionally by its effect on future responding; feedback is defined by its formal characteristics (information about some aspect of per- formance). The operation of either concept is neither nec- essary nor sufficient for the other. That is, reinforcement may occur in the absence of feedback, and feedback may occur without a reinforcement effect.

Sometimes Commonsense Language Is Better

The technical language of behavior analysis is complex, and mastering it is no simple matter. Beginning students of behavior analysis are not the only ones who commit

terminology errors. Well-trained practitioners, established researchers, and experienced authors also make mistakes now and then when speaking and writing about behavior analysis. Using behavioral concepts and principles—such as positive reinforcement—to confidently explain com- plex situations involving multiple processes and uncon- trolled and unknown variables is a mistake that catches the most attentive and conscientious of behavior analysts at times.

Instead of invoking the terminology and concepts of reinforcement to explain the influence of temporally distant consequences on behavior, it is probably wiser to follow Jack Michael’s (2004) advice and simply use everyday descriptive language and commonsense relations.

Incorrectly used technical language is worse than com- mon sense language because it suggests that the situa- tion is well understood, and it may displace serious attempts at further analysis. Until we are able to provide an accurate analysis of the various processes relevant to indirect effects [of reinforcement], we are better off using ordinary descriptive language. Thus, say “the successful grant application is likely to encourage future efforts in the same direct,” but don’t say it as though you had the science of behavior behind you. Stop referring to suc- cessful settlements of a labor dispute as reinforcement for striking, and successful election of a political candi- date as reinforcement for political activity. . . . Don’t talk about good grades as reinforcement for effective study behavior, although they are no doubt responsible for maintaining it in some cases. Just say that they’re re- sponsible for maintaining it. Restraint of this sort will deprive some of us of an opportunity to (incorrectly) dis- play our technical knowledge, but so much the better. (p. 165, emphasis in original)

and (2) They alter the momentary frequency of all be- havior that has been reinforced by those stimuli, ob- jects, or events (the behavior-altering effect). The value-altering effect, like response-reinforcement delay, is relevant to the effectiveness of the reinforcer at the time of conditioning, and stating that the conse- quence is a form of reinforcement implies that a rele- vant MO is in effect and at sufficient strength. (p. 31)

In other words, for a stimulus change to “work” as re- inforcement at any given time, the learner must already want it. This is a critical qualification in terms of the en- vironmental conditions under which the effects of rein- forcement will be seen. Michael (2004) explained this qualification as follows:

The behavior-altering effect is relevant to the increased future frequency of the reinforced behavior, and must be

added as a third qualification to the operant reinforce- ment relation: In a given stimulus situation (S) when a type of behavior (R) is followed immediately by rein- forcement (SR) there will be an increase in the future fre- quency of that type of behavior in the same or similar stimulus conditions, but the increased frequency will only be seen when the MO relevant to the reinforcement that was used is again in effect. (p. 31, emphasis in original)

Motivating operations take two forms. An MO that increases the current effectiveness of a reinforcer is called an establishing operation (EO) (e.g., food deprivation makes food more effective as a reinforcer); an MO that decreases the current effectiveness of a reinforcer is an abolishing operation (AO) (e.g., food ingestion reduces the effectiveness of food as a reinforcer).4

4Motivating operations are described in detail in chapter entitled “Moti- vating Operations.”

Positive Reinforcement

283

SR+

Cold water flows into cup

See colorful bird

Effect on Future Frequency of Similar Responses

in Presence of SD SD R

Blue tap on water cooler Hold cup under tap and

push lever

Chirping sound to left Turn head and look left

Figure 3 Three-term contingency illustrating positive reinforcement of a discriminated operant: A re- sponse (R) emitted in the presence of a discriminative stimulus (SD) is followed closely in time by a stimulus change (SR+) and results in an increased frequency of similar responses in the future when the SD is present. A discriminated operant is the product of a conditioning history in which responses in the presence of the SD

produced reinforcement while similar responses in the absence of the SD (a condition called stimulus delta [S�]) have not been reinforced (or resulted in a reduced amount or quality of reinforcement than in the SD

condition).

Adding the establishing operation (EO) to a dis- criminated operant results in a four-term contingency as shown in Figure 4. Spending several hours in a hot and stuffy room without water is an EO that (a) makes water more effective as a reinforcer and (b) in- creases the momentary frequency of all behaviors that have produced water in the past. Similarly, a park ranger stating prior to a hike that any hiker who de- scribes the coloring of the bird that makes a certain chirping sound will receive a $5 token for the gift shop is an EO that will (a) make seeing a bird that makes the chirping sound effective as reinforcement and (b) increase the frequency of all behaviors (e.g., turning one’s head and looking around) that have produced sim- ilar consequences (in this case, seeing the source of sounds) in the past.

In plain English, establishing operations (EOs) de- termine what an individual wants at any particular mo- ment. EOs are dynamic, always changing. The reinforcer value (the want) goes up with increasing lev- els of deprivation and goes down with levels of satiation. Vollmer and Iwata (1991) demonstrated how the rein- forcing effectiveness of three classes of stimuli—food, music, and social attention—varied under conditions of deprivation and satiation. Participants were five adults with developmental disabilities, and the dependent vari- able was the number of responses per minute on two motor tasks—pressing a switch or picking up small blocks from a container and putting them through the

hole in the top of another container. All sessions lasted 10 minutes and began with the experimenter saying, “Do this, [participant’s name],” and modeling the re- sponse. During baseline, participants’ responses re- ceived no programmed consequences. During the deprivation and satiation conditions, responses were fol- lowed by presentation of either food, music, or social at- tention. Initially each response was followed by the programmed consequence; this gradually shifted to every third, fifth, or tenth response being followed by the consequence.

Different procedures were used to create depriva- tion and satiation conditions for each stimulus class. With food, for example, baseline and deprivation con- dition sessions were conducted 30 minutes prior to a participant’s scheduled lunchtime; sessions during the satiation condition were conducted within 15 min- utes after the participant had eaten lunch. For social attention, baseline and deprivation condition sessions were conducted immediately following a 15-minute period in which the participant had either been alone or had been observed to have had no social inter- action with another person. Immediately prior to each session in the satiation condition, the experimenter pro- vided continuous social interaction (e.g., played a simple game, conversation) with the participant for 15 minutes.

All five participants responded at higher rates under the deprivation condition than during the satiation

Positive Reinforcement

284

SR+

Cold water flows into cup

See colorful bird

Effect on Future Frequency of Similar Responses

under Similar Conditions in the Future

SD R

Blue tap on water cooler

Hold cup under tap and push lever

Chirping sound to left

Turn head and look left

Water deprived for 2 hours in hot, stuffy room

Park ranger says, “After our hike, anyone who describes

the bird that makes this chirping sound will receive a $5 token for the gift shop.”

Figure 4 Four-term contingency illustrating positive reinforcement of a discriminated operant made current by a motivating operation: An establishing operation (EO) increases the momentary effectiveness of a stimulus change as a reinforcer, which in turn makes the SD more likely to evoke behavior that has been reinforced by that stimulus change in the past.

condition. Figure 5 shows the effects of deprivation and satiation of social attention on the effectiveness of social attention as a reinforcer for two of the study’s participants, Donny and Sam. Other researchers have re- ported similar findings concerning the effects of depri-

vation and satiation of various stimuli and events as mo- tivating operations that affect the relative effectiveness of reinforcement (e.g., Gewirtz & Baer, 1958; Klatt, Sher- man, & Sheldon, 2000; North & Iwata, 2005; Zhou, Iwata, & Shore, 2002).

Baseline Deprivation Satiation Deprivation

Donny

Baseline Deprivation Satiation Deprivation

Sam

Satiation

10 20 30 40

Sessions

R es

p o

n se

s p

er M

in u

Figure 5 Responses per minute by two students during baseline and when social attention was used as rein- forcement under deprivation and satiation conditions for social attention. From “Establishing Operations and Reinforcement Effects” by T. R. Vollmer and B. A. Iwata, 1997, Journal of Applied Behavior Analysis, 24, p. 288. Copyright 1991 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Positive Reinforcement

285

Automaticity of Reinforcement A reinforcing connection need not be obvious to the in- dividual reinforced.

—B. F. Skinner (1953, p. 75)

The fact that a person does not have to understand or ver- balize the relation between his actions and a reinforcing consequence, or for that matter even be aware that a con- sequence has occurred, for reinforcement to occur is known as the automaticity of reinforcement. Skinner (1983) provided an interesting example of automaticity in the third and final volume of his autobiography, A Mat- ter of Consequences. He described an incident that took place at a meeting of distinguished scholars who had been invited to discuss the role of intention in political activ- ity. At one point during the meeting, the psychologist Erich Fromm began to argue that “people were not pi- geons,” perhaps implying that an operant analysis based on positive reinforcement could not explain human be- havior, which is the product of thought and free will. Skinner recounted what happened next:

I decided that something had to be done. On a scrap of paper I wrote, “Watch Fromm’s left hand. I am going to shape [reinforce by successive approximations] a chop- ping motion” and passed it down the table to Halleck [a member of the group]. Fromm was sitting directly across the table and speaking mainly to me. I turned my chair slightly so that I could see him out of the corner of my eye. He gesticulated a great deal as he talked, and whenever his left hand came up, I looked straight at him. If he brought the hand down, I nodded and smiled. Within five minutes he was chopping the air so vigor- ously that his wristwatch kept slipping out over his hand. (p. 150–151, words in brackets added)

Arbitrariness of the Behavior Selected

So far as the organism is concerned, the only important property of the contingency is temporal.

—B. F. Skinner (1953, p. 85)

No logical or adaptive connection between behavior and a reinforcing consequence is necessary for reinforcement to occur. In other words, reinforcement will strengthen any behavior that immediately precedes it. This arbitrary nature of the behavior selected is critical to understand- ing reinforcement. All other relations (e.g., what’s logi- cal, desirable, useful, appropriate) must compete with the critical temporal relation between behavior and conse- quence. “To say that reinforcement is contingent upon a response may mean nothing more than that it followed the response . . . conditioning takes place presumably be-

5It is a mistake to assume that all superstitious behavior is the direct result of adventitious reinforcement. Many superstitious behaviors are probably the result of following cultural practices. For example, high school base- ball players may wear their caps inside out and backwards when a late- inning rally is needed because they have seen major leaguers don such “rally caps” in the same situation.

cause of the temporal relation only, expressed in terms of the order and proximity of response and reinforce- ment” (Skinner, 1948, p. 168).

Skinner (1948) demonstrated the arbitrary nature of the behaviors selected by reinforcement in one of his most famous experimental papers, “‘Superstition’ in the Pigeon.” He gave pigeons a small amount of food every 15 seconds, “with no reference whatsoever to the bird’s behavior” (p. 168). The fact that reinforcement will strengthen whatever behavior it immediately follows was soon evident. Six of the eight birds developed idiosyn- cratic behaviors “so clearly defined, that two observers could agree perfectly in counting instances” (p. 168). One bird walked counterclockwise around the cage; another repeatedly thrust its head into one of the upper corners of the cage. Two birds acquired a “pendulum motion of the head and body, in which the head was extended forward and swung from right to left with a sharp movement fol- lowed by a somewhat slower return” (p. 169). The pi- geons had exhibited none of those behaviors at “any noticeable strength” during adaptation to the cage or be- fore the food was periodically presented.

Whatever behavior the pigeons happened to be exe- cuting when the food hopper appeared tended to be re- peated, which made it more likely to be occurring when food appeared the next time. That is, reinforcement was not contingent (in the sense of, dependent) on the be- havior; it was only a coincidence that reinforcement sometimes followed the behavior. Such accidentally re- inforced behavior is called “superstitious” because it has no influence on whether reinforcement follows. Humans engage in many superstitious behaviors. Sports provide countless examples: A basketball player tugs on his shorts before shooting a foul shot, a golfer carries his lucky ball marker, a batter goes through the same sequence of ad- justing his wristbands before each pitch, a college foot- ball fan wears a goofy-looking necklace made of inedible nuts to bring good luck to his team.5

The importance of understanding the arbitrariness of reinforcement goes far beyond providing a possible ex- planation for the development of harmless superstitious and idiosyncratic behaviors. The arbitrary nature of se- lection by reinforcement may explain the acquisition and maintenance of many maladaptive and challenging be- haviors. For example, a caregiver’s well-meaning social attention provided in an attempt to console or divert a person who is hurting himself may help shape and main- tain the very behavior the caregiver is trying to prevent or

Positive Reinforcement

286

eliminate. Kahng, Iwata, Thompson, and Hanley (2000) documented with a functional analysis that social rein- forcement maintained the self-injurious behavior (SIB) and aggression of three adults with developmental dis- abilities. Kahng and colleagues’ data support the hy- pothesis that aberrant behaviors may have been selected and maintained by social attention because of the arbi- trariness of reinforcement.

Automatic Reinforcement

Some behaviors produce their own reinforcement inde- pendent of the mediation of others. For example, scratch- ing an insect bite relieves the itch. Behavior analysts use the term automatic reinforcement to identify a behavior– reinforcement relation that occurs without the presenta- tion of consequences by other people (Vaughan & Michael, 1982; Vollmer, 1994, 2006). Automatic rein- forcement occurs independent of social mediation by oth- ers. Response products that function as automatic reinforcement are often in the form of a naturally pro- duced sensory consequence that “sounds good, looks good, tastes good, smells good, feels good to the touch, or the movement itself is good” (Rincover, 1981, p. 1).

Persistent, nonpurposeful, repetitive self-stimulatory behaviors (e.g., flipping fingers, head rolling, body rock- ing, toe walking, hair pulling, fondling body parts) may produce sensory stimuli that function as automatic rein- forcement. Such “self-stimulation” is thought to be a fac- tor in the maintenance of self-injurious behavior (Iwata, Dorsey, Slifer, Bauman, & Richman, 1994), stereotypic repetitive movements, and “nervous habits” such as hair pulling (Rapp, Miltenberger, Galensky, Ellingson, & Long, 1999), nail biting, chewing on the mouth or lips, and object manipulation such as continually twirling a pencil or fondling jewelry (Miltenberger, Fuqua, & Woods, 1998).

The response product that functions as automatic re- inforcement may be an unconditioned reinforcer or a once neutral stimulus that, because it has been paired with other forms of reinforcement, has become a condi- tioned reinforcer. Sundberg, Michael, Partington, and Sundberg (1996) described a two-stage conditioning his- tory that may account for this type of conditioned auto- matic reinforcement.

For example, a person may persist in singing or hum- ming a song while coming home for a movie despite no obvious direct reinforcement for singing. In order for this behavior to occur as automatically reinforced be- havior, a special two-stage conditioning history is neces- sary. In stage one, some stimulus (e.g., a song) must be paired with an existing form of conditioned or uncondi- tioned reinforcement (e.g., an enjoyable movie, popcorn,

relaxation). As a result, the new stimulus can become a form of conditioned reinforcement (e.g., hearing the song may now be a new form of conditioned reinforce- ment). In stage two, the emission of a response (for whatever reason) produces a response product (i.e., the auditory stimuli produced by singing the song) that has topographical similarity to that previously neutral stimu- lus (e.g., the song), and may now have self-strengthening properties. (pp. 22–23)

Several theorists have suggested that automatic re- inforcement may help to explain the extensive babbling of infants and how babbling shifts naturally, without ap- parent intervention from others, from undifferentiated vocalizations to the speech sounds of their native lan- guage (e.g., Bijou & Baer, 1965; Mowrer, 1950; Skin- ner, 1957; Staats & Staats, 1963; Vaughan & Michael, 1982). Caregivers frequently talk and sing while hold- ing, feeding, and bathing a baby. As a result of repeated pairing with various reinforcers (e.g., food, warmth), the sounds of a caregiver’s voice may become conditioned reinforcers for the baby. The baby’s babbling is auto- matically reinforced when it produces sounds that match or closely approximate the caregiver’s. At that point, “The young child alone in the nursery may automatically re- inforce his own exploratory vocal behavior when he pro- duces sounds that he has heard in the speech of others” (Skinner, 1957, p. 58).

Although the idea that automatic reinforcement is a factor in early language acquisition has been proposed for many years, experimental analyses of the phenome- non have appeared in the literature only recently (e.g., Miguel, Carr, & Michael, 2002; Sundberg et al., 1996; Yoon & Bennett, 2000). Sundberg and colleagues (1996) reported the first study showing the effects of a stimulus– stimulus pairing procedure on the frequency with which children emitted new vocal sounds without direct reinforcement or prompts to respond. Five children, ages 2 to 4 and representing a broad range of language abilities, served as subjects. During the prepairing (base- line) condition, the parents and in-home trainers sat a few feet away from the child and recorded each word or vocal sound emitted by the child as he played with a train set and several toys. Data were collected in consecutive 1-minute intervals. The adults did not interact with the subject during the prepairing baseline. The stimulus– stimulus pairing procedure consisted of a familiar adult approaching the child, emitting a target vocal sound, word, or phrase, and then immediately delivering a stim- ulus that had been established previously as a form of re- inforcement for the child (e.g., tickles, praise, bouncing in a parachute held by adults). This stimulus–stimulus pairing procedure was repeated 15 times per minute for 1 or 2 minutes. The adult used a variety of pitches and in- tonations when voicing the target sound, word, or phrase.

Positive Reinforcement

287

C u

m u

la ti

ve V

o ca

l R

es p

o n

se s

Subject 2

Prepairing PostpairingPairing

New Word “Apple”

Other Vocalization

1 2 3 4 5 6 7 8 9 10 11 12 13 Minutes

Figure 6 Cumulative number of times a 4-year-old child with autism vocalized “apple” before and after “apple” had been paired repeatedly with an established form of reinforce- ment. Automatic reinforcement may explain the increased frequency of the child’s vocal- izing “apple” after pairing. From “Automatic Reinforcement” by M. L. Sundberg, J. Michael, J. W. Partington, and C. A. Sundberg, 1996, “Repertoire-Altering Effects of Remote Contingencies” The Analysis of Verbal Behavior, 13, p. 27. Copyright 1996 by the Association for Behavior Analysis, Inc. Used by permission.

6Mands, tacts, and intraverbals—three elementary verbal operants first de- scribed by Skinner (1957)—are explained in chapter entitled “Verbal Behavior.”

During the postpairing condition, which began immedi- ately after the stimulus–stimulus pairings, the adult moved away from the child and conditions were the same as during the prepairing condition.

The stimulus–stimulus pairing of a vocal sound, word, or phrase with an established reinforcer was fol- lowed by an increased frequency of the targeted word during the postpairing condition for all five children. Figure 6 shows the results of a representative sample of one of three pairings conducted with Subject 2, a 4-year-old boy with autism. Subject 2 had a verbal reper- toire of more than 200 mands, tacts, and intraverbals, but rarely emitted spontaneous vocalizations or engaged in vocal play.6 During the prepairing condition, the child did not say the target word and emitted four other vocal- izations at a mean rate of 0.5 per minute. The stimulus– stimulus pairing procedure consisted of pairing the word apple with tickles approximately 15 times in 60 seconds. Immediately after the pairing, the subject said “apple” 17 times in 4 minutes, a rate of 4.25 responses per minute. In addition, the child said “tickle” four times within the

first minute of the postpairing condition. Sundberg and colleagues’ results provide evidence that the children’s vocal response products may have functioned as auto- matic conditioned reinforcement after being paired with other forms of reinforcement.

Kennedy (1994) noted that applied behavior analysts use two meanings of the term automatic reinforcement. In the first instance, automatic reinforcement is deter- mined by the absence of social mediation (Vollmer, 1994; 2006). In the second instance, when functional behavior assessments do not identify a reinforcer for a persistent behavior, some behavior analysts hypothesize that auto- matic reinforcement is the controlling variable. When SIB occurs in the absence of social attention or any other known form of reinforcement, automatic reinforcement is often assumed to be involved (e.g., Fisher, Lindauer, Alterson, & Thompson, 1998; Ringdahl, Vollmer, Marcus, & Roane, 1997; Roscoe, Iwata, & Goh, 1998). Determining that a behavior may be maintained by automatic reinforcement, and when possible isola- ting or substituting the source of that reinforcement (e.g., Kennedy & Souza, 1995; Shore, Iwata, DeLeon, Kahng, & Smith, 1997), has important implica- tions or designing interventions to either capitalize on

Positive Reinforcement

288

the automatic reinforcing nature of the behavior or coun- teract it.

In summing up the uses and limitations of automatic reinforcement as a concept, Vollmer (2006) suggested that:

• Practitioners should recognize that not all rein- forcement is planned or socially mediated.

• Some behaviors maintained by automatic reinforce- ment (e.g., self-stimulation, stereotypy) may not be reduced or eliminated with certain procedures (e.g., timeout, planned ignoring, or extinction).

• Affixing the label automatic reinforcement to an observed phenomenon too quickly may limit our analysis and effectiveness by precluding fur- ther efforts to identify the actual reinforcer- maintaining behavior.

• When socially mediated contingencies are difficult to arrange or simply not available, practitioners might consider automatic reinforcement as a po- tential aim.

Classifying Reinforcers In this section we review the technical classification of re- inforcers by their origin as well as several practical cat- egories by which practitioners and researchers often describe and classify reinforcers by their formal charac- teristics. The reader should recognize, however, that all reinforcers, regardless of type or classification, are the same in their most important (i.e., defining) characteris- tic: All reinforcers increase the future frequency of be- havior that immediately precedes them.

Classification of Reinforcers by Origin

There are two basic types of reinforcers by origin—that is, whether a reinforcer is the product of the evolution of the species (an unconditioned reinforcer) or the result of the learning history of the individual (a conditioned reinforcer).

Unconditioned Reinforcers

A stimulus change that functions as reinforcement even though the learner has had no particular learning history with it is called an unconditioned reinforcer. (Some au- thors use the terms primary reinforcer and unlearned re- inforcer as synonyms for unconditioned reinforcers.) Because unconditioned reinforcers are the product of the evolutionary history of a species (phylogeny), all bio- logically intact members of a species are more or less susceptible to reinforcement by the same unconditioned

7Remember, it is the environment, not the learner, that does the pairing. The learner does not have to “associate” the two stimuli.

reinforcers. For example, food, water, oxygen, warmth, and sexual stimulation are examples of stimuli that do not have to undergo a learning history to function as re- inforcers. Food will function as an unconditioned rein- forcer for a human deprived of sustenance; water will function as an unconditioned reinforcer for a person de- prived of liquid, and so forth.

Human touch may also be an unconditioned rein- forcer (Gewirtz & Pelaez-Nogueras, 2000). Pelaez- Nogueras and colleagues (1996) found that infants preferred face-to-face interactions that included touch stimulation. Two conditioning treatments were imple- mented in alternated counterbalanced order. Under the touch condition, infants’ eye-contact responses were fol- lowed immediately by adult attention (eye contact), smil- ing, cooing, and rubbing the infants’ legs and feet. Eye-contact responses during the no-touch condition were followed by eye contact, smiles, and coos from the adult, but no touching. All of the babies in the study emit- ted eye contact for longer durations, smiled and vocal- ized at higher rates, and spent less time crying and protesting in the contingent condition that included touch. From these results and several related studies, Pelaez- Nogueras and colleagues concluded that “these results suggest that . . . touch stimulation can function as a pri- mary reinforcer for infant behavior” (p. 199).

Conditioned Reinforcers

A conditioned reinforcer (sometimes called a secondary reinforcer or learned reinforcer) is a previously neutral stimulus change that has acquired the capability to func- tion as a reinforcer through stimulus–stimulus pairing with one or more unconditioned reinforcers or condi- tioned reinforcers. Through repeated pairings, the previ- ously neutral stimulus acquires the reinforcement capability of the reinforcer(s) with which it has been paired.7 For example, after a tone has been paired re- peatedly with food, when food is delivered as a reinforcer, the tone will function as a reinforcer when an EO has made food a currently effective reinforcer.

Neutral stimuli can also become conditioned rein- forcers for humans without direct physical pairing with another reinforcer through a pairing process Alessi (1992) called verbal analog conditioning.

For example, a class of preschool children who have been receiving M&M candies for good school work might be shown pieces of cut up yellow construction

Positive Reinforcement

289

paper and told, “These pieces of yellow paper are what big kids work for” (Engelmann, 1975, pp. 98–100). Many children in the group immediately refuse M&Ms, and work extra hard, but accept only pieces of yellow paper as their rewards.

We might say that the pieces of yellow paper act as “learned reinforcers.” Laboratory research tells us that neutral stimuli become reinforcers only through direct pairing with primary reinforcers (or other “learned rein- forcers”). Yellow paper was not paired with any rein- forcer and certainly not with the primary (M&Ms) reinforcers. Yellow paper acquired reinforcing properties even more powerful than the primary M&Ms rein- forcers, as demonstrated by the children’s refusal to ac- cept M&Ms, demanding instead pieces of yellow paper. (For the sake of this example, assume that the children had not been satiated with M&Ms just before the ses- sion.) (p. 1368)

It is sometimes thought that the “power” of a condi- tioned reinforcer is determined by the number of times it has been paired with other reinforcers. However, a state- ment such as, “The more often the tone is paired with food, the more reinforcing the tone will become” is not completely accurate. Although numerous pairings will increase the likelihood that the tone will function as a conditioned reinforcer in the first place (though a single pairing is sometimes sufficient), the momentary effec- tiveness of the tone as a reinforcer will be a function of the relevant EO for the reinforcer(s) with which the con- ditioned reinforcer has been paired. A tone that has been paired only with food will function as an effective rein- forcer for a food-deprived learner, but the tone will have little effect as a reinforcer if the learner has just consumed a lot of food, regardless of the number of times it has been paired with food.

A generalized conditioned reinforcer is a condi- tioned reinforcer that as a result of having been paired with many unconditioned and conditioned reinforcers does not depend on a current EO for any particular form of reinforcement for its effectiveness. For example so- cial attention (proximity, eye contact, praise) is a gener- alized conditioned reinforcer for many people because it occurs simultaneously with many reinforcers. The more reinforcers with which a generalized conditioned rein- forcer has been paired, the greater is the likelihood that it will be effective at any given time. Because it can be ex- changed for a nearly limitless variety of backup rein- forcers, money is a generalized conditioned reinforcer whose effectiveness is usually independent of current es- tablishing operations.

It is sometimes thought that a conditioned reinforcer is called a generalized conditioned reinforcer because it can function as reinforcement across a wide range of be- haviors. But this is not so—any reinforcer is capable of strengthening any behavior that immediately precedes its

occurrence. A conditioned reinforcer is called a general- ized conditioned reinforcer because it is effective as re- inforcement across a wide range of EO conditions. Because of their versatility across EO conditions, gener- alized conditioned reinforcers offer great advantages for practitioners, who often have limited control of the EOs for particular reinforcers.

Generalized conditioned reinforcers provide the basis for implementing a token economy, a reinforcement-based system capable of improving multiple behaviors of mul- tiple participants (e.g., Higgins, Williams, & McLaugh- lin, 2001; Phillips, Phillips, Fixen, & Wolf, 1971). In a token economy, participants receive tokens (e.g., points, check marks, poker chips) contingent on a variety of tar- get behaviors. Participants accumulate the tokens and ex- change them at specific times for their choices from a menu of backup reinforcers (e.g., free time, computer time, snacks).

Classification of Reinforcers by Formal Properties

When applied behavior analysts describe reinforcers by their physical properties—a practice that can enhance communication among researchers, practitioners, and the agencies and people they serve—reinforcers are typically classified as edible, sensory, tangible, activity, or social.

Edible Reinforcers

Researchers and practitioners have used bites of preferred foods, snacks, or candy, and sips of drinks, as reinforcers. One interesting and important use of edibles as rein- forcers is in the treatment of chronic food refusal in chil- dren. For example, Riordan, Iwata, Finney, Wohl, and Stanley (1984) used “highly preferred food items” as re- inforcers to increase the food intake of four children at a hospital treatment facility. The treatment program con- sisted of dispensing the high-preference food items (e.g., cereal, yogurt, canned fruit, ice cream) contingent on the consumption of a target food item (e.g., vegetables, bread, eggs).

Edible reinforcers were also used by Kelley, Piazza, Fisher, and Oberdorff (2003) to increase cup drinking by Al, a 3-year-old boy who had been admitted to a day treatment program for food refusal and bottle depen- dency. The researchers measured the percentage of trials in which Al consumed 7.5 ml of three different liquids from the cup. During baseline, when Al was praised if he consumed the drink, his consumption averaged 0%, 44.6% and 12.5% of trials for orange juice, water, and a

Positive Reinforcement

290

chocolate drink, respectively. During the positive rein- forcement component of the cup-drinking intervention, each time Al consumed the drink the therapist praised him (as was done in baseline) and delivered a level spoon of peaches (a preferred food) to his mouth. Al consumed all three beverages on 100% of the trials during the pos- itive reinforcement condition.

Sensory Reinforcers

Various forms of sensory stimulation such as vibration (e.g., massager), tactile stimulation (e.g., tickles, strokes with a feather boa), flashing or sparkling lights, and music have been used effectively as reinforcers (e.g., Bailey & Meyerson, 1969; Ferrari & Harris, 1981; Gast et al., 2000; Hume & Crossman, 1992; Rincover & Newsom, 1985; Vollmer & Iwata, 1991).

Tangible Reinforcers

Items such as stickers, trinkets, school materials, trading cards, and small toys often serve as tangible reinforcers. An object’s intrinsic worth is irrelevant to its ultimate ef- fectiveness as a positive reinforcer. Virtually any tan- gible item can serve as a reinforcer. Remember Engelmann’s (1975) kindergarten students who worked for yellow slips of paper!

Activity Reinforcers

When the opportunity to engage in a certain behavior serves as reinforcement, that behavior may be called an activity reinforcer. Activity reinforcers may be everyday activities (e.g., playing a board game, leisure reading, lis- tening to music), privileges (e.g., lunch with the teacher, shooting baskets in the gym, first in line), or special events (e.g., a trip to the zoo).

McEvoy and Brady (1988) evaluated the effects of contingent access to play materials on the completion of math worksheets by three students with autism and be- havior disorders. During baseline, the teacher told the students to complete the problems as best that they could, and that they should either complete other unfinished as- signments or “find something else to do” if they finished the worksheets before a 6-minute timing elapsed. No other prompts or instructions were given for completing the worksheets. The teacher praised the completion of the worksheets.

On the first day of intervention for each student, he was taken to another room and shown a variety of toys and play materials. The teacher told the student he would have approximately 6 minutes to play with the materials

if he met a daily criterion for completing math problems. Figure 7 shows the results. During baseline, the rate at which all three students correctly completed problems was either low (Dicky) or highly variable (Ken and Jimmy). When contingent access to the play activities was introduced, each student’s completion rate increased and eventually exceeded criterion levels.

Premack (1959) hypothesized that activity rein- forcers can be identified by looking at the relative dis- tribution of behaviors in a free operant situation. Premack believed that behaviors themselves could be used as re- inforcers and that the relative frequency of behavior was an important factor in determining how effective a given behavior might be as a reinforcer if the opportunity to engage in the behavior is contingent on another behav- ior. The Premack principle states that making the op- portunity to engage in a behavior that occurs at a relatively high free operant (or baseline) rate contingent on the occurrence of low-frequency behavior will func- tion as reinforcement for the low-frequency behavior. For a student who typically spends much more time watching TV than doing homework, a contingency based on the Premack principle (informally known as “Grandma’s Law”) might be, “When you have finished your homework, you can watch TV.”

Building on Premack’s concept, Timberlake and Al- lison (1974) proposed the response-deprivation hy- pothesis as a model for predicting whether access to one behavior (the contingent behavior) will function as rein- forcement for another behavior (the instrumental re- sponse) based on the relative baseline rates at which each behavior occurs and whether access to the contingent be- havior represents a restriction compared to the baseline level of engagement. Restricting access to a behavior pre- sumably acts as a form of deprivation that serves as an EO, thus making the opportunity to engage in the re- stricted behavior an effective form of reinforcement (Al- lison, 1993; Iwata & Michael, 1994).

Iwata and Michael (1994) cited a series of three studies by Konarski and colleagues as demonstrating the veracity and applied implications of the response- deprivation hypothesis. In the first study, when students were given access to coloring (a high-probability be- havior contingent on completing math problems (a low- probability behavior), they spent more time doing math, but only if the reinforcement schedule represented a re- striction of the amount of time spent coloring compared to baseline (Konarski, Johnson, Crowell, & Whitman, 1980). The researchers found that a contingency in which students could earn more time coloring than they did in baseline for completing math problems was ineffective. These basic findings were reproduced in a subsequent study in which access to reading (or math, depending on the subject) was contingent on math (or reading)

Positive Reinforcement

291

6.0

5.0

4.0

3.0

2.0

1.0

Dicky

Baseline Contingent Access

Problems completed correctly Problems completed incorrectly

6.0

5.0

4.0

3.0

2.0

1.0

Ken

(6.9)

N u

m b

er p

er M

in u

6.0

5.0

4.0

3.0

2.0

1.0

Jimmy

Sessions 5 10 15 20 25 30

(6.3) (6.3)

Figure 7 Number of math problems com- pleted correctly and incorrectly per minute by three special education students during base- line and contingent access to play materials. Dashed horizontal lines indicate criteria. From “Contingent Access to Play Materials as an Academic Motivator for Autistic and Behavior Disordered Children” by M. A. McEvoy and M. P. Brady, 1988, Education and Treatment of Children, 11, p. 15. Copyright 1998 by the Editorial Review Board of Education and Treatment of Children. Used by permission.

(Konarski, Crowell, Johnson, & Whitman (1982). In the third study, Konarski, Crowell, and Duggan (1985) took the response-deprivation hypothesis a step further by ex- amining the “reversibility of reinforcement” within sub- jects; that is, engaging in either of two activities—reading or math—could serve as reinforcement for increased per- formance in the other activity, in a response-deprivation condition for the contingent activity. Response depriva- tion for writing as the contingent response resulted in in- creases in math (instrumental response); conversely, response deprivation for math as the contingent response produced increases in reading. In all three studies, re- sponse restriction was the key factor in determining whether access to the contingent response would be reinforcing.

Iwata and Michael (1994) concluded that the collec- tive results of Konarski and colleagues’ studies illustrate each of three predictions based on the response- deprivation hypothesis (assume the ratio of baseline rates of doing homework to watching TV is 1:2 in the follow- ing examples):

• Reinforcement of a low-rate target behavior when access to a high-rate contingent behavior is re- stricted below baseline levels (e.g., 30 minutes of homework gets access to 30 minutes of TV).

• Nonreinforcement of a low-rate behavior when ac- cess to a high-rate contingent behavior is not re- stricted below baseline levels (e.g., 30 minutes of homework gets access to 90 minutes of TV).

• Reinforcement of a high-rate target behavior when access to the low-rate behavior is restricted be- low baseline levels (e.g., 30 minutes of TV yields 5 minutes of homework).

Although recognizing that practitioners seldom de- sign reinforcement programs to increase the rate of be- haviors such as TV watching that already occur at high rates, Iwata and Michael (1994) noted that:

There are a number of instances in which one may wish to produce highly accelerated performance (e.g., as in superlative academic or athletic performance that is

Positive Reinforcement

292

good to begin with). In such cases, one need not find an- other activity that occurs at a higher rate to serve as rein- forcement if one could arrange a suitable deprivation schedule with an activity that occurs at a relatively low rate. (p. 186)

As with all other descriptive categories of reinforcers, there is no a priori list that reveals what activities will or will not function as reinforcers. An activity that serves as effective reinforcement for one learner might have quite another effect on the behavior of another learner. For example, in Konarski, Crowell, and colleagues’ (1982) study, access to math functioned as reinforcement for doing more reading for three students, whereas get- ting to read was the reinforcer for completing math prob- lems for a fourth student. Many years ago, a classic cartoon brought home this crucial point very well. The cartoon showed two students dutifully cleaning the chalk- board and erasers after school. One student said to the other, “You’re cleaning erasers for punishment!? I get to clean erasers as a reward for completing my homework.”

Social Reinforcers

Physical contact (e.g., hugs, pats on the back), proximity (e.g., approaching, standing, or sitting near a person), at- tention, and praise are examples of events that often serve as social reinforcers. Adult attention is one of the most powerful and generally effective forms of reinforcement for children. The nearly universal effects of contingent social attention as reinforcement has led some behavior analysts to speculate that some aspects of social atten- tion may entail unconditioned reinforcement (e.g., Gewirtz & Pelaez-Nogueras, 2000; Vollmer & Hacken- berg, 2001).

The original experimental demonstrations and dis- covery of the power of adults’ social attention as rein- forcement for children’s behavior took place in a series of four studies designed by Montrose Wolf and carried out by the preschool teachers at the Institute of Child De- velopment at the University of Washington in the early 1960s (Allen, Hart, Buell, Harris, & Wolf, 1964; Harris, Johnston, Kelly, & Wolf, 1964; Hart, Allen, Buell, Har- ris, & Wolf, 1964; Johnston, Kelly, Harris, & Wolf, 1966). Describing those early studies, Risley (2005) wrote:

We had never seen such power! The speed and magni- tude of the effects on children’s behavior in the real world of simple adjustments of something so ubiquitous as adult attention was astounding. Forty years later, so- cial reinforcement (positive attention, praise, “catching them being good”) has become the core of most Ameri- can advice and training for parents and teachers—mak- ing this arguably the most influential discovery of modern psychology. (p. 280)

8The first volume of the Journal of Applied Behavior Analysis (1968) is a treasure trove of classic studies in which simple and elegant experimental designs revealed the powerful effects of operant conditioning and contin- gency management. We strongly encourage any serious student of applied behavior analysis to read it from cover to cover.

Because of the profound importance of this long- known but underused phenomenon, we describe a sec- ond study showing the effects of contingent attention as reinforcement for children’s behavior. The first volume of the Journal of Applied Behavior Analysis included no fewer than seven studies building on and extending Wolf and colleagues’ pioneering research on social reinforce- ment.8 R. Vance Hall and colleagues conducted two of those studies. Like the Hall, Lund, and Jackson (1968) study, from which we selected the example of a teacher’s use of positive reinforcement with Robbie that introduced this chapter, the three experiments reported by Hall, Pa- nyan, Rabon, and Broden (1968) continue to serve as powerful demonstrations of the effects of teacher atten- tion as social reinforcement.

A first-year teacher whose class of 30 sixth-graders exhibited such high rates of disruptive and off-task be- haviors that the school principal described the class as “completely out of control” participated in one of the ex- periments. Throughout the study, Hall, Panyan, and col- leagues (1968) measured teacher attention and students behavior during a continuous 30-minute observation pe- riod in the first hour of the school day. The researchers used a 10-second partial-interval observation and record- ing procedure to measure study behavior (e.g., writing the assignment, looking in the book, answering the teacher’s question) and nonstudy behavior (e.g., talking out, being out of seat, looking out the window, fighting or poking a classmate). The observers also recorded the occurrence of teacher attention in each interval. Each in- stance of teacher verbal attention, defined as a comment directed to a student or group of students, was recorded with a “+” if it followed appropriate study behavior, and with a “�” if it followed nonstudy behavior.

During baseline the class had a mean percentage of intervals of study behavior of 44%, and the teacher made an average of 1.4 comments following study behavior per session (see Figure 8). “Almost without exception those [comments] that followed study behavior were ap- proving and those that followed nonstudy behavior were in the form of a verbal reprimand” (Hall, Panyan et al., 1968, p. 316). The level of study behavior by the class was 90% on one day when the helping teacher presented a demonstration lesson (see data points marked by solid arrow). On three occasions during baseline (data points marked by open arrows), the principal met with the teacher to discuss his organizational procedures in an ef- fort to improve the students’ behavior. These counseling

Positive Reinforcement

293

100

P er

ce n

ta ge

o f

T im

e Sp

en t

St u

d y in

Baseline

5 10 15 20 25 30 35 40 45 50 55

Reinforcement1 Reversal Reinforcement2 Post

0 5 10 15 20 25 30 35 40 45 50 55

Sessions

In te

rv al

s o

f T

ea ch

er A

tt en

ti o

n t

o S

tu d

Figure 8 A record of class study behavior and teacher attention for study behavior during reading period in a sixth-grade classroom. Baseline = before experimental proce- dures; reinforcement 1 = increased teacher attention for study; reversal = removal of teacher attention for study; reinforcement 2 = return to increased teacher attention for study. Post- follow-up checks occurred up to 20 weeks after termination of the experimental procedures. From “Instructing Beginning Teachers in Reinforcement Procedures Which Improve Classroom Control” by R. V. Hall, M. Panyan, D. Rabon, and M. Broden, 1968, Journal of Applied Behavior Analysis, 1, p. 317. Copyright by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

sessions resulted in the teacher writing all assignments on the board (after the first meeting) and changing the seat- ing chart (after the third meeting). Those changes had no apparent effect on the students’ behavior.

Prior to the first day of the reinforcement condition, the teacher was shown baseline data on the class study be- havior and the frequency of teacher attention following study behavior. The teacher was instructed to increase the frequency of positive comments to students when they were engaged in study behavior. After each session dur- ing this condition the teacher was shown data on the level of class study behavior and the frequency of his com- ments that followed study behavior. During the first re- inforcement phase, teacher comments following study behavior increased to a mean frequency of 14.6, and the mean level of study behavior was 72%. The teacher, prin- cipal, and data collectors reported that the class was under better control and that noise had decreased significantly.

During a brief return of baseline conditions, the teacher provided “almost no reinforcement for study be- havior,” and a sharp downward trend in class study be- havior was observed. The teacher, principal, and data

collectors all reported that disruptive behavior and high noise levels had returned. The reinforcement conditions were then reinstated, which resulted in a mean frequency of 14 teacher comments following study behavior, and a mean level of 76% of intervals of study behavior.

Identifying Potential Reinforcers

In the laboratory, we had learned to use a simple test: Place a candy in the palm of our hand, show it to the child, close our fist fairly tightly around the candy, and see if the child will try to pull away our fingers to get at the candy. If he or she will do that, even against in- creasingly tightly held fingers, the candy is obviously a reinforcer.

—Murray Sidman (2000, p. 18)

The success of many behavior change programs requires an effective reinforcer that the practitioner or researcher can control. Fortunately, identifying effective and

Positive Reinforcement

294

accessible reinforcers for most learners is relatively easy. Sidman (2000) described a quick and simple method for determining whether candy would likely function as a reinforcer. However, every stimulus, event, or activity that might function as a reinforcer cannot be held in the palm of the hand.

Identifying robust and reliable reinforcers for many learners with severe and multiple disabilities poses a major challenge. Although many common events serve as effective reinforcers for most people (e.g., praise, music, free time, tokens), these stimuli may not serve as reinforcers for all learners. Time, energy, and resources would be lost if planned interventions were to fail be- cause a practitioner used a presumed, instead of an ac- tual, reinforcer.

Also, reinforcer preferences shift, and the transitory nature of preference has been reported repeatedly in the literature (Carr, Nicholson, & Higbee, 2000; DeLeon et al., 2001; Kennedy & Haring, 1993; Logan & Gast, 2001; Ortiz & Carr, 2000; Roane, Vollmer, Ringdahl, & Mar- cus, 1998). Preference assessments may change with the person’s age, interest level, time of day, social interac- tions with peers, and the presence of certain establish- ing operations (EOs) (Gottschalk, Libby, & Graff, 2000). What a teacher asks in September to determine preferences may have to be repeated a month later (or sooner). Likewise, a therapist who asks a client what is reinforcing during a morning session may find that this stimulus is not stated as a preferred item in an afternoon session.

After reviewing 13 published studies that evaluated preferences and reinforcers for people with profound multiple disabilities, Logan and Gast (2001) concluded that preferred stimuli do not always function as rein- forcers, and preferred stimuli at one point in time changed later. Additionally, people with severe-to-profound de- velopmental disabilities may engage in activities for such

a limited time that it is difficult to clearly determine whether a stimulus change is a reinforcer.

To meet the challenge of identifying effective rein- forcers, researchers and practitioners have developed a variety of procedures that fall under the twin headings of stimulus preference assessment and reinforcer assess- ment. Stimulus preference assessment and reinforcer as- sessment are often conducted in tandem, as described by Piazza, Fisher, Hagopian, Bowman, and Toole (1996):

During preference assessments, a relatively large num- ber of stimuli are evaluated to identify preferred stimuli. The reinforcing effects of a small subset of stimuli (i.e., the highly preferred stimuli) are then evaluated during reinforcer assessment. Although the preference assess- ment is an efficient procedure identifying potential rein- forcers from a large number of stimuli, it does not evaluate the reinforcing effects of the stimuli. (pp. 1–2)

Stimulus preference assessment identifies stimuli that are likely to serve as reinforcers, and reinforcer as- sessment puts the potential reinforcers to a direct test by presenting them contingent on occurrences of a behavior and measuring any effects on response rate. In this sec- tion we describe a variety of techniques developed by re- searchers and practitioners for conducting stimulus preference assessments and reinforcer assessment (see Figure 9). Together these methods form a continuum of approaches ranging from simple and quick to more complex and time-consuming.

Stimulus Preference Assessment

Stimulus preference assessment refers to a variety of procedures used to determine (a) the stimuli that the per- son prefers, (b) the relative preference values of those stimuli (high preference versus low preference), and (c) the conditions under which those preference values

Person

Significant others

Pre-task choice

Contrived observation

Naturalistic observation

Single stimulus

Paired stimuli

Multiple stimuli

Multiple schedules

Progressive ratio

schedules

Concurrent schedules

Ask Free-Operant Trial-Based

Stimulus Preference Assessment Reinforcer Assessment

Figure 9 Stimulus preference assessment and reinforcer assessment methods for identifying potential reinforcers.

Positive Reinforcement

295

change when task demands, deprivation states, or sched- ules of reinforcement are modified. Generally speaking, stimulus preference assessment is usually conducted using a two-step process: (1) A large pool of stimuli that might be used as reinforcers is gathered, and (2) those stimuli are presented to the target person systematically to identify preference. It is essential for practitioners to narrow the field of possible stimuli to those that have good odds of functioning as reinforcers.

In more specific terms, stimulus preference assess- ments can be conducted using three basic methods: ask- ing the person (or his or her significant others) to identify preferred stimuli; observing the person interacting or en- gaging with various stimuli in a free operant situation; and measuring the person’s responses to trial-based tests of paired or multiply presented stimuli. In choosing which method to use, practitioners must balance two competing perspectives: (a) gaining the maximum amount of pref- erence assessment data within the least amount of time, but without false positives (i.e., believing a stimulus is preferred when it is not), versus (b) conducting a more time- and labor-intensive assessment that will delay in- tervention, but may yield more conclusive results.

Asking about Stimulus Preferences

A person’s preference for various stimuli might be de- termined by merely asking what she likes. Asking can greatly reduce the time needed for stimulus preference assessment, and it often yields information that can be integrated in an intervention program. Several variations of asking exist: asking the target person, asking signifi- cant others in the person’s life, or offering a pretask choice assessment.

Asking the Target Person. A straightforward method for determining stimulus preference is to ask the target person what he likes. Typical variations in- clude asking open-ended questions, providing the per- son with a list of choices or asking him to rank-order a list of choices.

• Open-ended questions. Depending on the learner’s language abilities, an open-ended assessment of stimulus preference can be done orally or in writ- ing. The person may be asked to name preferences among general categories of reinforcers—for ex- ample, What do you like to do in your free time? What are your favorite foods and drinks? Are there any types of music or performers whose music you like? An open-ended assessment can be accom- plished simply by asking the learner to list as many favorite activities or items as possible. She should list not only everyday favorite things and

activities, but also special items and activities. Is simply a sheet with numbered lines on which family members identify potential rewards they would like to earn by completing tasks on contin- gency contracts.

• Choice format. This format could include asking questions such as the following: “Which would you do a lot of hard work to get? Would you rather get things to eat, like chips, cookies, popcorn, or get to do things, like art projects, play computer games, or go to the library?” (Northup, George, Jones, Broussard, & Vollmer, 1996, p. 204)

• Rank-ordering. The learner can be given a list of items or stimuli and instructed to rank-order them from most to least preferred.

For learners with limited language skills, pictures of items, icons, or, preferrably, the actual stimuli can be pre- sented. For example, a teacher, while pointing to an icon, might ask a student, “Do you like to drink juice, use the computer, ride the bus, or watch TV?” Students simply nod yes or no.

Surveys have been developed to assess students’ preferences. For example, elementary school teachers might use the Child Reinforcement Survey, which in- cludes 36 rewards in four categories: edible items (e.g., fruit, popcorn), tangible items (e.g., stickers), activities (e.g., art projects, computer games), and social attention (e.g., a teacher or friend saying, “I like that”) (Fantuzzo, Rohrbeck, Hightower, & Work, 1991). Other surveys are the School Reinforcement Survey Schedule for students in grades 4 through 12 (Holmes, Cautela, Simpson, Motes, & Gold, 1998) and the Reinforcement Assessment for In- dividuals with Severe Disabilities (Fisher, Piazza, Bow- man, & Almari, 1996).

Although asking for personal preferences is relatively uncomplicated, the procedure is not foolproof with re- spect to confirming that a preferred choice will later serve as a reinforcer. “Poor correspondence between verbal self-reports and subsequent behavior has been long noted and often demonstrated” (Northup, 2000, p. 335). Al- though a child might identify watching cartoons as a pre- ferred event, watching cartoons may function as a reinforcer only when the child is at home on Saturday mornings, but not at Grandma’s house on Sunday night.

Further, surveys may not differentiate accurately be- tween what children claim to be high-preference and low- preference items for reinforcers. Northup (2000) found that preferences of children with attention-deficit/hyper- activity disorder (ADHD) did not rise beyond chance lev- els when survey results were later compared to reinforcer functions. “The relatively high number of false positives and low number of false negatives again suggest that sur- veys may more accurately identify stimuli that are not

Positive Reinforcement

296

reinforcers than those that are” (p. 337). Merely asking children their preferences once might lead to false posi- tives (i.e., children may choose an event or stimulus as a reinforcer, but it may not be reinforcing).

Asking Significant Others. A pool of potential re- inforcers can be obtained by asking parents, siblings, friends, or caregivers to identify the activities, items, foods, hobbies, or toys that they believe the learner prefers. For example, the Reinforcer Assessment for In- dividuals with Severe Disabilities (RAISD) is an inter- view protocol that asks caregivers to identify preferred stimuli across visual, auditory, olfactory, edible, tactile, and social domains (Fisher et al., 1996). Significant others then rank-order the selected preferences based on likely high- versus low-preference items. Finally, significant others are asked to identify the conditions under which they predict that specific items might function as reinforcers (e.g., cookies with milk versus just cookies alone). Again, although stimuli that are identified as highly preferred by significant others are not always effective as reinforcers, they often are.

Offering a Pretask Choice. In this method the practitioner asks the participant to choose what he wants to earn for doing a task. The participant then chooses one item from two or three options presented (Piazza et al., 1996). All of the stimuli presented as pre- task choices will have been identified as preferred stim- uli by other assessment procedures. For instance, a teacher might make the following statement: “Robyn, when you finish your math problems, you may have 10 minutes to play Battleship with Martin, read quietly, or help Ms. Obutu prepare the social studies poster. Which activity do you want to work for?” A learner’s choice of a consequence will not necessarily be a more effective reinforcer than one selected by the researcher or practi- tioner (Smith, Iwata, & Shore, 1995).

Free Operant Observation

The activities that a person engages in most often when able to choose freely from among behaviors will often serve as effective reinforcers when made contingent on engaging in low-probability behaviors. Observing and recording what activities the target person engages in when she can choose during a period of unrestricted ac- cess to numerous activities is called free operant obser- vation. A total duration measure of the time the person engages with each stimulus item or activity is recorded. The longer the person engages with an item, the stronger the inference that the item is preferred.

Procedurally, the person has unconstrained and si- multaneous access to a predetermined set of items or ac- tivities or to the materials and activities that are naturally available in the environment. There are no response re- quirements, and all stimulus items are available and within the person’s sight and reach. An item is never removed after engagement or selection. According to Ortiz and Carr (2000), free operant responding is less likely to pro- duce aberrant behavior that might otherwise be observed if a stimulus is removed. Free operant observations can be contrived or conducted in naturalistic settings.

Contrived Free Operant Observation. Practitioners use contrived observation to determine whether, when, how, and the extent to which the person engages with each of a predetermined set of activities and materials. The observation is contrived because the researcher or practitioner “salts” the environment with a variety of items that may be of interest to the learner.

Free operant assessment presupposes that the per- son has had sufficient time to move about and explore the environment and has had the chance to experience each of the stimuli, materials, or activities. Just prior to the free operant observation period, the learner is pro- vided brief noncontingent exposure to each item. All of the items are then placed within view and easy access to the learner, who then has the opportunity to sample and choose among them freely. Observers record the total duration of time that the learner engages with each stim- ulus item or activity.

Naturalistic Free Operant Observation. Natural- istic observations of free operant responding are con- ducted in the learner’s everyday environment (e.g., playground, classroom, home). As unobtrusively as possible, the observer notes how the learner allocates his time and records the number of minutes the learner devotes to each activity. For instance, Figure 10 shows how a teenager, Mike, distributed his time during 2 hours of free time each day after school. Mike’s par- ents collected these data by keeping a chart of the total number of minutes their son was engaged in each activ- ity. The summary chart for the week shows that Mike played computer video games, watched television, and talked on the phone to his friends every day. On two different days Mike spent 10 minutes reading a library book, and he played with a new construction toy for a brief time on Wednesday. Two activities—watching television and playing video games—occurred the most often and for the longest duration. If Mike’s parents wanted to apply the Premack principle introduced ear- lier in this chapter to increase the amount of time he spends reading for pleasure or playing with the con- struction toy (i.e., low-probability behaviors), they

Positive Reinforcement

297

Figure 10 Number of minutes Mike spent engaged in activities during 2 hours of free time after school.

Activity Mon Tue Wed Thu Fri Total Leisure reading — 10 — 10 — 20 Watch TV 35 50 60 30 30 205 Phone with friends 15 15 10 20 10 70 Play video games 70 45 40 60 80 295 Play with construction toy — — 10 — — 10 Minutes observed 120 120 120 120 120 600

might make watching television or playing video games (i.e., high-probability behaviors) contingent on a cer- tain amount of time spent leisure reading or playing with the construction toy.

Trial-Based Methods

In trial-based methods of stimulus preference assessment, stimuli are presented to the learner in a series of trials and the learner’s responses to the stimuli are measured as an index of preference. One or more of three measures of the learner’s behavior are recorded in trial-based stimu- lus preference assessment: approach, contact (DeLeon & Iwata, 1996), and engagement with the stimulus (DeLeon, Iwata, Conners, & Wallace, 1999; Hagopian, Rush, Lewin, & Long, 2001; Roane et al., 1998). Approach responses typically include any detectable movement by the person toward the stimulus (e.g., eye gaze, head turn, body lean, hand reach), a contact is tal- lied each time the person touches or holds the stimulus, and engagement is a measure of the total time or per- centage of observed intervals in which the person inter- acts with the stimulus (e.g., in which the person held a massager against her leg). An assumption is made that the more frequently the person approaches, touches or holds, or engages with a stimulus, the more likely it is that the stimulus is preferred. As DeLeon and colleagues (1999) stated, “duration of item contact is a valid index of reinforcer value” (p. 114).

Preferred stimuli are sometimes labeled as high- preference (HP), medium-preference (MP), or low- preference (LP) stimuli based on predetermined criteria (e.g., stimuli chosen 75% or more of the time are HP) (Carr, Nicolson, & Higbee, 2000; Northup, 2000; Pace, Ivancic, Edwards, Iwata, & Page, 1985; Piazza et al., 1996). An implicit, but testable, assumption is that a highly preferred stimulus will serve as a reinforcer. Al- though this assumption does not always hold (Higbee, Carr, & Harrison, 2000), it has proven to be an efficient assumption with which to begin.

The many variations of trial-based stimulus prefer- ence assessment can be grouped by presentation method as single stimulus (successive choice), paired stimuli (forced choice), and multiple stimuli.

Single Stimulus. A single-stimulus presentation method, also called a “successive choice” method, rep- resents the most basic assessment available for deter- mining preference. Simply stated, a stimulus is presented and the person’s reaction to it is noted. Pre- senting one stimulus at a time “may be well suited for individuals who have difficulty selecting among two or more stimuli” (Hagopian et al., 2001, p. 477).

Target stimuli across all sensory systems (i.e., visual, auditory, vestibular, tactile, olfactory, gustatory, and mul- tisensory) are presented one at a time in random order, and the person’s reaction to each stimulus is recorded (Logan, Jacobs et al., 2001; Pace et al., 1985). Approach or rejection responses are recorded in terms of occur- rence (yes or no), frequency (e.g., number of touches per minute), or duration (i.e., time spent engaged with an item). After recording, the next item in the sequence is presented. For example, a mirror might be presented to determine the duration of time the person gazes into it, touches it, or rejects the mirror (i.e., pushes it away). Each item should be presented several times, and the order of presentation should be varied.

Paired Stimuli. Each trial in the paired-stimuli pre- sentation method, also sometimes called the “forced choice” method, consists of the simultaneous presentation of two stimuli. The observer records which of the two stimuli the learner chooses. During the course of the as- sessment, each stimulus is matched randomly with all other stimuli in the set (Fisher et al., 1992). Data from a paired-stimuli assessment show how many times each stimulus is chosen. The stimuli are then rank-ordered in terms of high, medium, or low preference. Piazza and colleagues (1996) used 66 to 120 paired-stimuli trials to determine high, middle, and low preferences. Pace and colleagues (1985) found that paired-stimuli presentations

Positive Reinforcement

298

yielded more accurate distinctions between high- and low-preference items than did single-stimulus presenta- tions. Paired-stimuli sometimes outperform single- stimulus presentation formats with respect to ultimately identifying reinforcers (Paclawskyj & Vollmer, 1995).

Because every possible pair of stimuli must be pre- sented, paired-stimuli assessing may take more time than the simultaneous presentation of an array of multiple stimuli (described in the next section). However, DeLeon and Iwata (1996) argued that ultimately the paired-stimuli method may be more time efficient because “the more consistent results produced by the PS method may indi- cate that stable preferences can be determined in fewer, or even single, sessions” (p. 520).

Multiple Stimuli. The multiple-stimuli presenta- tion method is an extension of the paired-stimuli proce- dure developed by Fisher and colleagues (1992). The person chooses a preferred stimulus from an array of three or more stimuli (Windsor, Piche, & Locke, 1994). By presenting multiple stimuli together, assessment time is reduced. For example, instead of presenting a series of trials consisting of all possible pairs of stimuli from a group of six stimuli and continuing until all pairs have been presented, all six stimuli are presented simultaneously.

The two major variations of the multiple-stimuli pref- erence assessment are multiple stimuli with replacement and multiple stimuli without replacement. The difference between the two is which stimuli are removed or replaced after the person indicates a preference among the dis- played items in preparation for the next trial. In the mul- tiple stimuli with replacement procedure, the item chosen by the learner remains in the array and items that were not selected are replaced with new items. In the multiple stim- uli without replacement procedure, the chosen item is re- moved from the array, the order or placement of the remaining items is rearranged, and the next trial begins with a reduced number of items in the array.

In any case, each trial begins by asking the person, “Which one do you want the most?” (Higbee et al., 2000) “Choose one” (Ciccone, Graff, & Ahearn, 2005) and then continuing until all items from the original array, or the gradually reducing array, have been selected. The entire sequence is usually repeated several times, although a single round of trials may identify stimuli that function as reinforcers (Carr et al., 2000).

The stimuli presented in each trial might be tangible objects themselves, pictures of the items, or verbal de- scriptions. Higbee, Carr, and Harrison (1999) provided a variation of the multiple-stimuli procedure that included stimulus preference selection based on a tangible object versus a picture of the object. The tangible objects pro-

duced greater variation and distribution of preferences than the picture objects did. Cohen-Almeida, Graff, and Ahearn (2000) found that the tangible object assessment was about as effective as a verbal preference assessment, but the clients completed the verbal preference assess- ment in less time.

DeLeon and Iwata (1996) used an adaptation of the multiple-stimuli and paired-stimuli presentations they de- scribed as a brief stimulus assessment to reduce the time needed to determine stimulus preference. Basically, in the brief stimulus assessment, once a particular stimulus item is chosen, that item is not returned to the array. Subse- quent trials present a reduced number of items from which to choose (Carr et al., 2000; DeLeon et al., 2001; Roane et al., 1998). DeLeon and Iwata (1996) found that multi- ple stimuli without replacement identified preferred items in approximately half the time that the multiple stimuli with replacement procedure did. According to Higbee and colleagues (2000), “With a brief stimulus preference pro- cedure, practitioners have a method for reinforcer identi- fication that is both efficient and accurate” (pp. 72–73).

Guidelines for Selecting and Using Stimulus Preference Assessments

Practitioners can combine assessment procedures to com- pare single versus paired, paired versus multiple, or free operant versus trial-based methods (Ortiz & Carr, 2000). In day-to-day practice, brief stimulus presentations using comparative approaches might facilitate reinforcer iden- tification, thereby speeding up possible interventions using those reinforcers. In summary, the goal of stimulus preference assessments is to identify stimuli that are most likely to function as reinforcers. Each method for as- sessing preference has advantages and limitations with respect to identifying preferences (Roane et al., 1998). Practitioners may find the following guidelines helpful when conducting stimulus preference assessments (DeLeon & Iwata, 1996; Gottschalk et al., 2000; Higbee et al., 2000; Ortiz & Carr, 2000; Roane et al., 1998; Roscoe, Iwata, & Kahng, 1999):

• Monitor the learner’s activities during the time pe- riod before the stimulus preference assessment ses- sion to be aware of EOs that may affect the results.

• Use stimulus preference assessment options that balance the cost-benefit of brief assessments (but possible false positives) with more prolonged as- sessments that may delay reinforcer identification.

• Balance using a stimulus preference method that may yield a ranking of preferred stimuli against an assessment method that occurs without rankings,

Positive Reinforcement

299

but occurs more frequently, to counteract shifts in preference.

• When time is limited, conduct a brief stimulus preference assessment with fewer items in an array.

• When possible, combine data from multiple as- sessment methods and sources of stimulus prefer- ence (e.g., asking the learner and significant others, free operant observation, pretask choice, and trial- based methods).

Reinforcer Assessment The only way to tell whether or not a given event is rein- forcing to a given organism under given conditions is to make a direct test.

—B. F. Skinner (1953, pp. 72–73)

Highly preferred stimuli may not always function as re- inforcers (Higbee et al., 2000); even the candy that a child pried out of Sidman’s hand may not have func- tioned as reinforcement under certain conditions. Con- versely, least preferred stimuli might serve as reinforcers under some conditions (Gottschalk et al., 2000). The only way to know for sure whether a given stimulus serves as a reinforcer is to present it immediately fol- lowing the occurrence of a behavior and note its effects on responding.

Reinforcer assessment refers to a variety of direct, data-based methods used to present one or more stimuli contingent on a target response and then measuring the future effects on the rate of responding. Researchers and practitioners have developed reinforcer assessment meth- ods to determine the relative effects of a given stimulus as reinforcement under different and changing conditions and to assess the comparative effectiveness of multiple stimuli as reinforcers for a given behavior under specific conditions. Reinforcer assessment is often accomplished by presenting stimuli suspected of being reinforcers con- tingent on responding within concurrent, multiple, or progressive-ratio reinforcement schedules.9

Concurrent Schedule Reinforcer Assessment

When two or more contingencies of reinforcement oper- ate independently and simultaneously for two or more behaviors a concurrent schedule of reinforcement is in effect. When used as a vehicle for reinforcer assessment, a concurrent schedule arrangement essentially pits two

9These and other types of schedules of reinforcement and their effects on behavior are described in chapter entitled “Schedules of Reinforcement.”

stimuli against each other to see which will produce the larger increase in responding when presented as a con- sequence for responding. If a learner allocates a greater proportion of responses to one component of the con- current schedule over the other, the stimulus used as a contingent consequence for that component is the more effective reinforcer. For example, using a concurrent schedule in this way shows the relative effectiveness of high-preference (HP) and low-preference (LP) stimuli as reinforcers (Koehler, Iwata, Roscoe, Rolider, & O’Steen, 2005; Piazza et al., 1996).

Concurrent schedules may also be used to determine differences between relative and absolute reinforcement effects of stimuli. That is, will an LP stimulus now pre- sented contingently in the absence of the HP stimulus serve as a reinforcer? Roscoe and colleagues (1999) used concurrent schedules to compare the effects of HP and LP stimuli as reinforcers for eight adults with developmen- tal disabilities. Following the preference assessments, a concurrent schedule of reinforcement was established using the high-preference and low-preference stimuli. The target response was pressing either of two micro switch panels. Each panel was a different color. Pressing a panel would illuminate a small light in the center of the panel. A training condition took place prior to baseline to establish panel pressing in the subjects’ repertoires and to expose them to the consequences of responding. During baseline, pressing either panel resulted in no programmed consequences. During the reinforcement phase, an HP stimulus was placed on a plate behind one of the panels and an LP stimulus was placed on a plate behind another panel. All responses to either panel resulted in the par- ticipant immediately receiving the item on the plate be- hind the respective panel (i.e., an FR 1 schedule of reinforcement).

Under the concurrent schedule of reinforcement that enabled participants to choose reinforcers on the same FR 1 schedule, the majority of participants allocated most of their responding to the panel that produced the HP stimulus as reinforcement (e.g., see results for Sean, Peter, and Matt on Figure 11). However, these same participants, when later presented with the opportunity to obtain LP stimuli as reinforcers on a single-schedule contingency (i.e., only one panel to push), showed in- creased levels of responding over baseline similar to those obtained with the HP stimuli in the concurrent schedule. The study by Roscoe and colleagues (1999) demonstrated how concurrent schedules may be used to identify the relative effects of stimuli as reinforcers. The study also showed that the potential effects of a stimulus as a rein- forcer may be masked or overshadowed when that stim- ulus is pitted against another stimulus on a concurrent schedule. In such cases, a potentially reinforcing stimu- lus might be abandoned prematurely.

Positive Reinforcement

300

Concurrent BL Sr+ BL Sr+

Single

Sean

5 10 15 20 25 30 35

Concurrent BL Sr+ BL Sr+

Single

Peter

10 15 205 25

Concurrent BL Sr+ BL Sr+

Single

Matt

10 155 20

Concurrent BL Sr+

Mike

10 155 20 Sessions

R es

p o

n se

s p

er M

in u

Figure 11 Responses per minute during concur- rent-schedule and single-schedule baseline and rein- forcement conditions for four adults with mental retardation. From “Relative versus Absolute Reinforcement Effects: Implications for Preference Assessments” by E. M. Roscoe, B. A. Iwata, and S. Kahng, 1999, Journal of Applied Behavior Analysis, 32, p. 489. Copyright 1999 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Multiple Schedule Reinforcer Assessment

A multiple schedule of reinforcement consists of two or more component schedules of reinforcement for a single response with only one component schedule in effect at any given time. A discriminative stimulus (SD) signals the presence of each component schedule, and that stim- ulus is present as long as the schedule is in effect. One way that a multiple schedule could be used for reinforcer assessment would be to present the same stimulus event contingent (i.e., response dependent) on each occurrence of the target behavior in one component of the multiple

schedule and on a fixed-time schedule (i.e., response in- dependent) in the other component. For example, if a practitioner wanted to use a multiple schedule to assess whether social attention functioned as a reinforcer, she would provide social attention contingent on occurrences of cooperative play when one component of the multiple schedule is in effect, and during the other component the practitioner would present the same amount and kind of social attention except on a fixed-time schedule, in- dependent of cooperative play (i.e., noncontingent reinforcement). The teacher could apply the response- dependent schedule during the morning play period, and the response-independent schedule during the afternoon play period. If social attention functioned as reinforce- ment, cooperative play would likely increase over its baseline rate in the morning periods, and because of no relationship with cooperative play, attention would likely have no effect in the afternoon period. This situation fol- lows a multiple schedule because there is one class of behavior (i.e., cooperative play), a discriminative stim- ulus for each contingency in effect (i.e., morning and af- ternoon play periods), and different conditions for reinforcement (i.e., response dependent and response in- dependent).

Progressive-Ratio Schedule Reinforcer Assessment

Stimulus preference assessments with low response re- quirements (e.g., FR 1) may not predict the effectiveness of the stimulus as a reinforcer when presented with higher response requirements (e.g., on an FR 10 schedule, a stu- dent must complete 10 problems to obtain reinforcement). As DeLeon, Iwata, Goh, and Worsdell (1997) stated:

Current assessment methods may make inaccurate pre- dictions about reinforcer efficacy when the task used in training regimens requires either more responses or more effort before the delivery of reinforcement. . . . for some classes of reinforcers, simultaneous increases in sched- ule requirements may magnify small differences in pref- erences that are undetected when requirements are low. In such cases, a stimulus preference assessment involv- ing low response requirements (FR1) schedules does not accurately predict the relative potency of reinforcers under increased response requirements. (pp. 440, 446)

Progressive-ratio schedules provide a framework for assessing the relative effectiveness of a stimulus as re- inforcement as response requirements increase. In a progressive-ratio schedule of reinforcement the response requirements for reinforcement are increased systemati- cally over time independent of the participant’s behav- ior. In a progressive-ratio schedule, the practitioner gradually requires more responses per presentation of the

Positive Reinforcement

301

14 12 10 8 6 4 2 0

5 10 15 20 30 40 4525 35

Elaine

FR 1 FR 2 FR 5 FR 10 FR 20 FR 1

Drink

Balloon

0 5 10 15 20 30 40 4525 35

Rick

FR 1 FR 2 FR 5 FR 10 FR 20

Massager

5 10 15 20 30 40 4525 35

Elaine

FR 1 FR 2 FR 5 FR 10 FR 1

Chip

Pretzel

14 12 10 8 6 4 2 0

0 5 10 15 20 30 40 4525 35

Rick

FR 1 FR 2 FR 5 FR 10 Cookie

Cracker

N u

m b

er o

f R

ei n

fo rc

er s

E ar

n ed

Sessions

Figure 12 Responses per minute during concurrent-schedule and single-schedule baseline and reinforcement conditions for four adults with mental retardation. From “Emergence of Reinforcer Preference as a Function of Schedule Requirements and Stimulus Similarity” by I. G. DeLeon, B. A. Iwata, H. Goh, and A. S. Worsdell, 1997, Journal of Applied Behavior Analysis, 30, p. 444. Copyright 1997 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

preferred stimulus until a breaking point is reached and the response rate declines (Roane, Lerman, & Vorndran, 2001). For example, initially each response produces re- inforcement (FR 1), then reinforcement is delivered after every second response (FR 2), then perhaps after every fifth, tenth, and twentieth response (FR 5, FR 10, and FR 20). At some point, a preferred stimulus may no longer function as reinforcement (Tustin, 1994).

DeLeon and colleagues (1997) used a progressive ratio within a concurrent schedule to test the relative ef- fectiveness of two similarly preferred stimuli (e.g., cookie and cracker) and two dissimilar stimuli (e.g., drink and balloon) as reinforcers for micro switch panel pressing for Elaine and Rick, two adults with mental retardation. One panel was blue and one was yellow. The experimenters placed two reinforcers on separate plates, and put one plate behind each of the panels. Each trial (24 per ses- sion for Rick; 14 per session for Elaine) consisted of the subject pushing either one of the panels and immediately

receiving the item on the plate behind that panel. During the first phase, an FR 1 schedule was used (i.e., each re- sponse produced the item on the plate). Later, the re- sponse requirement for obtaining the items was gradually increased (FR 2, FR 5, FR 10, and FR 20).

Elaine and Rick made responses that produced the two dissimilar items at roughly the same rates during the FR 1 phase (see the top two graphs in Figure 12). As response requirements for receiving the dissimilar stim- uli increased, Elaine and Rick continued to evenly allo- cate responding between the two panels. However, when initially equivalent and similar reinforcers (food in FR 1) were compared under increasing schedule require- ments, the differences in responses rates on the two pan- els revealed clear and consistent preferences (see the bottom two graphs in Figure 12). For example, when Elaine needed to work more to receive food, she allo- cated the majority of her responses to the panel that pro- duced chips rather than the one that produced pretzels. In

Positive Reinforcement

302

the same way, as the number of responses required to re- ceive reinforcement increased, Rick showed a clear pref- erence for cookies over crackers. These results suggest that “for some classes of reinforcers, simultaneous in- creases in schedule requirements may magnify small differences in preference that are undetected when requirements are low” (DeLeon et al., 1997, p. 446).

Increasing response requirements within a concur- rent schedule may reflect the effects of increasing re- sponse requirements on the choice between reinforcers and also reveal whether and under what conditions two reinforcers are substitutable for each other. If two rein- forcers serve the same function (i.e., are made effective by the same establishing operation), an increase in the price (i.e., response requirement) for one of the rein- forcers will lead to a decreased consumption of that item if a substitutable reinforcer is available (Green & Freed, 1993). DeLeon and colleagues (1997) used a hypotheti- cal person with a slight preference for Coke over Pepsi as an analogy to explain the results shown in Figure 12.

Assuming that Coke and Pepsi are both available for $1.00 per serving and that a person has only a slight preference for Coke, the individual may allocate choices rather evenly, perhaps as a function of periodic satiation for the preferred item, but with slightly more overall se- lections of Coke. Now assume that the cost of each is increased to $5.00 per serving. At this price, the prefer- ence for Coke is likely to be expressed. By contrast, a similar arrangement involving Coke and bus tokens may produce different results. Again, at $1.00 per item, roughly equal selection between the two options would not be surprising, assuming that the establishing opera- tion for each dictates that both are momentarily equally valuable. However, these items serve distinctly different functions and are not substitutable; that is, the person is not free to trade one for the other and to continue to re- ceive functionally similar reinforcement at the same rate. The person is more likely to continue choosing equally, even when the price for both reinforcers in- creases substantially.

The same might be said for the results obtained in the present study. When choices involved two substi- tutable items, such as a cookie and a cracker, concurrent increases in the cost of each may have “forced” the expression of slight preference for one of the items. However, when reinforcers that were unlikely to be sub- stitutes, such as a cookie and a massager, were concur- rently available and equally preferred, increases in cost had little effect on preference. (pp. 446–447)

Although Stimulus X and Stimulus Y may each func- tion as reinforcers when task demands are low, or when the reinforcement schedule is dense, when the task de- mands increase or when the schedule becomes leaner (i.e., more responses required per reinforcement), participants

may choose only Stimulus Y. DeLeon and colleagues (1997) pointed out that practitioners who are alert to these relationships might be more skeptical in believing that original preferences will be sustained under changing en- vironmental conditions, and judicious in how they plan re- inforcement delivery relative to task assignment once intervention is underway. That is, it might be better to save some types of preferred stimuli for when task de- mands are high rather than substituting them for other equally preferred stimuli when task demands are low.

Control Procedures for Positive Reinforcement Positive reinforcement control procedures are used to manipulate the contingent presentation of a potential re- inforcer and observe any effects on the future frequency of behavior. Control, as the term is used here, requires an experimental demonstration that the presentation of a stimulus contingent on the occurrence of a target response functions as positive reinforcement. Control is demon- strated by comparing response rates in the absence and presence of a contingency, and then showing that with the absence and presence of the contingency the behav- ior can be turned on and off, or up and down (Baer, Wolf, & Risley, 1968). Historically, researchers and practition- ers have used the reversal technique as the major control technique for positive reinforcement. Briefly, the rever- sal technique includes two conditions and a minimum of four phases (i.e., ABAB). In the A condition, the behav- ior is measured over time until it achieves stability in the absence of the reinforcement contingency. The absence of the contingency is the control condition. In the B con- dition, the reinforcement contingency is presented; the same target behavior continues to be measured to assess the effects of the stimulus change. The presence of the re- inforcement contingency is the experimental condition. If the rate of responding increases in the presence of the contingency, the analyst then withdraws the reinforce- ment contingency and returns to the A and B conditions to learn whether the absence and presence of the contin- gency will turn the target behavior down and up.

However, using extinction as the control condition during the reversal phase presents practical and concep- tual problems. First, withdrawing reinforcement may re- sult in extinction-produced side effects (e.g., an initial increase in response rate, emotional responses, aggres- sion) that affect the demonstration of control. Second, in some situations it may be impossible to withdraw the reinforcement contingency completely (Thompson & Iwata, 2005). For example, it is unlikely that a teacher could completely remove teacher attention

Positive Reinforcement

303

during the A condition. In addition to these problems, Thompson and Iwata (2005) noted that, although

extinction has often been successful in reversing the be- havioral effects of positive reinforcement, its use as a control procedure presents interpretive difficulties. Es- sentially, extinction does not adequately isolate the rein- forcement contingency as the variable controlling the target response, because mere stimulus presentation can- not be ruled out as an equally viable explanation. (p. 261, emphasis added)

According to Thompson and Iwata (2005), “the ideal control procedure for positive reinforcement eliminates the contingent relation between the occurrence of the target re- sponse and the presentation of the stimulus while control- ling for the effects of stimulus presentation alone” (p. 259). They reviewed the effectiveness of three variations of the re- versal technique as control procedures for determining re- inforcement: noncontingent reinforcement (NCR), differential reinforcement of other behavior (DRO), and differential reinforcement of alternative behavior (DRA).10

Noncontingent Reinforcement

Noncontingent reinforcement (NCR) is the presentation of a potential reinforcer on a fixed-time (FT) or variable- time (VT) schedule independent of the occurrence of the target behavior. The response-independent presentation of the potential reinforcer eliminates the contingent relation between the target behavior and the stimulus presenta- tion while allowing any effects of the stimulus presenta- tion alone to be detected. Thus, NCR meets Thompson and Iwata’s (2005) criteria for an ideal control procedure for positive reinforcement.

The NCR reversal technique should entail a mini- mum of five phases (ABCBC): A is a baseline condition; B is an NCR condition, where the potential reinforcer is presented on a fixed- or variable-interval schedule inde- pendent of the target behavior; and C is a condition in which the potential reinforcer is presented contingent on the occurrence of the target behavior. The B and C con- ditions are then repeated to learn whether the level of re- sponding decreases and increases as a function of the absence and presence of the response-consequence con- tingency. The quality, amount, and rate of reinforcement should be approximately the same during the contingent and noncontingent B and C conditions of the analysis.

NCR often produces persistent responding, perhaps because of accidental reinforcement that sometimes oc- curs with a response-independent schedule, or because similar EOs and antecedent stimulus conditions evoke the persistent responding. Whatever the cause, persistent 10Chapter entitled “Reversal and Alternating Treatments Designs” presents the ABAB, NCR, DRO, and DRA control techniques in the context of sin- gle-case experimental designs.

11Chapter entitled “Differential Reinforcement” describes the use of DRO and DRA as behavior change tactics for decreasing the frequency of un- desirable behavior.

responding is a limitation of the NCR control procedure because it makes achieving a reversal effect (reduced re- sponding) more time-consuming than the reversal tech- nique with extinction. Achieving the effect may require lengthy contact with the NCR schedule.

Differential Reinforcement of Other Behavior

A practitioner using differential reinforcement of other be- havior (DRO) delivers a potential reinforcer whenever the target behavior has not occurred during a set time inter- val. The DRO reversal technique includes a minimum of five phases (i.e., ABCBC): A is a baseline condition; B is a reinforcement condition, in which the potential reinforcer is presented contingent on the occurrence of the target be- havior; and C is the DRO control condition in which the potential reinforcer is presented contingent on the absence of the target behavior. The analyst then repeats the B and C conditions to determine whether the level of respond- ing decreases and increases as a function of the absence and presence of the response-consequence contingency.

The DRO schedule allows for the continued presen- tation of the reinforcement contingency during the re- versal phases of the control procedure. In one condition, the contingency is active with occurrences of the target behavior. In another condition, the contingency is active for the omission of the target behavior. The DRO control procedure may produce the reversal effect in less time than the NCR schedule, perhaps because of the elimina- tion of accidental reinforcement of the target behaviors.

Differential Reinforcement of Alternative Behavior

When differential reinforcement of an alternative be- havior (DRA) is used as a control condition, the potential reinforcer is presented contingent on occurrences of a de- sirable alternative to the target behavior.11 The DRA re- versal technique includes a minimum of five phases (i.e., ABCBC): A is a baseline condition; B is a reinforcement condition, in which the potential reinforcer is presented contingent on the occurrence of the target behavior; and C is a condition in which the potential reinforcer is pre- sented contingent on the occurrence of an alternative be- havior (i.e., DRA). The analyst will then repeat phases B and C to ascertain whether the level of responding de- creases and increases as a function of the absence and presence of the response–consequence contingency.

Thompson and Iwata (2005) summarized the limi- tations of using DRO and DRA as control conditions pro- cedures to test for positive reinforcement:

Positive Reinforcement

304

[DRO and DRA] introduce a new contingency that was not present in the original experimental arrangement. As a result, reductions in the target response under a contin- gency reversal might be attributed to either (a) termina- tion of the contingency between the target response and the reinforcer or (b) introduction of reinforcement for the absence of the target response or for the occurrence of a competing response. In addition, given that rein- forcement is provided contingent on some characteristic of responding during the contingency reversal, it may be difficult to control for the rate of stimulus presentation across experimental and control conditions. If respond- ing is not quickly reduced (DRO) or reallocated toward responses that produce reinforcement (DRA), the rate of reinforcement in the control condition may be low rela- tive to the rate of reinforcement in the experimental con- ditions. When this occurs, the contingency-reversal strategy is functionally similar to the conventional ex- tinction procedure. (p. 267)

Given the considerations for the reversal technique with extinction, and its three variations, Thompson and Iwata (2005) concluded that NCR offers the most thor- ough and unconfounded demonstration of the effects of positive reinforcement.

Using Reinforcement Effectively We offer practitioners nine guidelines for applying pos- itive reinforcement effectively. These guidelines come from three main sources: the research literatures of the experimental analysis of behavior, applied behavior analysis and our personal experiences.

Set an Easily Achieved Initial Criterion for Reinforcement

A common mistake in applications of reinforcement is setting the initial criterion for reinforcement too high, which prohibits the learner’s behavior from contacting the contingency. To use reinforcement effectively, prac- titioners should establish an initial criterion so that the participant’s first responses produce reinforcement, and then increase the criterion for reinforcement gradually as performance improves. Heward (1980) suggested the fol- lowing method for establishing initial criteria for rein- forcement based on the learner’s level of responding during baseline (see Figure 13).

Figure 13 Examples of using data from a learner’s baseline performance to set an initial criterion for reinforcement.

The criterion-setting formulas are For increasing behaviors:

baseline average < initial criterion ≤ highest performance during baseline

For decreasing behaviors: baseline average > initial criterion ≥ lowest performance during baseline

Examples

Performance Baseline Range for Target Behavior Goal Lowest Highest Average Initial Criterion

Playing alone Increase 2 min. 14 min. 6 min. 7–14 min.

Identifying letters Increase 4 letters 9 letters 5 letters 6–9 letters of the alphabet

Number of leg Increase 0 22 8 9–22 exercises completed

Percentage of math Increase 25% 60% 34% 40–60% problems correctly solved

Number of typing Decrease 16 28 22 16–21 errors in one letter

Number of calories Decrease 2,260 3,980 2,950 2,260–2,900 consumed per day

Positive Reinforcement

305

For a behavior you wish to increase, set the initial crite- rion higher than the child’s average baseline perfor- mance and lower than or equal to his best performance during baseline. For a behavior you want to decrease in frequency, the initial criterion for reinforcement should be set below the child’s average performance during baseline and greater than or equal to his lowest (or best) baseline performance. (p. 7)

Use High Quality Reinforcers of Sufficient Magnitude

Reinforcers that maintain responding on simple tasks may not have the potency to produce similar levels of re- sponding on more difficult or longer tasks. Practitioners will likely need to use a reinforcer of higher quality for behaviors that require more effort or endurance. A highly preferred stimulus chosen during preference assessments sometimes functions as a high-quality reinforcer. Neef and colleagues (1992), for example, found that behaviors that received a lower reinforcer rate but a higher quality reinforcer increased in frequency, whereas behaviors that received a higher reinforcer rate with a lower quality re- inforcer decreased in frequency. Reinforcer quality is rel- ative also to other consequences for responding currently available to the learner.

Applied behavior analysts define the magnitude (or amount) of a reinforcer as (a) the duration of time for ac- cess to the reinforcer, (b) the number of reinforcers per unit of time (i.e., reinforcer rate), or (c) the intensity of the reinforcer. Increases in reinforcer magnitude may cor- relate with an increased effectiveness of the behavior- reinforcer relation. However, the effects of reinforcer magnitude are not well understood because “few applied studies have examined the effects of magnitude on re- sponding in a single-operant arrangement” (Lerman, Kelly, Vorndran, Kuhn, & LaRue, 2002, p. 30). Consid- eration of how much reinforcement to use should follow the maxim, “Reinforce abundantly, but don’t give away the store.” We suggest that the amount of reinforcement be proportional to the quality of the reinforcer and the effort required to emit the target response.

Use Varied Reinforcers to Maintain Potent Establishing Operations

Reinforcers often decrease in potency with frequent use. Presenting an overabundance of a specific reinforcer is likely to diminish the momentary effectiveness of the re- inforcer due to satiation. Practitioners can minimize sa- tiation effects by using a variety of reinforcers. If reading a specific book on sports functions as a reinforcer and the teacher relies solely on this reinforcer, ultimately read- ing that book may no longer produce reinforcement. Con-

versely, known reinforcers that are not always available may have increased effectiveness when they are reintro- duced. If a teacher has demonstrated that “being first in line” is a reinforcer, but uses this reinforcer only once per week, the reinforcement effect will be greater than if “being first in line” is used frequently.

Varying reinforcers may enable less preferred stim- uli to function as reinforcers. For example, Bowman et al. (1997) found that some learners responded better to a va- riety of less preferred stimuli as compared to a continu- ous access to a single, more highly preferred stimulus. Also, using a variety of reinforcers may keep the potency of any one particular reinforcer higher. For example, Egel (1981) found that students’ correct responding and on- task behavior were higher when they had access to one of three randomly selected reinforcers across trials versus a constant reinforcement condition in which one of the stimuli was presented following each successful trial. Even within a session, teachers could let students select a variety of consequences from a menu. Similarly, vary- ing a property of a reinforcer may keep its reinforcing potency for a longer time. If comic books are used as re- inforcers, having several different genres of comic books available is likely to maintain their potency.

Use Direct Rather than Indirect Reinforcement Contingencies When Possible

With a direct reinforcement contingency, emitting the target response produces direct access to the reinforcer; the contingency does not require any intervening steps. With an indirect reinforcement contingency, the response does not produce reinforcement directly. The practitioner presents the reinforcer. Some research suggests that di- rect reinforcement contingencies may enhance perfor- mance (Koegel & Williams, 1980; Williams, Koegel, & Egel, 1981). Thompson and Iwata (2000), for example, linked the definitions of direct and indirect contingen- cies to the difference between automatic reinforcement (i.e., direct) and socially mediated reinforcement (i.e., indirect) and summarized their research on response ac- quisition under direct and indirect contingencies of re- inforcement this way:

Under both contingencies, completion of identical tasks (opening one of several types of containers) produced access to identical reinforcers. Under the direct contin- gency, the reinforcer was placed inside the container to be opened; under the indirect contingency, the therapist held the reinforcer and delivered it to the participant upon task completion. One participant immediately per- formed the task at 100% accuracy under both contin- gencies. Three participants showed either more immediate or larger improvements in performance

Positive Reinforcement

306

under the direct contingency. The remaining two partici- pants showed improved performance only under the di- rect reinforcement contingency. Data taken on the occurrence of “irrelevant” behaviors under the indirect contingency (e.g., reaching for the reinforcer instead of performing the task) provided some evidence that these behaviors may have interfered with task performance and that their occurrence was a function of differential stimulus control. (p. 1)

Whenever possible, practitioners should use direct reinforcement contingency, especially with learners with limited behavioral repertoires.

Combine Response Prompts and Reinforcement

Response prompts are supplementary antecedent stimuli used to occasion a correct response in the presence of an SD that will eventually control the behavior. Applied be- havior analysts give response prompts before or during the performance of a target behavior. The three major forms of response prompts are verbal instructions, mod- eling, and physical guidance.

Concerning verbal instructions, sometimes describ- ing the contingency (i.e., the verbal instructions) may function as a motivating operation for learners with ver- bal skills, thereby, making it more likely that the learner will more quickly contact the reinforcer. For example, Mayfield and Chase (2002) explained their reinforcement contingency to college students who were learning five basic algebra rules.

The reinforcement procedures were described to the par- ticipants in the general instructions administered at the beginning of the study. Participants earned money for correct answers on all the tests and were not penalized for incorrect answers. During the session following a test, participants were presented with a record of their total earnings on the test. This was the only feedback provided concerning their performance on the tests. (p. 111)

Bourret, Vollmer, and Rapp (2004) used verbal re- sponse prompts during an assessment of the vocal verbal mand repertoires of three participants with autism.

Each vocalization assessment session consisted of 10 trials, each 1 min in duration. A nonspecific prompt [de- scribing the contingency] was delivered (e.g., “If you want this, ask me for it”) 10 s after the onset of the trial. A prompt including a model of the complete targeted utterance (e.g., “If you want this, say ‘chip’ ”) was de- livered 20 s into the trial. The participant was prompted to say just the first phoneme of the targeted response (e.g., “If you want this, say ‘ch’ ”) 30 s after the initia- tion of the trial.

Reinforce Each Occurrence of the Behavior Initially

Provide reinforcement for each occurrence of the target behavior (i.e., continuous reinforcement) to strengthen behavior, primarily during the initial stages of learning a new behavior. After the behavior is established, gradu- ally thin the rate of reinforcement so that some but not all occurrences of the behavior are reinforced (i.e., inter- mittent reinforcement). For example, a teacher might ini- tially reinforce each correct response to sight words printed on flash cards and then use a ratio schedule to thin reinforcement. To firm the responses after the initial learning, provide reinforcement following two correct re- sponses for a few trials, then following each set of four correct responses, and so on. Hanley and colleagues (2001) gradually moved from a very dense fixed interval (FI) 1-second schedule of reinforcement (on an FI sched- ule, the first target response following the end of the in- terval produced reinforcement) to thinner schedule with the following increments of intervals: 2 s, 4 s, 8 s, 16 s, 25 s, 35 s, 46 s, and finally to an FI-58 seconds. For ex- ample, target responses that occurred before the end of the FI 58 seconds were not reinforced, but the first re- sponse after 58 seconds was reinforced.

Use Contingent Attention and Descriptive Praise

As discussed earlier in this chapter, social attention and praise are powerful reinforcers for many people. How- ever, behavioral improvements following praise often in- volve something more, or altogether different from, the direct effects of reinforcement. Michael (2004) discussed the common conceptual mistake of assuming that in- creased responding following praise and attention are a function of reinforcement.

Consider the common use of descriptive praise, provid- ing some general sign of social approval (a smile plus some comment such as “Good work!”) and, in addition, a brief description of the behavior that is responsible for the approval (“I like the way you’re . . .!”). When such praise is provided to a normally verbal person over 5 or 6 years of age, it probably functions as a form of in- struction or as a rule, much as if the praiser had said, “If you want my continued approval you have to . . .” For example, a factory supervisor walks up to an employee who is cleaning up an oil spill on the factory floor,

Positive Reinforcement

307

smiles broadly, and says, “George, I really like the way you’re cleaning up that spill before anyone steps in it. That’s very considerate of you.” Now suppose that George cleans up spills from that time forward—a rather larger change in behavior considering that it was fol- lowed by only a single instance of reinforcement. We might suspect that the praise functioned not simply as reinforcement but rather as a form of rule or instruction, and that George, for various reasons, provided himself with similar instruction every time another spill oc- curred. (pp. 164–165, emphasis in original)

A study by Goetz and Baer (1973) investigating the effects of teacher praise on preschool children’s creative play with building blocks used descriptive praise in one condition of the study. “The teacher remarked with in- terest, enthusiasm, and delight every time that the child placed and/or rearranged the blocks so as to create a form that had not appeared previously in that session’s con- struction(s). . . . ‘Oh, that’s very nice—that’s different’!)” (p. 212). The three 4-year-old girls increased the con- struction of block form diversity during each phase of contingent descriptive praise. Goetz and Baer did not conduct a component analysis to determine how much of the girls’ improved performance could be attributed to reinforcement in the form of positive attention (“That’s very nice!”) or to the feedback they received (“That’s dif- ferent!”), which enabled them to create a rule to follow (“Building different things with the blocks gets the teacher’s attention.”). The authors surmised that

for some children, either [reinforcing attention and de- scriptive praise] will be sufficient without the other, but that for other children, the mix of the two will be more effective than either alone. If so, then for applied pur- poses a package of positive attention and descriptive praise is probably the best technique to apply to chil- dren in general. (p. 216, words in brackets added)

We recommend that, in the absence of data showing that attention and praise have produced counter thera- peutic effects for a given learner, practitioners incorporate contingent praise and attention into any intervention en- tailing positive reinforcement.

Gradually Increase the Response-to-Reinforcement Delay

We recommended in the previous guideline that practi- tioners reinforce each occurrence of a target behavior during the initial stages of learning, and then thin the de- livery of reinforcers by switching to an intermittent sched- ule of reinforcement. Because the consequences that maintain responding in natural environments are often delayed, Stromer, McComas, and Rehfeldt (2000) re- minded us that using continuous and intermittent sched-

12Moving from continuous reinforcement to an intermittent schedule of re- inforcement is sometimes described as a means of increasing reinforcer delay (e.g., Alberto & Troutman, 2006; Kazdin, 2001). However, an inter- mittent schedule of reinforcement does not entail “delayed reinforcement” unless specified. Although only some occurrences of the target behavior are reinforced on an intermittent schedule of reinforcement (see chapter enti- tled “Schedules of Reinforcement”), reinforcement is delivered immediately following the response that meets the contingency. For example, on a fixed ratio 10 schedule of reinforcement, every tenth response produces imme- diate reinforcement. Delay-to-reinforcement or reinforcement delay de- scribes the time lapse between the response and delivery of the reinforcer after the contingency has been met (e.g., the reinforcer was delivered 45 sec- onds after every tenth response).

ules of reinforcement might be just the first steps of pro- gramming consequences for everyday situations. “Es- tablishing the initial instances of a behavioral repertoire typically requires the use of programmed consequences that occur immediately after the target response occurs. However, the job of the applied behavior analyst also in- volves the strategic use of delayed reinforcement. Be- haviors that yield delayed reinforcement are highly adaptive in everyday life, but they may be difficult to es- tablish and maintain” (p. 359).12

Examples of the tactics that applied behavior ana- lysts have used to help people learn to respond effectively for delayed consequences include: (a) a delay-to- reinforcement time interval that begins with a short delay and is then gradually increased the extent of the delay (Dixon, Rehfeldt, & Randich, 2003; Schweitzer & Sulzer- Azaroff, 1988); (b) a gradual increase in work require- ments during the delay (Dixon & Holcomb, 2000); (c) an activity during the delay to “bridge the gap” between the behavior and reinforcer (Mischel, Ebbesen, & Zeiss, 1972); and, importantly, (d) verbal instruction in the form of an as- surance that the reinforcer will be available following a delay (e.g., “The calculator will show the amount of money to be placed in a savings account for you. You will be given all the nickels in your savings account on [day]” (Neef, Mace, & Shade, 1993, p. 39).

Gradually Shift from Contrived to Naturally Occurring Reinforcers

We end this chapter with an extract from Murray Sid- man’s (2000) insightful and thought-provoking account of what he learned in the “early days” of applying be- havioral principles to human behavior. In describing a project from 1965 to 1975 that emphasized the use of positive reinforcement with boys between the ages of 6 and 20 years who were diagnosed with mental retarda- tion and living in a state institution, Sidman recollected how introducing tokens as generalized conditioned rein- forcers eventually led to praise from the project staff,

Positive Reinforcement

308

and later to learning itself, becoming powerful reinforcers for the boys.

We began with tokens, which had the advantage of being visible and easily handled. Later, after the boys had learned to save tokens and to understand numbers, we were able to introduce points. For some, points led eventually to money. As the boys saw how pleased we were when they earned the tokens and points that brought them other reinforcers, our pleasure also be- came important to them, and we became able to use praise as a reinforcer. As they learned more and more,

many of the boys found that what they learned permitted them to deal more effectively with their gradually en- larging world. For them, learning itself became reinforcing. (p. 19)

Success in manipulating the environment may be the ultimate naturally occurring reinforcer. As Skinner (1989) pointed out, this powerful reinforcer “does not need to be contrived for instructional purposes; it is unrelated to any particular kind of behavior and hence always avail- able. We call it success.” (p. 91)

Summary Definition and Nature of Positive Reinforcement

1. Positive reinforcement is a functional relation defined by a two-term contingency: A response is followed immedi- ately by the presentation of a stimulus, and, as a result, similar responses occur more frequently in the future.

2. The stimulus change responsible for the increase in re- sponding is called a reinforcer.

3. The importance of the immediacy of reinforcement must be emphasized; a response-to-reinforcement delay of just 1 second can diminish intended effects because the be- havior temporally closest to the presentation of the rein- forcer will be strengthened by its presentation.

4. The effects of long-delayed consequences on human behavior should not be attributed to the direct effect of reinforcement.

5. A misconception held by some is that reinforcement is a circular concept. Circular reasoning is a form of faulty logic in which cause and effect are confused and not in- dependent of each other. Reinforcement is not a circular concept because the two components of the response– consequence relation can be separated and the conse- quence manipulated to determine whether it increases the frequency of the behavior it follows.

6. In addition to increasing the future frequency of the be- havior it follows, reinforcement changes the function of antecedent stimuli. An antecedent stimulus that evokes be- havior because it has been correlated with the availability of reinforcement is called a discriminative stimulus (SD).

7. A discriminated operant is defined by a three-term con-

8. The momentary effectiveness of any stimulus change as reinforcement depends on an existing level of motivation with respect to that stimulus change. An establishing op- eration (EO) (e.g., deprivation) increases the current ef- fectiveness of a reinforcer; an abolishing operation (AO) (e.g., satiation) decreases the current effectiveness of a reinforcer.

9. A complete description of reinforcement of a discrimi-

10. Automaticity of reinforcement refers to the fact that a per- son does not have to understand or be aware of the relation between his behavior and a reinforcing consequence for re- inforcement to occur.

11. Reinforcement strengthens any behavior that immediately precedes it; no logical or adaptive connection between be- havior and the reinforcing consequence is necessary.

12. The development of superstitious behaviors that often ap- pear when reinforcement is presented on a fixed-time sched- ule irrespective of the subject’s behavior demonstrates the arbitrary nature of the behaviors selected by reinforcement.

13. Automatic reinforcement occurs when behaviors pro- duce their own reinforcement independent of the medi- ation of others.

Classifying Reinforcers

14. Unconditioned reinforcers are stimuli that function as re- inforcement without requiring a learning history. These stimuli are the product of phylogenic development, mean- ing that all members of a species are susceptible to the same properties of stimuli.

15. Conditioned reinforcers are previously neutral stimuli that function as reinforcers as a result of prior pairing with one or more other reinforcers.

16. A generalized conditioned reinforcer is a conditioned re- inforcer that as a result of having been paired with many unconditioned and conditioned reinforcers does not de- pend on a current EO for any particular form of rein- forcement for its effectiveness.

17. When reinforcers are described by their physical proper- ties, they are typically classified as edible, sensory, tangi- ble, activity, or social reinforcers.

18. The Premack principle states that making the opportunity to engage in a high-probability behavior contingent on the

Positive Reinforcement

tingency of SD → R → SR+

nated operant entails a four-term contingency: EO → SD → R → SR+

309

occurrence of low-frequency behavior will function as re- inforcement for the low-frequency behavior.

19. The response-deprivation hypothesis is a model for pre- dicting whether contingent access to one behavior will function as reinforcement for engaging in another behav- ior based on whether access to the contingent behavior represents a restriction of the activity compared to the baseline level of engagement.

Identifying Potential Reinforcers

20. Stimulus preference assessment refers to a variety of pro- cedures used to determine (a) the stimuli that a person prefers, (b) the relative preference values (high versus low) of those stimuli, and (c) the conditions under which those preferences values remain in effect.

21. Stimulus preference assessments can be performed by ask- ing the target person and/or significant others what the tar- get person prefers, conducting free operant observations, and conducting trial-based assessments (i.e., single-, paired-, or multiple-stimulus presentations).

22. Preferred stimuli do not always function as reinforcers, and stimulus preferences often change over time.

23. Reinforcer assessment refers to a variety of direct, data- based methods for determining the relative effects of a given stimulus as reinforcement under different and changing conditions or the comparative effectives of mul- tiple stimuli as reinforcers for a given behavior under spe- cific conditions. Reinforcer assessment is often conducted with concurrent schedules of reinforcement, multiple schedules of reinforcement, and progressive reinforce- ment schedules.

Control Procedures for Positive Reinforcement

24. Positive reinforcement control procedures are used to ma- nipulate the presentation of a potential reinforcer and ob-

serve any effects on the future frequency of behavior. Pos- itive reinforcement control procedures require a believ- able demonstration that the contingent presentation following the occurrence of a target response functions as positive reinforcement. Control is demonstrated by com- paring rates of responding in the absence and presence of a contingency, and then showing that with the absence and presence of the contingency the behavior can be turned on and off, or up and down.

25. In addition to a reversal design using the withdrawal of the reinforcement contingency (i.e., extinction) as the con- trol condition, noncontingent reinforcement (NCR), dif- ferential reinforcement of other behavior (DRO), and differential reinforcement of alternative behavior (DRA) can be used as control conditions for reinforcement.

Using Reinforcement Effectively

26. Guidelines for increasing the effectiveness of positive re- inforcement interventions include:

• Set an easily achieved initial criterion for reinforcement

• Use high quality reinforcers of sufficient magnitude

• Use varied reinforcers

• Use a direct rather than indirect reinforcement contin- gency whenever possible

• Combine response prompts and reinforcement

• Reinforce each occurrence of the behavior initially, then gradually thin reinforcement schedule

• Use contingent praise and attention

• Gradually increase the response-to-reinforcement delay

• Gradually shift from contrived to naturally occurring reinforcers

Positive Reinforcement

310

Negative Reinforcement

Key Terms

avoidance contingency conditioned negative reinforcer discriminated avoidance

escape contingency free-operant avoidance

negative reinforcement unconditioned negative reinforcer

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 3: Principles, Processes, and Concepts

3-3 Define and provide examples of (positive and) negative reinforcement.

Content Area 9: Behavior Change Procedures

9-2 Use (positive and) negative reinforcement:

(a) Identify and use (negative) reinforcers.

(b) Use appropriate parameters and schedules of (negative) reinforcement.

(d) State and plan for the possible unwanted effects of the use of (negative) reinforcement.

This chapter was written by Brian A. Iwata and Richard G. Smith.

311

Negative Reinforcement

The use of positive reinforcement in educational and therapeutic programs is so commonplace that the terms positive reinforcement and reinf-

orcement have become almost synonymous; in fact, the usual lay term for reinforcement is simply reward. Posi- tive reinforcement involves an increase in responding as a function of stimulus presentation. In a complementary way, responding can lead to the termination of a stimulus, as in turning off an alarm clock in the morning, which re- sults in the cessation of noise. When responding increases as a result of stimulus termination, learning has occurred through negative reinforcement. This chapter expands the discussion of operant contingencies to include negative reinforcement. We define negative reinforcement, distin- guish between escape and avoidance contingencies, de- scribe events that may serve as the basis for negative reinforcement, illustrate ways negative reinforcement may be used to strengthen behavior, and discuss ethical issues that arise when using negative reinforcement. Readers in- terested in more in-depth discussions of basic and applied research on negative reinforcement are referred to reviews by Hineline (1977) and Iwata (1987).

Definition of Negative Reinforcement A negative reinforcement contingency is one in which the occurrence of a response produces the removal, ter- mination, reduction, or postponement of a stimulus, which leads to an increase in the future occurrence of that re- sponse. A full description of negative reinforcement re- quires specification of its four-term contingency (see Figure 1): (a) The establishing operation (EO) for be- havior maintained by negative reinforcement is an an- tecedent event in whose presence escape (termination of the event) is reinforcing, (b) the discriminative stimulus (SD) is another antecedent event in whose presence a re- sponse is more likely to be reinforced, (c) the response is the act that produces reinforcement, and (d) the reinforcer is the termination of the event that served as the EO.

Positive versus Negative Reinforcement

Positive and negative reinforcement have a similar effect on behavior in that both produce an increase in respond- ing. They differ, however, with respect to the type of stim- ulus change that follows behavior, as illustrated in Figure 2. In both examples, a stimulus change (consequence) strengthens the behavior that preceded it: Asking the sib- ling to make a sandwich is strengthened by obtaining food; carrying rain protection is strengthened by block- ing the rain. However, behavior maintained by positive re- inforcement produces a stimulus that was absent prior to responding, whereas behavior maintained by negative re- inforcement terminates a stimulus that was present prior to responding: Food was unavailable prior to asking for it but available after (positive reinforcement); rain was landing on one’s clothing before raising the newspaper but not after (negative reinforcement).

Thus, the key distinction between positive and nega- tive reinforcement is based on the type of stimulus change that occurs following a response. Many stimulus changes have discrete onsets and offsets and involve an “all-or- none” operation. For example, one can readily see the ef- fect of turning on a television (positive reinforcement) or turning off the light in a bedroom (negative reinforce- ment). Other stimulus changes exist on a continuum from less to more, such as turning up the volume of a stereo to hear it better (positive reinforcement) or turning it down when it is too loud (negative reinforcement). Sometimes, however, it is difficult to determine whether an increase in responding resulted from positive or negative reinforce- ment because the stimulus change is ambiguous. For ex- ample, although a change in temperature can be measured quantitatively so we know that it either increased or de- creased following behavior, it is unclear whether turning on a heater when the temperature is 40° F is an example of positive reinforcement because the response “produced heat” or negative reinforcement because the response “re- moved cold.” Another example can be found in a classic study by Osborne (1969) on the use of free time as a re- inforcer in the classroom. During baseline, students were observed to get out of their seats frequently during long

Ask roommate to close window

Noise subsidesRoommate nearby Loud noise from

construction outside

R SR-SDEO

Effect on Frequency of Similar Responses

under Similar Conditions in the Future

Figure 1 Four-term contingency illustrating negative reinforcement.

312

Negative Reinforcement

Ask sibling to make sandwich

Sibling makes sandwich, child

consumes

Older sibling present when

child gets home

Child missed lunch at school

R SR+SDEO

Effect on Frequency of Similar Responses

under Similar Conditions in the Future

Buy newspaper and use as cover

Effects of rain on clothes diminished

Newsstand nearby Rain starts to fall and

lands on clothes

R SR-SDEO

Effect on Frequency of Similar Responses

under Similar Conditions in the Future

Negative Reinforcement

Positive Reinforcement

Figure 2 Four-term contingency illustrating similarities and difference between positive and negative reinforcement.

work periods. During treatment, the students were given 5 minutes of free time if they remained in their seats dur- ing 10-minute work periods, and in-seat behavior in- creased. At first glance, the free-time contingency appears to involve negative reinforcement (termination of the in- seat requirement contingent on appropriate behavior). As Osborne noted, however, activities (games, social inter- action, etc.) to which the students had access during free time may have functioned as positive reinforcement.

Given the ambiguous nature of some stimulus changes, Michael (1975) suggested that the distinction between positive and negative reinforcement, based on whether a stimulus is presented or removed, may be un- necessary. Instead, he emphasized the importance of specifying the type of environmental change produced by a response in terms of key stimulus features that com- prised both the “prechange” and “postchange” condi- tions. This practice, he proposed, would eliminate the necessity of describing the transition between prechange and postchange conditions as one involving the presen- tation or removal of stimuli and would facilitate a more complete understanding of functional relations between environment and behavior.

Little has changed since the publication of Michael’s (1975) article; the distinction between positive and nega- tive reinforcement continues to be emphasized in every text on learning principles, and citations to the term negative reinforcement have even increased in applied re- search (Iwata, 2006). In an attempt to renew the discus- sion, Baron and Galizio (2005) reiterated Michael’s position and included some additional points of emphasis.

1The term aversive is not meant to describe an inherent characteristic of a stimulus but, rather, a stimulus whose presentation functions as punish- ment or whose removal functions as negative reinforcement.

This terminological issue is a complex one that can be considered from several perspectives—conceptual, pro- cedural, and historical—and is unresolved at the current time. Readers interested in the topic are referred to a series of reactions to Baron and Galizio (Chase, 2006; Iwata, 2006; Lattal & Lattal, 2006; Marr, 2006; Michael, 2006; Sidman, 2006) and their rejoinder (Baron & Galizio, 2006).

Negative Reinforcement versus Punishment

Negative reinforcement is sometimes confused with pun- ishment for two reasons. First, because the lay term for pos- itive reinforcement is reward, people mistakenly consider negative reinforcement as the technical term for the oppo- site of reinforcement (punishment). The terms positive and negative, however, do not refer to “good” and “bad” but to the type of stimulus change (presentation versus termina- tion) that follows behavior (Catania, 1998). A second source of confusion stems from the fact that the stimuli involved in both negative reinforcement and punishment are con- sidered “aversive” to most people.1 Although it is true that the same stimulus may serve as a negative reinforcer in one context and as a punisher in a different context, both the nature of the stimulus change and its effect on behavior dif- fer. In a negative reinforcement contingency, a stimulus that was present is terminated by a response, which leads to an

313

Negative Reinforcement

increase in responding; in a punishment contingency, a stim- ulus that was absent is presented following a response, which leads to a decrease in responding. Thus, a response that terminates loud noise would increase as a function of negative reinforcement, but one that produces loud noise would decrease as a function of punishment.

Escape and Avoidance Contingencies In its simplest form, negative reinforcement involves an escape contingency, in which a response terminates (pro- duces escape from) an ongoing stimulus. An early study by Keller (1941) illustrates typical laboratory research on escape. When a rat was placed in an experimental chamber, and a bright light was turned on, the rat quickly learned to a press lever, which turned off the light. Os- borne’s (1969) study on free-time contingencies may also serve as an example of escape learning in an applied con- text. To the extent that the important feature of the con- tingency was the termination of work requirements, in-seat behavior during 10-minute work periods produced 5 minutes of escape.

Although situations involving escape are commonly encountered in everyday life (e.g., we turn off loud noises, shield our eyes from the sun, flee from an aggressor), most behavior maintained by negative reinforcement is characterized by an avoidance contingency, in which a response prevents or postpones the presentation of a stim- ulus. Returning to the previous laboratory example, an experimenter can add to the escape contingency an arrangement in which another stimulus such as a tone precedes the presentation of the bright light, and a re- sponse in the presence of the tone eliminates the presen- tation of the light or postpones it until the tone is next

presented. This type of arrangement has been called dis- criminated avoidance, in which responding in the pres- ence of a signal prevents the onset of a stimulus from which escape is a reinforcer. Because responses in the presence of the tone are reinforced, whereas those in the absence of the tone have no effect, the tone is a discrim- inative stimulus (SD) in whose presence there is an in- creased likelihood of reinforcement for responding.

Avoidance behavior also can be acquired in the ab- sence of a signal. Suppose the experimenter arranges a schedule in which the bright light turns on for 5 seconds every 30 seconds, and a response (or some number of re- sponses) at any time during the interval resets the clock to zero. This type of arrangement is known as free- operant avoidance because the avoidance behavior is “free to occur” at any time and will delay the presenta- tion of the bright light.

Each of the three types of contingencies described earlier was illustrated in an ingenious study by Azrin, Rubin, O’Brien, Ayllon, and Roll (1968) on postural slouching (see Figure 3). Participants wore an apparatus that closed an electrical circuit when slouching occurred. Closure of the switch produced an audible click, which was followed 3 seconds later by a 55-db tone. Postural correction in the presence of the tone turned off the tone (escape) but prevented the tone if correction occurred during the 3 seconds following the click (discriminated avoidance). Furthermore, maintenance of correct posture prevented the click (free-operant avoidance). A hypo- thetical example involving homework management also illustrates these contingencies. A parent who sends a child to his or her room immediately following school and al- lows the child to leave the room only after completing homework has arranged an escape contingency: Home- work completion produces escape from the bedroom. A parent who first delivers a warning (e.g., “If you don’t

Figure 3 Three types of negative reinforcement contingencies used by Azrin and colleagues (1968) to maintain correct posture.

Free-Operant Avoidance

Correct posture maintained ➞ Avoid click and tone

Discriminated Avoidance

Slouching (incorrect posture) ➞ Audible click

Correct posture within 3 seconds of click ➞ Avoid tone

Escape

Slouching (incorrect posture) ➞ Audible click

Posture not corrected within 3 seconds ➞ 55-db tone

Posture corrected ➞ Tone turns off

314

Negative Reinforcement

start your homework in 10 minutes, you’ll have to do it in your bedroom”) has arranged a discriminated avoid- ance contingency: Starting homework following the warning avoids having to do it in the bedroom. Finally, the parent who waits until later in the evening to impose the in- room requirement has arranged a free-operant avoidance contingency: Homework completion at any time after school avoids having to do it in the bedroom later.

Characteristics of Negative Reinforcement Responses Acquired and Maintained by Negative Reinforcement

It a well-known fact that aversive stimulation produces a variety of responses (Hutchinson, 1977). Some of these may be respondent behaviors (as in reflexive actions to intense stimuli), but the focus in this chapter is on oper- ant behaviors. Recall that the presentation of an aversive stimulus serves as an EO for escape and occasions be- havior that has produced escape from similar stimulation in the past. Any response that successfully terminates the stimulation will be strengthened; as a result, a wide range of behaviors may be acquired and maintained by negative reinforcement. All of these behaviors are adap- tive because they enable one to interact effectively with the environment; some behaviors, however, are more so- cially appropriate than others. As will be seen later in the chapter, negative reinforcement may play an impor- tant role in the development of academic skills, but it also can account for the development of disruptive or dangerous behavior.

Events That Serve as Negative Reinforcers

In discussing the types of stimuli that can strengthen be- havior through negative reinforcement, a problem arises when attempting to use the same terminology as that ap- plied to the description of positive reinforcers. It is quite common to refer to positive reinforcers by listing things such as food, money, praise, and so on. It is, however, the presentation of the stimulus that strengthens behav- ior: Food presentation, and not food per se, is a positive reinforcer. Nevertheless, we often simply list the stimu- lus and assume that “presentation” is understood. In a similar way, to say that negative reinforcers include shock, noise, parental nagging, and so on, is an incom- plete description. It is important to remember that a stim-

ulus described as a negative reinforcer refers to its re- moval because, as noted previously, the same stimulus serves as an EO when presented prior to behavior and as punishment when presented following behavior.

Learning History

As is the case with positive reinforcers, negative rein- forcers influence behavior because (a) we have the in- herited capacity to respond to them or (b) their effects have been established through a history of learning. Stim- uli whose removal strengthens behavior in the absence of prior learning are unconditioned negative rein- forcers. These stimuli are typically noxious events such as shock, loud noise, intense light, extremely high or low temperature, or strong pressure against the body. In fact, any source of pain or discomfort (e.g., a headache) will occasion behavior, and any response that successfully eliminates the discomfort will be reinforced. Other stim- uli are conditioned negative reinforcers, which are pre- viously neutral events that acquire their effects through pairing with an existing (unconditioned or conditioned) negative reinforcer. A bicyclist, for example, usually heads for home when seeing a heavily overcast sky be- cause dark clouds have been highly correlated with bad weather. Various forms of social coercion, such as parental nagging, are perhaps the most commonly en- countered conditioned negative reinforcers. For exam- ple, reminding a child to clean his bedroom may have little effect on the child’s behavior unless failure to re- spond is followed by another consequence such as hav- ing to stay in the room until it is clean. To the extent that the nagging is reliably “backed up” by sending the child to his room, the child will eventually respond simply to stop or prevent the nagging. It is interesting to note that, in the case of negative reinforcement, neutral events (dark sky, nagging) function as both (a) discriminative stimuli because responding in their presence constitutes avoid- ance of another consequence and (b) conditioned nega- tive reinforcers because, due to their pairing with another consequence, they become stimuli to avoid or escape.

The Source of Negative Reinforcement

Another way to classify negative reinforcers is based on how they are removed (i.e., their source). A distinction was made between socially mediated reinforcement, in which the consequence results from the action of an- other person, and automatic reinforcement, in which the consequence is produced directly by a response in- dependent of the actions of another. This distinction

315

Negative Reinforcement

also applies to negative reinforcement. Returning to the example in Figure 1, we can see that termination of the construction noise was an instance of social negative re- inforcement (the roommate’s action closed the window). The person “being bothered” by the noise, however, could simply have walked across the room and closed the win- dow (automatic negative reinforcement). This example illustrates the fact that many reinforcers can be removed or terminated either way: One can consult a physician when experiencing a headache (social) or take a pain medication (automatic), ask the teacher for help with a difficult problem (social) or persist until it is solved (au- tomatic), and so on.

Consideration of the source of negative reinforce- ment may facilitate the design of behavior change inter- ventions by determining the focus of intervention. For example, when faced with a perplexing work task, an em- ployee may finish it incorrectly just to get it out of the way (automatic reinforcement) or ask for help (social re- inforcement). Aside from reassigning the employee, the quickest solution would be to reinforce the employee’s seeking help by offering assistance. Ultimately, however, the supervisor would want to teach the employee the nec- essary skills to complete the work tasks independently.

Identifying the Context of Negative Reinforcement

There are several ways to identify positive reinforcers; the difference with negative reinforcers is that equal em- phasis must be placed on the antecedent event (EO) as well as on the reinforcing consequence because, once the behavior occurs, the negative reinforcer may be gone and cannot be observed. The identification of EOs may be difficult with people who have limited verbal ability and cannot tell someone they are experiencing aversive stim- ulation. These people may engage in other behaviors, such as tantrums, attempts to leave the situation, de- structive behavior, aggression, or even self-injury. Weeks and Gaylord-Ross (1981), for example, observed students with severe disabilities when no task, an easy task, and a difficult task were presented. Little or no problem be- havior occurred during the no-task condition, and prob- lem behavior occurred somewhat more often in the difficult-task condition than in the easy-task condition. These results suggested that the students’ problem be- havior was maintained by escape from task demands and that difficult tasks were more “aversive” than were easy tasks. However, because the consequences that followed problem behavior were unknown, it is possible that the behaviors were maintained by some other consequence, such as attention, which would be a positive reinforcer.

Iwata, Dorsey, Slifer, Bauman, and Richman (1994) developed a method for identifying the types of contin- gencies that maintain problem behavior by observing peo- ple under a series of conditions that differed with respect to both antecedent and consequent events. One condition involved the presentation of task demands (EO) and the removal of demands (escape) if problem behavior oc- curred; higher rates of problem behavior under this con- dition relative to others indicated that problem behavior was maintained by negative reinforcement.

Smith, Iwata, Goh, and Shore (1995) extended the findings of Weeks and Gaylord-Ross (1981) and Iwata et al. (1994) by identifying some characteristics of task de- mands that make them aversive. After first determining that their participants’ (people with severe disabilities) problem behavior was maintained by escape from task demands, Smith and colleagues examined several di- mensions along which tasks might differ: task novelty, duration of the work session, and rate of demand pre- sentation. Results of one of these analyses are shown in Figure 4, which depicts frequency distributions and cu- mulative records of problem behavior from the begin- ning to the end of sessions. These data illustrate the importance of individualized assessments in identifying the basis for negative reinforcement because two partic- ipants (Evelyn and Landon) showed increasing rates of problem behavior as work sessions progressed, whereas two other participants (Milt and Stan) showed the oppo- site trend.

Factors That Influence the Effectiveness of Negative Reinforcement

The factors that determine whether a negative reinforce- ment contingency will be effective in changing behav- ior are similar to those that influence positive reinforcement and are related to (a) the strength of the contingency and (b) the presence of competing contin- gencies. In general, negative reinforcement for a given re- sponse will be more effective under the following conditions:

1. The stimulus change immediately follows the oc- currence of the target response.

2. The magnitude of reinforcement is large, referring to the difference in stimulation present before and after the response occurs.

3. Occurrence of the target response consistently pro- duces escape from or postponement of the EO.

4. Reinforcement is unavailable for competing (non- target) responses.

316

Negative Reinforcement

12 10 8 6 4 2 0

Stan

Milt

Larry

Landon

Evelyn Responses per Interval

250

200

150

100

800

600

400

200

400

300

200

100

300

200

100

200

100

Cumulative Responses

Consecutive 30-Second Intervals

Figure 4 Frequency distributions (left column) and cumulative records (right column) summed over sessions of self- injurious behavior (SIB) by five adults with develop- mental disabilities as work sessions progressed. From “Analysis of Establishing Operations for Self-Injury Maintained by Escape” by R. G. Smith, B. A. Iwata, H. Goh, and B. A. Shore, 1995, Journal of Applied Behavior Analysis, 28, p. 526. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Applications of Negative Reinforcement Negative reinforcement is a fundamental principle of learning that has been studied extensively in basic re- search (Hineline, 1977). Although many examples of es- cape and avoidance learning can be found in everyday life, research in applied behavior analysis has heavily emphasized the use of positive reinforcement over neg- ative reinforcement, mostly for ethical reasons, which are noted in the final section of this chapter. Still, negative re- inforcement has been used as one means of establishing a variety of behaviors. This section illustrates several ther- apeutic uses of negative reinforcement, as well as the un-

intended role it may play in strengthening problem be- havior.

Acquisition and Maintenance of Appropriate Behavior

Chronic Food Refusal

Pediatric feeding problems are common and are espe- cially prevalent among children with developmental dis- abilities. These disorders may take a variety of forms, including selective eating, failure to consume solid foods, and complete food refusal, and may be serious enough to require tube feeding or other artificial means to ensure

317

Negative Reinforcement

adequate nutritional intake. A large proportion of feeding problems cannot be attributed to a medical cause but, in- stead, appear to be learned responses most likely main- tained by escape or avoidance.

Results from a number of studies have shown that operant learning-based interventions can be highly ef- fective in treating many childhood feeding disorders, and a study by Ahearn, Kerwin, Eicher, Shantz, and Swearingin (1996) illustrated the use of negative rein- forcement as a form of intervention. Three children ad- mitted to a hospital who had histories of chronic food refusal were first observed under a baseline (positive re- inforcement) condition in which food was presented and access to toys was available contingent on accepting food. Food refusal, however, produced escape in that it termi- nated a trial. Subsequently, the experimenters compared the effects of two interventions. One treatment condition (nonremoval of the spoon) involved presenting food and keeping the spoon positioned at the child’s lower lip until the bite was accepted. The other treatment (physical guid- ance) involved presenting food and, if the child did not ac- cept, opening the child’s mouth so that the food could be delivered. Both treatments involved a negative rein- forcement contingency because food acceptance termi- nated the trial by producing removal of the spoon or avoidance of the physical guidance.

Figure 5 shows the results obtained for the three children. All children exhibited low rates of acceptance during baseline in spite of the availability of positive re-

inforcement. The two interventions were implemented in a multiple baseline across subjects design and were com- pared in a multielement design. As can be seen in the sec- ond phase of the study, both interventions produced immediate and large increases in food acceptance. These results showed that positive reinforcement for appropri- ate behavior may have limited effects if other behaviors (refusal) produce negative reinforcement, and that nega- tive reinforcement that maintains problem behavior can be used to establish alternative behavior.

Error-Correction Strategies

Positive reinforcement is a basic motivational compo- nent of effective instruction. Teachers commonly deliver praise, privileges, and other forms of reward contingent on correct performance. Another common procedure, but one that has received less attention than positive re- inforcement, involves the correction of student errors by repeating a learning trial, having the student practice correct performance, or giving the student additional work. To the extent that correct performance avoids these remedial procedures, improvements may be just as much a function of negative reinforcement as positive reinforcement.

Worsdell and colleagues (2005) examined the relative contributions of these contingencies during behavioral acquisition. The learning task involved reading words presented on flash cards, and the intervention of interest

100

Differential SR+ of

Acceptance

Alternating Treatments

Withdrawal of NS

Physical Guidance

Physical Guidance (PG)

Nonremoval of Spoon (NS)

SR+

Calvin

100

0 Pam

Donna

Sessions 0 15 30 45 60

100

P er

ce n

ta ge

o f

T ri

al s

w it

h A

cc ep

ta n

Figure 5 Percentage of trials in which three children with histories of chronic food refusal accepted bites during a baseline condition of positive reinforcement and two treatment conditions, nonremoval of the spoon and physical guidance, both of which involved a negative reinforcement contingency. From “An Alternating Treatments Comparison of Two Intensive Interven- tions for Food Refusal” by W. H. Ahearn, M .E. Kerwin, P. S. Eicher, J. Shantz, and W. Swearingin, 1996, Journal of Applied Behavior Analysis, 29, p. 326. Copyright 1996 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

318

Negative Reinforcement

was the correct repetition of misread words. As noted by the authors, the procedure provided additional practice of correct responses but also represented an avoidance contingency. To separate these effects (in Study 3), the au- thors implemented two error correction conditions. In the “relevant” condition, which combined the effects of prac- tice and negative reinforcement, students were prompted to pronounce the misread word correctly five times con- tingent on an error. In the “irrelevant” condition, students were prompted to repeat an unrelated, nontarget word five times contingent on an error. The irrelevant condition contained only the negative reinforcement contingency because repetition of irrelevant words provided no prac- tice in correctly reading misread words.

Figure 6 shows the results of Study 3, expressed as the cumulative number of words mastered by the 9 partici- pants. All participants’ performance improved during both error-correction conditions relative to baseline, when no error-correction procedure was in effect. Performance by 3 participants (Tess, Ariel, and Ernie) was better during rel- evant error correction. However, Mark’s performance was clearly superior during irrelevant error correction, and per- formance of the remaining 5 participants (Hayley, Becky, Kara, Maisy, and Seth) was similar in both conditions. Thus, all participants showed improvement in reading performance even when they practiced irrelevant words, and most par- ticipants (6 of 9) did just as well or better practicing irrele- vant rather than relevant words. These results suggest that the

100

Tess

BL Error Correction

Relevant

Irrelevant

Becky

BL Error Correction 60

Seth

BL Error Correction

Ariel

Kara

Ernie

Hayley

Maisy

100

Mark

10 20 30 40 50 60 10 20 30 40 50 60 10 20 30 40 50 60 Sessions

C u

m u

la ti

ve N

u m

b er

o f

W o

rd s

M as

te re

Figure 6 Cumulative number of words read correctly during baseline (positive reinforcement for correct responses) and during two error- correction conditions; one in which correct responses avoided repeated practice of the misread (relevant) word, and another in which correct responses avoided practice of an unrelated (irrelevant) word. Improvements in performance in the “irrel- evant” condition indicated that negative reinforce- ment plays a role in error- correction procedures. From: “Analysis of Response Repetition as an Error-Correction Strategy During Sight-Word Reading” by A. S. Worsdell, B. A. Iwata, C. L. Dozier, A. D. Johnson, P. L. Neidert, and J. L. Thomason, 2005, Journal of Applied Behavior Analysis, 38, p. 524. Copyright 2005 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

319

Negative Reinforcement

success of many remedial (error-correction) procedures may be due at least in part to negative reinforcement.

Acquisition and Maintenance of Problem Behavior

Well-designed instructional procedures maintain a high degree of on-task behavior and lead to improved learn- ing. Occasionally, however, the presentation of task de- mands may function as an EO for escape behavior due to the difficult or repetitive nature of the work requirements. Initial forms of escape may include lack of attention or mild forms of disruption. To the extent that positive rein- forcement for compliance is less than optimal, attempts to escape may persist and may even escalate to more severe forms of problem behavior. In fact, research on the as- sessment and treatment of problem behaviors has shown that escape from task demands is a common source of negative reinforcement for property destruction, aggres- sion, and even self-injury.

O’Reilly (1995) conducted an assessment of a per- son’s episodic aggressive behavior. The participant was an adult with severe mental retardation who attended a vocational day program. To determine whether aggres- sive behavior was maintained by positive versus nega- tive reinforcement, O’Reilly observed the participant under two conditions, which were alternated in a multi- element design. In one condition (attention), a therapist ignored the participant (EO) except to deliver reprimands following aggression (positive reinforcement). In the sec- ond condition (demand), a therapist presented difficult tasks to the participant (EO) and briefly terminated the trial following aggression (negative reinforcement).

As Figure 7 shows, aggressive behavior occurred more often in the demand condition, indicating that it was

maintained by negative reinforcement. Because anecdo- tal reports suggested that the participant also was more likely to be aggressive following nights when he had not slept well, the data for both conditions were further di- vided based on whether the participant slept for more or less than 5 hours the previous night. The highest rates of aggression occurred following sleep deprivation. These data are particularly interesting in that they illustrate the influence of two antecedent events on behavior maintained by negative reinforcement: Work tasks functioned as EOs for escape but even more so in the absence of sleep.

Behavioral Replacement Strategies

Problem behaviors maintained by negative reinforce- ment can be treated in a number of ways. One strategy is to strengthen a more socially appropriate replacement behavior using negative reinforcement, as illustrated in a study by Durand and Carr (1987). After determining that the “stereotypic” behaviors of four special educa- tion students were maintained by escape from task de- mands, the authors taught the students an alternative response (“Help me”), which was followed by assistance with the task at hand. As can be seen in Figure 8, all students engaged in moderate-to-high levels of stereo- typy during baseline. After being taught to use the phrase “Help me,” the students began to exhibit that behavior, and their stereotypy decreased.

Results of the Durand and Carr (1987) study showed that an undesirable behavior could be replaced with a desirable one; however, the replacement behavior might be considered less than ideal because it did not neces- sarily facilitate better task performance. This was shown

100

P er

ce n

ta ge

o f

In te

rv al

s o

f A

gg re

ss io

1 5 10 15 Sessions

Demand (Home) Attention (Voc Facility)

Attention (Home) Demand (Voc Facility)

Demand (Home) Attention (Voc Facility)

Demand < 5 hrs sleep Demand > 5 hrs sleep Attention < 5 hrs sleep Attention > 5 hrs sleep

Figure 7 Data showing that the effects of work tasks as EOs for escape- maintained aggression by an adult male with severe mental retardation were exac- erbated by sleep deprivation. From “Functional Analysis and Treatment of Escape- Maintained Aggression Correlated with Sleep Deprivation” by M. F. O’Reilly, 1995, Journal of Applied Behavior Analysis, 28, p. 226. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

320

Negative Reinforcement

R o

ck in

Baseline Treatment

R o

ck in

Baseline TreatmentKen

Jim

0H an

d F

la p

p in

Baseline TreatmentBob

0H an

d F

la p

p in

Baseline TreatmentLen

P er

ce n

ta ge

o f

In te

rv al

1 5 10 15 20 25 Sessions

Figure 8 Percentage of intervals of stereotypic behaviors maintained by escape from task demands of four special education students during baseline and treatment in which the students were taught an alternative response (“Help me”) to attain assistance with the task at hand. Shaded bars show students’ use of the “Help me” response. From “Social Influences on ‘Self- Stimulatory’ Behavior: Analysis and Treatment Application” by V. M. Durand and E. G. Carr, E. G., 1987, Journal of Applied Behavior Analysis, 20, 128. Copyright 1987 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

in a subsequent study by Marcus and Vollmer (1995). After collecting baseline data on a young girl’s compli- ant and disruptive behavior, the authors compared the effects of two treatments in a reversal design. In one con- dition, which was called DNR (differential negative re- inforcement) communication, the girl was given a brief

break from the task when she said “Finished.” In the sec- ond condition, called DNR compliance, the girl was given a break after complying with an instruction (the criterion for a break was later increased to compliance with three instructions). The results of this comparison (see Figure 9) showed that both treatments produced

4.0

3.0

2.0

1.0

D is

ru p

ti o

n s

p er

M in

u te

100

5 10 15 20 25 30 35 40 Sessions

DNR (Communication)

Compliance

Disruption

DNR (Compliance)

P er

ce n

ta ge

o f

In st

ru ct

io n

s w

it h

C o

m p

lia n

Figure 9 Disruptions and compli- ance by a 5-year-old girl during baseline and two differential negative rein- forcement conditions. From “Effects of Differential Negative Reinforcement on Disruption and Compliance” by B. A. Marcus and T. R. Vollmer, 1995, Journal of Applied Behavior Analysis, 28, p. 230. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

321

Negative Reinforcement

Summary Definition of Negative Reinforcement

1. Negative reinforcement involves the termination, reduc- tion, or postponement of a stimulus contingent on the oc- currence of a response, which leads to an increase in the future occurrence of that response.

2. A negative reinforcement contingency involves (a) an es- tablishing operation (EO) in whose presence escape is re- inforcing, (b) a discriminative stimulus (SD) in whose presence a response is more likely to be reinforced, (c) the response that produces reinforcement, and (d) termination of the event that served as the EO.

3. Positive and negative reinforcement are similar in that both lead to an increase in responding; they differ in that posi- tive reinforcement involves contingent stimulus presenta- tion, whereas negative reinforcement involves contingent stimulus termination.

4. Negative reinforcement and punishment differ in that (a) negative reinforcement involves contingent stimulus ter- mination, whereas punishment involves contingent stim- ulation, and (b) negative reinforcement leads to an increase in responding, whereas punishment leads to a decrease in responding.

Escape and Avoidance Contingencies

5. An escape contingency is one in which responding termi- nates an ongoing stimulus. An avoidance contingency is one in which responding delays or prevents the presenta- tion of a stimulus.

6. In discriminated avoidance, responding in the presence of a signal prevents stimulus presentation; in free-operant avoidance, responding at any time prevents stimulus pre- sentation.

Characteristics of Negative Reinforcement

7. Any response that successfully terminates aversive stim- ulation will be strengthened; as a result, a wide range of be- haviors may be acquired and maintained by negative reinforcement.

8. Negative reinforcement may play an important role in the development of academic skills, but it also can account for the development of disruptive or dangerous behavior.

9. Unconditioned negative reinforcers are stimuli whose re- moval strengthens behavior in the absence of prior learn- ing. Conditioned negative reinforcers are stimuli whose removal strengthens behavior as a result of previous pair- ing with other negative reinforcers.

10. Social negative reinforcement involves stimulus termina- tion through the action of another person. Automatic neg- ative reinforcement involves stimulus termination as a direct result of a response.

11. Identification of negative reinforcers requires the specifi- cation of the stimulus conditions in effect prior to and fol- lowing responding.

12. In general, negative reinforcement for a given response will be more effective when (a) the stimulus change im- mediately follows the occurrence of the target response, (b) the magnitude of reinforcement is large, (c) the target re- sponse consistently produces escape from or postpone- ment of the EO, and (d) reinforcement is unavailable for competing responses.

Applications of Negative Reinforcement

13. Although negative reinforcement is a fundamental prin- ciple of learning that has been studied extensively in basic research, applied behavior analysis has heavily

marked reductions in disruptive behavior. However, only the DNR compliance condition produced an increase in task performance.

Ethical Considerations in the Use of Negative Reinforcement Ethical concerns about the use of positive and negative re- inforcement are similar and arise from the severity of the antecedent event (EO) that occasions behavior. Most EOs for behavior maintained by positive reinforcement can be characterized as deprivation states, which, if severe, can constitute undue restriction of rights. By contrast, most

EOs for behavior maintained by negative reinforcement can be viewed as aversive events. Extremely noxious events, when presented as antecedent stimuli, cannot be justified as part of a typical behavior change program.

Another concern with negative reinforcement is that the presence of aversive stimuli can itself generate be- haviors that compete with the acquisition of desired be- havior (Hutchinson, 1977; Myer, 1971). For example, a socially withdrawn child, when placed in the midst of others, may simply scream and run away instead of play- ing with the peers, and running away is incompatible with social interaction. Finally, undesirable side effects typi- cally associated with punishment might also be observed when implementing behavior change programs based on negative reinforcement.

322

Negative Reinforcement

cially appropriate replacement behavior through nega- tive reinforcement.

Ethical Considerations in the Use of Negative Reinforcement

18. Ethical concerns about the use of positive and negative re- inforcement are similar and arise from the severity of the antecedent event (EO) that occasions behavior. Most EOs for behavior maintained by negative reinforcement can be viewed as aversive events. Extremely noxious events, when presented as antecedent stimuli, cannot be justified as part of a typical behavior change program.

19. Another concern with negative reinforcement is that the presence of aversive stimuli can itself generate behaviors that compete with the acquisition of desired behavior.

emphasized the use of positive reinforcement over neg- ative reinforcement.

14. Applied researchers have explored the therapeutic uses of negative reinforcement in treating pediatric feeding problems.

15. Improvements in student performance as a result of error correction that involve repeating a learning trial, having the student practice correct performance, or giving the student additional work may be a function of negative reinforcement.

16. The presentation of task demands during instruction may function as an EO for escape; initial forms of escape may include lack of attention or mild forms of disruption. To the extent that positive reinforcement for compliance is less than optimal, escape behaviors may persist and may even escalate.

17. One strategy for treating problem behaviors maintained by negative reinforcement is to strengthen a more so-

323

Schedules of Reinforcement

Key Terms

adjunctive behaviors alternative schedule (alt) chained schedule of

reinforcement (chain) compound schedule of

reinforcement concurrent schedule (conc) conjunctive schedule (conj) continuous reinforcement (CRF) differential reinforcement of

diminishing rates (DRD)

differential reinforcement of high rates (DRH)

differential reinforcement of low rates (DRL)

fixed interval (FI) fixed ratio (FR) intermittent schedule of

reinforcement (INT) limited hold matching law mixed schedule (mix)

multiple schedule (mult) postreinforcement pause progressive schedule of

reinforcement ratio strain schedule of reinforcement schedule thinning tandem schedule (tand) variable interval (VI) variable ratio (VR)

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 9: Behavior Change Procedures

9-2 (b) Use appropriate parameters and schedules of reinforcement.

9-6 Use differential reinforcement.

9-24 Use the matching law and recognize factors influencing choice.

324

A schedule of reinforcement is a rule that de- scribes a contingency of reinforcement, those environmental arrangements that determine

conditions by which behaviors will produce reinforce- ment. Continuous reinforcement and extinction provide the boundaries for all other schedules of reinforcement. A schedule of continuous reinforcement (CRF) pro- vides reinforcement for each occurrence of behavior. For example, a teacher using a continuous schedule of rein- forcement would praise a student each time she identified a sight word correctly. Examples of behaviors that tend to produce continuous reinforcement include turning on a water faucet (water comes out), answering a telephone after it rings (a voice is heard), and putting money into a vending machine (a product is obtained). During extinc- tion (EXT), no occurrence of the behavior produces re- inforcement.

Intermittent Reinforcement Between continuous reinforcement and extinction many intermittent schedules of reinforcement (INT) are pos- sible in which some, but not all, occurrences of the be- havior are reinforced. Only selected occurrences of behavior produce reinforcement with an intermittent schedule of reinforcement. CRF is used to strengthen be- havior, primarily during the initial stages of learning new behaviors. Applied behavior analysts use intermittent re- inforcement to maintain established behaviors.

Maintenance of Behavior

Maintenance of behavior refers to a lasting change in be- havior. Regardless of the type of behavior change tech- nique employed or the degree of success during treatment, applied behavior analysts must be concerned with sustaining gains after terminating a treatment pro- gram. For example, Mary is in the seventh grade and tak- ing French, her first foreign language class. After a few weeks, the teacher informs Mary’s parents that she is fail- ing the course. The teacher believes that Mary’s prob- lems in French have resulted from lack of daily language practice and study. The parents and teacher decide that Mary will record a tally on a chart kept on the family bul- letin board each evening that she studies French for 30 minutes. Mary’s parents praise her practice and study ac- complishments and offer encouragement. During a fol- low-up meeting 3 weeks later, the parents and teacher decide that Mary has done so well that the tally procedure can be stopped. Unfortunately, a few days later Mary is once again falling behind in French.

A successful program was developed to establish daily French language practice. However, gains did not maintain after removing the tally procedure. The par- ents and the teacher did not establish intermittent rein- forcement procedures. Let us review what happened and what could have happened. Continuous reinforcement was used correctly to develop daily study behavior. However, after the study behavior was established and the tally procedure removed, the parents should have continued to praise and encourage daily practice and gradually offer fewer encouragements. The parents could have praised Mary’s accomplishments after every second day of daily practice, then every fourth day, then once per week, and so on. With the intermittent praise, Mary might have continued daily practice after remov- ing the tally procedure.

Progression to Naturally Occurring Reinforcement

A major goal of most behavior change programs is the de- velopment of naturally occurring activities, stimuli, or events to function as reinforcement. It is more desirable for people to read because they like to read, rather than to obtain contrived reinforcement from a teacher or par- ent; to engage in athletics for the enjoyment of the activ- ity, rather than for a grade or because of a physician’s directive; to help around the house for the personal sat- isfaction it brings, rather than to earn an allowance. In- termittent reinforcement is usually necessary for the progression to naturally occurring reinforcement. Even though some individuals spend hours each day practicing a musical instrument because they enjoy the activity, chances are good that this persistent behavior developed gradually. At first the beginning music student needs a great deal of reinforcement to continue the activity: “You really practiced well today,” “I can’t believe how well you played,” “Your mother told me you received a first place in the contest—that’s super!” These social conse- quences are paired with other consequences from teach- ers, family members, and peers. As the student develops more proficiency in music, the outside consequences occur less frequently, intermittently. Eventually, the stu- dent spends long periods making music without receiv- ing reinforcement from others because making music has itself become a reinforcer for doing that activity.

Some might explain the transition of our music stu- dent from an “externally reinforced person” to a “self- reinforced musician” as the development of intrinsic motivation, which seems to imply that something inside the person is responsible for maintaining the behavior. This view is incorrect from a behavioral standpoint. Ap- plied behavior analysts describe intrinsic motivation as reinforcement that is received by manipulating the

Schedules of Reinforcement

325

physical environment. Some individuals ride bicycles, go backpacking, read, write, or help others because manip- ulations of the environment provide reinforcement for engaging in those activities.

Defining Basic Intermittent Schedules of Reinforcement Ratio and Interval Schedules

Applied behavior analysts directly or indirectly embed ratio and interval intermittent schedules of reinforce- ment in most treatment programs, especially ratio sched- ules (Lattal & Neef, 1996). Ratio schedules require a number of responses before one response produces re- inforcement. If the ratio requirement for a behavior is 10 correct responses, only the 10th correct response pro- duces reinforcement. Interval schedules require an elapse of time before a response produces reinforcement. If the interval requirement is 5 minutes, reinforcement is pro- vided contingent on the first correct response that oc- curs after 5 minutes has elapsed since the last reinforced response.

Ratio schedules require a number of responses to be emitted for reinforcement; an elapse of time does not change the number contingency. The participant’s re- sponse rate, however, determines the rate of reinforce- ment. The more quickly the person completes the ratio requirement, the sooner reinforcement will occur. Con- versely, interval schedules require an elapse of time be- fore a single response produces reinforcement. The total number of responses emitted on an interval schedule is ir- relevant to when and how often the reinforcer will be de- livered. Emitting a high rate of response during an interval schedule does not increase the rate of reinforcement. Re- inforcement is contingent only on the occurrence of one response after the required time has elapsed. The avail- ability of reinforcement is time-controlled with interval schedules, and rate of reinforcement is “self-controlled” with ratio schedules, meaning that the more quickly the individual completes the ratio requirement, the sooner reinforcement will occur.

Fixed and Variable Schedules

Applied behavior analysts can arrange ratio and interval schedules to deliver reinforcement as a fixed or a vari- able contingency. With a fixed schedule, the response ratio or the time requirement remains constant. With a variable schedule, the response ratio or the time require-

ment can change from one reinforced response to an- other. The combinations of ratio or interval and fixed or variable contingencies define the four basic schedules of intermittent reinforcement: fixed ratio, variable ratio, fixed interval, and variable interval.

The following sections define the four basic sched- ules of intermittent reinforcement, provide examples of each schedule, and present some well-established sched- ule effects derived from basic research.

Fixed Ratio Defined

A fixed ratio (FR) schedule of reinforcement requires the completion of a number of responses to produce a re- inforcer. For example, every fourth correct (or target) re- sponse produces reinforcement on an FR 4 schedule. An FR 15 schedule means that 15 responses are required to produce reinforcement. Skinner (1938) conceptualized each ratio requirement as a response unit. Accordingly, the response unit produces the reinforcer, not just the last response of the ratio.

Some business and industrial tasks are paid on an FR schedule (e.g., piecework). A worker might receive a pay credit after completing a specified number of tasks (e.g., assembling 15 pieces of equipment or picking a box of or- anges). A student might receive either a happy face after learning 5 new sight words or a certain number of points after completing 10 math problems.

De Luca and Holborn (1990) reported a comparison of obese and nonobese children’s rate of pedaling an ex- ercise bicycle under baseline and FR schedules of rein- forcement. The baseline and FR conditions used the same duration of exercise. After establishing a stable rate of pedaling during baseline, De Luca and Holborn intro- duced an FR schedule that matched the rate of rein- forcement produced during baseline. All participants increased their rate of pedaling with the introduction of the FR schedule.

Fixed Ratio Schedule Effects

Consistency of Performance

FR schedules produce a typical pattern of responding: (a) After the first response of the ratio requirement, the participant completes the required responses with little hesitation between responses; and (b) a postrein- forcement pause follows reinforcement (i.e., the partic- ipant does not respond for a period of time following reinforcement). The size of the ratio influences the duration of the postreinforcement pause: Large ratio

Schedules of Reinforcement

326

requirements produce long pauses; small ratios produce short pauses.

Rate of Response

FR schedules often produce high rates of response. Quick responding on FR schedules maximizes the delivery of re- inforcement because the quicker the rate of response, the greater the rate of reinforcement. People work rapidly with a fixed ratio because they receive reinforcement with the completion of the ratio requirements. Computer key- boarders (typists) who contract their services usually work on an FR schedule. They receive a specified amount for the work contracted. A typist with a 25-page manu- script to complete is likely to type at the maximum rate. The sooner the manuscript is typed, the sooner payment is received, and the more work the typist can complete in a day.

The size of the ratio can influence the rate of re- sponse on FR schedules. To a degree, the larger the ratio requirement, the higher the rate of response. A teacher could reinforce every third correct answer to arithmetic facts. With this ratio requirement, the student might com- plete 12 problems within the specified time, producing re- inforcement four times. The student might complete more problems in less time if the teacher arranged reinforce- ment contingent on 12 correct answers rather than 3. The higher ratio is likely to produce a higher rate of response. The rate of response decreases, however, if the ratio re- quirements are too large. The maximum ratio is deter- mined in part by the participant’s past FR history of reinforcement, motivating operations, the quality of the reinforcer, and the procedures that change the ratio re- quirements. For example, if ratio requirements are raised gradually over an extended period of time, extremely high ratio requirements can be reached.

Figure 1 summarizes the schedule effects typically produced by FR schedules of reinforcement.

Variable Ratio Defined

A variable ratio (VR) schedule of reinforcement re- quires the completion of a variable number of responses to produce a reinforcer. A number representing the aver- age (e.g., mean) number of responses required for rein- forcement identifies the VR schedule. For example, with a VR 10 schedule every tenth correct response on the av- erage produces reinforcement. Reinforcement can come after 1 response, 20 responses, 3 responses, 13 responses, or n responses, but the average number of responses re- quired for reinforcement is 10 (e.g., 1 + 20 + 3 + 13 + 18 = 55; 55/5 = 10).

The operation of a slot machine, the one-armed ban- dit, provides a good example of a VR schedule. These machines are programmed to pay off only a certain pro- portion of the times they are played. A player cannot pre- dict when the next operation of the machine will pay off. The player might win 2 or 3 times in succession and then not win again for 20 or more plays.

De Luca and Holborn (1992) examined the effects of a VR schedule on three obese and three nonobese chil- dren’s rate of pedaling an exercise bicycle. The children could use the exercise bicycle Monday to Friday during each week of the analysis, but received no encourage- ment to do so. The participants received the instruction to “exercise as long as you like” to initiate the baseline con- dition. De Luca and Holborn introduced the VR schedule of reinforcement after establishing a stable baseline rate of pedaling. They calculated the baseline mean number of pedal revolutions per minute and programmed the first VR contingency at approximately 15% faster pedaling than the baseline mean. The children received points on

Schedules of Reinforcement

Definition: Reinforcement delivered contingent on emission of a specified number of responses.

Stylized Graphic Curve of Cumulative Responses:

Schedule Effects: After reinforcement a postreinforcement pause occurs. After the pause the ratio requirement is completed with a high rate of response and very little hesitation between responses. The size of the ratio influences both the pause and the rate.

R es

p o

n se

b a

Time

a = postreinforcement pause

b = high rate of response "run"

c = reinforcer delivered upon emission of nth

response Figure 1 Summary of FR schedule effects during ongoing reinforcement.

327

Baseline VR 80 VR 115 VR 130 VR 130BL

SCOTT Nonobese

140

120

100

160

80 60 40

Baseline VR 85 VR 115 VR 125 VR 125BL

SHAWN Nonobese

130 110 90

150

70 50 40

Baseline VR 85 VR 115 VR 125 VR 125BL

STEVE Nonobese

130 110 90

150

70 50 40

Baseline VR 70 VR 95 VR 100 VR 100BL

PETER Obese

120

100

140

Baseline VR 80 VR 105 VR 120 VR 120BL

PAUL Obese

120 100

140

80 60 40

Baseline VR 70 VR 90 VR 110 VR 110BL

PERRY Obese

120 100

140

80 60 40

Sessions

1 5 10 15 20 25 30 35 40

M ea

n R

ev o

lu ti

o n

s p

er M

in u

Figure 2 Mean revolutions per minute during baseline, VR 1 (VR range, 70 to 85), VR 2 (VR range 90 to 115), VR 3 (VR range 100 to 130), return to baseline, and return to VR 3 phases for obese and nonobese subjects. From “Effects of a Variable-Ratio Reinforcement Schedule with Changing Criteria on Exercise in Obese and Nonobese Boys” by R. V. De Luca and S. W. Holborn, 1992, Journal of Applied Behavior Analysis, 25 p. 674. Copyright 1992 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

the VR schedule to exchange for backup reinforcers. De Luca and Holborn increased the VR schedule in two additional increments by approximately 15% per incre- ment. All participants had systematic increases in their rate of pedaling with each VR value, meaning that the larger the variable ratio, the higher the rate of response. De Luca and Holborn reported that the VR schedule pro- duced higher rates of response than did the FR schedule in their previous study (De Luca & Holborn, 1990).

Figure 2 presents the participants’ performances under baseline and VR (i.e., VR ranges 70 to 85, 90 to 115, 100 to 130) conditions.

Student behaviors usually produce reinforcement following the completion of variable ratios. Usually a student cannot predict when the teacher will call on him to give an answer, and receive reinforcement. Good grades, awards, promotions—all may come after an unpredictable number of responses. And in checking

Schedules of Reinforcement

328

Definition: Reinforcer is delivered after the emission of a variable number of responses.

Stylized Graphic Curve of Cumulative Responses:

Schedule Effects: Ratio requirements are completed with a very high rate of response and little hesitation between responses. Postreinforcement pauses are not a characteristic of the VR schedule. Rate of response is influenced by the size of the ratio requirements.

R es

p o

n se

Time

a = high, steady rate of responding

b = reinforcement delivered after a varying number of required responses are emitted

Figure 3 Summary of VR schedule effects during ongoing reinforcement.

seatwork, the teacher might reinforce a student’s work after the completion of 10 tasks, another student’s work after 3 tasks, and so on.

Variable Ratio Schedule Effects

Consistency of Performance

VR schedules produce consistent, steady rates of re- sponse. They typically do not produce a postreinforce- ment pause, as do FR schedules. Perhaps the absence of pauses in responding is due to the absence of information about when the next response will produce reinforce- ment. Responding remains steady because the next re- sponse may produce reinforcement.

Rate of Response

Like the FR schedule, the VR schedule tends to produce a quick rate of response. Also similar to the FR schedule, the size of the ratio influences the rate of response. To a de- gree, the larger the ratio requirement, the higher the rate of response. Again like FR schedules, when variable ratio re- quirements are thinned gradually over an extended period of time, participants will respond to extremely high ratio requirements. Figure 3 summarizes the schedule effects typically produced by VR schedules of reinforcement.

Variable Ratio Schedules in Applied Settings

Basic researchers use computers to select and program VR schedules of reinforcement. VR schedules used in ap- plied settings are seldom implemented with a planned and systematic approach. In other words, the reinforcer is de- livered by chance, hit or miss in most interventions. This

nonsystematic delivery of reinforcement is not an effec- tive use of VR schedules. Teachers can select and preplan VR schedules that approximate the VR schedules used in basic research. For example, teachers can plan variable ratios by (a) selecting a maximum ratio for a given activ- ity (e.g., 15 responses) and (b) using a table of random numbers to produce the specific variable ratios for the schedule of reinforcement. A table of random numbers might produce the following sequence of ratios: 8, 1, 1, 14, 3, 10, 14, 15, and 6, producing a VR 8 schedule of rein- forcement (on the average each 8th response produces the reinforcer) with the ratios ranging from 1 to 15 responses.

Teachers can apply the following VR procedures as individual or group contingencies of reinforcement for academic or social behavior:

Tic-Tac-Toe VR Procedure 1. The teacher establishes a maximum number for the

individual student or group. The larger the maxi- mum number selected, the greater the odds against meeting the contingency. For example, 1 chance out of 100 has less chance of being selected than 1 chance out of 20.

2. The teacher gives the individual or group a tic- tac-toe grid.

3. Students fill in each square of the grid with a num- ber no greater than the maximum number. For ex- ample, if the maximum number is 30, the score sheet might look like this:

Schedules of Reinforcement

329

4. The teacher fills a box or some other type of con- tainer with numbered slips of paper (with numbers no higher than the maximum number). Each num- ber should be included several times; for example, five 1s, five 2s, five 3s.

5. Contingent on the occurrence of the target behav- ior, students withdraw one slip of paper from the box. If the number on the paper corresponds with a number on the tic-tac-toe sheet, the students mark out that number on the grid.

6. The reinforcer is delivered when students have marked out three numbers in a row—horizontally, vertically, or diagonally.

For example, a student might withdraw one slip of paper for each homework assignment completed. Se- lecting an activity from the class job board (e.g., teacher’s helper, collecting milk money, running the projector) could serve as the consequence for marking out three numbers in a row.

Classroom Lottery VR Procedure 1. Students write their names on index cards after

successfully completing assigned tasks.

2. Students put signature cards into a box located on the teacher’s desk.

3. After an established interval of time (e.g., 1 week), the teacher draws a signature card from the box and declares that student the winner. The lottery can have first, second, and third place, or any number of winners. The more cards students earn, the greater is the chance that one of their cards will be picked.

Teachers have used classroom lotteries with a vari- ety of student accomplishments, such as nonassigned book reading. For example, for each book read, students write their names and the titles of the book they have read on a card. Every 2 weeks the teacher picks one card from the box and gives the winning student a new book. To make the book an especially desirable consequence, the teacher lets students earn the privilege of returning the book to the school, inscribed with the student’s name, class, and date (e.g., Brian Lee, fifth grade, donated this book to the High Street Elementary School Library on May 22, 2007).

Desk Calendar VR Procedure 1. Students receive desk calendars with loose-leaf

date pages secured to the calendar base.

2. The teacher removes the loose-leaf date pages from the calendar base.

3. The teacher establishes a maximum ratio for the students.

4. The teacher numbers index cards consecutively from 1 to the maximum ratio. Multiple cards are

included for each number (e.g., five 1s, five 2s). If a large average ratio is desired, the teacher in- cludes more large numbers; for small average ra- tios, the teacher uses smaller numbers.

5. The teacher uses a paper punch to punch holes in the index cards for attaching the cards to the cal- endar base.

6. The teacher or student shuffles the index cards to quasi-randomize the order and attaches the index cards to a calendar base face down.

7. Students produce their own VR schedules by turn- ing over one index card at a time. After meeting that ratio requirement, students flip the second card to produce the next ratio, and so on.

Students can use the desk calendar base to program VR schedules for most curriculum area (e.g., arithmetic facts). For example, after receiving an arithmetic work- sheet, the student flips the first card. It has a 5 written on it. After completing five problems, she holds up her hand to signal her teacher that she has completed the ratio re- quirement. The teacher checks the student’s answers, pro- vides feedback, and presents the consequence for correct problems. The student flips the second card; the ratio re- quirement is 1. After completing that single problem, she receives another consequence and flips the third card. This time the ratio is 14. The cycle continues until all of the cards in the stack are used. New cards can then be added or old cards reshuffled to create a new sequence of numbers. The average of the numbers does not change in the reshuffling.

Fixed Interval Schedules

A fixed interval (FI) schedule of reinforcement provides reinforcement for the first response following a fixed du- ration of time. With an FI 3-minute schedule, the first re- sponse following the elapse of 3 minutes produces the reinforcer. A common procedural misunderstanding with the FI schedule is to assume that the elapse of time alone is sufficient for the delivery of a reinforcer, assuming that the reinforcer is delivered at the end of each fixed interval of time. However, more time than the fixed interval can elapse between reinforced responses. The reinforcer is available after the fixed time interval has elapsed, and it remains available until the first response. When the first response occurs sometime after the elapse of a fixed in- terval, that response is immediately reinforced, and the timing of another fixed interval is usually started with the delivery of the reinforcer. This FI cycle is repeated until the end of the session.

Actual examples of FI schedules in everyday life are difficult to find. However, some situations do approxi- mate and in reality function as FI schedules. For

Schedules of Reinforcement

330

example, mail is often delivered close to a fixed time each day. An individual can make many trips to the mailbox to look for mail, but only the first trip to the mailbox fol- lowing the mail delivery will produce reinforcement. Many textbook examples of FI schedules, such as the mail example, do not meet the definition of an FI sched- ule; but the examples do appear similar to an FI schedule. For example, receiving a paycheck as wages for work by the hour, day, week, or month is contingent on the first re- sponse on payday that produces the paycheck. Of course, receiving the paycheck requires many responses during the interval that eventually lead to receiving the paycheck. In a true FI schedule, responses during the interval do not influence reinforcement.

FI schedules are relatively easy to use in applied settings. A teacher could make reinforcement available on an FI 2-minute schedule for correct answers on an arithmetic worksheet. The teacher or student could use an electronic timer with a countdown function to sig- nal the elapse of the 2-minute interval. The student’s first correct answer following the interval produces re- inforcement, and then the teacher resets the timer for another 2-minute interval. Similarly, the teacher could use small timing instruments such as the Gentle Re- minder (dan@gentlereminder.com) and MotivAiders (www.habitchange.com) that vibrate to signal the elapse of an interval.

Fixed Interval Schedule Effects

Consistency of Performance

FI schedules typically produce a postreinforcement pause in responding during the early part of the interval. An ini- tially slow but accelerating rate of response is evident to-

ward the end of the interval, usually reaching a maximum rate just before delivery of the reinforcer. This gradually accelerating rate of response toward the end of the inter- val is called an FI scallop because of the rounded curves that are shown on a cumulative graph (see Figure 4).

FI postreinforcement pause and scallop effects can be seen in many everyday situations. When college students are assigned a term paper, they typically do not rush to the library and start to work on the paper immediately. More often they wait a few days or weeks before starting to work. However, as the due date approaches, their work on the assignment increases in an accelerating fashion, and many are typing the final draft just before class. Cram- ming for a midterm or final examination is another ex- ample of the FI scallop effect.

These examples with the reinforcement pause and scallop effects appear to be produced by FI schedules of reinforcement. They are not, however, because like the paycheck example, college students must complete many responses during the interval to produce the term paper or a good grade on the examinations, and the term paper and examinations have deadlines. With FI schedules, re- sponses during the interval are irrelevant, and FI sched- ules have no deadlines for the response.

Why does an FI schedule produce a characteristic pause and scallop effect? After adjustment to an FI sched- ule, participants learn (a) to discriminate the elapse of time and (b) that responses emitted right after a reinforced response are never reinforced. Therefore, extinction dur- ing the early part of the interval might account for the postreinforcement pause. The effects of FI and FR sched- ules of reinforcement are similar in that both schedules produce postreinforcement pauses. However, it is im- portant to recognize the different characteristics of be- havior that emerge under each schedule. Responses under

Schedules of Reinforcement

Definition: The first correct response after a designated and constant amount of time produces the reinforcer.

Stylized Graphic Curve of Cumulative Responses:

Schedule Effects: FI schedules generate slow to moderate rates of responding with a pause in responding following reinforcement. Responding begins to accelerate toward the end of the interval.

R es

p o

n se

Time

a = postreinforcement pause b = increase in response

rates as interval progresses and reinforcer becomes available

c = reinforcer delivered contingent on first correct response after interval

Figure 4 Summary of FI schedule effects during ongoing reinforcement.

331

Definition: The first correct response following varying intervals of time produces the reinforcer.

Stylized Graphic Curve of Cumulative Responses:

Schedule Effects: A VI schedule generates a slow to moderate response rate that is constant and stable. There are few, if any, post- reinforcement pauses with VI schedules.

R es

p o

n se

Time

a = steady response rate; few, if any, postreinforcement pauses

b = reinforcer delivered Figure 5 Summary of VI schedule effects during ongoing reinforcement.

an FR schedule are emitted at a consistent rate until com- pleting the ratio requirement, whereas responses under an FI schedule begin at a slow rate and accelerate toward the end of each interval.

Rate of Responding

Overall, FI schedules tend to produce a slow to moder- ate rates of response. The duration of the time interval influences the postreinforcement pause and the rate of response; to a degree, the larger the fixed interval re- quirement, the longer the postreinforcement pause and the lower the overall rate of response.

Variable Interval Schedules

A variable interval (VI) schedule of reinforcement pro- vides reinforcement for the first correct response fol- lowing the elapse of variable durations of time. The distinguishing feature of VI schedules is that “the inter- vals between reinforcement vary in a random or nearly random order” (Ferster & Skinner, 1957, p. 326). Be- havior analysts use the average (i.e., mean) interval of time before the opportunity for reinforcement to describe VI schedules. For example, in a VI 5-minute schedule the average duration of the time intervals between rein- forcement and the opportunity for subsequent reinforce- ment is 5 minutes. The actual time intervals in a VI 5-minute schedule might be 2 minutes, 5 minutes, 3 min- utes, 10 minutes, or n minutes (or seconds).

An example of VI reinforcement in everyday situa- tions occurs when one person telephones another person whose line is busy. This is a VI schedule because a vari- able interval of time is necessary for the second person to conclude the telephone conversation and hang up so that another call can be connected. After that interval the first dialing of the second person’s number will probably pro-

duce an answer (the reinforcer). The number of responses (attempts) does not influence the availability of rein- forcement in a VI schedule; no matter how many times the busy number is dialed, the call will not be completed until the line is free. And the time interval is unpredictable in a VI schedule: The busy signal may last for a short or long time.

Variable Interval Schedule Effects

Consistency of Performance

A VI schedule of reinforcement tends to produce a con- stant, stable rate of response. The slope of the VI sched- ule on a cumulative graph appears uniform with few pauses in responding (see Figure 5). A VI schedule typ- ically produces few hesitations between responses. For example, pop quizzes at unpredictable times tend to oc- casion more consistent study behavior from students than do quizzes scheduled at fixed intervals of time. Further- more, students are less apt to engage in competing off- task behaviors during instructional and study periods when a pop quiz is likely. The pop quiz is used often as an example of a VI schedule because the performance ef- fect is similar to a VI performance. The pop quiz does not represent a true VI schedule, however, because of the required responses during the interval, and the deadline for receiving reinforcement.

Rate of Responding

VI schedules of reinforcement tend to produce low to moderate rates of response. Like the FI schedule, the av- erage duration of the time intervals on VI schedules in- fluences the rate of response; to a degree, the larger the average interval, the lower the overall rate of response. Figure 5 summarizes the schedule effects typically pro- duced by VI schedules during ongoing reinforcement.

Schedules of Reinforcement

332

Variable Interval Schedules in Applied Settings

Basic researchers use computers to select and program VI schedules of reinforcement, as they do with VR sched- ules. Teachers seldom apply VI schedules in a planned and systematic way. For example, a teacher might set an electronic countdown timer with varied intervals of time ranging from 1 minute to 10 minutes without any prior plan as to which intervals or which order will be used. This set-them-as-you-go selection of intervals approxi- mates the basic requirements for a VI schedule; however, it is not the most effective way of delivering reinforce- ment on a VI schedule. A planned, systematic applica- tion of varied intervals of time should increase the effectiveness of a VI schedule.

For example, applied behavior analysts can select the maximum time interval, whether in seconds or min- utes, that will maintain performance and still be appro- priate for the situation. Preferably, applied behavior analysts will use data from a direct assessment to guide the selection of the maximum VI interval, or at the least clinical judgment based on direct observation. Analysts can use a table of random numbers to select the varied in- tervals between 1 and the maximum interval, and then identify the VI schedule by calculating an average value for the VI schedule. The VI schedule may need adjust- ments following the selection of time intervals. For ex- ample, if a larger average interval of time appears reasonable, the teacher can replace some of the smaller intervals with larger ones. Conversely, if the average ap- pears too large, the teachers can replace some of the higher intervals with smaller ones.

Interval Schedules with a Limited Hold

When a limited hold is added to an interval schedule, reinforcement remains available for a finite time follow- ing the elapse of the FI or VI interval. The participant will miss the opportunity to receive reinforcement if a targeted response does not occur within the time limit. For example, on an FI 5-minute schedule with a limited hold of 30 seconds, the first correct response following the elapse of 5 minutes is reinforced, but only if the re- sponse occurs within 30 seconds after the end of the 5- minute interval. If no response occurs within 30 seconds, the opportunity for reinforcement has been lost and a new interval begins. The abbreviation LH identifies interval schedules using a limited hold (e.g., FI 5-minute LH 30- second, VI 3-minute LH 1-minute). Limited holds with interval schedules typically do not change the overall re- sponse characteristics of FI and VI schedules beyond a possible increase in rate of response.

Martens, Lochner, and Kelly (1992) used a VI sched- ule of social reinforcement to increase the academic en- gagement of two 8-year-old boys in a third-grade classroom. The classroom teacher reported that the boys had serious off-task behaviors. The experimenter wore an earphone connected to a microcassette recorder con- taining a 20-second fixed-time cueing tape. The cueing tape was programmed for a VI schedule of reinforcement in which only some of the 20-second intervals provided the opportunity for reinforcement in the form of verbal praise for academic engagement. If the boys were not academically engaged when the VI interval timed out, they lost that opportunity for reinforcement until the next cue. Thus, this VI schedule entailed a very short limited hold for the availability of reinforcement. Following base- line, the experimenter delivered contingent praise on a VI 5-minute or VI 2-minute schedule that alternated daily on a quasi-random basis. Both boys’ academic engage- ment on the VI 5-minute schedule resembled their base- line engagement. Both students had a higher percentage of academic engagement on the VI 2-minute schedule than they had during baseline and VI 5-minute condi- tions. Figure 6 presents percentages of academic en- gagement across baseline and VI conditions.

Thinning Intermittent Reinforcement Applied behavior analysts often use one of two proce- dures for schedule thinning. First, they thin an existing schedule by gradually increasing the response ratio or the duration of the time interval. If a student has answered addition facts effectively and responded well to a CRF schedule for two or three sessions, the teacher might thin the reinforcement contingency slowly from one correct addition fact (CRF) to a VR 2 or VR 3 schedule. The stu- dent’s performance should guide the progression from a dense schedule (i.e., responses produce frequent rein- forcement) to a thin schedule (i.e., responses produce less frequent reinforcement). Applied behavior analysts should use small increments of schedule changes during thinning and ongoing evaluation of the learner’s perfor- mance to adjust the thinning process and avoid the loss of previous improvements.

Second, teachers often use instructions to clearly communicate the schedule of reinforcement, facilitating a smooth transition during the thinning process. Instruc- tions include rules, directions, and signs. Participants do not require an awareness of environmental contin- gencies for effective intermittent reinforcement, but in- structions may enhance the effectiveness of interventions when participants are told what performances produce reinforcement.

Schedules of Reinforcement

333

100

P er

ce n

t E

n ga

ge m

en t

Baseline Treatment

2 4 6 8 10 12 14 16 18 20 22 24

Sessions

V15

V12

Bob

100

P er

ce n

t E

n ga

ge m

en t

Baseline Treatment

2 4 6 8 10 12 14 16 18 20 22 24

Sessions

V15

V12

Mark

Figure 6 Percentage of academic engagement for each child across all conditions in Experiment 2. From “The Effects of Variable-Interval Reinforcement on Academic Engage- ment: A Demonstration of Matching Theory” by B. K. Martens, D. G. Lochner, and S. Q. Kelly, 1992, Journal of Applied Behavior Analysis, 25, p. 149. Copyright 1992 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Ratio strain can result from abrupt increases in ratio requirements when moving from denser to thinner rein- forcement schedules. Common behavioral characteris- tics associated with ratio strain include avoidance, aggression, and unpredictable pauses in responding. Ap- plied behavior analysts should reduce the ratio require- ment when ratio strain is evident. The analyst can again gradually thin ratio requirements after recovering the be- havior. Small and gradual increases in ratio requirements help to avoid the development of ratio strain. Ratio strain will occur also when the ratio becomes so large that the reinforcement cannot maintain the response level or the response requirement exceeds the participant’s physio- logical capabilities.

Variations on Basic Intermittent Schedules of Reinforcement Schedules of Differential Reinforcement of Rates of Responding

Applied behavior analysts frequently encounter behav- ior problems that result from the rate that people perform certain behaviors. Responding too infrequently, or too often, may be detrimental to social interactions or aca-

demic learning. Differential reinforcement provides an intervention for behavior problems associated with rate of response. Differential reinforcement of particular rates of behavior is a variation of ratio schedules. Delivery of the reinforcer is contingent on responses occurring at a rate either higher than or lower than some predetermined criterion. The reinforcement of responses higher than a predetermined criterion is called differential reinforce- ment of high rates (DRH). When responses are rein- forced only when they are lower than the criterion, the schedule provides differential reinforcement of low rates (DRL). DRH schedules produce a higher rate of responding. DRL schedules produce a lower rate of responding.

Applied behavior analysts use three definitions of DRH and DRL schedules. The first definition states that reinforcement is available only for responses that are sep- arated by a given duration of time. This first definition is sometimes called spaced-responding DRH or spaced- responding DRL. An interresponse time (IRT) identifies the duration of time that occurs between two responses. IRT and rate of response are functionally related. Long IRTs produce low rates of responding; short IRTs pro- duce high rates of responding. Responding on a DRH schedule produces reinforcement whenever a response occurs before a time criterion has elapsed. If the time cri- terion is 30 seconds, the participant’s response produces reinforcement only when the IRT is 30 seconds or less.

Schedules of Reinforcement

334

Under the DRL schedule, a response produces rein- forcement when it occurs after a time criterion has elapsed. If the stated DRL time criterion is again 30 sec- onds, a response produces reinforcement only when the IRT is 30 seconds or greater.

This first definition of DRH and DRL as IRT sched- ules of reinforcement has been used almost exclusively in laboratory settings. There are two apparent reasons for its lack of application in applied settings: (a) Most ap- plied settings do not have sufficient automated equip- ment to measure IRT and to deliver reinforcement using an IRT criterion; and (b) reinforcement is delivered usu- ally, but not necessarily, following each response that meets the IRT criterion. Such frequent reinforcement would disrupt student activity in most instructional set- tings. However, with increased use of computers for tu- torial and academic response practice, opportunities increasingly should become available for using IRT-based schedules of reinforcement to accelerate or decelerate academic responding. Computers can monitor the pauses between academic responses and provide consequences for each response meeting the IRT criterion, with little disruption in instructional activity.

Based on the laboratory procedures for programming DRL schedules presented previously, Deitz (1977) la- beled and described two additional procedures for using differential reinforcement of rates of responding in ap- plied settings: full-session DRH or DRL and interval DRH or DRL. Deitz initially used the full-session and interval procedures as a DRL intervention for problem behaviors. The full-session and interval procedures, how- ever, apply also for DRH.

A DRH full-session schedule provides reinforcement if the total number of responses during the session meets or exceeds a number criterion. If the participant emits fewer than the specified number of responses during the session, the behavior is not reinforced. The DRL full- session schedule is procedurally the same as the DRH schedule, except reinforcement is provided for respond- ing at or below the criterion limit. If the participant emits more than the specified number of responses during the session, reinforcement is not delivered.

The interval definition for DRH and DRL schedules states that reinforcement is available only for responses that occur at a minimum or better rate of response over short durations of time during the session. To apply an in- terval DRH schedule, the applied behavior analyst orga- nizes the instructional session into equal intervals of time and dispenses a reinforcer at the end of each interval when the student emits a number of responses equal to, or greater than, a number criterion. The interval DRL schedule is procedurally like the DRH interval schedule, except that reinforcement is provided for responding at or below the criterion limit.

The differential reinforcement of diminishing rates (DRD) schedule provides reinforcement at the end of a predetermined time interval when the number of re- sponses is less than a criterion that is gradually decreased across time intervals based on the individual’s perfor- mance (e.g., fewer than five responses per 5 minutes, fewer than four responses per 5 minutes, fewer than three responses per 5 minutes, etc.). Deitz and Repp (1973) used a group DRD contingency to reduce off-task talk- ing of 15 high school senior girls. They set the first DRD criterion limit at five or fewer occurrences of off-task talking during each 50-minute class session. The DRL criterion limits were then gradually reduced to three or fewer, one or fewer, and finally no responses. The stu- dents earned a free Friday class when they kept off-task talking at or below the DRD limit Monday through Thursday.

The previous example of a DRD schedule used an identical procedure as described for the full-session DRL. DRD is also a procedural variation on interval DRL schedules described by Deitz (1977) and Deitz and Repp (1983). The typical procedure for using an inter- val DRL as an intervention for problem behavior pro- vided reinforcement contingent on emitting one or no responses per brief interval. After the problem behavior stabilizes at the initial criterion, the applied behavior an- alyst maintains the maximum criterion of one or no re- sponses per interval, but increases the duration of the session intervals to further diminish the behavior. In- creasing the duration of session intervals continues grad- ually until the problem behavior achieves a terminal low rate of responding.

Later Deitz and Repp (1983) programmed the inter- val DRL with a criterion greater than one response per in- terval, then gradually diminished the maximum number of responses per interval while the duration of the inter- val remained constant (e.g., fewer than five responses per 5 minutes, fewer than four responses per 5 minutes, fewer than three responses per 5 minutes, etc.). The DRD sched- ule and the interval DRL schedule that use a maximum number criterion greater than one per interval are differ- ent terms for the same procedure. Full-session and inter- val DRL have a long history of application in applied behavior analysis. DRD offers applied behavior analysts a new, and perhaps improved, label for the interval DRL procedure.

Progressive Schedules of Reinforcement

A progressive schedule of reinforcement systematically thins each successive reinforcement opportunity inde- pendent of the participant’s behavior. Progressive ratio

Schedules of Reinforcement

335

(PR) and progressive interval (PI) schedules of rein- forcement change schedule requirements using (a) arith- metic progressions to add a constant amount to each successive ratio or interval or (b) geometric progressions to add successively a constant proportion of the preced- ing ratio or interval (Lattal & Neef, 1996). Progressive schedules of reinforcement are often used for reinforcer assessment and behavioral intervention as described in the following sections.

Using Progressive Schedules for Reinforcer Assessment

Applied behavior analysts typically use a dense sched- ule of reinforcement (e.g., CRF) during reinforcer as- sessment while presenting preferred stimuli to increase or maintain existing behavior. However, Roane, Lerman, and Vorndran (2001) cautioned that “reinforcement ef- fects obtained during typical reinforcer assessments may have limited generality to treatment efficacy when sched- ule thinning and other complex reinforcement arrange- ments are used” (p. 146). They made an important clinical point by showing that two reinforcers could be equally ef- fective for dense schedules of reinforcement, but differ- entially effective when the schedule of reinforcement requires more responses per reinforcement. Progressive schedules of reinforcement provide an assessment pro- cedure for identifying reinforcers that will maintain treat- ment effects across increasing schedule requirements. During the session, progressive schedules are typically thinned to the “breaking point,” when the participant stops responding. Comparing the breaking points and corresponding number of responses associated with each reinforcer can identify relative reinforcement effects.

Using Progressive Schedules for Intervention

Applied behavior analysts have used progressive sched- ules to develop self-control (e.g., Binder, Dixon, & Ghezzi, 2000; Dixon & Cummins, 2001). For example, Dixon and Holcomb (2000) used a progressive sched- ule to develop cooperative work behaviors and self- control of six adults dually diagnosed with mental retardation and psychiatric disorders. The adults partic- ipated in two groups comprised of three men in Group 1 and three women in Group 2. During a natural baseline condition, the groups received instruction to exchange or share cards to complete a cooperative task of sorting playing cards into piles by categories (i.e., hearts with hearts, etc.). Dixon and Holcomb terminated a natural baseline session for the group when one of the adults quit sorting cards.

The groups received points for working on the card- sorting task during the choice baseline condition and the self-control training condition. Groups exchanged their points earned for items such as soda pop or cassette play- ers, ranging in values from 3 points to 100 points.

During the choice baseline conditions, the group’s participants could choose an immediate 3 points before doing the card sorting or a delayed 6 points after sorting the cards. Both groups chose the immediate smaller num- ber of points rather than the larger amount following a delay to reinforcement.

During self-control training, the participants were asked while working on a cooperative task, “Do you want 3 points now, or would you like 6 points after sorting the cards for Z minutes and seconds?” (pp. 612–613). The delay was initially 0 seconds for both groups. The pro- gressive delay to reinforcement ranged from an increase of 60 seconds to 90 seconds following each session that the group performance met the exact criterion for num- ber of seconds of task engagement. The terminal goals for the delay to reinforcement were 490 seconds for Group 1 and 772 seconds for Group 2. Both groups achieved these delay-to-reinforcement goals. Following the introduction of the progressive delay procedure, both groups improved their cooperative work engagement and the self-control necessary to select progressively larger delays to reinforcement that resulted in more points earned. Figure 7 shows the performance of both groups of adults during natural baselines, choice baselines, and self-control training conditions.

Compound Schedules of Reinforcement Applied behavior analysts combine the elements of con- tinuous reinforcement (CRF), the four intermittent sched- ules of reinforcement (FR, VR, FI, VI), differential reinforcement of various rates of responding (DRH, DRL), and extinction (EXT) to form compound sched- ules of reinforcement. Elements from these basic sched- ules can occur

• successively or simultaneously;

• with or without discriminative stimuli; and

• as a reinforcement contingency for each element in- dependently, or a contingency formed by the combi- nation of all elements (Ferster & Skinner, 1957).

Concurrent Schedules

A concurrent schedule (conc) of reinforcement occurs when (a) two or more contingencies of reinforcement (b) operate independently and simultaneously (c) for two

Schedules of Reinforcement

336

520 480 440 400 360 320 280 240 200 160 120 80 40 0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

900 800 700 600 500 400 300 200 100

N.B. C.B. S.C.T.

Peter left study

Goal

G ro

u p

E n

ga ge

m en

t (s

ec o

n d

Sessions

Goal

Group 1

Group 2

= no reinforcement

= failed attempt at large reinforcer

= small reinforcer

= large reinforcer

Figure 7 Number of seconds of engagement in the concurrent delay activity of cooperative card sorting during natural baseline (N.B.), choice baseline (C.B.), and self-control training (S.C.T.) for each group of participants. Filled circles represent performance at exactly the criterion level, and X data points represent the number of seconds of engagement below the criterion. From “Teaching Self-Control to Small Groups of Dually Diagnosed Adults” by M. R. Dixon and S. Holcomb, 2000, Journal of Applied Behavior Analysis, 33, p. 613. Copyright 1992 by the Society for the Experimental Analysis of Behavior Inc. Reprinted by permission.

or more behaviors. People in the natural environment have opportunities for making choices among concur- rently available events. For example, Sharon receives a weekly allowance from her parents contingent on doing daily homework and cello practice. After school she can choose when to do homework and when to practice the cello, and she can distribute her responses between these two simultaneously available schedules of reinforcement. Applied behavior analysts use concurrent schedules for reinforcer assessment and for behavioral interventions.

Using Concurrent Schedules for Reinforcer Assessment

Applied behavior analysts have used concurrent sched- ules extensively to provide choices during the assessment of consequence preferences and the assessment of re- sponse quantities (e.g., force, amplitude) and reinforcer quantities (e.g., rate, duration, immediacy, amount). Re- sponding to concurrent schedules provides a desirable assessment procedure because (a) the participant makes choices, (b) making choices during assessment approxi- mates the natural environment, (c) the schedule is effec- tive in producing hypotheses about potential reinforcers operating in the participant’s environment, and (d) these assessments require the participant to choose between stimuli rather than indicating a preference for a given stimulus (Adelinis, Piazza, & Goh, 2001; Neef, Bicard, & Endo, 2001; Piazza et al., 1999).

Roane, Vollmer, Ringdahl, and Marcus (1998) pre- sented 10 items to a participant, 2 items at a time. The

participant had 5 seconds to select 1 item by using a reaching response to touch the selected item. As a con- sequence for the selection, the participant received the item for 20 seconds. The analyst verbally prompted a re- sponse if the participant did not respond within 5 sec- onds, waiting another 5 seconds for the occurrence of a prompted response. Items were eliminated from the as- sessment (a) if they were not chosen during the first five presentations or (b) if they were chosen two or fewer times during the first seven presentations. The partici- pant made a total of 10 choices among the remaining items. The number of selections out of the 10 opportuni- ties served as a preference index.

Using Concurrent Schedules for Intervention

Applied behavior analysts have used concurrent sched- ules extensively for improving vocational, academic, and social skills in applied settings (e.g., Cuvo, Lerch, Leurquin, Gaffaney, & Poppen, 1998; Reid, Parsons, Green, & Browning, 2001; Romaniuk et al., 2002). For example, Hoch, McComas, Johnson, Faranda, and Guen- ther (2002) arranged two concurrent response alterna- tives for three boys with autism. The boys could play in one setting with a peer or sibling, or play alone in an- other area. Hoch and colleagues manipulated the dura- tion of access to toys (i.e., reinforcer magnitude) and preference (i.e., reinforcer quality). In one condition, the magnitude and quality of the reinforcer was equal in both settings. In the other condition, the magnitude and qual- ity of the reinforcer was greater for play in the setting

Schedules of Reinforcement

337

100

P er

ce n

ta ge

o f

R es

p o

n se

s A

ll o

ca te

d t

o P

ee r

Equal Magnitude

Unequal Magnitude— Paired

Equal Magnitude

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39

100

P er

ce n

ta ge

o f

R es

p o

n se

s A

ll o

ca te

d t

o N

o ve

l P

ee rs

Equal Magnitude

Unequal Magnitude— Paired

Equal Magnitude

1 2 3 4 5 6 7 8 9 10 11

Robbie

100

P er

ce n

ta ge

o f

R es

p o

n se

s A

ll o

ca te

d t

o S

ib li

n g

Unequal Quality— Unpaired

Unequal Quality— Paired

Equal Quality (High)

2 4

Equal Quality (Low)

6 8 10 12 14 16 18 20

Abe

Sessions

Figure 8 Percentage of responses allo- cated to the play area with the peer across experimental sessions (top panel) and in natural-setting probes with different peers in the classroom (middle panel) for the analysis of magnitude of reinforcement with Robbie, and the percentage of responses allocated to the play area with the sibling across experimental ses- sions for the analysis of quality of reinforcement with Abe (bottom panel). From “The Effects of Magnitude and Quality of Reinforcement on Choice Responding During Play Activities” by H. Hoch, J. J. McComas, L. Johnson, N. Faranda, and S. L. Guenther, 2002, Journal of Applied Behavior Analysis, 35, p. 177. Copyright 1992 by the Society for the Experimental Analysis of Behavior Inc. Reprinted by permission.

with a peer or sibling than in the play-alone setting. With the introduction of the condition with greater magnitude and quality of the reinforcer, the boys allocated more play responses to the setting with the peer or sibling, rather than playing alone. The magnitude and quality of the re- inforcer influenced choices made by the three boys. Figure 8 reports the percentage of responses allocated to the concurrent play areas.

Concurrent Performances: Formalizing the Matching Law

Cuvo and colleagues (1998) reported that concurrent schedules typically produce two response patterns. With concurrent interval schedules (conc VI VI, conc FI FI), participants “typically do not allocate all of their re- sponses exclusively to the richer schedule [i.e., the sched- ule producing the higher rate of reinforcement]; rather, they distribute their responding between the two sched- ules to match or approximate the proportion of rein- forcement that is actually obtained on each independent schedule” (p. 43). Conversely, with concurrent ratio

schedules (conc VR VR, conc FR FR), participants are sensitive to the ratio schedules and tend to maximize re- inforcement by responding primarily to the ratio that pro- duces the higher rate of reinforcement.

Williams (1973) identified three types of interactions found with concurrent schedules. First, when similar re- inforcement is scheduled for each of the concurrent re- sponses, the response receiving the higher frequency of reinforcement will increase in rate whereas a corre- sponding decrease will occur in the response rate of the other behavior. Second, when one response produces re- inforcement and the other produces punishment, re- sponses associated with punishment will decrease in occurrence. That decrease may produce a higher rate of response for the behavior producing reinforcement. Third, with a concurrent schedule programmed for one response to produce reinforcement and the other response to pro- duce avoidance of an aversive stimulus, the rate of avoid- ance responding will accelerate with an increase in the intensity or the frequency of the aversive stimulus. As avoidance responding accelerates, typically responding on the reinforcement schedule will then decrease.

Schedules of Reinforcement

338

Schedules of Reinforcement

The characteristics of performance on concurrent schedules as detailed previously by Cuvo and colleagues and Williams are consistent with the relationships formal- ized by Herrnstein (1961, 1970) as the matching law. The matching law addresses response allocation to choices avail- able with concurrent schedules of reinforcement. Basically, the rate of responding typically is proportional to the rate of reinforcement received from each choice alternative.

Discriminative Schedules of Reinforcement

Multiple Schedules

A multiple schedule (mult) presents two or more basic schedules of reinforcement in an alternating, usually ran- dom, sequence. The basic schedules within the multiple schedule occur successively and independently. A dis- criminative stimulus is correlated with each basic sched- ule, and that stimulus is present as long as the schedule is in effect.

Academic behaviors can become sensitive to the con- trol of multiple schedules of reinforcement. A student might respond to basic arithmetic facts with her teacher, and also with her tutor. With the teacher, the student re- sponds to arithmetic facts during small-group instruction. The tutor then provides individual instruction and prac- tice on the facts. This situation follows a multiple sched- ule because there is one class of behavior (i.e., math facts), a discriminative stimulus for each contingency in effect (i.e., teacher/tutor, small group/individual), and dif- ferent conditions for reinforcement (i.e., reinforcement is less frequent in group instruction). In another everyday example of the multiple schedule, Jim helps his mother and father clean house on Friday afternoons and Saturday mornings. Jim cleans his grandmother’s bedroom and bathroom on Friday afternoons and the family room and downstairs bathroom on Saturday mornings. Jim receives $5 per week for cleaning his grandmother’s rooms but does not receive money for cleaning the family room or downstairs bathroom. Again, there is one class of behav- iors of interest (i.e., cleaning the house), a cue for each contingency in effect (i.e., grandmother’s rooms on Fri- days or other rooms on Saturdays), and different sched- ules of reinforcement associated with the different cues (i.e., $5 for grandmother’s rooms and no money for the other rooms).

Chained Schedules

A chained schedule (chain) is similar to a multiple schedule. The multiple and chained schedules have two or more basic schedule requirements that occur succes- sively, and have a discriminative stimulus correlated with

each independent schedule. A chained schedule differs from a multiple schedule in three ways. First, the basic schedules in a chain schedule always occur in a specific order, never in the random or unpredictable order of mul- tiple schedules. Second, the behavior may be the same for all elements of the chain, or different behaviors may be required for different elements in the chain. Third, conditioned reinforcement for responding in the first el- ement in a chain is the presentation of the second ele- ment; conditioned reinforcement for responding in the second element is presentation of the third element, and so on until all elements in the chain have been completed in a specific sequence. The last element normally pro- duces unconditioned reinforcement in a laboratory set- ting, or unconditioned or conditioned reinforcement in applied settings.

The following example shows an elaborate sequence of different behaviors that must occur in a specific order. To service a bicycle headset, the mechanic will complete a chain with 13 components: (1) Disconnect the front brake cable; (2) remove handlebar and stem; (3) remove front wheel; (4) remove locknut; (5) unscrew adjusting race; (6) take fork out of frame; (7) inspect races; (8) grease and replace bearing balls for lower stack; (9) grease and replace bearing balls for upper race; (10) grease threads of steering column; (11) put fork into frame and thread the screwed race; (12) return lock washer; (13) adjust and lock the headset. The final out- come (i.e., a clean, greased, and adjusted bicycle headset) is contingent on the completion of all 13 components.

Nondiscriminative Schedules of Reinforcement

Mixed Schedules

The mixed schedule (mix) uses a procedure identical to the multiple schedules, except the mixed schedule has no discriminative stimuli correlated with the independent schedules. For example, with a mix FR 10 FI 1 sched- ule, reinforcement sometimes occurs after the comple- tion of 10 responses and sometimes occurs with the first correct response after a 1-minute interval from the pre- ceding reinforcement.

Tandem Schedules

The tandem schedule (tand) uses a procedure identical to the chained schedule, except, like the mix schedule, the tandem schedule does not use discriminative stimuli with

339

Schedules of Reinforcement

the elements in the chain. After a participant makes 15 re- sponses on a tand FR 15 FI 2, then the first correct re- sponse following an elapse of 2 minutes produces reinforcement.

Antecedent stimuli appear to relate functionally to most occurrences of behaviors in natural environments. Perhaps, therefore, the mixed and tandem schedules have little applied application at this time. However, basic re- search has produced considerable data concerning the ef- fects of mixed and tandem schedules on behavior. It may become more apparent how applied behavior analysts can effectively apply mixed and tandem schedules in as- sessment, intervention, and analysis as the knowledge base of applied behavior analysis continues to develop.

Schedules Combining the Number of Responses and Time

Alternative Schedules

An alternative schedule (alt) provides reinforcement whenever the requirement of either a ratio schedule or an interval schedule—the basic schedules that comprise the alt—is met, regardless of which of the component schedule’s requirements is met first. With an alt FR 50 FI 5-minute schedule, reinforcement is delivered whenever either of these two conditions has been met: (a) 50 cor- rect responses, provided the 5-minute interval of time has not elapsed; or (b) the first response after the elapse of 5 minutes, provided that fewer than 50 responses have been emitted.

For instance, a teacher using an alt FR 25 FI 3-minute schedule of reinforcement assigns 25 math problems and assesses the student’s correct and incorrect answers fol- lowing the elapse of 3 minutes. If the student completes the 25 problems before the elapse of 3 minutes, the teacher checks the student’s answers and provides a con- sequence consistent with the FR 25 schedule. However, if the ratio requirement of 25 math problems has not been completed after an elapse of 3 minutes, the first correct answer following the 3 minutes produces reinforcement. The alternative schedule offers the advantage of a sec- ond chance for reinforcement if the student has not met the FR requirement in a reasonable amount of time. The FI provides reinforcement for one response, and that one reinforced response might encourage continued re- sponding with the new start of the FR requirement.

Conjunctive Schedules

A conjunctive schedule (conj) of reinforcement is in ef- fect whenever reinforcement follows the completion of response requirements for both a ratio schedule and an in-

terval schedule of reinforcement. For example, a student behavior produces reinforcement when at least 2 minutes have elapsed and 50 responses have been made. This arrangement is a conj FI 2 FR 50 schedule of reinforce- ment. With the conjunctive schedule of reinforcement, the first response following the conclusion of the time interval produces reinforcement if the criterion number of responses has been completed.

A 14-year-old boy with autism had higher rates of aggression with two of his four therapists during in- struction. The higher rates of aggression were directed toward the two therapists who previously worked with the boy at a different treatment facility. Progar and col- leagues (2001) intervened to reduce the levels of ag- gression with the therapists from the different facility to the levels that occurred with the other two therapists in the current setting. The boy’s aggression occurred in de- mand situations (e.g., making his bed) and was escape maintained. The initial intervention used three conse- quences: (1) a 10-minute chair time-out for attempts to choke, (2) escape extinction, and (3) differential-rein- forcement-other-behavior for the omission of aggression during the 10-minute sessions. This intervention was identical to the treatment used with the boy at the other facility. It was ineffective in reducing the boy’s aggres- sion in the current setting.

Because of the ineffectiveness of the initial inter- vention, Progar and colleagues added a conj FR VI-DRO schedule of reinforcement to their initial intervention. They delivered edible reinforcers contingent on com- pleting a three-component task such as dusting or straightening objects (i.e., an FR 3 schedule) and the omission of aggression for an average of every 2.5 min- utes (i.e., the VI-DRO 150-second). An occurrence of ag- gression reset the conj schedule. (Note: Resetting this conj schedule used a standard procedure because any oc- currence of the problem behavior during a DRO interval immediately resets the time to the beginning of the in- terval.) Progar and colleagues demonstrated that the conj FR VI-DRO schedule produced a substantial reduction in aggression directed toward the two therapists previ- ously from the other treatment facility.

Duvinsky and Poppen (1982) found that human per- formance on a conjunctive schedule is influenced by the ratio and interval requirements. When task requirements are high in relationship to the interval requirements, peo- ple are likely to work steadily on the task throughout the time available. However, people are likely to engage in behaviors other than the task requirements when there is a large time interval and a low ratio requirement.

Table 1 provides a summary of the characteristics of compound schedules of reinforcement.

340

Schedules of Reinforcement

Table 1 Summary and Comparison of Basic Dimensions Defining Compound Schedules of Reinforcement

Compound Schedule Name

Dimension Concurrent Multiple Chained Mixed Tandem Alternative Conjunctive

Number of basic 2 or more 2 or more 2 or more 2 or more 2 or more 2 or more 2 or more schedules of rein- forcement in effect

Number of response 2 or more 1 1 or more 1 1 or more 1 1 classes involved

Discriminative stimuli Possible Yes Yes No No Possible Possible or cues associated with each compo- nent schedule

Successive presen- No Yes Yes Yes Yes No No tation of basic schedules

Simultaneous pre- Yes No No No No Yes Yes sentation of basic schedules

Reinforcement lim- No No Yes No Yes No Yes ited to final compo- nent of basic schedule

Reinforcement for Yes Yes No Yes No Yes No independent compo- nents of basic schedule

Perspectives on Applying Schedules of Reinforcement in Applied Settings Applied Research with Intermittent Schedules

Basic researchers have systematically analyzed the ef- fects of intermittent schedules of reinforcement on the performance of organisms (e.g., Ferster & Skinner, 1957). Their results have produced well-established schedule effects. These schedule effects have strong gen- erality across many species, response classes, and labo- ratories. However, a review of the applied literature on schedule effects (e.g., Journal of Applied Behavior Analysis, 1968 to 2006) will show that applied behavior analysts have not embraced the analysis of schedule ef- fects with enthusiasm, as have basic researchers. Con- sequently, schedule effects have not been documented clearly in applied settings. Uncontrolled variables in ap- plied settings, such as the following, influence a partic-

ipant’s sensitivity and insensitivity to the schedule of re- inforcement:

1. Instructions given by the applied behavior analyst, self-instructions, and environmental aids (e.g., cal- endars, clocks) make human participants resistant to temporal schedule control.

2. Past histories of responding to intermittent sched- ules of reinforcement can affect current schedule sensitivity or insensitivity.

3. Immediate histories from schedules of reinforce- ment may affect current schedule performances more than remote past histories.

4. Sequential responses required in many applied ap- plications of intermittent schedules of reinforce- ment (e.g., work leading to the paycheck, studying for a pop quiz) are uncommon applications of schedules of reinforcement, particularly with inter- val schedules.

5. Uncontrolled establishing operations in conjunc- tion with schedules of reinforcement in applied settings will confound schedule effects.

341

Some well-established schedule effects found in basic research were presented earlier in this chapter. Ap- plied behavior analysts, however, should use caution in extrapolating these effects to applied settings, for the fol- lowing reasons:

1. Most applied applications of schedules of rein- forcement only approximate true laboratory sched- ules of reinforcement, especially the interval schedules that may occur rarely in natural environ- ments (Nevin, 1998).

2. Many uncontrolled variables in applied settings will influence a participant’s sensitivity and insen- sitivity to the schedule of reinforcement (Madden, Chase, & Joyce, 1998).

Applied Research with Compound Schedules

Applied researchers have seldom analyzed the effects of compound reinforcement schedules, with the notable ex- ceptions of concurrent schedules and, to a lesser degree, chained schedules. Applied researchers should include the analysis of compound schedules in their research agendas. A better understanding of the effects of com- pound schedules on behavior will advance the develop- ment of applied behavior analysis and its applications. This perspective is important because compound sched- ules of reinforcement act directly on human behavior, and they influence behavior also by interacting with other environmental variables (e.g., antecedent stimuli, moti- vating operations) (Lattal & Neef, 1996).

Applied Research with Adjunctive Behavior

This chapter has stressed the effects of schedules of re- inforcement on the specific behaviors that produce rein- forcement. Other behaviors can occur when an individual responds to a given contingency of reinforcement. These other behaviors occur independently of schedule control. Typical examples of such behaviors include normal time fillers, such as doodling, smoking, idle talking, drinking. Such behaviors are called adjunctive behaviors, or schedule-induced behaviors, when the frequency of these time-filling behaviors increases as a side effect of other behaviors maintained by a schedule of reinforcement (Falk, 1961, 1971).

A substantial body of experimental literature has developed on many types of adjunctive behaviors with nonhuman subjects (see reviews, Staddon, 1977; Wether- ington, 1982) and some basic research with human sub- jects (e.g., Kachanoff, Leveille, McLelland, & Wayner 1973; Lasiter, 1979). Common diverse examples of ad- junctive behaviors observed in laboratory experiments include aggression, defecation, pica, and wheel running. Some common excessive human problem behaviors might develop as adjunctive behaviors (e.g., the use of drugs, tobacco, caffeine, and alcohol; overeating; nail bit- ing; self-stimulation; and self-abuse). These potentially excessive adjunctive behaviors are socially significant, but the possibility that such excesses are developed and maintained as adjunctive behaviors has been essentially ignored in applied behavior analysis.

Foster (1978), in an extended communication to the readership of the Journal of Applied Behavior Analysis, reported that applied behavior analysts have neglected the potentially important area of adjunctive behavior. He stated that applied behavior analysis does not have a data or knowledge base for adjunctive phenomena. Similarly, Epling and Pierce (1983) called for applied behavior an- alysts to extend the laboratory-based findings in adjunc- tive behavior to the understanding and control of socially significant human behavior. To our knowledge, Lerman, Iwata, Zarcone, and Ringdahl’s (1994) article provides the only research on adjunctive behavior published in the Journal of Applied Behavior Analysis from 1968 through 2006. Lerman and colleagues provided an assessment of stereotypic and self-injurious behavior as adjunctive re- sponses. Data from this preliminary study suggest that intermittent reinforcement did not induce self-injury, but with some individuals, stereotypic behavior showed char- acteristics of adjunctive behavior.

Foster (1978) and Epling and Pierce (1983) cautioned that many teachers and therapists may apply interventions directly to adjunctive behaviors rather than to the variables functionally related to their occurrence. These direct in- terventions may be futile and costly in terms of money, time, and effort because adjunctive behaviors appear re- sistant to interventions using operant contingencies.

The condition under which adjunctive behaviors are developed and maintained is a major area for future re- search in applied behavior analysis. Applied research di- rected to adjunctive behaviors will advance the science of applied behavior analysis and will provide an important foundation for improved practices in therapy and instruction.

Schedules of Reinforcement

342

Summary Intermittent Reinforcement

1. A schedule of reinforcement is a rule that establishes the probability that a specific occurrence of a behavior will produce reinforcement.

2. Only selected occurrences of behavior produce reinforce- ment with an intermittent schedule of reinforcement.

3. Applied behavior analysts use continuous reinforcement during the initial stages of learning and for strengthen- ing behavior.

4. Applied behavior analysts use intermittent reinforcement to maintain behavior.

Defining Basic Intermittent Schedules of Reinforcement

5. A fixed ratio schedule requires a specified number of re- sponses before a response produces reinforcement.

6. A variable ratio requires a variable number of responses before reinforcement is delivered.

7. A fixed interval schedule provides reinforcement for the first response following the elapse of a specific, constant duration of time since the last reinforced response.

8. A variable interval schedule provides reinforcement for the first response following the elapse of variable duration of time since the last reinforced response.

9. When a limited hold is added to an interval schedule, re- inforcement remains available for a finite time following the elapse of the FI or VI interval.

10. Each basic schedule of reinforcement has unique response characteristics that determine the consistency of re- sponding, the rate of responding, and performance during extinction.

Thinning Intermittent Reinforcement

11. Applied behavior analysts often use one of two procedures to thin schedules of reinforcement. An existing schedule is thinned by gradually increasing the response ratio or by gradually increasing the duration of the time interval.

12. Applied behavior analysts should use small increments of schedule changes during thinning and ongoing evaluation of the learner’s performance to adjust the thinning process and avoid the loss of previous improvements.

13. Ratio strain can result from abrupt increases in ratio re- quirements when moving from denser to thinner rein- forcement schedules.

Variations on Basic Intermittent Schedules of Reinforcement

14. DRH and DRL are variations of ratio schedules and spec- ify that reinforcement will be delivered contingent on re- sponses occurring above or below criterion response rates.

15. The differential reinforcement of diminishing rates sched- ule provides reinforcement at the end of a predetermined time interval when the number of responses is below a cri- terion. The criterion for the number of responses is grad- ually decreased across time intervals based on the individual’s performance.

16. Progressive schedules of reinforcement systematically thin each successive reinforcement opportunity independent of the participant’s behavior.

Compound Schedules of Reinforcement

17. Continuous reinforcement, the four simple intermittent schedules of reinforcement, differential reinforcement of rates of responding, and extinction, when combined, pro- duce compound schedules of reinforcement.

18. Compound schedules of reinforcement include concur- rent, multiple, chained, mixed, tandem, alternative, and conjunctive schedules.

Perspectives on Applying Schedules of Reinforcement in Applied Settings

19. Some well-established schedule effects found in basic re- search were presented in this chapter. Applied behavior analysts, however, should use caution in extrapolating these effects to applied settings.

20. Applied researchers should include an analysis of the basic intermittent schedules and the compound schedules in their research agendas. A better understanding of the schedule effects in applied settings will advance the development of applied behavior analysis and its applications.

21. The conditions under which adjunctive behaviors are de- veloped and maintained is an important area for future re- search in applied behavior analysis.

Schedules of Reinforcement

343

Punishment by Stimulus Presentation

Key Terms behavioral contrast conditioned punisher discriminative stimulus for

punishment generalized conditioned punisher

negative punishment overcorrection positive practice overcorrection positive punishment punisher

punishment response blocking restitutional overcorrection unconditioned punisher

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 3: Principles, Processes, and Concepts

3-5 Define and provide examples of positive and negative punishment.

3-6 Define and provide examples of conditioned and unconditioned punishment.

Content Area 9: Behavior Change Procedures

9-3 Use positive (and negative) punishment.

(a) Identify and use punishers.

(b) Use appropriate parameters and schedules of punishment.

344

Have you ever stubbed your toe while walking too fast in a darkened room and then walked slowly the rest of the way to the light switch?

Have you ever left a sandwich unattended at a beach party, watched a seagull fly away with it, and then re- frained from turning your back on the next treat you pulled from the picnic basket? If you have had these or similar experiences, you have been the beneficiary of punishment.

It may strike you as strange that we would refer to someone who stubbed his toe or lost his sandwich as ben- efiting from the experience, as opposed to referring to that person as a “victim” of punishment. Although many people consider punishment a bad thing—reinforcement’s evil counterpart—punishment is as important to learning as reinforcement. Learning from the consequences that produce pain or discomfort, or the loss of reinforcers, has survival value for the individual organism and for the species. Punishment teaches us not to repeat responses that cause us harm. Fortunately, it usually does not take too many stubbed toes or lost picnic treats to reduce the frequency of the behaviors that produced those outcomes.

Although punishment is a natural phenomenon that “occurs like the wind and the rain” (Vollmer, 2002, p. 469) and is one of the basic principles of operant conditioning, it is poorly understood, frequently misapplied, and its ap- plication can be controversial. At least some of the mis- understanding and controversy surrounding the use of punishment in behavior change programs derives from confusing punishment as an empirically derived princi- ple of behavior with the variety of everyday and legal connotations of the concept. One common meaning of punishment is the application of aversive consequences— such as physical pain, psychological hurt, and the loss of privileges, or fines—for the purpose of teaching a lesson to a person who has misbehaved so she will not repeat the misdeed. Punishment is sometimes used as an act of ret- ribution on the part of the person or agency administer- ing the punishment, or to provide “a lesson for others” on how to behave. Punishments meted out by the legal sys- tem, such as jail time and fines, are often considered a process by which convicted lawbreakers must repay their debt to society.

These everyday and legal notions of punishment, al- though having various degrees of validity within their own contexts, have little, if anything, to do with punish- ment as a principle of behavior. In the everyday conno- tation of punishment, most people would agree that a teacher who sends a student to the principal’s office for fooling around in class, or a police officer who issues a ticket to a speeding motorist, has punished the offender. However, as a principle of behavior, punishment is not about punishing the person; punishment is a response →

consequence contingency that suppresses the future fre- quency of similar responses. From the perspective of both the science and the practice of behavior analysis, the trip to the principal’s office did not punish fooling around in class unless the future frequency at which the student fools around in class decreases as a function of his trip to the principal’s office, and the police officer’s ticket did not punish speeding unless the motorist drives above the speed limit less often than she did before re- ceiving the ticket.

In this chapter we define the principle of punishment, discuss its side effects and limitations, identify factors that influence the effectiveness of punishment, describe examples of several behavior change tactics that incor- porate punishment, discuss ethical considerations in the use of punishment, and present guidelines for using pun- ishment effectively. In the chapter’s concluding section, we underscore the need for more basic and applied re- search on punishment and reiterate Iwata’s (1988) rec- ommendation that behavior analysts view punishment by contingent stimulation as a default technology to be im- plemented when other interventions have failed.

Definition and Nature of Punishment This section presents the basic functional relation that defines punishment, the two operations by which pun- ishment can be implemented, discrimination effects of punishment, recovery from punishment, unconditioned and conditioned punishers, factors that influence the ef- fectiveness of punishment, and possible side effects of and problems with punishment.

Operation and Defining Effect of Punishment

Like reinforcement, punishment is a two-term, behavior → consequence functional relation defined by its effects on the future frequency of behavior. Punishment has oc- curred when a response is followed immediately by a stimulus change that decreases the future frequency of similar responses (Azrin & Holz, 1966).

An early study by Hall and colleagues (1971) pro- vides a straightforward example of punishment. Andrea, a 7-year-old girl with hearing impairments, “pinched and bit herself, her peers, the teacher, and visitors to the class- room at every opportunity and was so disruptive the teacher reported that academic instruction was impossi- ble” (p. 24). During an initial 6-day baseline period, An- drea bit or pinched an average of 71.8 times per day. Whenever Andrea bit or pinched anyone during the

Punishment by Stimulus Presentation

345

100

N u

m b

er o

f B

it es

a n

d P

in ch

es p

er D

5 10 15 20 25 30 35

Baseline1

Pointed Finger, “No” for Bites, Pinches1 B2 Pointed Finger, “No”2

Days

Figure 1 Number of bites and pinches by a 7-year-old girl during baseline and punishment (“No” plus pointing) conditions. From “The Effective Use of Punishment to Modify Behavior in the Classroom” by R. V. Hall, S. Axelrod, M. Foundopoulos, J. Shellman, R. A. Campbell, and S. S. Cranston, 1971, Educational Technology, 11(4), p. 25. Copyright 1971 by Educational Technology. Used by permission.

intervention condition, her teacher immediately pointed at her with an outstretched arm and shouted “No!” On the first day of intervention, the frequency of Andrea’s biting and pinching decreased substantially (see Figure 1). Her aggressive behavior followed a downward trend during the initial intervention phase, ending in an average of 5.4 incidents per day. A 3-day return to base- line conditions resulted in Andrea’s biting and pinching at a mean rate of 30 times per day. When the teacher re- instated the intervention, pointing her finger and stating “No!” each time Andrea pinched or bit, Andrea’s problem behavior dropped to 3.1 incidents per day. During the second intervention phase, the teacher reported that An- drea’s classmates were no longer avoiding her, perhaps because their behavior of being near Andrea was pun- ished less often by bites and pinches.

It is important to point out that punishment is de- fined neither by the actions of the person delivering the consequences (in the case of socially mediated punish- ment) nor by the nature of those consequences.1 A de- crease in the future frequency of the occurrence of the behavior must be observed before a consequence-based intervention qualifies as punishment. The intervention that proved successful in reducing the frequency of An- drea’s biting and pinching—her teacher’s pointed finger and “No!”—is classified as a punishment-based treat- ment only because of its suppressive effects. If Andrea had continued to bite and pinch at the baseline level of re-

1Although the term automatic punishment is used infrequently in the be- havior analysis literature, it is similar to automatic reinforcement. Automatic punishment occurs when a punishing consequence (e.g., burned finger) is a socially unmediated, unavoidable outcome of a response (e.g., touching a hot stove).

sponding when the intervention was applied, her teacher’s pointing and “No!” would not have been punishment.

Because the presentation of punishers often evokes behavior incompatible with the behavior being punished, the immediate suppressive effects of punishment can eas- ily be overestimated. Michael (2004) explained and pro- vided a good example:

The decreased frequency of the punished response that is due to its having been followed by punishment will not be seen until after behavior evoked by the punishing stimulus changes has ceased. Because the evocative effect of the punishment stimulus change (as a respondent uncondi- tioned or conditioned elicitor or as operant SD or MO) is in the same direction as the future change due to punishment as a response consequence (weakening of the punished be- havior), the former can be easily misinterpreted as the lat- ter. For example, when a small child’s misbehavior is followed by a severe reprimand, the misbehavior will cease immediately, but primarily because the reprimand controls behavior incompatible with the misbehavior—at- tending to the adult who is doing the reprimanding, deny- ing responsibility for the misbehavior, emotional behavior such as crying, etc. This sudden and total cessation of the misbehaver does not imply, however, that the future fre- quency of its occurrence has been reduced, which would be the true effect of punishment. (pp. 36–37)

Another factor that contributes to the difficulty of determining the true effectiveness of punishment is that the reduction in response rate is often confounded by ex- tinction effects caused by withholding reinforcement for the problem behavior (something that should be part of a punishment-based intervention whenever possible) (Iwata, Pace, Cowdery, & Miltenberger, 1994).

Punishment by Stimulus Presentation

346

Positive Punishment and Negative Punishment

Like reinforcement, punishment can be accomplished by either of two types of stimulus change operations. Positive punishment occurs when the presentation of a stimulus (or an increase in the intensity of an already pre- sent stimulus) immediately following a behavior results in a decrease in the frequency of the behavior. Stubbing one’s toe on a chair leg is a form of positive punish- ment—if it suppresses the frequency of the behavior that preceded the toe stub—because the painful stimulation is best described as the presence of a new stimulus. Be- havior change tactics based on positive punishment involve the contingent presentation of a stimulus immediately following occurrences of the target behav- ior. The intervention used by Andrea’s teacher consti- tuted positive punishment; the teacher’s pointed finger and “No!” were stimuli presented or added to Andrea’s environment.

Negative punishment involves the termination of an already present stimulus (or a decrease in the intensity of an already present stimulus) immediately following a behavior that results in a decrease in the future frequency of the behavior. The beach party attendee’s behavior of turning his back on his food was negatively punished when the seagull flew off with his sandwich. For a stim- ulus change to function as negative punishment, which amounts to the removal of a positive reinforcer, a “moti- vating operation for the reinforcer must be in effect, oth- erwise removing it will not constitute punishment” (Michael, 2004, p. 36). A seagull’s flying off with a hun- gry person’s sandwich would function as punishment for his inattentiveness but perhaps have little effect on the behavior of a person who has eaten his fill and set the sandwich down.

Behavior change tactics based on negative punish- ment involve the contingent loss of available reinforcers immediately following a behavior (i.e., response cost, a procedure akin to a fine) or the removal of the opportu- nity to acquire additional reinforcers for a period of time (i.e., timeout from reinforcement, a procedure akin to being sidelined during a game).

A wide variety of terms are used in the behavioral literature to refer to the two types of consequence oper- ations for punishment. Positive punishment and negative punishment are sometimes identified as Type I punish- ment and Type II punishment, respectively (Foxx, 1982). Malott and Trojan Suarez (2004) use the term penalty principle to refer to negative punishment. The Behavior Analyst Certification Board (BACB, 2001) and some textbook authors (e.g., Baum, 1994; Catania, 1998; Michael, 2004; Miltenberger, 2001) use the terms positive punishment and negative punishment, paralleling the

2As with reinforcement, the terms positive punishment and negative pun- ishment are, as Michael (2004) noted, “quite susceptible to misunder- standing. Assuming that you must receive either positive or negative punishment, which would you prefer? As with reinforcement, you should certainly not decide until you know specifically what each consists of. Would you prefer negative reinforcement or positive punishment? Of course, negative reinforcement” (p. 37).

terms positive reinforcement and negative reinforcement. As with reinforcement, the modifiers positive and negative used with punishment connote neither the in- tention nor the desirability of the behavior change pro- duced; they specify only how the stimulus change that served as the punishing consequence was affected— whether it is best described as the presentation of a new stimulus (positive punishment) or the termination (or re- duction in intensity or amount) of an already present stim- ulus (negative punishment).2

Positive punishment and negative reinforcement are frequently confused. Because aversive events are associ- ated with positive punishment and with negative rein- forcement, the umbrella term aversive control is often used to describe interventions involving either or both of these two principles. Distinguishing between the two principles is difficult when the same aversive event is in- volved in concurrent positive punishment and negative reinforcement contingencies. For example, Baum (1994) described how the application and threat of physical beat- ings might condition the behavior of people living in a po- lice state.

If speaking out results in a beating, then speaking out is positively punished. If lying avoids a beating, then lying is negatively reinforced. The two tend to go hand-in- hand; if one action is punished, there is usually some al- ternative that avoids punishment. (p. 153)

The keys to identifying and distinguishing concur- rent positive punishment and negative reinforcement con- tingencies involving the same aversive stimulus event are (a) recognizing the opposite effects the two contingencies have on the future frequency of behavior, and (b) realiz- ing that two different behaviors must be involved because the same consequence (i.e., stimulus change) cannot serve as positive punishment and negative reinforcement for the same behavior. In a positive punishment contingency, the stimulus is absent prior to a response and is presented as a consequence; in a negative reinforcement contingency, the stimulus is present prior to a response and is removed as a consequence. For example, positive punishment and negative reinforcement contingencies operated concur- rently in a study by Azrin, Rubin, O’Brien, Ayllon, and Roll (1968) in which adults wore an apparatus throughout the normal working day that automatically produced a

Punishment by Stimulus Presentation

347

Punishment by Stimulus Presentation

55 dB tone contingent on slouching (sustained rounding of the shoulders or upper back for 3 seconds) and imme- diately terminated the tone when they straightened their shoulders. Slouching produced a tone (positive punish- ment), straightening the shoulders escaped the tone (neg- ative reinforcement), and nonslouching avoided a tone (negative reinforcement).

Threatening a person with punishment if she engages in a behavior should not be confused with punishment. Punishment is a behavior → consequence relation, and a threat of what might happen if a person subsequently be- haves in a certain way is an antecedent event to the be- havior. When the threat of punishment suppresses behavior, it may be due to the threat functioning as an establishing operation that evokes alternative behaviors that avoid the threatened punishment.

Discriminative Effects of Punishment

Punishment does not operate in a contextual vacuum. The antecedent stimulus situation in which punishment oc- curs plays an important role in determining the environ- mental conditions in which the suppressive effects of punishment will be observed. The three-term contingency for “the operant functional relation involving punishment can be stated much like that involving reinforcement: (1) In a particular stimulus situation (S), (2) some kinds of behavior (R), when followed immediately by (3) certain stimulus changes (SP), show a decreased future frequency of occurrence in the same or in similar situations” (Michael, 2004, p. 36).

If punishment occurs only in some stimulus condi- tions and not in others (e.g., a child gets scolded for reaching into the cookie jar before dinner only when an adult is in the room), the suppressive effects of punish- ment will be most prevalent under those conditions (Azrin & Holz, 1966; Dinsmoor, 1952). A discriminated operant for punishment is the product of a conditioning history in which responses in the presence of a certain stimulus have been punished and similar responses in the absence of that stimulus have not been punished (or have resulted in a reduced frequency or magnitude of punishment). Highway speeding is a discriminated op- erant in the repertoire of many motorists who drive within the speed limit in and around locations where they have been pulled over by the police for speeding but continue to drive above the speed limit on roads where they have never seen a police cruiser.

There is no standard term or symbol in the behavior analysis literature for an antecedent stimulus that acquires stimulus control related to punishment. Some authors have modified the shorthand symbol for a discriminative stim- ulus for reinforcement (SD) to indicate the antecedent stim- ulus in a three-term punishment contingency, such as the

SD- used by Sulzer-Azaroff and Mayer (1971). Other au- thors simply refer to an antecedent stimulus correlated with the presence of a punishment contingency as a punishment-based SD (e.g., Malott & Trojan Suarez, 2004; Michael, 2004). We have adopted SDp as the symbol for the discriminative stimulus for punishment as proposed by O’Donnell (2001). “An SDp can be defined as a stimu- lus condition in the presence of which a response has a lower probability of occurrence than it does in its absence as a result of response-contingent punishment delivery in the presence of the stimulus” (p. 262). Figure 2 shows three-term contingency diagrams for discriminated oper- ants for positive punishment and negative punishment.

Recovery from Punishment

When punishment is discontinued, its suppressive effects on responding are usually not permanent, a phenomenon often called recovery from punishment, which is analo- gous to extinction. Sometimes the rate of responding after punishment is discontinued will not only recover but also briefly exceed the level at which it was occurring prior to punishment (Azrin, 1960; Holz & Azrin, 1962). Recov- ery of responding to prepunished levels is more likely to occur when the punishment was mild or when the person can discriminate that the punishment contingency is no longer active. Although the response-weakening effects of punishment often wane when punishment is discon- tinued, so too are the response strengthening effects of reinforcement often fleeting when previously reinforced behavior is placed on extinction (Vollmer, 2002). Michael (2004) noted that

Recovery from punishment is sometimes given as an ar- gument against the use of punishment to cause a de- crease in behavior. “Don’t use punishment because the effect is only temporary.” But of course the same argu- ment could be made for reinforcement. The strengthen- ing effect of reinforcement decreases when the behavior occurs without the reinforcement. The weakening effect of punishment decreases when the behavior occurs with- out the punishment. (p. 38)

Virtually permanent response suppression may occur when complete suppression of behavior to a zero rate of responding has been achieved with intense punishment. In their review of basic research on punishment, Azrin and Holz (1966) noted that:

Intense punishment did not merely reduce responses to the unconditioned or operant level, but reduced them to an absolute level of zero. Since punishment is delivered only after a response occurs, there is no opportunity for the subject to detect the absence of punishment unless he responds. If punishment is severe as to completely eliminate the responses then the opportunity for detect- ing the absence of punishment no longer exists. (p. 410)

348

Punishment by Stimulus Presentation

Grandma scolds

Seagull flies away with sandwich

Effect on Future Frequency of Similar Responses

in Presence of SD SDp R

Grandma in the kitchen before dinner

Reach into cookie jar

Seagulls present at beach picnic

Leave sandwich unattended

Figure 2 Three-term contingencies illustrating positive and negative punishment of discriminated operants: A response (R) emitted in the presence of a discriminative stimulus (SDp) is followed closely in time by a stimulus change (SP) and results in a decreased frequency of similar responses in the future when the SDp is present. A discriminated operant for punishment is the product of a conditioning history in which responses in the presence of the SDp have been punished and similar responses in the absence of the SDp

have not been punished (or have resulted in a reduced frequency or magnitude of punishment than in the presence of the SDp).

Unconditioned and Conditioned Punishers

A punisher is a stimulus change that immediately fol- lows the occurrence of a behavior and reduces the future frequency of that type of behavior. The stimulus change in a positive punishment contingency may be referred to as a positive punisher, or more simply as a punisher. Like- wise the term negative punisher can be applied to a stim- ulus change involved in negative punishment. However, this usage is clumsy because it refers to a positive rein- forcer that is removed contingent on occurrences of the target behavior. Therefore, when most behavior analysts use the term punisher, they are referring to a stimulus whose presentation functions as punishment (i.e., a pos- itive punisher). As with reinforcers, punishers can be clas- sified as unconditioned or conditioned.

Unconditioned Punishers

An unconditioned punisher is a stimulus whose pre- sentation functions as punishment without having been paired with any other punishers. (Some authors use primary punisher or unlearned punisher as synonyms for an unconditioned punisher.) Because unconditioned pun- ishers are the product of the evolutionary history of a species (phylogeny), all biologically intact members of a species are more or less susceptible to punishment by the same unconditioned punishers. Painful stimulation such as that caused by physical trauma to the body, certain

odors and tastes, physical restraint, loss of bodily sup- port, and extreme muscular effort are examples of stim- ulus changes that typically serve as unconditioned punishers for humans (Michael, 2004). Like uncondi- tioned reinforcers, unconditioned punishers are “phylo- genically important events that bear directly on fitness” of the organism (Baum, 1994, p. 59).

However, virtually any stimulus to which an organ- ism’s receptors are sensitive—light, sound, and temper- ature, to name a few—can be intensified to the point that its delivery will suppress behavior even though the stim- ulus is below levels that actually cause tissue damage (Bijou & Baer, 1965). Unlike unconditioned reinforcers, such as food and water, whose effectiveness depends on a relevant establishing operation, under most conditions many unconditioned punishers will suppress any behav- ior that precedes their onset. For example, an organism does not have to be “deprived of electric stimulation” for the onset of electric shock to function as punishment. (However, the behavior of an organism that has just re- ceived many shocks in a short period of time, particu- larly shocks of mild intensity, may be relatively unaffected by another shock.)

Conditioned Punishers

A conditioned punisher is a stimulus change that func- tions as punishment as a result of a person’s conditioning history. (Some authors use secondary punisher or learned

349

Punishment by Stimulus Presentation

punisher as synonyms for a conditioned punisher.) A con- ditioned punisher acquires the capability to function as a punisher through stimulus–stimulus pairing with one or more unconditioned or conditioned punishers. For ex- ample, as a result of its onset at or very near the same time as an electric shock, a previously neutral stimulus change, such as an audible tone, will become a condi- tioned punisher capable of suppressing behavior that im- mediately precedes the tone when it occurs later in the absence of the shock (Hake & Azrin, 1965).3 If a condi- tioned punisher is repeatedly presented without the pun- isher(s) with which it was initially paired, its effectiveness as punishment will wane until it is no longer a punisher.

Previously neutral stimuli can also become condi- tioned punishers for humans without direct physical pair- ing with another punisher through a pairing process Alessi (1992) called verbal analog conditioning. This is similar to the example of verbal pairing for the condi- tioning of a conditioned reinforcer in which Engelmann (1975) showed cut up pieces of yellow construction paper to a group of preschool children and told them, “These pieces of yellow paper are what big kids work for” (pp. 98–100). From that point on, many children began working extra hard for yellow pieces of paper. Milt- enberger (2001) gave the example of a carpenter telling his apprentice that if the electric saw starts to make smoke, the saw motor may become damaged or the blade may break. The carpenter’s statement establishes smoke from the saw as a conditioned punisher capable of de- creasing the frequency of any behaviors that immediately precede the smoke (e.g., pushing too forcefully on the saw, holding the saw at an inappropriate angle).

A stimulus change that has been paired with numer- ous forms of unconditioned and conditioned punishers becomes a generalized conditioned punisher. For ex- ample, reprimands (“No!” “Don’t do that!”) and social disapproval (e.g., scowl, head shake, frowns) are gener- alized conditioned punishers for many people because they have been paired repeatedly with a wide range of unconditioned and conditioned punishers (e.g., burned finger, loss of privileges). As with generalized condi- tioned reinforcers, generalized conditioned punishers are free from the control of specific motivating conditions and will function as punishment under most conditions.

At the risk of redundancy, we will again stress the critical point that punishers, like reinforcers, are not de-

3A stimulus that becomes a conditioned punisher by being paired with an- other punisher does not have to be a neutral stimulus prior to the pairing. The stimulus could already function as a reinforcer under other conditions. For example, a blue light that has been paired repeatedly with reinforce- ment in one setting and with punishment in another setting is a conditioned reinforcer or a conditioned punisher depending on the setting.

(Morse & Kelleher, 1977). Even stimuli whose presen- tation under most conditions would function as uncon- ditioned reinforcers or punishers can have the opposite effect under certain conditions. For example, a bite of food will function as a punisher for a person who has eaten too much, and an electric shock may function as a conditioned reinforcer if it signals the availability of food for a food-deprived organism (e.g., Holz & Azrin, 1961). If a student receives smiley face stickers and praise for his academic work and his productivity decreases as a result, smiley face stickers and praise are punishers for that stu- dent. What might serve as a punisher at home might not be a punisher at school. What might be a punisher under one set of circumstances might not be a punisher under a different set of circumstances. Although common ex- periences mean that many of the same stimulus events function as conditioned punishers for most people, a pun- isher for one person may be a reinforcer for another.

Factors That Influence the Effectiveness of Punishment

Reviews of basic and applied research on punishment consistently identify the following variables as keys to the effectiveness of punishment: the immediacy of pun- ishment, the intensity of the punisher, the schedule or fre- quency of punishment, the availability of reinforcement for the target behavior, and the availability of reinforce- ment for an alternative behavior (e.g., Axelrod, 1990; Azrin & Holz, 1966; Lerman & Vorndran, 2002; Matson & Taras, 1989).

Immediacy

Maximum suppressive effects are obtained when the onset of the punisher occurs as soon as possible after the occurrence of a target response.

The longer the time delay between the occurrence of the response and the occurrence of the stimulus change, the less effective the punishment will be in changing the rel- evant response frequency, but not much is known about upper limits. (Michael, 2004, p. 36)

Intensity/Magnitude

Basic researchers examining the effects of the punishers of varying intensity or magnitude (in terms of amount or duration) have reported three reliable findings: (1) a pos- itive correlation between the intensity of the punishing

fined by their physical properties, but by their functions

350

Punishment by Stimulus Presentation

stimulus and response suppression: the greater the mag- nitude of the punishing stimulus, the more immediately and thoroughly it suppresses the occurrence of the be- havior (e.g., Azrin & Holtz, 1966); (2) recovery from punishment is negatively correlated with intensity of the punishing stimulus: the more intense the punisher, the less likely that responding will reoccur when punishment is terminated (e.g., Hake, Azrin, & Oxford, 1967); and (3) a high-intensity stimulus may be ineffective as pun- ishment if the stimulus used as punishment was initially of low intensity and gradually increased (e.g., Terris & Barnes, 1969). However, as Lerman and Vorndran (2002) pointed out, relatively few applied studies have exam- ined the relation between punishment magnitude and treatment efficacy, and that research has yielded incon- sistent results sometimes contradictory with basic re- search findings (e.g., Cole, Montgomery, Wilson, & Milan, 2000; Singh, Dawson, & Manning, 1981; Williams, Kirkpatrick-Sanchez, & Iwata, 1993). When selecting the magnitude of a punishing stimulus, the prac- titioner should ask: Will this amount of the punishing stimulus suppress occurrences of the problem behavior? Lerman and Vorndran (2002) recommended:

While the punishing stimulus needs to be intensive enough for an effective application, it should not be more intense than necessary. Until further applied re- search on magnitude is conducted, practitioners should select magnitudes that have been shown to be safe and effective in clinical studies, as long as the magnitude is considered acceptable and practical by those who will be implementing treatment. (p. 443)

Schedule

The suppressive effects of a punisher are maximized by a continuous schedule of punishment (FR 1) in which each occurrence of the behavior is followed by the punishing consequence. In general, the greater the proportion of re- sponses that are followed by the punisher, the greater the response reduction will be (Azrin, Holz, & Hake, 1963; Zimmerman & Ferster, 1962). Azrin and Holz (1966) summarized the comparative effects of punishment on continuous and intermittent schedules as follows:

Continuous punishment produces more suppression than does intermittent punishment for as long as the punish- ment contingency is maintained. However, after the pun- ishment contingency has been discontinued, continuous punishment allows more rapid recovery of the re- sponses, possibly because the absence of punishment can be more rapidly discriminated. (p. 415)

Intermittent punishment may be somewhat effective under some conditions (e.g., Clark, Rowbury, Baer, & Baer, 1973; Cipani, Brendlinger, McDowell, & Usher,

1991; Romanczyk, 1977). Results of a study by Lerman, Iwata, Shore, and DeLeon (1997) demonstrated that a gradual thinning of the punishment schedule might main- tain the suppressive effects of punishment that was ini- tially delivered on a continuous schedule (FR 1). Participants were five adults with profound mental retar- dation and histories of chronic self-injurious behavior (SIB) in the form of hand mouthing or head hitting. Treat- ment by punishment (timeout from reinforcement for one participant and contingent restraint for the other four) de- livered on a continuous (FR 1) schedule produced marked reductions in SIB from baseline levels for all five partic- ipants. (Figure 3 shows the results for three of the five participants.) The participants were then exposed to in- termittent schedules of punishment (either fixed-interval [FI] 120 seconds or FI 300 seconds). On the FI 120- second schedule, the therapist delivered punishment con- tingent on the first SIB response after 120 seconds had elapsed since the previous application of punishment or the start of the session. The frequency of SIB under the intermittent schedule of punishment for all but one par- ticipant (Wayne, not shown in Figure 3) increased to base- line levels.

After reestablishing low levels of SIB for each par- ticipant with FR 1 punishment, the researchers gradually thinned the punishment schedules. For example, schedule thinning for Paul consisted of increasing the duration of the fixed-interval duration by 30-second increments to FI 300 seconds (i.e., FR 30 seconds, FR 60 seconds, FR 90 seconds, etc.). With the exception of a few sessions, Paul’s SIB remained at low levels as the punishment schedule was progressively thinned over 57 sessions. During the final 11 sessions in which punishment was delivered on an FI 300-second schedule, his SIB occurred at a mean of 2.4% of observed intervals (compared to 33% in base- line). A similar pattern of success with a gradually thin- ning punishment schedule was obtained for another subject, Wendy (not shown in Figure 3).

The effectiveness of punishment on an FI 300- second schedule for three of the five participants—Paul, Wendy, and Wayne (whose SIB remained low even though punishment was abruptly changed from FR 1 to FI 120 seconds and then FI 300 seconds)—enabled an infrequent application of punishment. In practice, this would free therapists or staff from continuous monitor- ing of behavior.

For Melissa and Candace, however, repeated at- tempts to gradually thin the punishment schedule proved unsuccessful in maintaining the frequency of their SIB at levels attained by FR 1 punishment. Lerman and col- leagues (1997) speculated that one explanation for the ineffectiveness of punishment on a fixed-interval sched- ule was that after a person has experienced an FI sched- ule for some time, the delivery of punishment could

351

Punishment by Stimulus Presentation

100

% I

n te

rv al

s H

an d

M o

u th

in g

FI 3

00 "

FI 2

70 "

FI 2

40 "

FI 2

10 "

FI 1

80 "

FI 1

50 "

FI 9

FI 6

FI 3

FR 1

FI 1

20 "

20 40 60 80 100

Paul

BL FR 1 BL FI 120 FI 300 Schedule Thinned

100

% I

n te

rv al

s H

an d

M o

u th

in g

FR 1

20 40 60 80 100

Melissa

BL FR 1 BL FI 120 Schedule Thinned

120 FR

FR 1

FI 3

0" FI

4 5"

FI 1

FR 1

FI 1

FI 5

FI 2

0" FI

2 5"

FI 1

H ea

d H

it s

p er

M in

u te

20 40 60 80 100

Candace

Schedule Thinned

FR 1

120 140 160 Sessions

FR 1

FI 120"

FI 5

" FR

FI 1

FR 1

FI 1

20 "

FI 7

FI 4

5" FI

3 0"

FI 1

FI 5

FR 1

Figure 3 Self-injurious behavior (SIB) by three adults with profound mental retar- dation during baseline and punishment delivered on continuous (FR 1) and various fixed-interval schedules. From “Effects of Intermittent Punishment on Self-Injurious Behavior: An Evaluation of Schedule Thinning” by D. C. Lerman, B. A. Iwata, B. A. Shore, and I. G. DeLeon, 1997, Journal of Applied Behavior Analysis, 30, p. 194. Copyright 1997 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

function “as a discriminative stimulus for punishment- free periods, leading to a gradual overall increase in re- sponding under FI punishment” (p. 198).

Reinforcement for the Target Behavior

The effectiveness of punishment is modulated by the re- inforcement contingencies maintaining the problem be- havior. If a problem behavior is occurring at a frequency

sufficient to cause concern, it is presumably producing reinforcement. If the target response was never rein- forced, then “punishment would scarcely be possible since the response would rarely occur” (Azrin & Holz, 1966, p. 433).

To the extent that the reinforcement maintaining the problem behavior can be reduced or eliminated, punish- ment will be more effective. Of course, if all reinforce- ment for the problem behavior was withheld, the resulting

352

Punishment by Stimulus Presentation

extinction schedule would result in the reduction of the behavior independent of the presence of a punishment contingency. However, as Azrin and Holz (1966) pointed out:

The physical world often provides reinforcement contin- gences that cannot be eliminated easily. The faster we move through space, the quicker we get to where we are going, whether the movement be walking or driving an auto. Hence, running and speeding will inevitably be re- inforced. Extinction of running and speeding could be accomplished only by the impossible procedure of elim- inating all reinforcing events that result from movement through space. Some other reductive method, such as punishment must be used. (p. 433)

Reinforcement for Alternative Behaviors

Holz, Azrin, and Ayllon (1963) found that punishment was ineffective in reducing psychotic behavior when that behavior was the only means by which patients could at- tain reinforcement. However, when patients could emit an alternative response that resulted in reinforcement, pun- ishment was effective in reducing their inappropriate be- havior. Summing up laboratory and applied studies that have reported the same finding, Millenson (1967) stated:

If punishment is employed in an attempt to eliminate certain behavior, then whatever reinforcement the unde- sirable behavior had led to must be made available via a more desirable behavior. Merely punishing school chil- dren for “misbehavior” in class may have little perma- nent effect. . . . The reinforcers for “misbehavior” must be analyzed and the attainment of these reinforcers per- haps permitted by means of different responses, or in other situations. . . . But for this to happen, it appears important to provide a rewarded alternative to the pun- ished response. (p. 429)

A study by Thompson, Iwata, Conners, and Roscoe (1999) is an excellent illustration of how the suppressive effects of punishment can be enhanced by reinforcement for an alternative response. Four adults with developmen- tal disabilities who had been referred to a day treatment program for self-injurious behavior (SIB) participated in the study. Twenty-eight-year-old Shelly, for example, expelled saliva and then rubbed it onto her hands and other surfaces (e.g., tables, windows), which led to fre- quent infections; and Ricky, a 34-year-old man with deaf- blindness, frequently hit his head and body resulting in bruises or contusions. Previous interventions such as dif- ferential reinforcement of appropriate behavior, response blocking, and protective equipment had been ineffective in reducing the SIB of all four participants.

The results of a functional behavior analysis with each participant suggested that SIB was maintained by

automatic reinforcement. Reinforcer assessments were conducted to identify materials that produced the highest levels of contact or manipulation and the lowest levels of SIB (e.g., wooden stringing beads, a mirrored mi- croswitch that produced vibration and music, a balloon). The researchers then conducted a punisher assessment to determine the least intrusive consequences that produced at least a 75% reduction in SIB for each participant.

Thompson and colleagues (1999) analyzed the effects of punishment with and without reinforcement of alterna- tive behavior with an experimental design that combined alternating treatments, reversal, and multiple baseline across subjects’ elements. During the no-punishment con- dition, the therapist was present in the room but did not in- teract with the participant or provide any consequences for SIB. Immediately following each occurrence of SIB during the punishment condition, the therapist delivered the consequence that had previously been identified as a punisher for the participant. For example:

Each time Shelly expelled saliva, the therapist delivered a reprimand (“no spitting”) and briefly dried each of her hands (and any other wet surfaces) with a cloth. Ricky’s hands were held in his lap for 15 s each time he engaged in SIB. Donna and Lynn both received a verbal repri- mand and had their hands held across their chests for 15 s following SIB. (p. 321)

Within each no-punishment and punishment phase, sessions of reinforcement and no-reinforcement condi- tions were alternated. During reinforcement sessions, par- ticipants had continual access to leisure materials or activities previously identified as highly preferred; during no-reinforcement sessions, participants had no access to leisure materials. Because Ricky never independently ma- nipulated any of the leisure materials during the reinforcer assessment but approached M&M candies on the high- est percentage of assessment trials, during the reinforce- ment sessions he received edible reinforcement for each 2 seconds that he manipulated any of several items at- tached to a vest (e.g., Koosh ball, beads, fur).

Figure 4 shows the results of the study. During the no-punishment baseline phase, only Shelly’s SIB was consistently lower during the reinforcement sessions than during the no-reinforcement sessions. Although the in- troduction of punishment reduced SIB from baseline lev- els for all four participants, punishment was more effective during those sessions in which reinforcement for alternative behaviors was available. Also of interest to practitioners should be the finding that fewer punishers were delivered during the punishment condition sessions when reinforcement was available. (Ricky began to re- sist the hands-down restraint procedure during the pun- ishment with no reinforcement condition, and several

353

Punishment by Stimulus Presentation

1.5

0.5

R es

p o

n se

s p

er M

in u

te (S

IB )

No Punish Punish No Punish Punish

Shelly

R es

p o

n se

s p

er M

in u

te (S

IB )

Ricky

100

% I

n te

rv al

s (S

IB )

Donna

R es

p o

n se

s p

er M

in u

te (S

IB )

Lynn

No Sr+

Sr+

10 20 30 40 50 60 70 Sessions

Figure 4 Self-injurious behavior by four adults with developmental disabilities during alternating reinforcement and no reinforce- ment conditions across no-punishment and punishment phases. From “Effects of Reinforcement for Alternative Behavior during Punishment for Self-Injury,” by R. H. Thompson, B. A. Iwata, J. Conners, and E. M. Roscoe, 1999, Journal of Applied Be- havior Analysis, 32, p. 323. Copyright 1999 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

sessions were terminated early because the therapist could not implement the procedure safely. Therefore, the pun- ishment with no reinforcement condition was terminated after seven sessions and the punishment with reinforce- ment continued for six additional sessions.)

Thompson and colleagues (1999) summarized the main findings of their study with these conclusions:

Consistent with the recommendation made by Azrin and Holz (1966), results of this study indicate that the effects of punishment can be enhanced when reinforcement is provided for an alternative response. Furthermore, these results suggest a method for increasing the effectiveness of punishment through means other than increasing the

aversiveness of the punishing stimulus, thereby resulting in the development of more effective yet less restrictive interventions. (p. 326)

Possible Side Effects and Problems with Punishment

A variety of side effects and problems are often corre- lated with the applications of punishment, including the elicitation of undesirable emotional responses and ag- gression, escape and avoidance, and an increased rate of the problem behavior under nonpunishment conditions (e.g., Azrin & Holz, 1966; Hutchinson, 1977; Linscheid

354

Punishment by Stimulus Presentation

& Meinhold, 1990). Other problems noted include mod- eling undesirable behavior and overusing punishment be- cause of the negative reinforcement it provides for the punishing agent’s behavior.

Emotional and Aggressive Reactions

Punishment sometimes evokes emotional and aggressive reactions that can involve a combination of respondent and conditioned operant behaviors. Punishment, espe- cially positive punishment in the form of aversive stim- ulation, may evoke aggressive behavior with respondent and operant components (Azrin & Holz, 1966 ). For ex- ample, electric shock elicited reflexive forms of ag- gression and fighting in laboratory animals (Azrin, Hutchinson, & Hake, 1963; Ulrich & Azrin, 1962; Ulrich, Wolff, & Azrin, 1962). Such pain-elicited, or res- pondent aggression, is directed toward any nearby person or object. For example, a student who is punished se- verely may begin to throw and destroy materials within her reach. Alternatively, the student may try to attack the person delivering the punishment. Aggressive behavior following punishment that occurs because it has enabled the person to escape the aversive stimulation in the past is referred to as operant aggression (Azrin & Holz, 1966).

Although basic laboratory researchers, using intense and unavoidable punishers, have reliably produced re- spondent and operant aggression with nonhuman ani- mals, many applied studies of punishment report no evidence of aggression (e.g., Linscheid & Reichenbach, 2002; Risley, 1968).

Escape and Avoidance

Escape and avoidance are natural reactions to aversive stimulation. Escape and avoidance behaviors take a wide variety of forms, some of which may be a greater problem than the target behavior being punished. For example, a student who is admonished repeatedly for sloppy work or coming to class unprepared may stop coming to class al- together. A person may lie, cheat, hide, or exhibit other un- desirable behaviors to avoid punishment. Mayer, Sulzer, and Cody (1968) indicated that avoidance and escape need not always take form in the literal sense of those terms. People sometimes escape punishing environments by tak- ing drugs or alcohol, or by simply “tuning out.”

As the intensity of a punisher increases, so does the likelihood of escape and avoidance. For example, in a study evaluating the effectiveness of a specially designed cigarette holder that delivered an electric shock to the user when it was opened as an intervention for reducing cigarette smoking, Powell and Azrin (1968) found that

4Multiple schedules of reinforcement are described in Chapter entitled “Schedules of Reinforcement.”

“As the punishment intensity increased, the duration de- creased for which the subjects would remain in contact with the contingency; ultimately, an intensity was reached at which they refused to experience it altogether” (p. 69).

Escape and avoidance as side effects to punishment, like emotional and aggressive reactions, can be mini- mized or precluded altogether by providing the person with desirable alternative responses to the problem be- havior that both avoid the delivery of punishment and provide reinforcement.

Behavioral Contrast

Reynolds (1961) introduced the term behavioral con- trast to refer to the phenomenon in which a change in one component of a multiple schedule that increases or decreases the rate of responding on that component is ac- companied by a change in the response rate in the oppo- site direction on the other, unaltered component of the schedule.4 Behavioral contrast can occur as a function of a change in reinforcement or punishment density on one component of a multiple schedule (Brethower & Reynolds, 1962; Lattal & Griffin, 1972). For example, behavioral contrast for punishment takes the following general form: (a) Responses are occurring at similar rates on two components of a multiple schedule (e.g., a pigeon pecks a backlit key, which alternates between blue and green, reinforcement is delivered on the same schedule on both keys, and the bird pecks at roughly the same rate re- gardless of the key’s color); (b) responses on one com- ponent of the schedule are punished, whereas responses on the other component continue to go unpunished (e.g., pecks on the blue key are punished, and pecks on the green key continue to produce reinforcement at the prior rate); (c) rate of responding decreases on the punished component and increases on the unpunished component (e.g., pecks on the blue key are suppressed and pecks on the green key increase even though pecks on the green key produce no more reinforcement than before).

Here is a hypothetical applied example of a contrast effect of punishment. A child is eating cookies before dinner from the kitchen cookie jar at equal rates in the presence and absence of his grandmother. One day, Grandma scolds the child for eating a cookie before din- ner, which suppresses his rate of predinner cookie eating when she is in the kitchen (see Figure 2); but when Grandma’s not in the kitchen, the boy eats cookies from the jar at a higher rate than he did when unsupervised prior to punishment. Contrast effects of punishment can be minimized, or prevented altogether, by consistently

355

Punishment by Stimulus Presentation

punishing occurrences of the target behavior in all rele- vant settings and stimulus conditions, withholding or at least minimizing the person’s access to reinforcement for the target behavior, and providing alternative desirable behaviors. (With respect to our hypothetical case of the child eating cookies before dinner, we recommend sim- ply removing the cookie jar!)

Punishment May Involve Undesirable Modeling

Most readers are familiar with the example of the parent who, while spanking a child, says, “This will teach you not to hit your playmates!” Unfortunately, the child may be more likely to imitate the parent’s actions, not the par- ent’s words. More than two decades of research have found a strong correlation between young children’s ex- posure to harsh and excessive punishment and antisocial behavior and conduct disorders as adolescents and adults (Patterson, 1982; Patterson, Reid, & Dishion, 1992; Sprague & Walker, 2000). Although the proper use of be- havior change tactics based on the principle of punish- ment does not involve harsh treatment or negative personal interactions, practitioners should heed Bandura’s (1969) valuable counsel in this regard:

Anyone attempting to control specific troublesome re- sponses should avoid modeling punitive forms of behav- ior that not only counteract the effects of direct training but also increase the probability that on future occasions the individual may respond to interpersonal thwarting in an imitative manner. (p. 313)

Negative Reinforcement of the Punishing Agent’s Behavior

Negative reinforcement may be a reason for the wide- spread use of (too often ineffective and unnecessary) and reliance on (mostly misguided) punishment in child rear- ing, education, and society. When Person A delivers a reprimand or other aversive consequence to Person B for misbehaving, the immediate effect is often the cessation of the troubling behavior, which serves as negative rein- forcement for Person A’s behavior. Or, as Ellen Reese (1966) said so succinctly, “Punishment reinforces the punisher” (p. 37). Alber and Heward (2000) described how the natural contingencies at work in a typical class- room could strengthen a teacher’s use of reprimands for disruptive behavior while undermining her use of con- tingent praise and attention for appropriate behavior.

Paying attention to students when they are behaving in- appropriately (e.g., “Carlos, you need to sit down right now!”) is negatively reinforced by the immediate cessa- tion of the inappropriate behavior (e.g., Carlos stops running around and returns to his seat). As a result, the teacher is more likely to attend to student disruptions in

the future. . . . Although few teachers must be taught to reprimand students for misbehavior, many teachers need help increasing the frequency with which they praise student accomplishments. Teacher-praising behavior is usually not reinforced as effectively as teacher- reprimanding behavior. Praising a student for appropri- ate behavior usually produces no immediate effects—the child continues to do his work when praised. Although praising a student for working productively on an as- signment may increase the future likelihood of that be- havior, there are no immediate consequences for the teacher. By contrast, reprimanding a student often pro- duces an immediate improvement in the teacher’s world (if only temporary)—that functions as effective negative reinforcement for reprimanding. (pp. 178–179)

Even though the reprimands may be ineffective in suppressing future frequency of misbehavior, the imme- diate effect of stopping the annoying behavior is power- ful reinforcement that increases the frequency with which the teacher will issue reprimands when confronted with misbehavior.

Examples of Positive Punishment Interventions Interventions based on positive punishment take a wide variety of forms. We describe five in this section: repri- mands, response blocking, contingent exercise, overcor- rection, and contingent electric stimulation.

Reprimands

It may seem strange that immediately following a dis- cussion of teachers’ overreliance on and misuse of repri- mands our first example of a positive punishment intervention is reprimands. The delivery of verbal repri- mands following the occurrence of misbehavior is with- out doubt the most common form of attempted positive punishment. However, a number of studies have shown that a firm reprimand such as “No!” or “Stop! Don’t do that!” delivered immediately on the occurrence of a be- havior can suppress future responding (e.g., Hall et al., 1971, see Figure 1; Jones & Miller, 1974; Sajwaj, Culver, Hall, & Lehr, 1972; Thompson et al., 1999).

In spite of the widespread use of reprimands in an effort to suppress undesired behavior, surprisingly few studies have examined the effectiveness of reprimands as punishers. Results of a series of experiments by Van Houten, Nau, Mackenzie-Keating, Sameoto, and Colavec- chia (1982) designed to identify variables that increased the effectiveness of reprimands as punishers for disruptive behavior in the classroom found that (a) reprimands de- livered with eye contact and “a firm grasp of the student’s

356

Punishment by Stimulus Presentation

shoulders” were more effective than reprimands without those nonverbal components, and (b) reprimands deliv- ered in close proximity to the student were more effective than reprimands delivered from across the room.

The teacher who repeatedly admonishes her students in a mild fashion to “sit down” would be advised instead to state once strongly, ”SIT DOWN!” When the com- mand is issued once, students are more likely to follow the direction. If the command is given repeatedly, stu- dents may habituate to the increased frequency, and the reprimand will gradually lose its effect as a punisher. A loud reprimand, however, is not necessarily more effec- tive than a reprimand stated in a normal voice. An inter- esting study by O’Leary, Kaufman, Kass, and Drabman (1970) found that quiet reprimands that were audible only to the child being reprimanded were more effective in re- ducing disruptive behavior than loud reprimands that could be heard by many children in the classroom.

If the only way a child receives adult attention is in the form of reprimands, it should not be surprising that reprimands function as reinforcement for that child rather than as punishment. Indeed, Madsen, Becker, Thomas, Koser, and Plager (1968) found that the repeated use of a reprimand while students were out of seat served to in- crease, rather than reduce, the behavior. Consistent with research on other punishing stimuli, reprimands are more effective as punishers when motivation for the problem behavior has been minimized and the availability of an al- ternative behavior has been maximized (Van Houten & Doleys, 1983).

A parent or teacher does not want to be in a pattern of constantly reprimanding. Reprimands should be used thoughtfully and sparingly in combination with frequent praise and attention contingent on appropriate behavior. O’Leary and colleagues (1970) recommended that

An ideal combination would probably be frequent praise, some soft reprimands, and very occasional loud reprimands. . . . Combined with praise, soft reprimands might be very helpful in reducing disruptive behaviors. In contrast, it appears that loud reprimands lead one into a vicious cycle of more and more reprimands resulting in even more disruptive behavior. (p. 155)

Response Blocking

Response blocking—physically intervening as soon as the person begins to emit the problem behavior to pre- vent or “block” the completion of the response—has been shown to be effective in reducing the frequency of some problem behaviors such as chronic hand mouthing, eye poking, and pica (e.g., Lalli, Livezy, & Kates, 1996; Ler- man & Iwata, 1996; Reid, Parsons, Phillips, & Green,

1993). In addition to preventing the response from oc- curring by using the least amount of physical contact and restraint possible, the therapist might issue a verbal rep- rimand or prompt to stop engaging in the behavior (e.g., Hagopian & Adelinis, 2001).

Lerman and Iwata (1996) used response blocking in treating the chronic hand mouthing (contact between any part of the hand and the lips or mouth) of Paul, a 32- year-old man with profound mental retardation. Fol- lowing a baseline condition, in which Paul was seated in a chair with no one interacting with him and no leisure materials were available, response blocking on an FR 1 schedule was implemented. A therapist sat be- hind Paul and blocked his attempts to put his hand in his mouth. “Paul was not prevented from bringing his hand to his mouth; however, the therapist blocked the hand from entering the mouth by placing the palm of her hand about 2 cm in front of Paul’s mouth” (p. 232). Response blocking produced an immediate and rapid decrease in hand mouthing attempts to near-zero levels (see Figure 5).

Response blocking is often implemented as a treat- ment for SIB or self-stimulatory behavior when func- tional analysis reveals consistent responding in the absence of socially mediated consequences, which sug- gests the possibility that the behavior is maintained by automatic reinforcement by sensory stimuli produced by the response. Because response blocking prevents the learner from contacting the sensory stimuli that are nor- mally produced by the response, subsequent decreases in responding could be due to extinction. Lerman and Iwata (1996) presented their study as a potential method for distinguishing whether the suppressive effects of re- sponse blocking are due to punishment or extinction mechanisms. They explained their reasoning as follows:

Depending on the mechanism through which behavior is reduced (extinction vs. punishment), different schedules of reinforcement or punishment are in effect when a given proportion of responses is blocked. For example when every fourth response is blocked (.25), the behav- ior is exposed to either a fixed-ratio (FR) 1.3 schedule of reinforcement (if blocking functions as extinction) or an FR 4 schedule of punishment (if blocking functions as punishment); when three out of four responses are blocked (.75), the behavior is exposed to either an FR 4 schedule of reinforcement or an FR 1.3 schedule of pun- ishment. Thus, as larger proportions of responses are blocked, the reinforcement schedule becomes leaner and the punishment schedule becomes richer. If response blocking produces extinction, response rates should in- crease or be maintained as more responses are blocked (i.e., as the reinforcement schedule is thinned), until [the effects of] extinction [i.e., reduced response rate] occurs at some point along the progression. Conversely, if the

357

Punishment by Stimulus Presentation

R es

p o

n se

s p

er M

in u

20 40 60

Sessions

Baseline 1.0BL .50.25.50 .67 .751.0

Response Block

Figure 5 Rates of hand mouthing during baseline and varying schedules of response blocking. From “A Methodology for Distinguishing between Extinction and Punishment Effects Associated with Response Blocking,” by D. C. Lerman and B. A. Iwata, 1996, Journal of Applied Behavior Analysis, 29, p. 232. Copyright 1996 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

procedure functions as punishment, response rates should decrease as more responses are blocked (i.e., as the punishment schedule becomes richer). (pp. 231–232, words in brackets added)

A condition in which all responses are blocked might function as an extinction schedule (i.e., reinforcement in the form of sensory stimuli is withheld for all responses) or as a continuous (FR 1) schedule of punishment (i.e., all responses are followed by physical contact). As Lerman and Iwata (1996) explained, if only some responses are blocked, the situation may function as an intermittent schedule of reinforcement or as an intermittent schedule of punishment. Therefore, comparing response rates among conditions in which different proportions of re- sponses are blocked should indicate whether the effects are due to extinction or to punishment.

If response blocking functioned as extinction for Paul’s hand mouthing, an initial increase in response rate would be expected when the blocking procedure was im- plemented for every response; however, no such increase was observed.5 If response blocking functioned as pun- ishment, blocking every response would constitute a con- tinuous schedule of punishment and a rapid decrease in responding would be expected; and that is exactly what the results showed (see data for the first response block [1.0] phase in Figure 5).

On the other hand, if response blocking functioned as extinction for Paul’s hand mouthing, then blocking some but not all responses would place the hand

5When an extinction procedure is first implemented, an increase in re- sponding, called an extinction burst, is sometimes observed before the re- sponse rate begins to decline. The principle, procedure, and effects of extinction are detailed in Chapter entitled “Extinction.”

6It might be argued that the suppressive effects of response blocking can- not be due to punishment or to extinction on the grounds that response blocking occurs before the response has been emitted and punishment and extinction are both response → consequence relations (in the case of ex- tinction, the consequence is absence of the reinforcement that followed the behavior in the past). As Lalli, Livezy, and Kates (1996) noted, response blocking prevents the response → consequence cycle from occurring. How- ever, if the problem behavior response class is conceptualized to include the occurrence of any portion of a relevant response, then blocking a person’s hand after it begins to move toward contact with his head is a consequence whose suppressive effects can be analyzed in terms of extinction, punish- ment, or both.

mouthing on an intermittent schedule of reinforcement and responding would be expected to increase from base- line levels. And blocking an ever-larger proportion of re- sponses thins the reinforcement schedule further causing the response rate to rise even higher. Instead, as a greater proportion of responses were blocked, the suppressive effects on Paul’s SIB became more pronounced, a result expected as a punishment schedule becomes denser. Overall, therefore, the results of the experiment indicated that response blocking functioned as punishment for Paul’s hand mouthing.

On the other hand, a systematic replication of Ler- man and Iwata’s (1996) experiment, conducted by Smith, Russo, and Le (1999), found that the frequency of eye- poking by a 41-year-old woman treated with response blocking decreased gradually—a response pattern in- dicative of extinction. The authors concluded that, “whereas blocking may reduce one participant’s behav- ior via punishment, it may extinguish another partici- pant’s behavior” (p. 369).6

Although response blocking may be viewed as a less restrictive and more humane intervention than delivering aversive stimulation after a response has occurred, it must

358

Punishment by Stimulus Presentation

be approached with great care. Side effects such as ag- gression and resistance to the response blocking proce- dure have occurred in some studies (Hagopian & Adelinis, 2001; Lerman, Kelley, Vorndran, & Van Camp, 2003). Providing prompts and reinforcement for an al- ternative response can minimize resistance and aggres- sion. For example, the aggressive behavior by 26-year-old man with moderate mental retardation and bipolar dis- order during treatment with response blocking for pica (ingesting paper, pencils, paint chips, and human feces) was reduced by supplementing response blocking with a prompt and redirection to engage in an alternative be- havior, in this case moving to an area of the room where popcorn was available (Hagopian & Adelinis, 2001).

Contingent Exercise

Contingent exercise is an intervention in which the per- son is required to perform a response that is not topo- graphically related to the problem behavior. Contingent exercise has been found effective as punishment for var- ious self-stimulatory, stereotypic, disruptive, aggressive, and self-injurious behaviors (e.g., DeCatanzaro & Bald- win, 1978; Kern, Koegel, & Dunlap, 1984; Luce & Hall,

7Increasing the effort or force required to perform a behavior can be an ef- fective tactic for reducing responding (Friman & Poling, 1995). There is no consensus as to whether punishment accounts for the reduced re- sponding. As with response blocking, one perspective from which increased response effort can be conceptualized as a punishment procedure is to con- sider the movement necessary to come into contact with the increased ef- fort requirement as a member of the target behavior response class. In that case, the increased effort required to continue the response to completion is (a) a consequence for the response that brought the learner into contact with it, and (b) aversive stimulation that functions as punishment as the frequency of future responding decreases.

1981; Luiselli, 1984).7 In perhaps the most frequently cited example of contingent exercise as a punisher, Luce, Delquadri, and Hall (1980) found that the repetition of mild exercise contingent on aggressive behavior by two boys with severe disabilities reduced it to near-zero lev- els. Figure 6 shows the results for Ben, a 7-year-old who frequently hit other children at school. Each time Ben hit someone, he was required to stand up and sit down 10 times. Initially, Ben had to be physically prompted to stand; an assistant held the child’s hand while pulling his upper body forward. Physical prompts were accompanied by verbal prompts of “stand up” and “sit down.” Soon, whenever hitting occurred, the nearest adult simply said, “Ben, no hitting. Stand up and sit down 10 times” and the verbal prompts alone were sufficient

N u

m b

er o

f H

it s

5 10 15 20 25 30 35 40 Days

Baseline Contingent

ExerciseBaseline Baseline Contingent

Exercise

Figure 6 Number of times a 7-year-old boy hit other children during 6-hour school days during baseline and contingent exercise. Xs represent response measures recorded by a second observer. From “Contingent Exercise: A Mild but Powerful Procedure for Suppressing Inappropriate Verbal and Aggressive Behavior,” by S. C. Luce, J. Delquadri, and R. V. Hall, 1980, Journal of Applied Behavior Analysis, 13, p. 587. Copyright 1980 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

359

Punishment by Stimulus Presentation

for Ben to complete the exercise. If a hitting episode oc- curred during the contingent exercise, the procedure was reinstated.

Overcorrection

Overcorrection is a behavior reduction tactic in which, contingent on each occurrence of the problem behavior, the learner is required to engage in effortful behavior that is directly or logically related to the problem. Orig- inally developed by Foxx and Azrin (1972, 1973; Foxx & Bechtel, 1983) as a method for decreasing disruptive and maladaptive behaviors of adults with mental retar- dation in institutional settings, overcorrection combines the suppressive effects of punishment and the educative effects of positive practice. Overcorrection includes ei- ther or both of two components: restitution and positive practice.

In restitutional overcorrection, contingent on the problem behavior, the learner is required to repair the damage caused by the problem behavior by returning the environment to its original state and then to engage in additional behavior that brings the environment to a con- dition vastly better than it was prior to the misbehavior. A parent applying restitutional overcorrection with a child who repeatedly tracks mud onto the kitchen floor might require the child to first wipe up the mud and clean his shoes and then to overcorrect the effects of his misbe- havior by mopping and waxing a portion of the floor and polishing his shoes.

Azrin and Foxx (1971) used restitutional overcorrec- tion in their toilet training program by requiring a person who had an accident to undress, wash her clothes, hang them up to dry, shower, dress in clean clothing, and then clean up a portion of the lavatory. Azrin and Wesolowki (1975) eliminated stealing of food by hospitalized adults with mental retardation by requiring residents to return not only the stolen food, or the portion that remained un- eaten, but also to purchase an additional item of that food at the commissary and give it to the victim.

Azrin and Besalel (1999) differentiated a procedure they called simple correction from overcorrection. With simple correction, the learner is required, subsequent to the occurrence of an inappropriate behavior, to restore the environment to its previous state. For example, a sim- ple correction procedure is in effect when requiring a stu- dent who cuts to the head of the lunch line to go to the back of the line. Requiring the student to wait until all others get in line and are served before reentering the line would constitute a form of overcorrection in this instance. Azrin and Besalel recommended that simple correction be used to reduce behaviors that are not severe, occur infre- quently, are not deliberate, and do not severely interfere with or annoy other people.

Correction is not possible if the problem behavior produces an irreversible effect (e.g., a one-of-a-kind dish is broken) or if the corrective behavior is beyond the per- son’s means or skills. In such instances, Azrin and Besalel (1999) recommended that the person be required to cor- rect as much of the damage his behavior caused as pos- sible, be present at all points in the correction, and assist with any parts of the correction that he is able to perform. For example, a child who broke a neighbor’s expensive window should clean up the pieces of glass, measure the window, contact the store for a replacement pane, be pre- sent while the new window pane is installed, and assist with every step.

In positive practice overcorrection, contingent on an occurrence of the problem behavior, the learner is re- quired to repeatedly perform a correct form of the be- havior, or a behavior incompatible with the problem behavior, for a specified duration of time or number of re- sponses. Positive practice overcorrection entails an ed- ucative component in that it requires the person to engage in an appropriate alternative behavior. The parent whose son tracks mud into the house could add a positive prac- tice component by requiring him to practice wiping his feet on the outside door mat and entering the house for 2 minutes or 5 consecutive times. Overcorrection that in- cludes restitution and positive practice helps teach what to do in addition to what not to do. The child who breaks an irreplaceable dish could be required to gently and slowly wash a number dishes, perhaps with an exagger- ated carefulness.

Researchers and practitioners have used positive practice overcorrection to reduce the frequency of prob- lem behaviors such as toilet training (Azrin & Foxx, 1971), self-stimulation and stereotypic behavior (Azrin, Kaplan, & Foxx, 1973; Foxx & Azrin, 1973), pica (Singh & Winton, 1985), bruxism (Steuart, 1993), sibling ag- gression (Adams & Kelley, 1992), and classroom dis- ruptions (Azrin & Powers, 1975). Positive practice overcorrection has been used for academic behaviors (Lenz, Singh, & Hewett, 1991), most often to decrease oral reading and spelling errors (e.g., Ollendick, Matson, Esveldt-Dawson, & Shapiro, 1980; Singh & Singh, 1986; Singh, Singh, & Winton, 1984; Stewart & Singh, 1986).

Positive practice overcorrection can also be applied to reduce or eliminate behaviors that do not create per- manent response products that can be repaired or restored to their original state. For example, Heward, Dardig, and Rossett (1979) described how parents used positive prac- tice overcorrection to help their teenage daughter stop making a grammatical error in her speech. Eunice fre- quently used the contraction “don’t” instead of “doesn’t” with the third person singular (e.g., “He don’t want to go.”). A positive reinforcement program in which Eunice earned points she could redeem for preferred activities

360

Punishment by Stimulus Presentation

each time she used “doesn’t” correctly had little effect on her speech. Eunice agreed with her parents that she should speak correctly but claimed the behavior was a habit. Eunice and her parents then decided to supplement the reinforcement program with a mild punishment pro- cedure. Each time Eunice or her parents caught her using “don’t” incorrectly in her speech, she was required to say the complete sentence she had just spoken 10 times in a row using correct grammar. Eunice wore a wrist counter that reminded her to listen to her speech and to keep track of the number of times she employed the positive prac- tice procedure.

When positive practice effectively suppresses the problem behavior, it is not clear what behavioral mech- anisms are responsible for the behavior change. Punish- ment may result in decreased frequency of responding, because the person engages in effortful behavior as a con- sequence of the problem behavior. A reduction in the fre- quency of the problem behavior as an outcome of positive practice may also be a function of an increased frequency of an incompatible behavior, the correct behavior that is strengthened in the person’s repertoire as the result of the intensive, repeated practice. Azrin and Besalel (1999) suggested that the reason why positive practice is effec- tive varies depending upon the whether the problem be- havior is “deliberate” or the result of a skill deficit:

Positive practice may be effective because of the incon- venience and effort involved, or because it provides ad- ditional learning. If the child’s errors are caused by a deliberate action, the extra effort involved in positive practice will discourage future misbehaviors. But if the misbehavior is the result of insufficient learning, the child will stop the misbehavior—or error—because of the intensive practice of the correct behavior. (p. 5)

Although specific procedures for implementing over- correction vary greatly depending upon the problem be- havior and its effects on the environment, the setting, the desired alternative behavior, and the learner’s current skills, some general guidelines can be suggested (Azrin & Besalel, 1999; Foxx & Bechtel, 1983; Kazdin, 2001; Miltenberger & Fuqua, 1981):

1. Immediately upon the occurrence of the problem behavior (or the discovery of its effects), in a calm, unemotional tone of voice, tell the learner that he has misbehaved and provide a brief explanation for why the behavior must be corrected. Do not criti- cize or scold. Overcorrection entails a logically re- lated consequence to reduce future occurrences of the problem behavior; criticism and scolding do not enhance the tactic’s effectiveness and may harm the relationship between the learner and practitioner.

2. Provide explicit verbal instructions describing the overcorrection sequence the learner must perform.

3. Implement the overcorrection sequence as soon as possible after the problem behavior has occurred. When circumstances prevent immediately com- mencing the overcorrection sequence, tell the learner when the overcorrection process will be conducted. Several studies have found that overcor- rection conducted at a later time can be effective (Azrin, & Powers, 1975; Barton & Osborne, 1978).

4. Monitor the learner throughout the overcorrection activity. Provide the minimal number of response prompts, including gentle physical guidance, needed to ensure that the learner performs the en- tire overcorrection sequence.

5. Provide the learner with minimal feedback for correct responses. Do not give too much praise and attention to the learner during the overcorrec- tion sequence.

6. Provide praise, attention, and perhaps other forms of reinforcement to the learner each time he “spon- taneously” performs the appropriate behavior dur- ing typical activities. (Although technically not part of the overcorrection procedure, reinforcing an alternative behavior is a recommended comple- ment to all punishment-based interventions.)

Although “the results of a few minutes of corrective training after the undesired behavior have often led to rapid and long-lasting therapeutic effects” (Kazdin, 2001, p. 220), practitioners should be aware of several potential problems and limitations associated with overcorrection. First, overcorrection is a labor intensive, time-consuming procedure that requires full attention of the practitioner implementing it. Implementing overcorrection usually requires the practitioner to monitor the learner directly throughout the overcorrection process. Second, for over- correction to be effective as punishment, the time that the learner spends with the person monitoring the overcor- rection sequence must not be reinforcing. “If it is, it just might be worth waxing the entire kitchen floor if Mom chats with you and provides a milk and cookies break” (Heward et al., 1979, p. 63).

Third, a child who misbehaves frequently may not execute a long list of “cleanup” behaviors just because he was told to do so. Azrin and Besalel (1999) recom- mended three strategies to minimize the likelihood of re- fusal to perform the overcorrection sequence: (1) remind the learner what the more severe disciplinary action will be and, if the refusal persists, then impose that discipline; (2) discuss the need for correction before the problem behavior occurs; and (3) establish correction as an ex- pectation and routine habit for any disruptive behavior. If

361

Punishment by Stimulus Presentation

the child resists too strongly or becomes aggressive, over- correction may not be a viable treatment. Adult learners must voluntarily make the decision to perform the over- correction routine.

Contingent Electric Stimulation

Contingent electric stimulation as punishment involves the presentation of a brief electrical stimulus immedi- ately following an occurrence of the problem behavior. Although the use of electric stimulation as treatment is controversial and evokes strong opinions, Duker and Seys (1996) reported that 46 studies have demonstrated that contingent electric stimulation can be a safe and highly effective method for suppressing chronic and life- threatening self-injurious behavior (SIB). One of the most rigorously researched and carefully applied procedures for implementing punishment by electric stimulation for self-inflicted blows to the head or face is the Self- Injurious Behavior Inhibiting System (SIBIS) (Linscheid, Iwata, Ricketts, Williams, & Griffin, 1990; Linscheid, Pejeau, Cohen, & Footo-Lenz, 1994; Linscheid & Reichenbach, 2002). The SIBIS apparatus consists of a sensor module (worn on the head) and a stimulus mod- ule (worn on the leg or arm) that contains a radio receiver, a 9 V battery, and circuitry for the generation and timing of the electric stimulus. Above–threshold blows to the face or to the head trip an impact detector in the sensor module that transmits a radio signal to a receiver in the stimulus module that in turn produces an audible tone followed by electric stimulation (84 V, 3.5 mA) for 0.08 seconds. “Subjectively, the experience has been described at its extremes as imperceptible (low) and similar to hav- ing a rubber band snapped on the arm (high)” (Linscheid et al., 1990, p. 56).

Linscheid and colleagues (1990) evaluated SIBIS as treatment for five people with “longstanding, severe, and previously unmanageable” self-injurious behavior. One of the participants was Donna, a 17-year-old girl with pro- found mental retardation, no language, and no indepen- dent feeding or toileting skills. Donna’s parents reported that she began hitting herself with sufficient force to pro- duce lesions on her face and head more than 10 years prior to the study. Numerous treatments to stop Donna from hurting herself had been tried without success, in- cluding differential reinforcement of other behaviors (DRO), “gentle teaching,” redirection, and response pre- vention. “For example, in order to prevent head hitting in bed, her parents had to hold her arms each night until she fell asleep. This sometimes required 3 to 4 hours of un- divided attention, which the parents felt they were un- able to continue” (pp. 66–67). Although Donna was ambulatory, when the study began she was spending most

of the day with her wrists restrained to the arms of a wheelchair to prevent her SIB.

The effects of SIBIS on Donna’s SIB were evaluated with a reversal design that included SIBIS-active and SIBIS-inactive conditions. All sessions lasted for 10 min- utes or until Donna hit her head 25 times. During SIBIS- inactive sessions, Donna wore the sensor and stimulus modules but the stimulus module was inoperative. Dur- ing six sessions each of initial baseline and SIBIS-inactive phases, Donna hit her head at a mean rate of at least once per second (68.1 and 70.2 responses per minute, respec- tively) (see Figure 7). In the first session in which SIBIS was applied, Donna’s SIB decreased to 2.4 head hits per minute. The mean rate of head hitting during all SIBIS- active sessions was 0.5 responses per minute (a range of 0 to 5.6), compared to rates of 50.2 (1.7 to 78.9) and 32.5 (0 to 48.0) responses per minute for all baseline and SIBIS-inactive sessions combined, respectively. SIBIS reduced Donna’s head hitting 98.9% from baseline lev- els. Across all of the SIBIS-active sessions combined, Donna received 32 electric stimulations lasting a com- bined 2.6 seconds. No sessions during the SIBIS-active condition had to be terminated early because of risk, whereas 100% of the baseline sessions and 64% of the SIBIS-inactive sessions were terminated because Donna hit her head 25 times.

Similar results were obtained for all five partici- pants: SIBIS treatment produced an immediate and al- most complete cessation of SIB. Although legal, ethical, and moral issues and concerns surround the use of elec- tric aversive stimulation, Linscheid and Reichenbach (2002) offered the following perspective: “While the de- cision to use an aversive treatment must be made in con- sideration of numerous factors, it is suggested that the speed and degree of suppression of SIB must be among these considerations” (p. 176). Formal data and anecdo- tal reports indicated an absence of negative side effects and the occurrence of some positive side effects. For ex- ample, Donna’s parents reported a general improvement in her overall adaptive functioning and that she no longer had to be restrained in bed at night. Donna’s teacher re- ported that:

Since the introduction of SIBIS, it is like we have a to- tally new girl in the classroom. Donna no longer has to have her hands restrained. She is walking around the classroom without the wrestling helmet or cervical col- lar. She smiles more frequently and fusses a lot less. She pays more attention to what is going on in the class- room. She reaches out for objects and people more than she did. (p. 68)

Although the maintenance of suppressive effects has not been universally demonstrated (Ricketts, Goza, &

362

Punishment by Stimulus Presentation

100

R es

p o

n se

s p

er M

in u

Baseline SIBIS

Inactive SIBIS SIBIS

Inactive BL SIBIS

Inactive SIBIS Donna

10 20 30 40 50 Sessions

Figure 7 Number of head hits per minute by a 17-year-old with a 10-year history of self-injurious behavior during baseline and SIBIS-inactive and SIBIS- active conditions. From “Clinical Evaluation of SIBIS: The Self-Injurious Behavior Inhibiting System” by T. R. Linscheid, B. A. Iwata, R. W. Ricketts, D. E. Williams, and J. C. Griffin, 1990, Journal of Applied Behavior Analysis, 23, p. 67. Copyright 1990 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Matese, 1993), some have reported the long-term effec- tiveness of SIBIS (Linscheid & Reichenbach, 2002). For example, Salvy, Mulick, Butter, Bartlett, and Linscheid (2004) reported that the SIB of a 3-year-old female with an 18-month history of chronic and repetitive head bang- ing remained at virtually zero levels at a 1-year post- SIBIS follow-up.

Guidelines for Using Punishment Effectively Research and clinical applications have demonstrated that punishment can yield rapid, long-lasting suppression of problem behavior. Increasingly, agency policies, human subject review procedures, and historic practices have lim- ited the use of punishment for research and treatment in clinic-based settings (Grace, Kahng, & Fisher, 1994). Pun- ishment, however, may be a treatment of choice when: (a) the problem behavior produces serious physical harm and must be suppressed quickly, (b) reinforcement-based treatments have not reduced the problem behavior to so- cially acceptable levels, or (c) the reinforcer maintaining

the problem behavior cannot be identified or withheld (Lerman & Vorndran, 2002).

If the decision is made to use a punishment-based treatment, steps should be taken to ensure that the pun- ishing stimulus is as effective as possible. The guide- lines that follow will help practitioners apply punishment with optimal effectiveness while minimizing undesir- able side effects and problems. We based these guide- lines on the assumption that the analyst has conducted a functional behavior assessment to identify the variables that maintain the problem behavior, that the problem be- havior has been defined in a manner that minimizes am- biguity concerning its parameters and that the participant cannot avoid or escape the punisher.

Select Effective and Appropriate Punishers

Conduct Punisher Assessments

a parallel process can be applied to As with the reinforcer assessment methods,

363

Punishment by Stimulus Presentation

identify stimuli that are likely to function as punish- ers. Fisher and colleagues (1994), who differentially applied a combination of empirically derived punishers and reinforcers to reduce the pica behavior of three children referred to an inpatient treatment facility, iden- tified two advantages of conducting punisher assess- ments. First, the sooner an effective punisher can be identified, the sooner it can be applied to treat the prob- lem behavior. Second, data from punisher assessments might reveal the magnitude or intensity of the punisher necessary for behavioral suppression, enabling the practitioner to deliver the lowest intensity punisher that is still effective.

Punisher assessment mirrors stimulus preference and/or reinforcement assessment, except that instead of measuring engagement or duration of contact with each stimulus, the behavior analyst measures negative ver- balizations, avoidance movements, and escape attempts associated with each stimulus. Data from the punisher assessments are then used to develop a hypothesis on the relative effectiveness of each stimulus change as a punisher.

The decision of which among several potentially ef- fective punishers to choose should be based on the rela- tive degree of intrusiveness the punisher creates and the ease with which it can be consistently and safely deliv- ered by the therapists, teachers, or parents who will implement the punishment procedure in the clinic, classroom, or home. Subsequent observation might re- veal that a consequence that is less intrusive, time- consuming, or difficult to apply might be used, as in the following experience reported by Thompson and col- leagues (1999):

We conducted brief evaluations to identify an ef- fective punishment procedure for each participant. Procedures were chosen for evaluation based on topographies of SIB, an apparent degree of minimal intrusiveness, and the ability of the experimenter to safely and efficiently implement the procedure. Dur- ing this phase, we used brief AB designs to evaluate several procedures and chose the least restrictive pro- cedure that resulted in a 75% or greater decrease in SIB. For example, we initially evaluated a 15-s man- ual restraint with Shelly. During this procedure, the therapist delivered a verbal reprimand, held Shelly’s hands in her lap for 15 s, and then dried her hands with a cloth. This procedure reduced SIB to the crite- rion level. Subsequently, however, we observed a comparable decrease in SIB when the therapist sim- ply delivered a reprimand (e.g., “no spitting”) and dried Shelly’s hands (without holding her hands in her lap). Therefore, we chose to implement the repri- mand and hand-drying procedure. (p. 321)

Use Punishers of Sufficient Quality and Magnitude

The quality (or effectiveness) of a punisher is relative to a number of past and current variables that affect the par- ticipant. For example, Thompson and colleagues found that 15 seconds of physical restraint was a high-quality punisher for Ricky, Donna, and Lynn, but the physical restraint was not influential for Shelly. Although stimuli that reliably evoke escape and avoidance behaviors often function as high-quality punishers, practitioners should recognize that a stimulus change that effectively sup- presses some behaviors may not affect other behaviors and that highly motivated problem behaviors may only be suppressed by a particularly high-quality punisher.

Generally, basic and applied researchers have found that the greater the intensity (magnitude, or amount) of a punishing stimulus, the greater the suppression of be- havior. This finding is conditional on delivering the pun- ishing stimulus at its optimum level of magnitude initially, rather than gradually increasing the level over time (Azrin & Holtz, 1966). For example, Thompson and colleagues (1999) used the previously described punisher assessment procedure to determine an optimum level of magnitude for the punishing stimulus: a magnitude that produced a 75%, or greater, decrease in self-injurious be- haviors (SIB) from baseline occurrences of four adult participants. The punishing stimuli meeting their crite- rion included: “Ricky’s hands were held in his lap for 15 s each time he engaged in SIB. Donna and Lynn both received a verbal reprimand and had their hands held across their chests for 15 s following SIB” (p. 321).

Beginning with a punishing stimulus of sufficient magnitude is important, because participants may adapt to the punishing stimulus when levels of magnitude are increased gradually. For example, it is possible that 15 seconds of movement restriction would have been inef- fective as punishment for Ricky, Donna, and Lynn’s SIB if Thompson and colleagues had begun with a punishing stimulus of 3 seconds and gradually increased its mag- nitude in 3-second intervals.

Use Varied Punishers

The effectiveness of a punishing stimulus can decrease with repeated presentations of that stimulus. Using a vari- ety of punishers may help to reduce habituation effects. In addition, using various punishers may increase the effec- tiveness of less intrusive punishers. For example, Charlop, Burgio, Iwata, and Ivancic (1988) compared various pun- ishers to a single presentation of one the punishers (i.e., a stern “No!” overcorrection, time-out with physical re- straint, a loud noise). Three-, 5-, and 6-year-old children

364

Punishment by Stimulus Presentation

with developmental disabilities served as participants. Their problem behaviors included aggression (Child 1), self-stimulation and destructive behavior (Child 2), and aggression and out-of-seat (Child 3). The varied-punisher condition was slightly more effective than the single-pre- sentation condition and enhanced the sensitivity of the be- havior to less intrusive punishing stimuli. Charlop and colleagues concluded, “It appears that by presenting a var- ied format of commonly used punishers, inappropriate be- haviors may further decrease without the use of more intrusive punishment procedures” (p. 94; see Figure 8).

Deliver the Punisher at the Beginning of a Behavioral Sequence

Punishing an inappropriate behavior as soon as it begins is more effective than waiting until the chain of behavior has been completed (Solomon, 1964). Once the sequence of responses that make up the problem behavior is initi- ated, powerful secondary reinforcers associated with com- pleting each step of the chain may prompt its continuation, thereby counteracting the inhibiting or suppressing effects of the punishment that occurs at the end of the sequence. Therefore, whenever practical, the punishing stimulus should be presented early in the behavioral sequence

rather than later. For example, if violent arm swinging is a reliable precursor to self-injurious eye poking, then pun- ishment (e.g., response blocking, restraint) should be de- livered as soon as arm swinging starts.

Punish Each Instance of the Behavior Initially

Punishment is most effective when the punisher follows each response. This is especially important when a pun- ishment intervention is first implemented.

Gradually Shift to an Intermittent Schedule of Punishment

Although punishment is most effective when the punisher immediately follows each occurrence of the problem be- havior, practitioners may find a continuous schedule of punishment unacceptable because they lack the resources and time to attend to each occurrence of the behavior (O’Brien & Karsh, 1990). Several studies have found that, after responding has been reduced by a continuous sched- ule of punishment, an intermittent schedule of punish- ment may be sufficient to maintain the behavior at a

100

P er

ce n

t In

te rv

al s

No Treatment

"No"

Varied

Time-Out

5 10 15 20 25 30 35 40 45 50 Sessions

Overcorrection

Baseline Treatment Baseline Treatment Baseline Treatment Treatment

Figure 8 Percentage of intervals of occurrence of self-stimulation and destructive behavior for a 6-year-old girl with autism. From “Stimulus Variation as a Means of Enhancing Punishment Effects,” by M. H. Charlop, L. D. Burgio, B. A. Iwata, and M. T. Ivancic, 1988, Journal of Applied Behavior Analysis, 21, p. 92. Copyright 1988 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

365

Punishment by Stimulus Presentation

socially acceptable frequency (e.g. Clark, Rowbury, Baer, & Baer, 1973; Lerman et al., 1997; Romanczyk, 1977).

We recommend two guidelines for using intermittent punishment. First—and this is especially important—a con- tinuous (FR 1) schedule of punishment should be used to diminish the problem behavior to a clinically acceptable level before gradually thinning to an intermittent schedule of punishment. Second, combine intermittent punishment with extinction. It is unlikely that reduced responding will continue under intermittent punishment if the reinforcer that maintains the problem behavior cannot be identified and withheld. If these two guidelines for intermittent pun- ishment are met and the frequency of the problem behav- ior increases to an unacceptable level, return to a continuous schedule of punishment and then, after recovery of accept- ably low rates of responding, gradually shift to an inter- mittent schedule of punishment that is denser than used previously (e.g., VR 2 rather than a VR 4).

Use Mediation with a Response-to-Punishment Delay

As with reinforcers, punishers that occur immediately following a response are more effective than are punish- ers presented after a period of time has elapsed since the response occurred. In the context of reinforcement, Stromer, McComas, and Rehfeldt (2000) reminded us that using continuous and intermittent schedules of rein- forcement might be just the first steps of programming consequences for everyday situations, because the con- sequences that maintain responding in natural environ- ments are often delayed:

Establishing the initial instances of a behavioral repertoire typically requires the use of programmed consequences that occur immediately after the target response occurs. However, the job of the applied behavior analyst also in- volves the strategic use of delayed reinforcement. (p. 359)

The job of the applied behavior analyst could also involve the strategic programming of contingencies in- volving a delay-to-punishment interval. As Lerman and Vorndran, (2002) noted:

Consequences for problem behavior are frequently de- layed in the natural environment. Caregivers and teach- ers often are unable to monitor behavior closely or to deliver lengthy punishers (e.g., 15-min contingent work) immediately following instances of problem behavior. Punishment also may be delayed when the individual actively resists application of the programmed conse- quences by struggling with the punishing agent or run- ning away. In some cases, problem behavior occurs primarily in the absence of the punishing agent, neces- sarily delaying programmed consequences until the be- havior is detected. (pp. 443-444)

Generally, applied behavior analysts have avoided programming a delay-to-punishment interval. In their re- view of the basic and applied punishment literature, Ler- man and Vorndran (2002) found just two applied studies that addressed the variables related to the effective use of a delay-to-punishment (Rolider & Van Houten, 1985; Van Houten & Rolider, 1988), and noted that these ap- plied behavior analysts used a variety of techniques to mediate the delay between the occurrence of the prob- lem behavior and the punishing consequences.

[They] demonstrated the efficacy of delayed punishment using various mediated consequences with children with emotional and developmental disabilities. One form of mediation involved playing audiotape recordings of the child’s disruptive behavior that were collected earlier in the day. The punishing consequence (physical restraint, verbal reprimands) then was delivered. In some cases, the tape recorder was clearly visible to the child while the recordings were being collected, and a verbal expla- nation of its role in the delivery of delayed punishment was provided. These factors may have served to bridge the temporal gap between inappropriate behavior and its consequence (e.g., by functioning as discriminative stimuli for punishment). (p. 444)

Supplement Punishment with Complementary Interventions

Applied behavior analysts typically do not use punish- ment as a single intervention: they supplement punish- ment with other interventions, primarily, differential reinforcement, extinction, and a variety of antecedent in- terventions. Basic and applied researchers consistently find that the effectiveness of punishment is enhanced when the learner can make other responses for rein- forcement. In most circumstances applied behavior ana- lysts should incorporate differential reinforcement of alternative behavior (DRA), differential reinforcement of incompatible behavior (DRI), or differential reinforce- ment of other behaviors (DRO) into a treatment program to supplement a punishment. When used as a reductive procedure for problem behavior, differential reinforce- ment consists of two components: (1) providing rein- forcement contingent on the occurrence of a behavior other than the problem behavior, and (2) withholding re- inforcement for the problem behavior (i.e., extinction). The study by Thompson and colleagues (1999) presented previously in this chapter (see Figure 4) provides an ex- cellent example of how reinforcement of an alternative behavior can enhance the suppressive effects of punish- ment and enable “relatively benign punishment proce- dures” to be effective even for chronic problem behaviors that have been resistant to change.

366

Punishment by Stimulus Presentation

We recommend that practitioners reinforce alterna- tive behaviors copiously. Additionally, the more rein- forcement the learner obtains by emitting appropriate behaviors, the less motivated he will be to emit the prob- lem behavior. In other words, heavy and consistent doses of reinforcement for alternative behaviors function as an abolishing operation that weakens (abates) the frequency of the problem behavior.

There is still another important reason for recom- mending the reinforcement of alternative behaviors. Applied behavior analysts are in the business of build- ing repertoires by teaching the clients and students they serve new skills and more effective ways of control- ling their environments and achieving success. Pun- ishment (with the exception of some overcorrection procedures) eliminates behaviors from a person’s reper- toire. Although the person will be better off without those behaviors in her repertoire, punishment only teaches her what not to do, it does not teach her what to do instead.

The effectiveness of antecedent interventions such as functional communication training (FCT), high- probability (high-p) request sequence, and noncontin- gent reinforcement (NCR), which diminish the frequency of problem behaviors by decreasing the effectiveness of the reinforcers that maintain the problem behaviors, can be made more effective when combined with pun- ishment. For instance, Fisher and colleagues (1993) found that, although functional communication train- ing (FCT) did not reduce to clinically significant levels the destructive behaviors of four participants with se- vere retardation and communication deficits, a combi- nation of FCT and punishment produced the largest and most consistent reductions in problem behaviors.

Be Prepared for Negative Side Effects

It is difficult to predict the side effects that may result from punishment (Reese, 1966). The suppression of one problem behavior by punishment may lead to an in- crease in other undesirable behaviors. For example, pun- ishment of self-injurious behavior may produce increased levels of noncompliance or aggression. Pun- ishing one problem behavior may lead to a parallel de- crease in desirable behaviors. For example, requiring that a student rewrite an ill-conceived paragraph may result in his ceasing to produce any academic work. Al- though punishment may produce no undesirable side effects, practitioners should be alert to problems such as escape and avoidance, emotional outbursts, and behav- ioral contrast, and should have a plan for dealing with such events should they occur.

Record, Graph, and Evaluate Data Daily

Data collection in the initial sessions of a punishment- based intervention is especially critical. Unlike some be- havior reduction procedures, such as extinction and differential reinforcement of alternative behavior, whose suppressive effects are often gradual, the suppressive ef- fects of punishment are abrupt. In their classic review of the basic research on punishment, Azrin and Holz (1966) wrote:

Virtually all studies of punishment have been in complete agreement that the reduction of responses by punishment is immediate if the punishment is at all effective. When the data have been presented in terms of the number or responses per day, the responses have been drastically re- duced or eliminated on the very first day in which pun- ishment was administered. When the data have been presented in terms of moment-to-moment changes, the reduction of responses has results with the first few de- liveries of punishment or within a few minutes. (p. 411)

The first data points from the punishment conditions in all of the graphs presented in this chapter provide addi- tional empirical evidence for Azrin and Holz’s statement of the immediate effects of punishment. Because of the abrupt effect of punishment, the practitioner should pay particular attention to the data from the first session or two of intervention. If a noticeable reduction of the problem behavior has not occurred within two sessions of a pun- ishment-based intervention, we recommend that the prac- titioner make adjustments to the intervention.

Frequent inspection of the data from a punishment perspective reminds those involved of the purpose of the intervention and reveals whether the problem behavior is being reduced or eliminated as intended. When the data indicate that a clinically or socially significant change has occurred and has been maintained, punishment can be shifted to an intermittent schedule or perhaps terminated altogether.

Ethical Considerations Regarding the Use of Punishment Ethical considerations regarding the use of punishment revolve around three major issues: the client’s right to safe and humane treatment, the professional’s responsibility to use least restrictive procedures, and the client’s right to effective treatment.8

8Chapter entitled “Ethical Considerations for Applied Behavior Analysts” provides a detailed discussion of ethical issues and practices for applied be- havior analysts.

367

Punishment by Stimulus Presentation

Right to Safe and Humane Treatment

Dating from the Hippocratic Oath (Hippocrates, 460 BC–377 BC), the first ethical canon and responsibil- ity for any human services provider is to do no harm. Accordingly, any behavior change program, be it a punishment-based intervention to reduce life-threatening self-injurious behavior or an application of positive re- inforcement to teach a new academic skill, must be phys- ically safe for all involved and contain no elements that are degrading or disrespectful to participants.

Treatments are deemed safe when they put neither the caregiver nor the target individual at physical, psy- chological, or social risk (Favell & McGimsey, 1993). Although there exists no universally accepted definition of what constitutes humane treatment, a reasonable case could be made that humane treatments are (a) designed for therapeutic effectiveness, (b) delivered in a compas- sionate and caring manner, (c) assessed formatively to determine effectiveness and terminated if effectiveness is not demonstrated, and (d) sensitive and responsive to the overall physical, psychological, and social needs of the person.

Least Restrictive Alternative

A second canon of ethics for human services profes- sionals is to intrude on a client’s life only as necessary to provide effective intervention. The doctrine of the least restrictive alternative holds that less intrusive procedures should be tried and found to be ineffective before more intrusive procedures are implemented. Interventions can be viewed as falling along a continuum of restrictiveness from least to most. The more a treatment procedure af- fects a person’s life or independence, such as his ability to go about daily activities in his normal environment, the greater its restrictiveness. A completely unrestrictive intervention is a logical fallacy. Any treatment must af- fect the person’s life in some way to qualify as an inter- vention. At the other end of the continuum, absolute restrictiveness exists during solitary confinement, where personal independence is unavailable. All behavior change interventions developed and implemented by ap- plied behavior analysts fall within these extremes.

Selecting any punishment-based intervention es- sentially rules out as ineffective all positive or positive reductive approaches based on their demonstrated in- ability to improve the behavior. For example, Gaylord- Ross (1980) proposed a decision-making model for reducing aberrant behavior that suggested that practi- tioners rule out assessment considerations, inappropriate or ineffective schedules of reinforcement, ecological

variables, and curriculum modifications before punish- ment is implemented.

Some authors and professional organizations have advanced the position that all punishment-based proce- dures are inherently intrusive and should never be used (e.g., Association for Persons with Severe Handicaps, 1987; LaVigna & Donnellen, 1986; Mudford, 1995). Oth- ers have advanced the counter position that punishment- based procedures, because of their inherent level of intrusiveness, should be used only as a last resort (Gay- lord-Ross, 1980).

Most people would rate interventions based on pos- itive reinforcement as less restrictive than interventions based on negative reinforcement; reinforcement inter- ventions as less restrictive than punishment interventions; and interventions using negative punishment as less re- strictive than those using positive punishment. However, the intrusiveness or restrictiveness of an intervention can- not be determined by the principles of behavior on which it is based. Restrictiveness is a relative concept that de- pends on the procedural details and ultimately rests at the level of the person with whom it is applied. A posi- tive reinforcement intervention that requires deprivation may be more restrictive than a positive punishment pro- cedure in which a buzzer sounds each time an incorrect response is made, and what one person considers intru- sive may pose no discomfort for another person. Horner (1990) suggested that most people accept punishment in- terventions to reduce challenging behavior as long as those interventions: “(a) do not involve the delivery of physical pain, (b) do not produce effects that require med- ical attention, and (c) are subjectively judged to be within the typical norm of how people in our society should treat each other” (pp. 166-167).

In their review of response-reduction interventions based on the principle of punishment, Friman and Poling (1995) pointed out that punishment-based tactics, such as those that require that the person make an effortful re- sponse contingent on the occurrence of target behavior, meet Horner’s (1990) criteria for acceptable response- reduction interventions. They stated: “None of the pro- cedures caused pain or required medical attention, nor did they contrast with societal norms. For example, coaches often require disobedient players to run laps, and drill instructors make misbehaving recruits do push- ups” (p. 585).

Although least restrictive alternative practices as- sume that the less intrusive procedures are tried and found ineffective before a more restrictive interven- tion is introduced, practitioners must balance that ap- proach against an effectiveness standard. Gast and Wolery (1987) suggested that, if a choice must be made between a less intrusive but ineffective procedure

368

Punishment by Stimulus Presentation

and a more intrusive but effective procedure, the latter should be chosen. A Task Force on the Right to Ef- fective Treatment appointed by the Association for Be- havior Analysis provided the following perspective on the importance of judging the ultimate restrictiveness of treatment options on the basis of their degree of proven effectiveness:

Consistent with the philosophy of least restrictive yet effective treatment, exposure of an individual to restric- tive procedures is unacceptable unless it can be shown that such procedures are necessary to produce safe and clinically significant behavior change. It is equally unac- ceptable to expose an individual to a nonrestrictive inter- vention (or a series of such interventions) if assessment results or available research indicate that other proce- dures would be more effective. Indeed, a slow-acting but nonrestrictive procedure could be considered highly restrictive if prolonged treatment increases risk, signifi- cantly inhibits or prevents participation in needed train- ing programs, delays entry into a more optimal social or living environment, or leads to adaptation and the even- tual use of a more restrictive procedure. Thus, in some cases, a client’s right to effective treatment may dictate the immediate use of quicker-acting, but temporarily more restrictive, procedures.

A procedure’s overall level of restrictiveness is a combined function of its absolute level of restrictive- ness, the amount of time required to produce a clinically acceptable outcome, and the consequences associated with delayed intervention. Furthermore, selection of a specific treatment technique is not based on personal conviction. Techniques are not considered either “good” or “bad” according to whether they involve the use of antecedent rather than consequent stimuli or reinforce- ment rather than punishment. For example, positive rein- forcement, as well as punishment, can produce a number of indirect effects, some of which are undesirable.

In summary, decisions related to treatment selection are based on information obtained during assessment about the behavior, the risk it poses, and its controlling variables; on a careful consideration of the available treatment options, including their relative effectiveness, risks, restrictiveness, and potential side effects; and on examination of the overall context in which treatment will be applied. (Van Houten et al., 1988, pp. 113, 114)

Right to Effective Treatment

Ethical discussions regarding the use of punishment re- volve most often around its possible side effects and how experiencing the punisher may cause unnecessary pain and possible psychological harm for the person. Although each of these concerns deserves careful con- sideration, the right to effective treatment raises an equally important ethical issue, especially for persons

who experience chronic, life-threatening problem be- haviors. Some maintain that failing to use a punish- ment procedure that research has shown to be effective in suppressing self-destructive behavior similar to their client’s is unethical because doing so withholds a po- tentially effective treatment and risks maintaining a dangerous or uncomfortable state for the person. For example, Baer (1971) stated that, “[Punishment] is a legitimate therapeutic technique that is justifiable and commendable when it relieves persons of the even greater punishments that result from their own habitual behavior” (p. 111).

For some clients, punishment-based intervention may be the only means of reducing the frequency, duration, or magnitude of chronic and dangerous behaviors that have proven resistant to change using positive reinforcement, extinction, or positive reductive approaches such as dif- ferential reinforcement of alternative behavior. Such cir- cumstances “may justify, if not necessitate, the use of punishment” (Thompson et al., 1999, p. 317). As Iwata (1988) explained, if all other less-intrusive treatments with a sufficient research base to promise a legitimate chance of success have failed, the use of punishment pro- cedures is the only ethical option:

In the ideal world, treatment failures do not occur. But in the actual world failures do occur, in spite of our best efforts, leaving us with these options: continued occur- rence of the behavior problem toward some devastating endpoint, restraint, sedating drugs, or aversive contin- gencies. I predict that if we apply our skills to maximize the effectiveness of positive reinforcement programs, we will succeed often. After following such a course, my further prediction is that if we reach the point of having to decide among these ultimate default options, the client’s advocate, the parent, or, if necessary, the courts will select or use the option of aversive contingencies. Why? Because under such circumstances it is the only ethical action. (pp. 152-153)

Punishment Policy and Procedural Safeguards

Armed with knowledge from the experimental litera- ture, real-world variables and contingencies with which to contend (i.e., chronic, life-threatening problems), and practices and procedures rooted in ethical codes of con- duct, practitioners can consider punishment approaches, when necessary, to provide meaningful programs for the persons in their care. One mechanism for ensuring that best practice approaches are used (Peters & Heron, 1993) is to adopt and use dynamic policies and proce- dures that provide clear guidelines and safeguards to practitioners.

369

Punishment by Stimulus Presentation

Figure 9 Suggested components in an agency’s policy and procedures guidelines to help ensure the ethical, safe, and effective use of punishment.

Policy Requirements • Intervention must conform to all local, state, and federal statutes. • Intervention must conform to the policies and codes of ethical conduct of relevant

professional organizations. • Intervention should include procedures for strengthening and teaching alternative

behaviors. • Intervention must include plans for generalization and maintenance of behavior change

and criteria for eventual termination or reduction of punishment. • Informed consent must be obtained from the client or a parent or legal advocate before

intervention begins. Procedural Safeguards • Before intervention begins, all relevant staff must be trained in (a) the technical details of

properly administering the punishment procedure, (b) procedures for ensuring the physical safety of the client and staff and the humane treatment of the client, and (c) what to do in the event of negative side effects such as emotional outbursts, escape and avoidance aggression, and noncompliance.

• Supervision and feedback of staff administering the punishment intervention must be provided and, if necessary, booster training sessions provided.

Evaluation Requirements • Each occurrence of the problem behavior must be observed and recorded. • Each delivery of the punisher must be recorded and client’s reactions noted. • Periodic review of the data (e.g., daily, weekly) by a team of parent/advocates, staff, and

technical consultants must be conducted to ensure that ineffective treatment is not prolonged or effective treatment terminated.

• Social validity data must be obtained from client, significant others, and staff on (a) treatment acceptability and (b) the real and likely impact of any behavior change on the client’s current circumstances and future prospects.

Agencies can help protect and ensure their clients’ rights to safe, humane, least-restrictive, and effective treatment by developing policy and procedural guidelines that must be followed when any punishment-based in- tervention is implemented (Favell & McGimsey, 1993; Griffith, 1983; Wood & Braaten, 1983). Figure 9 pro- vides an outline and examples of the kinds of compo- nents that might be included in such a document.

Practitioners should also consult their local, state, or national professional association policy statements re- garding the use of punishment. For example, the Associ- ation for the Advancement of Behavior Therapy (AABT) provides guidelines that address treatment selection, in- cluding punishment (Favell et al., 1982). The Associa- tion for Behavior Analysis adheres to the American Psychological Association’s Ethics Code (2004), which, in turn, addresses treatment issues. Collectively, policy statements address implementation requirements, proce- dural guidelines and precautions, and evaluation meth-

Concluding Perspectives We conclude this chapter with brief comments on three complementary perspectives regarding the principle of punishment and the development and application of in- terventions involving punishment by contingent aversive stimulation. We believe that applied behavior analysis would be a stronger, more competent discipline, and its practitioners would be more effective if (a) punishment’s natural role and contributions to survival and learning are recognized and appreciated, (b) more basic and applied research on punishment is conducted, and (c) treatments featuring positive punishment are viewed as default tech- nologies to be used only when all other methods have failed.

ods that an agency should use when implementing any punishment-based intervention.

370

Punishment by Stimulus Presentation

Punishment’s Natural and Necessary Role in Learning Should Be Recognized

Behavior analysts should not shy away from punishment. Positive and negative punishment contingencies occur naturally in everyday life as part of a complex mix of concurrent reinforcement and punishment contingencies, as Baum (1994) illustrated so well in this example:

Life is full of choices between alternatives that offer different mixes of reinforcement and punishment. Going to work entails both getting paid (positive reinforcement) and suffering hassles (positive punish- ment), whereas calling in sick may forfeit some pay (negative punishment), avoid the hassles (negative rein- forcement), allow a vacation (positive reinforcement), and incur some workplace disapproval (positive punish- ment). Which set of relations wins out depends on which relations are strong enough to dominate, and that depends on both the present circumstances and the per- son’s history of reinforcement and punishment. (p. 60)

Punishment is a natural part of life. Vollmer (2002) suggested that the scientific study of punishment should continue because unplanned and planned punishment occur frequently and that planned, sophisticated appli- cations of punishment are within the scope of inquiry for applied behavior analysts. Whether punishment is so- cially mediated, planned or unplanned, or conducted by sophisticated practitioners, we agree with Vollmer that a science of behavior should study punishment.

Scientists interested in the nature of human behavior cannot ignore or otherwise obviate the study of punish- ment. There should be no controversy. Scientists and practitioners are obligated to understand the nature of punishment if for no other reason than because punish- ment happens. (p. 469)

More Research on Punishment Is Needed

Although many of the inappropriate and ineffective ap- plications of punishment are the result of misunder- standing, other misapplications no doubt reflect our incomplete knowledge of the principle as a result of in- adequate basic and applied research (Lerman & Vorn- dran, 2002).

Most of our knowledge about and recommendations for applying punishment are derived from basic research conducted more than 40 years ago. Although sound sci- entific data that describe still-relevant questions have an unlimited shelf life, much more basic research on the

mechanisms and variables that produce effective pun- ishment is needed.

Basic laboratory research allows for the control of variables that is difficult or impossible to attain in applied settings. Once mechanisms are revealed in basic research, however, applications to real-world challenges can be de- vised and adapted with more confidence regarding the degree of potential effectiveness and limitations. Practi- tioners must recognize when, how, why, and under what conditions punishment techniques produce behavioral suppression for the people with whom they work.

We support Horner’s (2002) call for practical appli- cations of punishment research in field settings. It is im- portant to determine how punishment works in situations in which environmental variables may not be as well con- trolled and in which the practitioner delivering the pun- isher might not be a trained professional. Educational and clinical applications of punishment also demand that be- havior analysts understand the individual, contextual, and environmental variables that produce effective applica- tion. Without a clear and refined knowledge base about these variables and conditions, applied behavior analysis as a science of behavior cannot make a legitimate claim to have achieved a comprehensive analysis of its own basic concepts (Lerman & Vorndran, 2002; Vollmer, 2002).

Interventions Featuring Positive Punishment Should Be Treated as Default Technologies

Iwata (1988) recommended that punishment-based in- terventions involving the contingent application of aver- sive stimulation be treated as default technologies. A default technology is one that a practitioner turns to when all other methods have failed. Iwata recommended that behavior analysts not advocate for the use of aversive technologies (because advocacy is not effective, not nec- essary, and not in the best interests of the field), but that they be involved in research and the development of ef- fective aversive technologies.

We must do the work because, whether or not we like it, default technologies will evolve whenever there is failure and, in the case of aversive stimulation, we are in a unique position to make several contributions. First, we can modify the technology so that it is effective and safe. Second, we can improve it by incorporating contingen- cies of positive reinforcement. Third, we can regulate it so that application will proceed in a judicious and ethical manner. Last, and surely most important, by studying the conditions under which default technologies arise, as well as the technologies themselves, we might eventu- ally do away with both. Can you think of a better fate for the field of applied behavior analysis? (p. 156)

371

Punishment by Stimulus Presentation

Summary Definition and Nature of Punishment

1. Punishment has occurred when a stimulus change imme- diately follows a response and decreases the future fre- quency of that type of behavior in similar conditions.

2. Punishment is defined neither by the actions of the per- son delivering the consequences nor by the nature of those consequences. A decrease in the future frequency of the occurrence of the behavior must be observed before a consequence-based intervention qualifies as punishment.

3. Positive punishment has occurred when the frequency of responding has been decreased by the presentation of a stimulus (or an increase in stimulus intensity) immediately following a behavior.

4. Negative punishment has occurred when the frequency of responding has been decreased by the removal of a stim- ulus (or a decrease in the stimulus intensity) immediately following a behavior.

5. Because aversive events are associated with positive pun- ishment and with negative reinforcement, the term aversive control is often used to describe interventions involving either or both of these two principles.

6. A discriminative stimulus for punishment, or SDp, is a stim- ulus condition in the presence of which a response class occurs at a lower frequency than it does in the absence of the SDp as a result of a conditioning history in which re- sponses in the presence of the SDp have been punished and similar responses in the absence of that stimulus have not been punished (or have resulted in a reduced frequency or magnitude of punishment).

7. A punisher is a stimulus change that immediately follows the occurrence of a behavior and reduces the future fre- quency of that type of behavior.

8. An unconditioned punisher is a stimulus whose presenta- tion functions as punishment without having been paired with any other punishers.

9. A conditioned punisher is a stimulus that has acquired its punishing capabilities by being paired with unconditioned or conditioned punishers.

10. A generalized conditioned punisher will function as pun- ishment under a wide range of motivating operations be- cause of its previous pairing with numerous unconditioned and conditioned punishers.

11. In general, the results of basic and applied research show that punishment is more effective when

• the onset of the punisher occurs as soon as possible after the occurrence of a target response,

• the intensity of the punisher is high,

• each occurrence of the behavior is followed by the pun- ishing consequence,

• reinforcement for the target behavior is reduced, and

• reinforcement is available for alternative behaviors.

12. Punishment sometimes causes undesirable side effects and problems, such as the following:

• Emotional and aggressive reactions to aversive stimulation

• Escape and avoidance behaviors

• Behavioral contrast: reduced responding from punish- ment in one situation may be accompanied by in- creased responding in situations in which responses go unpunished

• The modeling of undesirable behavior

• Overuse of punishment caused by the negative rein- forcement of the punishing agent’s behavior (i.e., the immediate cessation of problem behavior)

Examples of Positive Punishment Interventions

13. Reprimands: Used sparingly, a firm reprimand such as “No!” can suppress future responding.

14. Response blocking: When the learner begins to emit the problem behavior, the therapist physically intervenes to prevent or “block” the completion of the response.

15. Overcorrection is a punishment-based tactic in which, con- tingent on each occurrence of the problem behavior, the learner is required to engage in effortful behavior that is di- rectly or logically related to the problem.

16. In restitutional overcorrection, the learner must repair the damage caused by the problem behavior and then bring the environment to a condition vastly better than it was prior to the misbehavior.

17. In positive practice overcorrection, the learner repeatedly performs a correct form of the behavior, or a behavior in- compatible with the problem behavior, for a specified time or number of responses.

18. Contingent electric stimulation can be a safe and effective method for suppressing chronic and life-threatening self- injurious behavior.

Guidelines for Using Punishment Effectively

19. To apply punishment with optimal effectiveness while minimizing undesirable side effects, a practitioners should:

• Select effective and appropriate punishers: (a) conduct punisher assessments to identify the least intrusive pun- isher that can be applied consistently and safely; (b) use punishers of sufficient quality and magnitude; (c) use a variety of punishers to combat habituation and increase the effectiveness of less intrusive punishers.

• If problem behavior consists of a response chain, deliver the punisher as early in the response sequence as possible.

• Punish each occurrence of the behavior.

372

Punishment by Stimulus Presentation

• Gradually shift to an intermittent schedule of punish- ment if possible.

• Use mediation with a response-to-punishment delay. • Supplement punishment with complementary interven-

tions, in particular, differential reinforcement, extinc- tion, and antecedent interventions.

• Watch and be prepared for unwanted side effects. • Record, graph, and evaluate data daily.

Ethical Considerations Regarding the Use of Punishment

20. The first ethical responsibility for any human services pro- fessional or agency is to do no harm. Any intervention must be physically safe for all involved and contain no el- ements that are degrading or disrespectful to the client.

21. The doctrine of the least restrictive alternative holds that less intrusive procedures (e.g., positive reductive ap- proaches) must be tried first and found to be ineffective before more intrusive procedures are implemented (e.g., a punishment-based intervention).

22. A client’s right to effective treatment raises an important ethical issue. Some maintain that the failure to use a pun- ishment procedure that research has shown to suppress self-destructive behavior similar to the client’s is unethi- cal because it withholds a potentially effective treatment and may maintain a dangerous or uncomfortable state for the person.

23. Agencies and individuals providing applied behavior analysis services can help ensure that applications of punishment-based interventions are safe, humane, ethical, and effective by creating and following a set of policy stan- dards, procedural safeguards, and evaluation requirements.

Concluding Perspectives on Punishment

24. Applied behavior analysts should recognize and appreci- ate the natural role of punishment and the importance of punishment to learning.

25. Many misapplications of punishment reflect the field’s in- complete knowledge of the principle. More basic and ap- plied research on punishment is needed.

26. Iwata (1988) recommended that punishment-based in- terventions involving the contingent application of aver- sive stimulation be treated as default technologies; that is, as interventions to be used only when other methods have failed. He argued that applied behavior analysts should not advocate for the use of aversive technologies, but instead must be involved in conducting research on such interventions to (a) make them effective and safe, (b) improve them by incorporating contingencies of pos- itive reinforcement, (c) regulate their judicious and eth- ical application, and (d) study the conditions under which default technologies are used so as to eventually make them unnecessary.

373

Punishment by Removal of a Stimulus

Key Terms

bonus response cost contingent observation exclusion time-out hallway time-out

nonexclusion time-out partition time-out planned ignoring response cost

time-out from positive reinforcement

time-out ribbon

Behavior Analyst Certificiation Board® BCBA® & BCABA® Behavior Analyst Task List©, Third Edition

Content Area 3: Principles, Processes and Concepts

3-5 Define and provide examples of (positive and) negative punishment.

3-6 Define and provide examples of conditioned and unconditioned punishment.

Content Area 9: Behavior Change Procedures

9-3 Use (positive and) negative punishment:

(a) Identify and use punishers.

(b) Use appropriate parameters and schedules of punishment.

© 2006 The Behavior Analyst Certification Board, Inc.,®(BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and ques- tions about this document must be submitted directly to the BACB.

374

Punishment by the contingent removal of a stim- ulus is referred to as negative punishment. In negative punishment, an environmental change

occurs such that a stimulus is removed subsequent to the performance of a behavior, and the corresponding future frequency of the preceding behavior is reduced. By con- trast, in positive punishment, a stimulus is presented, and the corresponding future frequency of that behavior is re- duced. Table 1 shows the distinction between positive and negative punishment with respect to an environmental stimulus change.

Negative punishment occurs in two principal ways: time-out from positive reinforcement and response cost (see the shaded area of Table 1). This chapter defines and operationalizes time-out from positive reinforcement and response cost, provides examples of how these punish- ment procedures are used in applied settings, and offers guidelines to help practitioners implement time-out and response cost effectively.

Definition of Time-Out Time-out from positive reinforcement, or simply time- out, is defined as the withdrawal of the opportunity to earn positive reinforcement or the loss of access to pos- itive reinforcers for a specified time, contingent on the occurrence of a behavior. The effect of either of these procedures is the same: The future frequency of the tar- get behavior is reduced. Implicit in the definition of time- out are three important aspects: (a) the discrepancy between the “time-in” and the time-out environment, (b) the response-contingent loss of access to reinforce- ment, and (c) a resultant decrease in the future frequency of the behavior. Contrary to popular thinking, time-out is not exclusively defined, nor does it require only re- moving an individual to an isolated or secluded setting. Although such a removal procedure may accurately de- scribe isolation time-out (Costenbader & Reading- Brown, 1995), it is by no means the only way by which time-out can be used. Technically, time-out as a nega-

tive punishment procedure removes a reinforcing stim- ulus for a specified time contingent on the occurrence of a behavior, the effect of which is to decrease the fu- ture frequency of the behavior.

Time-out can be viewed from procedural, concep- tual, and functional perspectives. Procedurally, it is that period of time when the person is either removed from a reinforcing environment (e.g., a student is removed from the classroom for 5 minutes) or loses access to reinforcers within an environment (e.g., a student is ineligible to earn reinforcers for 5 minutes). Conceptually, the distinction between the time-in and the time-out environment is of paramount importance. The more reinforcing the time- in setting is, the more effective time-out is likely to be as a punisher. Stated otherwise, the greater the difference between the reinforcing value of time-in and absence of that reinforcing value in the time-out setting, the more effective time-out will be. From a functional perspective, time-out involves the reduced frequency of the future oc- currence of the behavior. Without a reduction in fre- quency of future behavior, time-out is not in effect, even if the person is procedurally removed from the setting or loses access to reinforcers. For instance, if a teacher re- moves a student from the classroom (presumably the time-in setting), but upon the student’s return, the prob- lem behavior continues, effective time-out cannot be claimed.

Time-Out Procedures for Applied Settings There are two basic types of time-out: nonexclusion and exclusion. Within each type, several variations allow the practitioner flexibility in deciding a course of action for reducing behavior. As a rule, nonexclusion time-out is the recommended method of first choice because practi- tioners are ethically bound to employ the most powerful, but least restrictive, alternative when deciding on eligible variations.

Table 1 Distinction Between Positive and Negative Punishment

Future frequency Stimulus change

Stimulus presented Stimulus removed

Behavior is reduced Positive punishment Negative punishment (e.g., “No!,” overcorrection) (e.g., time-out, response

cost)

Punishment by Removal of a Stimulus

375

Punishment by Removal of a Stimulus

Nonexclusion Time-Out

Nonexclusion time-out means that the participant is not completely removed physically from the time-in setting. Although the person’s position relative to that setting shifts, he or she remains within the environment. Nonex- clusion time-out occurs in any one of four ways: planned ignoring, withdrawal of a specific reinforcer, contingent observation, and the time-out ribbon. Each variation has a common element: Access to reinforcement is lost, but the person remains within the time-in setting. For in- stance, if a student misbehaves during outdoor recess, he would be required to stand next to the adult playground monitor for a period of time.

Planned Ignoring

Planned ignoring occurs when social reinforcers—usu- ally attention, physical contact, or verbal interaction— are removed for a brief period, contingent on the occurrence of an inappropriate behavior. Planned ignor- ing assumes that the time-in setting is reinforcing and that all extraneous sources of positive reinforcement can be eliminated.

Operationally, planned ignoring can involve system- atically looking away from the individual, remaining quiet, or refraining from any interaction whatsoever for a specific time (Kee, Hill, & Weist, 1999). Planned ignoring has the advantage of being a nonintrusive time-out procedure that can be applied quickly and conveniently.

For example, let us suppose that during a group ther- apy session for drug rehabilitation a client begins to express her fascination with stealing money to buy narcotics. If at that point the other members of the group break eye contact and do not respond to her verbaliza- tions in any way until her comments are more consistent with the group’s discussion, then planned ignoring is in effect. In this case, because the members of the group participated in the reductive procedure, it would be termed peer-mediated time-out (Kerr & Nelson, 2002).

Withdrawal of a Specific Positive Reinforcer

Bishop and Stumphauzer (1973) demonstrated that the withdrawal of a specific positive reinforcer contingent on an inappropriate behavior decreased the level of that behavior. In their study, the contingent termination of television cartoons successfully reduced the frequency of thumb sucking in three young children. Unknown to the children, a remote on/off switch was attached to the television. Baseline data indicated that each child emit- ted a high rate of thumb sucking while viewing the car- toons. During this variation of time-out, the television was immediately turned off when thumb sucking oc- curred and was turned back on when thumb sucking

stopped. The procedure was effective in reducing thumb sucking not only in the treatment location (an office), but also during a story period at school.

West and Smith (2002) reported on a novel group application of this time-out variation. They mounted a facsimile traffic signal light to a cafeteria wall and rigged the light with a sensor to detect noise at various thresh- old levels. For instance, if an acceptable level of conver- sation was evident, a green light registered on the traffic signal and prerecorded music played (i.e., the time-in condition was in effect). The music remained available as long as the students engaged in an appropriate level of conversation. As the conversational level increased, the traffic light changed from green to yellow, visibly warn- ing the students that any further increase in noise would lead to red light onset and music loss (i.e., time-out). The music was switched off automatically for 10 seconds after the sensor registered the noise above a certain decibel threshold shown by the red light. Under this group pro- cedure, the inappropriate behavior was reduced.

A group time-out contingency has several advan- tages. First, it can be used in an existing environment; special provisions for removing students are not needed. Second, using an electronic device automatically signals when time-in and time-out are in effect. In the West and Smith study, students soon discriminated that continued inappropriate loud conversation resulted in the loss of the reinforcer (e.g., music).

Contingent Observation

In contingent observation, the person is repositioned within an existing setting such that observation of ongoing activities remains, but access to reinforcement is lost. A teacher uses contingent observation when upon the oc- currence of an undesirable behavior the teacher redirects the offending student to sit away from the group, and re- inforcement for a specific time is withheld (Twyman, John- son, Buie, & Nelson, 1994). In short, the student is told to “sit and watch” (White & Bailey, 1990). When the con- tingent observation period ends, the student rejoins the group and is able to earn reinforcement for appropriate be- havior. Figure 1 shows the effects of White and Bailey’s (1990) use of contingent observation in reducing the num- ber of disruptive behaviors in two inclusive physical edu- cation classes. When contingent observation was in effect, the number of disruptive behaviors across two classrooms decreased and remained near zero levels.

Time-Out Ribbon

The time-out ribbon is defined as a colored band that is placed on a child’s wrist and becomes discriminative for receiving reinforcement (Alberto, Heflin, & Andrews,

376

Punishment by Removal of a Stimulus

Figure 1 Number of disruptive behaviors per 10- minute observation period. The numbers above the data points represent the number of times “sit and watch” was implemented during the class. From “Reducing Disruptive Behaviors of Elementary Physical Education Students with Sit and Watch” by A. G. White and J. S. Bailey, 1990, Journal of Applied Behavior Analysis, 23, p. 357. Copyright 1990 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

2002). When the ribbon is on the child’s wrist, she is el- igible to earn reinforcers. If the child misbehaves, the ribbon is removed, and all forms of social interaction with the child are terminated for a specific period (e.g., 2 minutes).

Foxx and Shapiro (1978) used the time-out ribbon to reduce the disruptive behavior of four elementary stu- dents. Contingent on an inappropriate behavior, the rib-

bon was removed from the student, and all forms of so- cial interaction with the offender were terminated for 3 minutes. The student was permitted to remain in the room, however. If inappropriate behavior still occurred after 3 minutes, the time-out was extended until the mis- behavior was finished. Figure 2 shows that when the time-out ribbon plus reinforcement was in effect, the mean percentage of disruptive behavior was markedly reduced for the four students.

Laraway, Snycerski, Michael, and Poling (2003) dis- cussed the removal of the time-out ribbon in the context of motivating operations (MOs). In their view, removing the time-out ribbon functioned as a punisher because of its relationship to the existing establishing operation (EO). In a retrospective analysis of Foxx and Shapiro’s (1978) study, Laraway and colleagues stated, “Thus, the EOs for the programmed reinforcers in time-in also es- tablished the punishing effect of the ribbon loss (i.e., functioned as EOs for ribbon loss as a punishing event) and abated misbehaviors that resulted in ribbon loss” (p. 410). With respect to the role of MOs as they relate to the time-out ribbon, Laraway and colleagues went on to state:

In commonsense terms, losing the opportunity to earn a consequence is only important if you currently “want” that consequence. Therefore, MOs that increase the rein- forcing effectiveness of particular objects or events also increase the punishing effectiveness of making those ob- jects or events unavailable (i.e., time-out). . . . a single environmental event can have multiple and simultaneous motivating effects. (p. 410)

Exclusion Time-Out

The distinguishing feature of exclusion time-out is that the person is removed from the environment for a spec- ified period, contingent on the occurrence of the targeted

Classes

P er

ce n

t o

f D

is ru

p ti

ve B

eh av

io r

Figure 2 The mean percentage of time spent in disruptive classroom behavior by four subjects. The horizontal broken lines indicate the mean for each condition. The arrow marks a 1 day probe (Day 39) during which the time-out contingency was suspended. A follow-up obser- vation occurred on Day 63. From “The Timeout Ribbon: A Nonexclusionary Timeout Procedure” by R. M. Foxx and S. T. Shapiro, 1978, Journal of Applied Behavior Analysis, 11, p. 131. Copyright 1978 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

200 180 160 140 120 100 80 60 40 20

Alternative Education Class

3 4 24 6 3 531 5 3 61

Base- line

(343)

Behavior Checklist Sit and Watch

200 180 160 140 120 100 80 60 40 20

Fourth- Grade Class

Baseline

N u

m b

er o

f D

is ru

p ti

ve B

eh av

io rs

p er

1 0 M

in u

te S

es si

o n

71 62 4 5 7 7 6

5 10 15 20 25 30 35 40 45 50 Days

377

inappropriate behavior. Exclusion time-out can be con- ducted in classroom settings in three ways: (a) The student can be removed to a time-out room, (b) the stu- dent can be separated from the rest of the group by a par- tition, or (c) the student can be placed in the hallway.

Time-Out Room

A time-out room is any confined space outside the par- ticipant’s normal educational or treatment environment that is devoid of positive reinforcers and in which the per- son can be safely placed for a temporary period. The time-out room should preferably be located near the time- in setting and should have minimal furnishing (e.g., a chair and a table). It should have adequate light, heat, and ventilation but should not have other potentially rein- forcing features available (e.g., pictures on the wall, tele- phone, breakable objects). The room should be secure, but not locked.

A time-out room has several advantages that make it attractive to practitioners. First, the opportunity to ac- quire reinforcement during time-out is eliminated or re- duced substantially because the time-out environment is physically constructed to minimize such an occurrence. Second, after a few exposures to the time-out room, stu- dents learn to discriminate this room from other rooms in the building. The room assumes conditioned aversive properties, thus increasing the probability that the time- in setting will be viewed as more desirable. Finally, the risk of a student hurting other students in the time-in set- ting is reduced when the offending student is removed to this space.

However, the practitioner must also weigh several disadvantages of using a time-out room. Foremost is the necessity of escorting the student to the time-out room. From the time the individual is informed that time-out is in effect until the time she is actually placed in the setting, resistance can be encountered. Practitioners should an- ticipate and be fully prepared to deal with emotional out- bursts when using this exclusion time-out variation. In addition, unlike the nonexclusion options mentioned pre- viously, removing an individual from the time-in envi- ronment prohibits that individual from access to ongoing academic or social instruction. Missed instructional time should be minimized, and in instances in which time-out has been used excessively and in which it has been shown to serve as a negative reinforcer for the teacher, it should be reconsidered altogether (Skiba & Raison, 1990). Fur- ther, in the time-out room the person may engage in be- haviors that should be stopped but that go undetected (e.g., self-destructive or self-stimulatory behaviors). Fi- nally, practitioners must be sensitive to the public’s per- ception of the time-out room. Even the most benign time-out room can be viewed with dread by persons who

might be misinformed about its purpose or place in an overall behavior management program.

Partition Time-Out

In partition time-out, the person remains within the time-in setting, but his view within the setting is restricted by a partition, wall, cubicle, or similar structure. A teacher would use partition time-out by directing a student, con- tingent on a misbehavior, to move from his assigned seat to a location behind an in-class cubicle for a specified time period. Although partition time-out has the advan- tage of keeping the student within the time-in setting— presumably to hear academic content and the teacher praising other students for appropriate behavior—it also can be disadvantageous. That is, the person may still be able to obtain covert reinforcement from other students. If he does receive reinforcement from students, it is un- likely that the disruptive behavior will decrease. Also, public perceptions must be taken into account with this form of exclusion. Even though the student remains in the room, and the partition time-out area might be called by an innocuous name (e.g., quiet space, office, personal space), some parents may view any type of separation from other members of the class as discriminatory.

Hallway Time-Out

Hallway time-out is a particularly popular method with teachers—and perhaps parents—when dealing with dis- ruptive behavior. In this method, the student is directed to leave the classroom and sit in the hallway. Although it shares the advantages of the variations just mentioned, this approach is not highly recommended for two rea- sons: (a) The student can obtain reinforcement from a multitude of sources (e.g., students in other rooms, indi- viduals walking in the hallway), and (b) there is increased likelihood of escape if the student is combative on the way to time-out. Even with the door of the classroom open, teachers are often too busy with activities in the time-in setting to monitor the student closely in the hall- way. This approach might be more beneficial for younger children who follow directions, but it is clearly inappro- priate for any student who lacks basic compliance skills.

Desirable Aspects of Time-Out

Ease of Application

Time-out, especially the nonexclusion variations, is rel- atively easy to apply. Even physically removing a stu- dent from the environment can be accomplished with comparative ease if the teacher acts in a businesslike fash- ion and does not attempt to embarrass the student. Issuing

Punishment by Removal of a Stimulus

378

a direction privately (e.g., “Deion, you have disrupted Monique twice; time-out is now in effect”) can help the teacher handle a student who has misbehaved but who does not want to leave the room. If the behavior warrants time-out, the teacher must insist that the student leave; however, that insistence should be communicated at close range so that the student is not placed in a position of challenging the teacher openly to save face with a peer group. Teachers should consult district policy on whether an administrator should be called to remove the student.

Acceptability

Time-out, especially nonexclusion variations, meets an acceptability standard because practitioners regard it as appropriate, fair, and effective. Even so, prior to imple- mentation, practitioners should always check with the appropriate administering body to ensure compliance with agency policy before applying time-out for major or minor infractions.

Rapid Suppression of Behavior

When effectively implemented, time-out usually sup- presses the target behavior in a moderate-to-rapid fash- ion. Sometimes only a few applications are needed to

achieve an acceptable reduction levels. Other reductive procedures (e.g., extinction, differential reinforcement of low rates) also produce decreases in behavior, but they can be time-consuming. Many times the practitioner does not have the luxury of waiting several days or a week for a behavior to decrease. In such instances, time-out mer- its strong consideration.

Combined Applications

Time-out can be combined with other procedures, ex- tending its usability in applied settings. When it is com- bined with differential reinforcement, desirable behavior can be increased and undesirable behavior can be de- creased (Byrd, Richards, Hove, & Friman, 2002).

Using Time-Out Effectively Effective implementation of time-out requires that the practitioner make several decisions prior to, during, and after time-out implementation. Figure 3 shows Powell and Powell’s (1982) time-out implementation checklist, which can aid in the decision-making process. The fol- lowing sections expand on the major points underlying many of these decisions.

Figure 3 Implementation checklist for time-out.

Date Teacher’s Step Task Completed Initials 1. Try less aversive techniques and document results. _____ _____ 2. Operationally define disruptive behaviors. _____ _____ 3. Record baseline on target behaviors. _____ _____ 4. Consider present levels of reinforcement (strengthen

if necessary). _____ _____ 5. Decide on time-out procedure to be used. _____ _____ 6. Decide on time-out area. _____ _____ 7. Decide on length of time-out. _____ _____ 8. Decide on command to be used to place child in time-out. _____ _____ 9. Set specific criteria that will signal discontinuance of time-out. _____ _____

10. Set dates for formal review of the time-out procedure. _____ _____ 11. Specify back-up procedures for typical time-out problems. _____ _____ 12. Write up the entire procedure. _____ _____ 13. Have the procedure reviewed by peers and supervisors. _____ _____ 14. Secure parental/guardian approval and include the written

program in the child’s IEP. _____ _____ 15. Explain procedure to the student and class (if appropriate). _____ _____ 16. Implement the procedure, take data, and review progress daily. _____ _____ 17. Formally review procedure as indicated. _____ _____ 18. Modify the procedure as needed. _____ _____ 19. Record results for future teachers/programs. _____ _____

From “The Use and Abuse of Using the Timeout Procedure for Disruptive Pupils” by T. H. Powell and I. Q. Pow- ell, 1982, The Pointer, 26 (2), p. 21. Reprinted by permission of the Helen Dwight Reid Educational Foundation. Published by Heldref Publications, 1319 Eighteenth St., N.W., Washington DC 20036-1802. Copyright © 1982.

Punishment by Removal of a Stimulus

379

Reinforcing and Enriching the Time-In Environment

The time-in environment must be reinforcing if time-out is to be effective. In making the time-in environment re- inforcing, the practitioner should seek ways to reinforce behaviors that are alternative or incompatible with be- haviors that lead to time-out (e.g., using differential reinforcement of alternative behavior, differential rein- forcement of incompatible behavior). Differential rein- forcement will facilitate the development of appropriate behaviors. Additionally, upon return from time-out, re- inforcement for appropriate behavior must be delivered as quickly as possible.

Defining Behaviors Leading to Time-Out

Before time-out is implemented, all of the appropriate parties must be informed of the behaviors that will lead to time-out. If a teacher decides to use time-out, she should describe in explicit, observable terms those be- haviors that will result in time-out. For instance, merely informing students that disruptive behavior will result in time-out is not sufficient. Providing specific examples and nonexamples of what is meant by disruptive behav- ior is a better course to follow. Readdick and Chapman (2000) discovered in post-time-out interviews that chil- dren are not always aware of the reasons time-out was implemented. Providing explicit examples and nonex- amples addresses this problem.

Defining Procedures for the Duration of Time-Out

In most applied settings such as schools and residential and day treatment centers, the initial duration of time- out should be short. A period of 2 to 10 minutes is suf- ficient, although a short duration may be ineffective initially if an individual has had a history of longer time- out periods. As a rule, time-out periods exceeding 15 minutes are not likely to be effective. Further, longer time-out periods are counterproductive for several rea- sons. First, the person may develop a tolerance for the longer duration and find ways to obtain reinforcement during time-out. This situation is likely to occur with people with a history of self-stimulatory behavior. The longer the duration of time-out, the more opportunity there is to engage in reinforcing activities (e.g., self- stimulation) and the less effective time-out becomes. Also, longer time-out periods remove the person from the educational, therapeutic, or family time-in environ- ment in which the opportunity to learn and earn re- inforcers is available. Third, given the undesirable

practical, legal, and ethical aspects of longer time-out periods, a prudent initial course of action is for the prac- titioner to use relatively short, but consistent, time-out periods. When time-out periods are short, student aca- demic achievement is not affected adversely (Skiba & Raison, 1990).

Defining Exit Criteria

If a person is misbehaving when time-out ends, it should be continued until the inappropriate behavior ceases (Brantner & Doherty, 1983). Thus, the decision to ter- minate time-out should not be based exclusively on the passage of time; an improved behavioral condition should also be used as the ultimate criterion for ending time-out. Under no conditions should time-out be ter- minated if any inappropriate behavior is occurring.

If the practitioner anticipates that the inappropriate behavior that led to time-out may occur at the point of scheduled termination, two strategies can be tried. First, the practitioner can inform the person that the scheduled time-out period (e.g., 5 minutes) will not begin until the inappropriate behavior ceases. The second alternative is to simply extend the time until the disruptive behavior stops.

Deciding on Nonexclusion or Exclusion Time-Out

School board policy or institutional constraints may be es- tablished that set the parameters for time-out variations that can be used in applied settings. In addition, physical factors within the building (e.g., a lack of available space) may prohibit exclusionary forms of time-out. In the main, nonexclusion time-out is the preferred method.

Explaining the Time-Out Rules

In addition to posting the behaviors that will lead to time- out, the teacher should also state the rules. Minimally, these rules should focus on the initial duration of time-out and the exit criteria (i.e., rules that determine when time- out is over and what happens if ongoing inappropriate behavior occurs when time-out is over).

Obtaining Permission

One of the most important tasks a practitioner must per- form prior to using time-out, especially the exclusionary variations, is to obtain permission. Because of the po- tential to misuse time-out (e.g., leaving the person in time-out too long, continuing to use time-out when it is not effective), practitioners must obtain administrative approval before employing it. However, interactions

Punishment by Removal of a Stimulus

380

happen so fast in most applied settings that obtaining per- mission on an instance-by-instance basis would be un- duly cumbersome. A preferred method is for the practitioner, in cooperation with administrators, to de- cide beforehand the type of time-out that will be imple- mented for certain offenses (e.g., nonexclusion versus exclusion), the duration of time-out (e.g., 5 minutes), and the behaviors that will demonstrate reinstatement to the time-in environment. Communicating these procedures and/or policies to parents is advisable.

Applying Time-Out Consistently

Each occurrence of the undesirable behavior should lead to time-out. If a teacher, parent, or therapist is not in a position to deliver the time-out consequence after each occurrence of the target behavior, it may be better to use an alternative reductive technique. Using time-out occa- sionally may lead to student or client confusion about which behaviors are acceptable and which are not.

Evaluating Effectiveness

Universally, educators and researchers call for the regu- lar evaluation of the use of time-out in applied settings (Reitman & Drabman, 1999). At the very least, data need to be obtained on the inappropriate behavior that initially led to time-out. If time-out was effective, the level of that behavior should be reduced substantially, and that re- duction should be noticed by other persons in the envi- ronment. For legal and ethical reasons, additional records should be kept documenting the use of time-out; the du- ration of each time-out episode; and the individual’s be- havior before, during, and after time-out (Yell, 1994).

In addition to collecting data on the target behav- ior, it is sometimes beneficial to collect data on collat- eral behaviors, unexpected side effects, and the target behavior in other settings. For instance, time-out may produce emotional behavior (e.g., crying, aggressive- ness, withdrawal) that might overshadow positive gains and spill over into other settings. Keeping a record of these episodes is helpful. Also, Reitman and Drabman (1999) suggested that for home-based applications of time-out, parents keep a record of the date, time, and duration of time-out as well as a brief anecdotal de- scription of the effects of time-out. When reviewed with skilled professionals, these anecdotal logs provide a session-by-session profile of performance, making ad- justments to the time-out procedure more informed. In their view,

The timeout record and discussions centered around it yielded a wealth of information including the type, fre- quency, context, and rate of misbehavior. . . . Graphing

1We additionally recommend consideration of the Gaylord-Ross decision- making model as an option for when to use punishment procedures, in- cluding time-out (see Gaylord-Ross, 1980).

these data provided visual feedback to both the thera- pists and the clients regarding the extent to which be- havior change had occurred and facilitated the setting of new goals (p. 143).

Considering Other Options

Although time-out has been shown to be effective in re- ducing behavior, it should not be the method of first choice. Practitioners faced with the task of reducing be- havior should initially consider extinction or positive re- ductive procedures (e.g., differential reinforcemont of other behavior, differential reinforcemont of incompati- ble ).1 Only when these less intrusive procedures have failed should time-out be considered.

Legal and Ethical Time-Out Issues

The use of time-out in applied settings has been much discussed. Although litigation surrounding the use of this procedure has focused primarily on institutionalized pop- ulations (e.g., Morales v. Turman, 1973; Wyatt v. Stickney, 1974), the rulings in these cases have had a profound ef- fect on the use of time-out in other settings. The issues be- fore the courts focus on the protection of client rights, whether time-out represents cruel and unusual punish- ment, and the degree of public acceptability of the time- out procedure.

The upshot of major court rulings has been that a per- son’s right to treatment includes the right to be free from unnecessary and restrictive isolation. However, rulings also include language that permits the use of time-out in a behavior change program as long as the program is closely monitored and supervised and is designed to pro- tect the individual or others from bodily harm.

At least two conclusions have been drawn from the court rulings. First, removing a person to a locked room is considered illegal unless it can be demonstrated that the seclusion is part of an overall treatment plan and un- less the program is carefully and closely monitored. Sec- ond, the duration of time-out is intended to be brief (i.e., less than 10 minutes). Extensions can be obtained but only from a duly constituted review committee.

Definition of Response Cost Teachers who need to reduce inappropriate behavior may find response cost a functional alternative to time-out from positive reinforcement because it is quick, avoids

Punishment by Removal of a Stimulus

381

confrontations with students, and offers a treatment op- tion that may be shorter than other reductive procedures (e.g., extinction).

Response cost is a form of punishment in which the loss of a specific amount of reinforcement occurs, con- tingent on an inappropriate behavior, and results in the de- creased probability of the future occurrence of the behavior. As negative punishment, response cost can be classified and is defined by function. Specifically, if the future frequency of the punished behavior is reduced by the response-contingent withdrawal of a positive rein- forcer, then response cost has occurred. However, if the removal of the reinforcer increases the level of the be- havior or has no effect on it, response cost has not oc- curred. This distinction between application and effect is the key aspect in the definition of response cost.

Response cost occurs any time a teacher reduces the number of minutes of recess, reclaims previously earned or awarded stickers, or otherwise “fines” the student for an infraction, with the results being a decrease in the fu- ture frequency of the targeted behavior. Each occurrence of the inappropriate behavior results in the loss of a spe- cific amount of positive reinforcement already held by the individual. Response cost usually involves the loss of generalized conditioned reinforcers (e.g., money, tokens), tangibles (e.g., stickers), or activities (e.g., minutes of lis- tening to music) (Keeney, Fisher, Adelinis, & Wilder, 2000; Musser, Bray, Kehle, & Jenson, 2001).

Desirable Aspects of Response Cost

Response cost has several features that make it desirable to use in applied settings: its moderate-to-rapid effects on decreasing behavior, its convenience, and its ability to be combined with other procedures.

Moderate-to-Rapid Decrease in Behavior

Like other forms of punishment, response cost usually produces a moderate-to-rapid decrease in behavior. The practitioner does not have to wait long to determine the suppressive effects of response cost. If the procedure is going to be effective in reducing behavior, a two- to three-session trial period is usually sufficient to note the effect.

Convenience

Response cost is convenient to implement and can be used in a variety of school- and home-based settings (Ashbaugh & Peck, 1998; Musser et al., 2001). For ex- ample, if students are expected to follow classroom pro-

cedures, rules, or contracts, then any rule infraction will produce a fine. Each occurrence of a misbehavior means that an explicit fine will be levied and that a positive re- inforcer will be lost. The fine further signals that future occurrences of the misbehavior will result in the same consequence.

Hall and colleagues (1970) found that the response- contingent removal of slips of paper bearing the stu- dent’s name had a reductive effect on the number of complaints an emotionally disturbed boy emitted. Prior to a reading and math session, the teacher placed five slips of paper bearing the boy’s name on his desk. Each time the boy cried, whined, or complained during read- ing or math, a slip was removed. Figure 4 shows that when response cost was in effect, disruptive behaviors decreased markedly; when response cost was not in ef- fect, these behaviors increased.

Keeney and colleagues (2000) compared response cost with baseline conditions and noncontingent rein- forcement after conducting a functional analysis of the destructive behavior of a 33-year-old woman with severe developmental disabilities. The purpose of the functional analysis was to determine the conditions that maintained the behavior. Figure 5 (top panel) shows that a func- tional analysis revealed that the demand (escape) and attention condition produced the highest levels of de- structive behavior. The bottom panel shows a baseline condition during which compliance to requests was praised, and episodes of destructive behavior resulted in a 30-second break from the task at hand (escape). Dur- ing noncontingent reinforcement, caretaker proximity, positive statements, or music was available to the woman. During response cost, the music or attention was avail- able at the beginning of a session. However, any episode of destructive behavior produced an immediate 30-sec- ond loss of music or attention.

Overall, the data show an ascending baseline, with mean percentages averaging approximately 30%. Non- contingent music did not result in a reduction in destruc- tive behavior. However, response cost (withdrawal of music for 30 seconds) produced an immediate and replic- able change in destructive behavior. An important key to this study was that the functional analysis provided the evidence of the controlling variable (demand), and a prior reinforcer assessment had identified music as a preferred reinforcer. In effect, the reinforcer preference assessment served as a mechanism to develop the response cost pro- gram as a treatment for the escape-maintained destruc- tive behavior.

At home, response cost is convenient to the extent that it can be incorporated into an existing allowance pro- gram. For instance, if a child earns an allowance of $7 per week ($1 per day) for cleaning his room and placing his dishes in the sink after eating, noncompliance with one

Punishment by Removal of a Stimulus

382

Baseline2

Days

N u

m b

er o

f C

ri es

, W

h in

es , C

o m

p la

in ts

Figure 4 The number of cries, whines, and complaints by Billy during the 30-minute reading and arithmetic periods. From “Modification of Disrupting and Talking-Out Behavior with the Teacher as Observer and Experimenter” by R. V. Hall, R. Fox, D. Williard, L. Goldsmith, M. Emerson, M. Owen, E. Porcia, and R. Davis, 1970, paper presented at the American Educational Research Association Convention, Minneapolis. Reprinted with permission.

1.5

0.5

0 1 3 5 7 9 11 13 15 17 19 21 23 25 27

Demand

Attention

Play Alone

D es

tr u

ct iv

e R

es p

o n

se s

p er

M in

u te

100

2 4 6 8 10 12 14 16 18 20 22 24 26 28

Baseline (BL)

P er

ce n

ta ge

o f

T ri

al s

w it

h D

es tr

u ct

iv e

B eh

av io

30 32 34 36

Noncontingent Reinforcement

Response Cost BL Response Cost

Attention

Music Withdrawal of Music

Withdrawal of Attention

Sessions

Figure 5 Rates of destructive behavior during functional analysis (top panel) and treatment analysis (bottom panel) of noncontingent reinforcement (NCR) and response cost. From “The Effects of Response Cost in the Treatment of Aberrant Behavior Maintained by Negative Reinforce- ment” by K. M. Keeney, W. W. Fisher, J. D. Adelinis, and D. A. Wilder, 2000, Journal of Applied Behavior Analysis, 33, p. 257. Copyright 2000 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

or both of these behaviors would lead to a monetary loss (e.g., 50¢ per infraction). In both home and school set- tings a response cost chart would provide the child feed- back on his status, while making the visual notation of progress convenient for teachers and parents (Bender & Mathes, 1995).

Figure 6 shows a sample response cost chart as it might be used in a classroom or tutorial setting. To im- plement the response cost procedure, the teacher writes a column of decreasing numbers on the board. Whenever a disruptive behavior occurs, the teacher crosses off the highest remaining number. If the teacher places the

Punishment by Removal of a Stimulus

383

Figure 6 A sample response cost chart showing the total number of minutes of free time remaining after three rule infractions.

34 32 30 28 26 24 22 20 18 16 14 12 10

8 6 4 2 0

5 10 15 20 25 30

Baseline Faded Bedtime/ Response Cost BL

Faded Bedtime/ Response Cost

24-Hour Periods

N u

m b

er o

f 1

5 -M

in u

te I

n te

rv al

s w

it h

D is

tr u

b ed

S le

Figure 7 Number of 15-minute intervals of disturbed sleep per 24-hour period during baseline and faded bed- time with response cost. From “Treatment of Sleep Problems in a Toddler” by R. Ashbaugh and S. Peck, 1998, Journal of Applied Behavior Analysis, 31, p. 129. Copyright 1998 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

numbers 15 through 0 on the board and three disruptive behaviors happened, the students had 12 minutes of free time left that day.

Combining with Other Approaches

Response cost can be combined with other behavior change procedures (Long, Miltenberger, Ellingson, & Ott, 1999). For instance, response cost was combined

with a faded bedtime procedure to treat sleep problems (Ashbaugh & Peck, 1998; Piazza & Fisher, 1991). In the Ashbaugh and Peck (1998) study, a systematic replication of Piazza and Fisher (1991), two principal phases were examined: baseline and faded bedtime plus response cost. During baseline, the typical family ritual of placing the child in bed was followed. That is, she was placed in bed when she appeared tired, and no attempts were made to awaken her if she slept during the day when she was sup- posed to be awake or if she was awake at times when she was supposed to be asleep (e.g., the middle of the night). If the child awoke during the night, she typically went to her parents’ room and slept with them.

During faded bedtime plus response cost, an aver- age “fall-asleep” time was calculated based on the base- line data; this time was adjusted each night as a function on the previous night’s fall-asleep time. For instance, an average sleep time during baseline may have been 10:30 P.M., adjusted to 10 P.M., if she fell asleep within 15 min- utes of the 10:30 P.M. time from the night before. Re- sponse cost consisted of having the child get out of her bed for 30 minutes if she was not asleep within 15 min- utes of being placed in bed. She was kept awake for the 30 minutes thereafter by playing with toys or talking to her parents. Also, if the child awoke during the night and attempted to sleep in her parents’ bed, she was escorted back to her own bed.

Figure 7 shows that during baseline 15-minute in- tervals of disruptive sleep averaged 24 intervals. How- ever, when faded bedtime plus response cost was implemented, the number of 15-minute intervals of dis- ruptive sleep fell to approximately 3 intervals. The child’s parents reported a better night’s sleep of their own, and gains were maintained a year later.

Punishment by Removal of a Stimulus

384

Response cost has been combined with differential reinforcement of alternative behavior (DRA) to treat the food refusal behavior of a 5-year-old child with devel- opmental delay. Kahng, Tarbox, and Wilke (2001) repli- cated the Keeney and colleagues (2000) study, except that response cost was combined with DRA. During a baseline condition, trials across various food groups were presented every 30 seconds. If food was accepted, praise was delivered. If food was refused, the behavior was ignored. Based on reinforcer assessments, during DRA and response cost, books and audiotapes were made available at the meal. If food was accepted, praise was delivered. If food was refused, the books and au- diotapes were removed for 30 seconds. The books and audiotapes were replaced during the next trial contin- gent on accepting food (i.e., DRA). Parents and grand- parents were also trained in the DRA plus response cost procedures. Results showed a substantial increase in the percentage of intervals of food acceptance and a dra- matic decrease in the percentage of intervals of food re- fusal and disruption.

In sum, the findings of studies directly combining response cost with other procedures have been positive. In any case, whether response cost is used alone or in combination with other procedures, practitioners who provide ample reinforcement during time-in occasions are more likely to be successful. Stated differently, re- sponse cost should not be used as the sole approach to modifying behavior because response cost, like any pun- ishment procedure, does not teach new behavior. Com- bining response cost procedures with behavior-building approaches (e.g., DRA) is a good plan.

Response Cost Methods Four methods for implementing response cost are as a direct fine, as bonus response cost, combined with posi- tive reinforcement, and within a group arrangement.

Fines

Response cost can be implemented by directly fining the person a specific amount of positive reinforcers. A stu- dent who loses 5 minutes of free time for each occur- rence of noncompliant behavior is an example. In this case, response cost is applied to stimuli known to be re- inforcing and available to the person. In some situations, however, removing unconditioned and conditioned re- inforcers (e.g., food, free time) from a person would be considered legally and ethically inappropriate or unde- sirable. For instance, it would be inadvisable to take structured leisure or free time from a person with de- velopmental disabilities. To avoid any potential prob-

lems, practitioners must obtain permission from the local human rights review committee or modify the response cost contingency.

Bonus Response Cost

Practitioners can make additional reinforcers available noncontingently to the participant, specifically for re- moval with a response cost contingency. For example, if the school district normally allocates 15 minutes of recess each morning for students at the elementary level, in a bonus response cost method, students would have an additional 15 minutes of recess available, subject to re- moval in prespecified amounts if classroom rules are vi- olated. The students retain the regularly scheduled recess, but each infraction or misbehavior reduces their bonus minutes by a certain number (e.g., 3 minutes). So, if a student emitted three misbehaviors, his recess that day would consist of his regularly scheduled time (i.e., 15 minutes) plus 6 additional bonus minutes. As students improve, gradually reduce the number of bonus minutes available.

Combining with Positive Reinforcement

Response cost can be combined with positive reinforce- ment. For example, a student might earn tokens for im- proved academic performance and simultaneously lose tokens for instances of inappropriate behavior. In another instance, a student could receive a point, a star, or a sticker for each of 10 academic assignments completed during the morning but lose a point, a star, or a sticker for each of 6 occurrences of calling out. In this scenario, the stu- dent would net a total of 4 points for that day (i.e., 10 points earned for academic performance minus 6 points for calling out equals a net of 4 points).

Musser and colleagues (2001) combined response cost and positive reinforcement to improve the compli- ant behavior of three students with emotional distur- bance and oppositional defiant behavior. Briefly, the procedures involved the teacher making a normal re- quest for compliance (e.g., “Take out your book please”). If the student complied within 5 seconds, praise was delivered. If a 30-minute period expired and no episodes of noncompliance were recorded, a sticker was delivered to the student. If the student failed to com- ply with a request within 5 seconds, the teacher waited 5 seconds and issued another directive (e.g., “Take out your book”). The teacher then waited an additional 5 seconds for compliance. If the student complied, praise was issued. If not, one sticker was deducted for

Punishment by Removal of a Stimulus

385

noncompliance. Stickers were later exchanged for a “mystery motivator” (e.g., a backup reinforcer from a grab bag). Results for all three students showed marked decreases in noncompliant behavior. Thus, a dual sys- tem, which was user friendly and convenient, proved to be effective insofar as the students earned stickers for in- tervals of compliant behavior, but lost stickers for non- compliant behavior (response cost).

This third method of response cost has at least two advantages: (a) If all of the tokens or points are not lost via response cost, the remaining ones can be traded for backup reinforcers, thus adding a reinforcement compo- nent that is incorporated into many programs; and (b) re- inforcers can be resupplied by performing appropriate behavior, thereby reducing legal and ethical concerns. The following vignette illustrates how a father might im- plement a bonus response cost procedure with his two sons to reduce their fighting at dinnertime.

Father: Boys, I’d like to talk to you. Tom and Pete: OK. Father: We have to find a way to stop the fighting and

squabbling at dinnertime. The way you two go at each other upsets everyone in the family, and I can’t eat my dinner in peace.

Tom: Well, Pete usually starts it. Pete: I do not! Tom: You do, too . . . Father: (Interrupting) That’s enough! This is just what I

mean, nit-picking at each other. It’s ridiculous and it needs to stop. I’ve given it some thought, and here’s what we’re going to do. Since each of you earns an allowance of $5 per week for doing household chores, I was tempted to take it away from you the next time either of you got into a fight. But I’ve de- cided against doing that. Instead, I’m going to make an additional $5 available to each of you, but for each sarcastic comment or squabble you get into at the dinner table, you’ll lose $1 of that extra money. So if you act properly, you’ll get an additional $5. If you have two fights, you’ll lose $2 and get only $3 Do you understand?

Pete: How about if Tom starts something and I don’t? Father: Whoever does any fighting or squabbling loses

$1. If Tom starts something and you ignore him, only he loses. Tom, the same thing holds true for you. If Pete starts something and you ignore him, only Pete loses. One more thing. Dinnertime starts when you are called to the table; it ends when you are excused from the table. Any other questions?

Tom: When do we begin? Father: We’ll begin tonight.

It is important to note that the father described the contingencies completely to his sons. He told them what would happen if no fights occurred, and he told them what would happen if one sibling attempted to start a fight but the other ignored him. In addition, he clearly defined when dinnertime would begin and end. Each of these explanations was necessary for a complete de- scription of the contingencies. Presumably, the father praised his sons for their appropriate mealtime behav- iors so that they were reinforced and strengthened.

Combining with Group Consequences

The final way that response cost can be implemented in- volves group consequences. That is, contingent on the inappropriate behavior of any member of a group, the whole group loses a specific amount of reinforcement. The Good Behavior Game serves as an illustration of how response cost can be applied to groups and presents this method of response cost in more depth.

Using Response Cost Effectively To use response cost effectively, the behavior targeted for response cost and the amount of the fine need to be ex- plicitly stated. Further, any rules that would apply for re- fusal to comply with the response cost procedure need to be explained. Included in the definition of target be- haviors should be the corresponding point loss for each behavior. However, in situations in which multiple be- haviors are subject to the response cost contingency or in which the degree of severity of the behavior determines the response cost, correspondingly greater fines should be associated with more severe behaviors. In effect, the mag- nitude of the punishment (response cost) should fit the offense.

According to Weiner (1962)—the originator of the term response cost—the magnitude of the response cost fine is important. It is likely that as the magnitude of the fine increases, larger reductions in the rate of the unde- sirable behavior may occur. However, if the loss of rein- forcers is too great and too rapid, the rate of the reduction of the inappropriate target behavior may be affected ad- versely. The person will soon be exhausted of his or her supply of reinforcers, and a deficit situation will exist. As a general rule, the fine should be large enough to sup- press the future occurrence of the behavior, but not so large as to bankrupt the person or cause the system to

Punishment by Removal of a Stimulus

386

lose its effectiveness. In short, if a client loses tokens or other reinforcers at too high a rate, he may give up and become passive or aggressive, and the procedure will be- come ineffective. Furthermore, fines should not be changed arbitrarily. If a 5-minute loss of recess is im- posed on each noncompliant behavior in the morning, the teacher should not impose a 15-minute fine on the same noncompliant behavior in the afternoon.

Determining the Immediacy of the Fines

Ideally, the fine should be imposed immediately after the occurrence of each undesirable behavior. The more quickly response cost is applied subsequent to the oc- currence of the behavior, the more effective the proce- dure becomes. For example, Ashbaugh and Peck (1998) applied a response cost contingency successfully with a young toddler with sleep disruptive behavior immedi- ately after a failure-to-sleep interval passed without sleep onset.

Response Cost or Bonus Response Cost?

Deciding which variation of response cost would be the most effective to reduce behavior is usually an empirical question. However, three considerations can help practi- tioners. First, the least aversive procedure should be at- tempted initially. Consistent with the principle of the least restrictive alternative, an effort should be made to ensure that the minimum loss of reinforcers occurs for the min- imum amount of time. Bonus response cost may be the less aversive of the two variations because the reinforcers are not deducted directly from the person; instead they are lost from a pool of potentially available (i.e., bonus) reinforcers.

The second consideration is similar to the first and can be stated in the form of a question: What is the po- tential for aggressive, emotional outbursts? From a so- cial validity perspective, students (and their parents) would probably find it more agreeable to lose reinforcers from an available reserve than from their earnings. Con- sequently, a bonus response cost procedure might be less likely to spark aggressive or emotional outbursts, or be of- fensive to students or their parents.

A third consideration is the need to reduce the be- havior quickly. Combative or noncompliant behavior may be more appropriately suppressed with response cost because the contingency directly reduces the stu- dent’s available reinforcers. The response-contingent

withdrawal of reinforcers serves to reduce the behavior swiftly and markedly.

Ensuring Reinforcement Reserve

Positive reinforcers cannot be removed from a person who does not have any. Prior to using response cost, the practitioner must ensure a sufficient reinforcer reserve. Without such a reserve, the procedure is unlikely to be successful. For example, if a teacher used response- contingent withdrawal of free time for each occurrence of inappropriate behavior in a highly disruptive class, the students could exhaust all of their available free time be- fore the end of the first hour, leaving the teacher won- dering what to do for the remainder of the day. Deducting free time for succeeding days would hardly be beneficial.

Two suggestions apply to reduce the likelihood of having no reinforcers available. First, the ratio of points earned to points lost can be managed. If baseline data indicate that the inappropriate behavior occurs at a high rate, more reinforcers can be programmed for removal. Also, determining the magnitude of the fine, and stating it explicitly beforehand, is helpful. Minor infractions may warrant relatively small fines, whereas major in- fractions may warrant substantially higher fines. Sec- ond, if all reinforcers are lost and another inappropriate behavior occurs, time-out might be applied. After that, when reinforcers have again been earned for appropriate behavior, response cost can be reinstituted.

To establish the initial number of available rein- forcers, practitioners collect baseline data on the oc- currence of inappropriate behavior during the day or session. The mean baseline figure can be increased for the number of reinforcers to ensure that all reinforcers are not lost when response cost is in effect. Although no empirically verifiable guidelines are available, a pru- dent approach is to increase the number of reinforcers 25% above the mean number of occurrences during baseline. For example, if baseline data indicate that the mean number of disruptions per day is 20, the practi- tioner might establish 25 minutes of free time (the pos- itive reinforcer) as the initial level (i.e., 20 × 1.25). If the practitioner calculates points instead of percentages, she could add an additional 10 to 20 points to ensure an ad- equate buffer.

Recognizing the Potential for Unplanned or Unexpected Outcomes

Two situations may require the implementation of a con- tingency plan. One occurs when the repeated imposition of response cost serves to reinforce, rather than punish, the

Punishment by Removal of a Stimulus

387

undesirable behavior. When this situation arises, the prac- titioner should stop using response cost and switch to an- other reductive procedure (e.g., time-out or DRA). The second situation occurs when the person refuses to give up her positive reinforcers. To reduce the likelihood of this event, the practitioner should clarify the consequences of such refusal beforehand and (a) make sure that an adequate supply of backup reinforcers is available, (b) impose an additional penalty fine for not giving up the reinforcer (e.g., the sticker) (Musser et al., 2001), and/or (c) reimburse the person with some fractional portion of the fine for complying with immediate payment.

Avoiding the Overuse of Response Cost

Response cost should be saved for those major undesir- able behaviors that call attention to themselves and need to be suppressed quickly. The teacher’s or parent’s pri- mary attention should always be focused on positive be- havior to reinforce; response cost should be a last resort and should be combined with other procedures to build adaptive behavior.

Keeping Records

Each occurrence of response cost and the behavior that occasioned it should be recorded. Minimally, the analyst should record the number of times fines are imposed, the persons to whom fines are issued, and the effects of the fines. Daily data collection helps to determine the effi- cacy of the response cost procedure. By graphing the ef- fects of the program, the behavior analyst can determine the suppressive effect of the procedure.

Response Cost Considerations Increased Aggression

Response-contingent withdrawal of positive reinforcers may increase student verbal and physical aggressiveness. The student who loses several tokens, especially within a short time, may verbally or physically assault the teacher. Emotional behaviors should be ignored when- ever possible if they accompany the implementation of response cost (Walker 1983). Still, teachers should an- ticipate this possibility and (a) preempt a decision to use response cost if they suspect that a worse condition will result in the aftermath of an emotional episode, or (b) be prepared to “ride out the storm.”

Avoidance

The setting in which response cost occurs or the person who administers it can become a conditioned aversive stim- ulus. If this situation occurs in school, the student may avoid the school, the classroom, or the teacher by being absent or tardy. A teacher can reduce the likelihood of becoming a conditioned aversive stimulus by contingently delivering positive reinforcement for appropriate behavior.

Collateral Reductions of Desired Behavior

The response-contingent withdrawal of positive rein- forcers for one behavior can affect the frequency of other behaviors as well. If the teacher fines Shashona 1 minute of recess each time she calls out during math class, the response cost procedure may reduce not only her call-outs but also her math productivity. Shashona may say to the teacher, “Since I lost my recess time for calling out, I am not going to do my math.” She could also just become passive-aggressive by just sitting in her seat with arms folded and not work. Teachers and other practitioners should anticipate such collateral behaviors and clearly explain the response cost rules, reinforce other classmates as models of appropriate behavior, and avoid face-to-face confrontations.

Calling Attention to the Punished Behavior

Response cost calls attention to the undesirable behavior. That is, upon the occurrence of an inappropriate behavior, the student is informed of reinforcer loss. The teacher’s at- tention—even in the form of notification of reinforcer loss—could serve as a reinforcing consequence. In effect, her attention may increase the frequency of future mis- behavior. For instance, a teacher may have difficulty with some students because every mark that the teacher places on the chalkboard indicates that a positive reinforcer has been lost (e.g., minutes of free time). Further, it calls at- tention to the undesirable behavior and may inadvertently reinforce it. In such a situation, the teacher should change his tactic, perhaps combining response cost with time-out. Also, to counteract the possibility of calling attention to inappropriate behavior, practitioners should ensure that the ratio of reinforcement to response cost contingencies favors reinforcement.

Unpredictability

As with other forms of punishment, side effects of re- sponse cost can be unpredictable. The effects of response

Punishment by Removal of a Stimulus

388

cost seem to be related to a number of variables that are not fully understood and have not been well investigated across participants, settings, or behaviors. These variables include the magnitude of the fine, the previous punish-

ment and reinforcement history of the individual, the fre- quency with which behaviors are fined, and the avail- ability of alternative responses that are eligible for reinforcement.

Summary Definition of Time-Out

1. Time-out from positive reinforcement, or simply time-out, is defined as the withdrawal of the opportunity to earn pos- itive reinforcement or the loss of access to positive rein- forcers for a specified time, contingent on the occurrence of a behavior.

2. Time-out is a negative punisher, and it has the effect of reducing the future frequency of the behavior that pre- ceded it.

Time-Out Procedures for Applied Settings

3. There are two basic types of time-out: nonexclusion and exclusion.

4. Within the nonexclusion type, planned ignoring, with- drawal of a specific positive reinforcer, contingent obser- vation, and time-out ribbons are the main methods.

5. Within exclusion time-out, the time-out room, partition time- out, and hallway time-out serve as the principal methods.

6. Time-out is a desirable alternative for reducing behavior because of its ease of application, acceptability, rapid sup- pression of behavior effects, and ability to be combined with other approaches.

Using Time-Out Effectively

7. The time-in environment must be reinforcing if time-out is to be effective.

8. Effective use of time-out requires that the behaviors lead- ing to, the duration of, and exit criteria for time-out be ex- plicitly stated. Further, practitioners must decide whether to use nonexclusion or exclusion time-out.

9. In most applications, permission is required before time- out can be implemented.

10. Practitioners should be aware of legal and ethical consid- erations before implementing time-out. As a punishment

procedure, it should be used only after positive reductive procedures have failed, and with planned monitoring, su- pervision, and evaluation considerations in place.

Definition of Response Cost

11. Response cost is a form of punishment in which the loss of a specific amount of reinforcement occurs, contingent on the performance of an inappropriate behavior, and re- sults in the decreased probability of the future occurrence of the behavior.

12. Four methods for implementing response cost are: attrac- tive procedure for practitioners, including moderate-to- rapid suppression of behavior, convenience, and its ability to be combined with other procedures.

Response Cost Methods

13. Four methods for implementing response cost are: as a di- rect fine, as a bonus response cost, combined with positive reinforcement, and within a group arrangement.

Using Response Cost Effectively

14. To use response cost effectively, practitioners should de- termine the immediacy of the fine, decide whether bonus response cost is a preferred option, ensure reinforcer re- serve, recognize the potential for unplanned or unexpected outcomes, avoid overusing response cost, and keep good records on its effects.

Response Cost Considerations

15. Implementing response cost may increase student aggres- siveness, produce avoidance responses, affect collateral reductions of desired behaviors, and call attention to the punished behavior. The effects of response cost can also be unpredictable.

Punishment by Removal of a Stimulus

389

Motivating Operations

Key Terms abative effect abolishing operation (AO) behavior-altering effect conditioned motivating operation

(CMO) discriminative stimulus (SD)

related to punishment establishing operation (EO) evocative effect

function-altering effect motivating operation (MO) recovery from punishment procedure reflexive conditioned motivating

operation (CMO-R) repertoire-altering effect reinforcer-abolishing effect reinforcer-establishing effect

surrogate conditioned motivating operation (CMO-S)

transitive conditioned motivating operation (CMO-T)

unconditioned motivating operation (UMO)

unpairing value-altering effect

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes, and Concepts

3-8 Define and provide examples of establishing operations.

This chapter was written by Jack Michael.

390

Motivating Operations

In commonsense psychology what people do at any particular moment is at least partly a function of what they want at that moment. From a be-

havioral perspective based on Skinner’s analysis of drive (1938, 1953), wanting something may be interpreted to mean that (a) the occurrence of what is wanted would func- tion as a reinforcer at that moment, and (b) the current fre- quency of any behavior that has previously been so reinforced will increase. This chapter surveys and classi- fies variables that have these two motivating effects.

Definition and Characteristics of Motivating Operations Basic Features

In their treatment of motivation, Keller and Schoenfeld (1950) identified the drive concept in terms of a relation between certain environmental variables, which they called establishing operations, and certain changes in be- havior. Although not conforming exactly to their usage, establishing operation (EO) was reintroduced in 1982 (Michael, 1982, 1993) as a term for any environmental variable that (a) alters the effectiveness of some stimulus, object, or event as a reinforcer; and (b) alters the current frequency of all behavior that has been reinforced by that stimulus, object, or event. The term establishing opera- tion (EO) is now commonly used in applied behavior analysis (e.g., Iwata, Smith, & Michael, 2000; McGill, 1999; Michael, 2000; Smith & Iwata, 1997; Vollmer & Iwata, 1991).

The term motivating operation (MO) has recently been suggested as a replacement for the term estab- lishing operation (Laraway, Snycerski, Michael, & Pol- ing, 2001), along with the terms value altering and behavior altering for the two defining effects described previously. This more recent arrangement will be pre- sented in this chapter.1

The value-altering effect is either (a) an increase in the reinforcing effectiveness of some stimulus, object, or event, in which case the MO is an establishing operation (EO); or (b) a decrease in reinforcing effectiveness, in which case the MO is an abolishing operation (AO). The behavior-altering effect is either (a) an increase in the current frequency of behavior that has been reinforced

1J. R. Kantor’s setting factor (1959d, p. 14) included motivating opera- tions as described, but also includes some events that do not fit the spe- cific definition in terms of the two effects. See Smith and Iwata (1997, pp 346–348) for a treatment of the setting factor as an antecedent influ- ence on behavior.

2The usefulness of the new term, abative, is described in detail in Laraway, Snycerski, Michael, and Poling (2001).

by some stimulus, object, or event, called an evocative ef- fect; or (b) a decrease in the current frequency of behav- ior that has been reinforced by some stimulus, object, or event, called an abative effect.2 These relations are shown in Figure 1.

For example, food deprivation is an EO that increases the effectiveness of food as a reinforcer and evokes all behavior that has been reinforced with food. Food in- gestion (consuming food) is an AO that decreases the ef- fectiveness of food as a reinforcer, and abates all behavior that has been followed by food reinforcement. These re- lations are shown in Figure 2.

An increase in painful stimulation is an EO that inc- reases the effectiveness of pain reduction as a reinforcer and evokes all behavior that has been reinforced by pain reduction. A decrease in painful stimulation is an AO that decreases the effectiveness of pain reduction as a rein- forcer and abates all behavior that has been followed by pain reduction. These relations are shown in Figure 3.

Most of the statements in this chapter about value- altering and behavior-altering effects refer to relations involving reinforcement rather than punishment. It is rea- sonable to assume that MOs also alter the effectiveness of stimuli, objects, and events as punishers, with either es- tablishing or abolishing effects; and also alter the current frequency of all behavior that has been punished by those stimuli, objects, or events, either abating or evoking such behavior. At the present time, however, motivation with respect to punishment is just beginning to be dealt with in applied behavior analysis. Later in the chapter a small section considers the role of UMOs for punishment, but it is likely that most of what is said about MOs and rein- forcement will be extended to punishment in future treat- ments of motivation.

Additional Considerations

Direct and Indirect Effects

Behavior-altering effects are actually more complex than has been implied so far. The alteration in frequency can be the result of (a) a direct evocative or abative effect of the MO on response frequency and (b) an indirect effect on the evocative or abative strength of relevant discrim- inative stimuli (SDs). One would also expect an MO to have a value-altering effect on any relevant conditioned reinforcers, which, in turn, would have a behavior-altering effect on the type of behavior that has been reinforced by those conditioned reinforcers. This relation will be considered later in connection with conditioned motivat- ing operations.

391

Figure 1 Motivating operations (MOs) and their two defining effects.

Establishing Operation (EO)

• Value-altering effect: An increase in the current effectiveness of some stimulus, object, or event as reinforcement.

• Behavior-altering effect: An increase in the current frequency of all behavior that has been reinforced by that stimulus, object, or event (i.e., an evocative effect).

Abolishing Operation (AO)

• Value-altering effect: A decrease in the current effectiveness of some stimulus, object, or event as reinforcement.

• Behavior-altering effect: A decrease in the current frequency of all behavior that has been reinforced by that stimulus, object, or event (i.e., an abative effect).

Not Just Frequency

In addition to frequency, other aspects of behavior can result from a change in an MO, such as response magni- tude (a more or less forceful response), response latency (a shorter or longer time from MO or SD occurrence until the first response), relative frequency (a larger or smaller proportion of responses to the total number of response opportunities), and others. However, because frequency is the best known measure of operant relations, it will be the behavioral alteration referred to in the remainder of this chapter.

A Common Misunderstanding

Sometimes the behavior-altering effect is interpreted as an alteration in frequency due to the organism’s encoun- tering a more or less effective form of reinforcement. This implies that the increase or decrease occurs only after the reinforcer has been obtained. The critical ob- servation contradicting this notion is that a strong relation exists between MO level and responding in extinction— when no reinforcers are being received (see Keller & Schoenfeld, 1950, pp. 266–267 and Figure 60). In terms of an organism’s general effectiveness, an MO should

Figure 2 Motivating operations related to food.

Food Deprivation as an Establishing Operation (EO)

• Value-altering effect: An increase in the reinforcing effectiveness of food. • Behavior-altering evocative effect: An increase in the current frequency of all behavior

that has been reinforced by food.

Food Ingestion as an Abolishing Operation (AO)

• Value-altering effect: A decrease in the reinforcing effectiveness of food. • Behavior-altering abative effect: A decrease in the current frequency of all behavior that

has been reinforced by food.

Figure 3 Motivating operations related to painful stimulation.

Increase in Painful Stimulation as an Establishing Operation (EO)

• Value-altering effect: An increase in the reinforcing effectiveness of pain reduction. • Behavior-altering evocative effect: An increase in the current frequency of all behavior

that has been reinforced by pain reduction.

Decrease in Painful Stimulation as an Abolishing Operation (AO)

• Value-altering effect: A decrease in the reinforcing effectiveness of pain reduction. • Behavior-altering abative effect: A decrease in the current frequency of all behavior that

has been reinforced by pain reduction.

Motivating Operations

392

Motivating Operations

evoke the relevant behavior even if it is not at first suc- cessful. At present, the two effects (value-altering and behavior-altering) will be considered independent in the sense that one does not derive from the other, although they are probably related at the neurological level.

Current versus Future Effects: Behavior- Altering versus Repertoire-Altering Effects

As a result of an environmental history, an organism has an operant repertoire of MO (motivating operation), SD

(discriminative stimulus), and R (response) relations. (Also present is a respondent repertoire of stimuli capa- ble of eliciting responses) MOs and SDs are components of the existing repertoire; they are antecedent variables that have behavior-altering effects. Antecedent events can evoke or abate responses, but their simple occurrence does not alter the organism’s operant repertoire of functional re- lations. These antecedent variables are in contrast to con- sequence variables, whose main effect is to change the organism’s repertoire of functional relations so that the organism behaves differently in the future. Consequence variables include reinforcers, punishers, and the occur- rence of a response without its reinforcer (extinction pro- cedure) or without its punisher (recovery from punishment procedure). This is what is meant when MOs and SDs are said to alter the current frequency of all behavior relevant to that MO; but reinforcers, punishers, and response occurrence without consequence alter the future frequency of whatever behavior immediately pre- ceded those consequences.

It is useful to have a different name for these two quite different effects of a behaviorally relevant event; thus repertoire-altering effect (Schlinger & Blakely, 1987) is contrasted with behavior-altering effect in this chapter. In this chapter the distinction between the two will usually be implied by referring to current frequency or to future frequency.

A Critical Distinction: Motivative versus Discriminative Relations MOs and SDs are both antecedent variables that alter the current frequency of some particular type of behavior. They are also both operant (rather than respondent) variables in that they control response frequency because of their rela- tion to reinforcing or punishing consequences (rather than to a respondent unconditioned stimulus). At this point, therefore, it may be useful to review the contrasting defi- nitions of these two types of antecedent variables.

An SD is a stimulus that controls a type of behavior because that stimulus has been related to the differential

availability of an effective reinforcer for that type of be- havior. Differential availability means that the relevant consequence has been available in the presence of, and unavailable in the absence of, the stimulus. Considering food deprivation and pain increase as putative SDs re- quires that food and pain removal be available in the pres- ence of those conditions. This is somewhat problematic in that both can and often do occur under conditions when the reinforcers are not available. A true SD consti- tutes at least a probabilistic guarantee that the relevant consequence will follow the response. Organisms may be food deprived or subjected to painful stimulation for considerable periods during which the condition cannot be alleviated.

More serious for the interpretation as an SD is that the unavailability of a reinforcer in the absence of the stimulus implies that the unavailable event would have been effective as a reinforcer if it had been obtained. It is with respect to this requirement that most motivative vari- ables fail to qualify as discriminative stimuli. It is true that food reinforcement is, in a sense, unavailable in the absence of food deprivation, and pain-reduction rein- forcement is unavailable in the absence of pain, but this is not the kind of unavailability that occurs in discrimi- nation training and that develops the evocative effect of a true SD (Michael, 1982). In the absence of food depri- vation or painful stimulation, no MO is making food or pain removal effective as a reinforcer; therefore, there is no reinforcer unavailability in the sense that is relevant to the discriminative relation.

However, food deprivation and painful stimulation easily qualify as MOs, as conditions that alter the ef- fectiveness of some stimuli, objects, or events as rein- forcers, and that simultaneously alter the frequency of the behavior that has been reinforced by those stimuli, objects, or events.

To summarize, a useful contrast can usually be made as follows: Discriminative stimuli are related to the dif- ferential availability of a currently effective form of re- inforcement for a particular type of behavior; motivative variables (MOs) are related to the differential reinforcing effectiveness of a particular type of environmental event.

Unconditioned Motivating Operations (UMOs) For all organisms there are events, operations, and stimu- lus conditions with value-altering motivating effects that are unlearned. Humans are born with the capacity to be more affected by food reinforcement as a result of food deprivation, or more affected by pain reduction reinforce- ment as a result of pain onset or increase. Thus, food de- privation and painful stimulation are called unconditioned

393

motivating operations (UMOs)3. By contrast, needing to enter a room through a locked door establishes the key to the door as an effective reinforcer, but this value-altering effect is clearly a function of a learning history involving doors and keys. MOs of this kind are called conditioned motivating operations (CMOs) and will be considered in detail later in this chapter.

Note that it is the unlearned aspect of the value- altering effect that results in an MO being classified as unconditioned. The behavior-altering effect of an MO is usually learned. Said another way, we are born with the capacity to be more affected by food reinforcement as a result of food deprivation, but we have to learn most of the behavior that obtains food—asking for some, going to where it is kept, and so forth.

Nine Main UMOs for Humans

Deprivation and Satiation UMOs

Deprivation with respect to food, water, oxygen, activ- ity, and sleep all have resultant reinforcer-establishing and evocative effects. By contrast, food and water in- gestion, oxygen intake, engaging in activity, and sleeping have reinforcer-abolishing and abative effects.

UMOs Relevant to Sexual Reinforcement

For many nonhuman mammals, hormonal changes in the female are related to time passage, ambient light condi- tions, daily average temperature, or other features of the environment that are related phylogenically to successful reproduction. These environmental features, or hormonal changes, or both, can be considered UMOs that cause contact with a male to be an effective reinforcer for the female. They may produce visual changes in some as- pect of the female’s body and elicit chemical (olfactory) attractants that function as UMOs for the male, estab- lishing contact with a female as a reinforcer and evoking any behavior that has produced such contact. The various hormonal changes may also evoke certain behaviors by the female (e.g., the assumption of a sexually receptive posture) that as stimuli function as UMOs for sexual be- havior by the male. Superimposed on this collection of UMOs and unconditioned elicitors is a deprivation effect that may also function as a UMO.

3The terms unconditioned and conditioned are used to modify MOs in the same way they modify respondent-eliciting stimuli and operant reinforcers and punishers. Unconditioned motivating operations, like unconditioned eliciting stimuli for respondent behavior and unconditioned reinforcers and punishers, have effects that are not dependent on a learning history. The ef- fects of conditioned motivating operations, like the effects of conditioned eliciting stimuli and conditioned reinforcers and punishers, are dependent on a learning history.

In the human, learning plays such a strong role in the determination of sexual behavior that the role of un- learned environment–behavior relations has been diffi- cult to determine. The effect of hormonal changes in the female on the female’s behavior is unclear, as is the role of chemical attractants on the male’s behavior. Other things being equal, both male and female seem to be af- fected by the passage of time since last sexual activity (deprivation), functioning as a UMO that establishes the effectiveness of sexual stimulation as a reinforcer and that evokes the behavior that has achieved that kind of reinforcement. Working in the opposite direction, sexual orgasm functions as a UMO that abolishes the effective- ness of sexual stimulation as a reinforcer and abates (de- creases the frequency of) behavior that has achieved that kind of reinforcement. In addition, tactile stimulation of erogenous regions of the body seems to function as a UMO in making further similar stimulation even more effective as reinforcement and in evoking all behavior that in the past has achieved such further stimulation.

Temperature Changes

Becoming uncomfortably cold is a UMO that establishes becoming warmer as a reinforcer, and evokes any be- havior that has had that effect. A return to a normal tem- perature condition is a UMO that abolishes an increase in warmth as a reinforcer and abates behavior that has had a warming effect. Becoming uncomfortably warm is a UMO that establishes a decrease in the temperature con- dition as an effective reinforcer and evokes any behavior that has resulted in a body-cooling effect. A return to a normal temperature condition abolishes being cooler as a reinforcer and abates body-cooling behavior.

These temperature-related UMOs could be combined by specifying becoming uncomfortable with respect to tem- perature as a UMO that establishes a change for the better as a reinforcer, and that evokes any behavior that has achieved such an effect. Returning to a normal condition would then have appropriate UMO abolishing and abative effects. Still, it seems better at this point to conceptualize the situation as involving different UMOs. Also, these UMOs could be grouped with the pain UMO and included under the broad category of aversive stimulation, but it will be clearer at this point to consider them separately.

Painful Stimulation

An increase in painful stimulation establishes pain re- duction as a reinforcer and evokes the behavior (called escape behavior) that has achieved such reduction. A de- crease in painful stimulation abolishes the effectiveness of pain reduction as a reinforcer and abates the behavior that has been reinforced by pain reduction.

Motivating Operations

394

Motivating Operations

Table 1 Nine Unconditioned Motivating Operations (UMOs) and Their Reinforcer-Establishing and Evocative Effects

Unconditioned Motivating Operation (UMO) Reinforcer-Establishing Effect Evocative Effect

Food deprivation Increases effectiveness of food ingestion as a Increases current frequency of all behavior reinforcer previously reinforced with food

Water deprivation Increases effectiveness of water ingestion as a Increases current frequency of all behavior reinforcer previously reinforced with water

Sleep deprivation Increases effectiveness of sleep as a reinforcer Increases current frequency of all behavior previously reinforced with being able to sleep

Activity deprivation Increases effectiveness of activity as a reinforcer Increases current frequency of all behavior previously reinforced with activity

Oxygen deprivation* Increases effectiveness of breathing as a Increases current frequency of all behavior reinforcer previously reinforced with being able to breathe

Sex deprivation Increases effectiveness of sex stimulation as Increases current frequency of all behavior a reinforcer previously reinforced with sexual stimulation

Becoming too warm Increases effectiveness of temperature decrease Increases current frequency of all behavior as a reinforcer previously reinforced with becoming cooler

Becoming too cold Increases effectiveness of temperature increase Increases current frequency of all behavior as a reinforcer previously reinforced with becoming warmer

Increase in painful Increases effectiveness of a decrease in pain Increases current frequency of all behavior stimulus as a reinforcer previously reinforced with a decrease in painful

stimulation

*It is not actually oxygen deprivation that functions as a UMO, but rather the buildup of carbon dioxide in the blood as a result of not being able to excrete carbon dioxide because of not being able to breathe or because of breathing in air that is as rich in carbon dioxide as the exhaled air.

4For a discussion of Skinner’s approach to emotional predispositions in the general context of MOs, see Michael (1993, p. 197).

In addition to establishing pain reduction as a rein- forcer and evoking the behavior that has produced pain re- duction, painful stimulation in the presence of another organism evokes aggressive behavior toward that organ- ism. In some organisms, including humans, some of this aggression may be the elicitative result of the pain func- tioning as a respondent unconditioned stimulus (US) (Ul- rich & Azrin, 1962). However, a case can be made for the painful stimulation also functioning as a UMO that makes events such as signs of damage to another organ- ism effective as reinforcers and evokes the behavior that has been reinforced by the production of such signs. Skin- ner made such a case (1953, pp. 162–170) in his analy- sis of anger, and extended the analysis to the emotions of love and fear.4

Review of UMO Effects

Table 1 summarizes the reinforcer-establishing and evocative effects of the nine UMOs for humans. Simi- larly, Table 2 shows the reinforcer-abolishing and aba- tive effects.

The Cognitive Misinterpretation

The behavior-altering effects of UMOs on human be- havior are generally understood to some degree. In- creased current frequency of behavior that has made us warmer as a result of becoming too cold is a part of our everyday experience, as is the cessation of this behavior when normal temperature returns. That water deprivation should evoke behavior that has obtained water, and that this behavior should cease when water has been obtained, seems only reasonable. However, the variables responsi- ble for these effects are often misconstrued.

The cognitive interpretation of the behavior-altering effects of UMOs on the behavior of a verbally sophisti- cated individual are in terms of that individual’s under- standing (being able to verbally describe) the situation and then behaving appropriately as a result of that un- derstanding. On the contrary, the behavioral interpretation that reinforcement automatically adds the reinforced be- havior to the repertoire that will be evoked and abated by the relevant UMO is not always well appreciated. From the behavioral perspective, the person, verbally sophisti- cated or not, does not have to “understand” anything for an MO to have value-altering and behavior-altering effects.

395

Table 2 UMOs that Decrease Reinforcer Effectiveness and Abate Relevant Behavior

Unconditioned Motivating Operation (UMO) Reinforcer-Abolishing Effect Abative Effect

Food ingestion (after Decreases effectiveness of food as a reinforcer Decreases current frequency of all behavior food deprivation) previously reinforced with food

Water ingestion (after Decreases effectiveness of water as a reinforcer Decreases current frequency of all behavior water deprivation) previously reinforced with water

Sleeping (after sleep Decreases effectiveness of sleep as a reinforcer Decreases current frequency of all behavior deprivation) previously reinforced with sleep

Being active (after Decreases effectiveness of activity as a reinforcer Decreases current frequency of all behavior activity deprivation) previously reinforced with activity

Breathing (after not Decreases effectiveness of breathing as a Decreases current frequency of all behavior being able to breathe) reinforcer previously reinforced with being able to breathe

Orgasm or sex stimula- Decreases effectiveness of sexual stimulation as Decreases current frequency of all behavior tion (after sex deprivation) a reinforcer previously reinforced with sexual stimulation

Becoming cooler (after Decreases effectiveness of temperature decrease Decreases current frequency of all behavior being too warm) as a reinforcer previously reinforced with becoming cooler

Becoming warmer (after Decreases effectiveness of temperature increase Decreases current frequency of all behavior being too cold) as a reinforcer previously reinforced with becoming warmer

Painful stimulation de- Decreases effectiveness of pain decrease as a Decreases current frequency of all behavior crease (while in pain) reinforcer previously reinforced with a decrease in pain

The cognitive misinterpretation of the behavior- altering effect encourages two forms of practical inef- fectiveness. First, insufficient effort may be made to teach appropriate behavior to individuals with limited verbal repertoires, on the grounds that they will not be able to understand the relevant environment–behavior relations. Second, there may be insufficient preparation for an in- crease in whatever behavior preceded the relevant rein- forcement (often some form of inappropriate behavior such as yelling and crying), again because it is not thought that the individual would understand the rela- tion between his behavior and any consequence.

Relevance of the MO to a Generality of Effects

In the applied area, the reinforcer-establishing effect of MOs seems to be increasingly understood and used. Ed- ibles may be temporarily withheld so that they will be more effective as reinforcers for behavior that is being taught; similarly with music, toys, and attention from an adult. However, it is not so widely recognized that the behavior being taught with these reinforcers will not occur in future circumstances, even if it was well learned and is a part of the learner’s repertoire, unless the relevant MO is in effect. This issue is considered in the case of punishment where the role of the MO in the occurrence of behavior in future circumstances is more complex and even more likely to be overlooked.

The importance of making the stimulus conditions during instruction similar to those present in the settings and situations in which generalization of the learned be- havior is desired is generally understood. However, the fact that the relevant MO must also be in effect for re- sponding to be generalized and maintained seems more easily overlooked.

Weakening the Effects of UMOs

For practical reasons it may be necessary to weaken the effects of an MO. Both reinforcer-establishing and evoca- tive effects of UMOs can be temporarily weakened by the relevant reinforcer-abolishing and abative operations. For example, food ingestion will have an abative effect on undesirable behavior being evoked by food deprivation, such as food stealing, but the behavior will return when deprivation is again in effect. In general it is not possible to permanently weaken the value-altering effects of UMOs. Water deprivation will always make water more effective as a reinforcer, and pain increase will always make pain decrease more effective as a reinforcer. But the behavior-altering effects are clearly based on a history of reinforcement, and such histories can be reversed by an extinction procedure—let the evoked response occur without its reinforcement. (And the abative effects of a punishment history can be reversed by allowing the response to occur without punishment—the recovery from punishment procedure.) With respect to the UMOs,

Motivating Operations

396

Motivating Operations

however, the relevant reinforcer must be obtainable in some acceptable way as the undesirable behavior is being extinguished. People cannot be expected to do without the various unconditioned reinforcers controlled by UMOs.

UMOs for Punishment

An environmental variable that alters the punishing ef- fectiveness of a stimulus, object, or event, and alters the frequency of the behavior that has been so punished, is an MO for punishment. If the value-altering effect does not depend on a learning history, then such a variable would qualify as a UMO.

Value-Altering Effect

An increase in painful stimulation functions as punish- ment as long as the current level of painful stimulation is not so high that an increase cannot occur. So the UMO must consist of a current pain level that is still capable of increase, or a change to such a level from one that is so high that further increase is not possible. This means that in general a pain increase will almost always function as an unconditioned punisher.5 This is also true for other kinds of stimulation that function as unconditioned pun- ishers—some sounds, odors, tastes, and so on.

Most of the punishers that affect humans, however, are effective because of a learning history; that is, they are conditioned rather than unconditioned punishers. If the learning history consisted of pairing conditioned pun- ishers with unconditioned punishers, then the UMOs for those unconditioned punishers are CMOs for the condi- tioned punishers. (This UMO–CMO relation is described in more detail in the following section on CMOs.) If they are punishers because of a historical relation to a reduced availability of reinforcers, then the MOs for those rein- forcers are the MOs for the conditioned punishers. Re- moving food as a punisher, or more commonly a change to a stimulus in the presence of which food has been less available, will only function as a punisher if food is cur- rently effective as a reinforcer. Thus, the MO for food re- moval as a punisher is food deprivation.

Social disapproval (expressed by a frown, a head shake, or a specific vocal response such as “No” or “Bad”) constitutes a stimulus condition in the presence

5Painful stimulation can be a conditioned punisher if it had been historically associated with some other punisher, such as when pain has been evidence that something more serious is wrong. On the other hand, painful stimula- tion can be a conditioned reinforcer if historically associated with some re- inforcer in addition to its own termination, such as when muscle pain is related to having had an effective exercise workout, or when painful stim- ulation has been paired with some form of sexual reinforcement.

of which the typical reinforcers provided by the disap- proving person have been withheld. However, such a stimulus condition will function as punishment only if the MOs for those withheld reinforcers are currently in effect. The punishment procedure called time-out from reinforcement is similar and will only function as pun- ishment if the reinforcers that are made unavailable— from which the person is timed out—are truly effective reinforcers at the time the person is being punished. Response cost—taking away objects such as tokens that can be exchanged for various reinforcers or imposing a monetary fine or a fine with respect to some kind of point or score bank—is a more complex type of procedure in- volving a time delay between when the punishment op- eration occurs and when the decrease in reinforcers takes effect. Still, unless the events that have been subjected to delayed removal (the things that the tokens, points, or money can be exchanged for) are effective as reinforcers at the time the response cost procedure occurs, punish- ment will not have occurred.

Behavior-Altering Effect

In general, observing a punishment effect is more com- plex than observing a reinforcement effect because one must also consider the status of the variable responsible for the occurrence of the punished behavior. This applies to the effects of MOs as well. The behavior-altering evocative effect of an MO for reinforcement consists of an increase in the current frequency of any behavior that had been so reinforced. For example, food deprivation evokes (increases the current frequency of) all behavior that had been reinforced with food. The behavior-altering effect of an MO for punishment is a decrease in the cur- rent frequency of all behavior that had been so punished. The onset of the MO will have an abative effect with re- spect to the type of behavior that had been punished. However, the observation of such an abative effect would not be possible unless the previously punished behavior was already occurring at a frequency sufficient that its decrease could be observed when the MO for punishment occurred. That is, the observation of the abative effect of an MO for punishment requires the evocative effect of an MO for reinforcement with respect to the punished behavior. In the absence of the MO for reinforcement, there would be no behavior to be abated by the MO for punishment, even though it had such an abative effect.

Suppose that a time-out procedure had been used to punish some behavior that was disruptive to the therapy situation. Only if the MOs relevant to the reinforcers available in the situation were in effect would one expect the time-out to function as punishment. And then only if those MOs were in effect would one expect to see the

397

abative effect of the punishment procedure on the dis- ruptive behavior. But only if the MO for the disruptive be- havior were also in effect would there be any disruptive behavior to be abated. These complex behavioral rela- tions have not received much attention in the conceptual, experimental, or applied literatures, but seem to follow naturally from existing knowledge of reinforcement, pun- ishment, and motivating operations. Behavior analysts should be aware of the possible participation of these be- havioral relations in any situation involving punishment.6

A Complication: Multiple Effects of the Same Variable

Any behaviorally important event typically has more than one effect, and it is important both conceptually and prac- tically that the various effects all be recognized and not confused with each other (Skinner, 1953, pp. 204–224). Multiple effects are apparent in the animal laboratory demonstration of a simple operant chain. A food-deprived rat is taught to pull a cord hanging from the ceiling of the chamber that turns on an auditory stimulus such as a buzzer. In the presence of the buzzer sound the rat is then taught to make a lever press that delivers a food pellet. The onset of the buzzer will now have two obvious op- erant effects: (a) It is an SD that evokes the lever-press response, and (b) it is a conditioned reinforcer that in- creases the future frequency of the cord-pulling response. The first is a behavior-altering evocative effect, and the second is a repertoire-altering reinforcement effect. These effects are in the same direction, an increase in current frequency and an increase in future frequency, although not necessarily for the same type of response.7

Similarly, a stimulus that functions as a discrim- inative stimulus (SD) related to punishment, will have abative effects on the current frequency of some type of response, and will function as a conditioned punisher that decreases the future frequency of the type of response that preceded its onset. Again, the effects are in the same direction, but in this case both are decreases.

6As an aside, it is sometimes argued that behavioral principles are too sim- ple for the analysis of the complexities of human behavior, and therefore some nonbehavioral—usually cognitive—approach is required. It may well be true that the behavioral repertoire of the people who make this argu- ment is too simple for the task. However, the principles themselves are not so simple, as can be seen from the preceding effort to understand the ef- fects of punishment MOs, or from footnote 5 with respect to the learned functions of painful stimulation. 7The buzzer sound will also function as a respondent conditioned eliciting stimulus for smooth muscle and gland responses typically elicited by food in the mouth, and will condition such responses to any other stimulus that is present at that time—an instance of higher order conditioning—but this chapter focuses on operant relations.

Environmental events that function as UMOs, like those that function as SDs, will (in their function as UMOs) typically have behavior-altering effects on the current frequency of a type of behavior, and (as conse- quences) have repertoire-altering effects on the future frequency of whatever behavior immediately preceded the onset of the event. An increase in painful stimulation will, as an MO, increase the current frequency of all be- havior that has alleviated pain, and as a behavioral con- sequence decrease the future frequency of whatever behavior preceded that instance of pain increase. In this case of multiple control, however, the effects will be in opposite directions.

In general, events that have a UMO evocative effect will also function as punishment for the response imme- diately preceding the onset of the event. This statement must be qualified somewhat for events that have such gradual onsets (like food deprivation) that they cannot easily function as response consequences. Table 3 shows these multiple effects for the UMOs that establish the effectiveness of certain events as reinforcers. Events that have a UMO abative effect with respect to current frequency generally do have onsets sufficiently sudden to function as behavioral consequences (e.g., food in- gestion) and will be reinforcers for the immediately pre- ceding behavior.

Applied Implications

Many behavioral interventions involve a manipulation chosen because of (a) its MO value-altering or behavior- altering effect, or (b) its repertoire-altering effect as a re- inforcer or a punisher. But irrespective of the purpose of the manipulation, it is important to be aware that an ef- fect in the opposite direction will also occur, which may or may not be a problem. The fact that reinforcement is also a form of satiation will not be a problem if the rein- forcer magnitude can be quite small. The fact that a sati- ation operation will also reinforce the behavior preceding it will not be a problem if that behavior is not undesirable. The fact that a deprivation operation with respect to an event that will be used productively as a reinforcer could also function as a punisher for the behavior that preceded the deprivation operation will not be a problem if the de- privation onset is very slow, or if the behavior that is pun- ished is not a valuable part of the person’s repertoire.

Table 3 shows that any UMO used to make some event more effective as reinforcement or to evoke the type of behavior that has been reinforced by that event will also function as a punisher for whatever behavior im- mediately preceded the manipulation. Restricting breath- ing ability, making the environment too cold or too warm, or increasing painful stimulation are not likely to

Motivating Operations

398

Motivating Operations

Table 3 Contrasting the Behavior-Altering and Repertoire-Altering Effects of Environmental Events as UMOs and as Punishers

Repertoire-Altering Effect on Future Behavior Environmental Event Evocative Effect on Current Behavior as a UMO as Punishment

Deprivation of food, water, Increases the current frequency of all behavior Should be punishment, but onset is too gradual sleep, activity, or sex that has been reinforced with food, water, sleep, to function as a behavioral consequence

activity, or sex

Oxygen deprivation Increases the current frequency of all behavior The sudden inability to breathe decreases the that has been reinforced with being able to future frequency of the type of behavior that breathe preceded that instance of being unable to

breathe

Becoming too cold Increases the current frequency of all behavior Decreases the future frequency of the type of that has been reinforced by becoming warmer behavior that preceded that instance of

becoming too cold

Becoming too warm Increases the current frequency of all behavior Decreases the future frequency of the type of that has been reinforced by becoming cooler behavior that preceded that instance of

becoming too warm

Increase in painful Increases the current frequency of all behavior Decreases the future frequency of the type of stimulation that has been reinforced with pain reduction behavior that preceded that instance of pain

increase

be used deliberately to control behavior in applied set- tings, but such changes could occur for other reasons (not under the behavior analyst’s control). For this reason it is important to be aware of the two different kinds of ef- fects they could have.

Similar opposite effects can be expected from ma- nipulations that worsen8 the person’s situation in any way, even though the worsening is related to a learning history. Such worsening will establish improvement as a reinforcer and will evoke any behavior that has been so reinforced. The nature of social attention as a reinforcer is unclear as to provenance (Michael, 2000, p. 404), but it would be safe to assume that any manipulation de- signed to increase the effectiveness of attention as a re- inforcer (attention deprivation, for example) would also function as a punisher for the behavior that preceded the manipulation.

Conversely, any operation (e.g., a time-out proce- dure) designed as punishment to decrease the future fre- quency of the behavior preceding the manipulation will

8Worsening here refers to any stimulus change that would function as pun- ishment for behavior that preceded it. The term punishment is not quite appropriate in describing this CMO because reference is not being made to a decrease in future frequency of any behavior. Similarly improvement is used to refer to a change that would function as reinforcement for be- havior that preceded it, but when no reference is being made to an increase in the future frequency of any behavior. Although worsening and improving are useful terms in this context, they are not presented here as technical terms.

also function as an MO in evoking any behavior that has escaped the condition produced by the manipulation.

The behavior analyst should also recognize that any reinforcer-abolishing operation designed to make some event less effective as reinforcement (e.g., a satiation pro- cedure) or to abate the type of behavior that has achieved that type of reinforcement will also function as rein- forcement for the behavior immediately preceding the operation. Food ingestion is an abolishing operation for food as reinforcement and abates any food-reinforced be- havior, but food ingestion also functions as reinforcement for the behavior immediately preceding the food inges- tion. Presenting a high level of noncontingent attention will have a reinforcer-abolishing and an abative effect, but will function as a reinforcer for whatever behavior precedes the operation. Obversely, any operation de- signed to function as reinforcement will also have MO reinforcer-abolishing and behavior-abating effects.

Aversive Stimuli

Environmental events with combined MO evocative ef- fects, repertoire-altering punishment effects, and re- spondent evocative effects with respect to certain smooth muscle and gland responses (heart rate increase, adrenal secretion, etc.) are often referred to as aversive stimuli, where the specific behavioral function [MO, uncondi- tioned stimulus (SP), unconditioned stimulus (US)] is not specified.

399

It is not clear at present just how close the correlation among these several functions is, nor is it clear that the advantages of an omnibus term of this sort outweigh the disadvantage of its lack of specificity. It is clear that some use of the term aversive stimulus is simply a behavioral translation of commonsense expressions for “unpleasant feelings,” “unpleasant states of mind,” and so on—a form of usage that is possibly fostered by the term’s lack of specificity. For these reasons, aversive stimulus has not been used in this chapter to refer to MO or to repertoire- altering variables.

Conditioned Motivating Operations (CMOs) Motivating variables that alter the reinforcing effective- ness of other stimuli, objects, or events, but only as a re- sult of the organism’s learning history, are called conditioned motivating operations (CMOs). As with UMOs, CMOs also alter the momentary frequency of all behavior that has been reinforced by those other events. In commonsense terms, some environmental variables, as a result of our experiences, make us want something different from what we wanted prior to en- countering those variables, and induce us to try to obtain what we now want.

There seem to be at least three kinds of CMOs, all of which were motivationally neutral stimuli prior to their relation to another MO or to a form of reinforcement or punishment. Depending on their relation to the behav- iorally significant event or condition, the three kinds of conditioned motivating operations are classified as surrogate, reflexive, or transitive. The surrogate CMO (CMO-S) accomplishes what the MO it was paired with accomplishes (is a surrogate for that MO), the reflexive CMO (CMO-R) alters a relation to itself (makes its own removal effective as reinforcement), and the transitive CMO (CMO-T) makes something else effective as re- inforcement (rather than altering itself).

Surrogate CMO (CMO-S): A Stimulus That Has Been Paired with Another MO

CMO-S Description

The respondent conditioned stimulus (CS), operant con- ditioned reinforcer (Sr), and operant conditioned punisher (Sp) are each stimuli that acquired a form of behavioral effectiveness by being paired with a behaviorally effec- tive stimulus. It is possible that stimuli that are paired

9The neutral stimulus could be correlated with a CMO rather than a UMO, with the same transfer of effects.

with a UMO9 will become capable of the same value- altering and behavior-altering effects as that UMO. With respect to its MO characteristics, such a stimulus will be called a surrogate CMO, or a CMO-S.

This relation would be illustrated if stimuli that had been temporally related to decreases in temperature would have MO effects similar to the temperature de- creases themselves. That is, in the presence of such stim- uli, a temperature increase would be a more effective reinforcer, and behavior that had produced such an in- crease would occur at a higher frequency than would be appropriate for the actual temperature. Research in this area is discussed at length in Michael, 1993 (pp. 199–202), but will not be reviewed here. Suffice it to say that the evidence for such an effect is not strong. Also, the existence of this type of learned motivating operation is somewhat problematic from an evolutionary perspective (Mineka, 1975). Behaving as though an MO were in ef- fect when it was not would seem to be opposed to the or- ganism’s best interest for survival. Trying to get warmer than is necessary for the existing temperature condition would not seem healthy, and such behavior might dis- place more important behavior. However, evolution does not always work perfectly.

With sexual motivation, MOs for aggressive behav- ior, and the other emotional MOs, the issue has not been addressed in terms specific to the CMO because its dis- tinction from CS, Sr, and Sp has not been previously em- phasized. The surrogate CMO is only just beginning to be considered within applied behavior analysis (see McGill, 1999, p. 396), but its effects could be quite prevalent. From a practical perspective, it may be helpful to con- sider the possibility of this type of CMO when trying to understand the origin of some puzzling or especially ir- rational behavior.

Weakening the Effects of Surrogate CMOs

Any relationship developed by a pairing procedure can generally be weakened by two kinds of unpairing, pre- senting the previously neutral stimulus without the pre- viously effective stimulus, or presenting the effective stimulus as often in the absence as in the presence of the previously neutral one. For example, if the CMO-S, let’s say a visual stimulus, that had often been paired with ex- treme cold now occurs frequently in normal temperature, its value-altering and behavior-altering effects would be weakened. Similarly, if extreme cold now occurred just as often in the absence of the CMO-S as in its presence, the effectiveness of the CMO-S would be reduced.

Motivating Operations

400

Motivating Operations

As mentioned earlier, the CMO-S is just beginning to be dealt with in applied behavior analysis (e.g., see McGill, 1999, p. 396), but its possible relation to problem behavior means that the behavior analyst should know how to weaken such CMOs.

Reflexive CMO (CMO-R): A Stimulus That Has Systematically Preceded Some Form of Worsening or Improvement

CMO-R Description

In the traditional “discriminated avoidance procedure,”10

an intertrial interval is followed by the onset of an ini- tially neutral warning stimulus, which is in turn followed by the onset of painful stimulation—usually electric shock. Some arbitrary response (one that is not part of the animal’s phylogenic pain–escape repertoire), such as lever pressing, terminates the painful stimulation (the an- imal escapes the pain) and restarts the intertrial interval. The same response, if it occurs during the warning stim- ulus, terminates that stimulus and the shock does not occur on that trial. The response at this phase of the pro- cedure is said to have avoided the pain and is called an avoidance response. As a result of exposure to this pro- cedure, many organisms learn to emit the relevant re- sponse during most of the warning stimulus occurrences, and thus receive very few of the shocks.

Recall the analysis of the role of the shock as an MO for the escape response, the reinforcement for which is shock termination. The warning stimulus has a similar function, except that its capacity to establish its own ter- mination as an effective form of reinforcement is of on- togenic provenance—as a result of the individual’s own history involving the relation of the warning stimulus to the onset of the painful stimulus. In other words, the warning stimulus evokes the so-called avoidance response as a CMO, just as the painful stimulation evokes the es- cape response as a UMO. In neither case is the relevant stimulus related to the availability of the response con- sequence, but rather to its reinforcing effectiveness.

In more general terms, any stimulus that systemati- cally precedes the onset of painful stimulation becomes a CMO-R, in that its own offset will function as a rein- forcer, and its occurrence will evoke any behavior that has been followed by such reinforcement. This set of functional relations is not limited to painful stimulation as a form of worsening (or even to worsening in general,

10The term discriminated arose so that this type of procedure could be dis- tinguished from an avoidance procedure with no programmed exterocep- tive stimulus except for the shock itself (also sometimes called avoidance without a warning stimulus).

as will be seen later). It is well known that organisms can learn to terminate stimuli that warn of stimulus changes other than the onset of pain—stimuli that warn of a low- ered frequency of food presentation, increased effort, a higher response ratio requirement, longer delays to food, and so forth. Such events have in common some form of worsening, and stimuli related to such events are often called conditioned aversive stimuli without specifying any particular behavioral function.

It may be useful to repeat the argument against such stimuli being considered discriminative stimuli (SDs). A discriminative stimulus is related to the current avail- ability of a type of consequence for a given type of be- havior. Availability has two components: (a) An effective consequence (one whose MO is currently in effect) must have followed the response in the presence of the stimu- lus; and (b) the response must have occurred without the consequence (which would have been effective as a re- inforcer if it had been obtained) in the absence of the stimulus. The relation between the warning stimulus and consequence availability does not meet the second com- ponent. In the absence of the warning stimulus, there is no effective consequence that could have failed to follow the response in an analog to the extinction responding that occurs in the absence of an SD. The fact that the avoidance response does not turn off the absent warning stimulus is in no sense extinction responding, but rather is behaviorally neutral, like the unavailability of food re- inforcement for a food-satiated organism.

Now consider a stimulus that is positively correlated with some form of improvement. Its CMO-R effects occur if the stimulus establishes its own offset as an ef- fective punisher and abates any behavior that has been so punished. The relation is quite plausible, although there seems to have been little directly relevant research.

Human Examples of the CMO-R

The CMO-R plays an important role in identifying a neg- ative aspect of many everyday interactions that might seem free from any deliberate aversiveness by one person toward another. Typically such interactions are interpreted as a sequence of discriminative stimuli, a sequence of op- portunities for each participant to provide some form of positive reinforcement to the other person.

Reactions to Mands. Imagine that a stranger asks you where a particular building on campus is lo- cated, or asks for the time. (Questions are usually mands for verbal action.) The appropriate response is to give the information quickly or to say that you don’t know. Typically the person who asked will smile and thank you for the information. Also your question may be reinforced by the knowledge that a

401

fellow human being has been helped. In a sense the question is an opportunity to obtain these reinforcers that were not available before the question. However, the question also begins a brief period that can be con- sidered a warning stimulus, and if a response is not made soon, a form of social worsening will occur. The asker may repeat the question, stating it more clearly or more loudly, and will certainly think you are strange if you do not respond quickly. You too would consider it socially inappropriate for you to provide no answer. Even when no clear threat for nonresponding is implied by the person who asked, our social history under such conditions implies a form of worsening for continued inappropriate behavior. Many such situations probably involve a mixture of the positive and negative compo- nents, but in those cases in which answering the ques- tion is an inconvenience (e.g., the listener is in a hurry), the asker’s thanks is not a strong reinforcer, nor is help- ing one’s fellow human being. The reflexive CMO is probably the main controlling variable.

Complication with Respect to Stimulus Termination

In the typical laboratory avoidance procedure, the re- sponse terminates the warning stimulus. In extending this type of analysis to the human situation, it must be recog- nized that the warning stimulus is not simply the event that initiated the interaction. In the previous example, the reflexive CMO is not the vocal request itself, which is too brief to be actually terminated. It is, instead, the more complex stimulus situation consisting of having been asked and not having made a response during the time when such a response would be appropriate. The termi- nation of that stimulus situation is the reinforcement for the response. Some social interactions with a stimulus— a facial expression, an aggressive posture—are more like the warning stimulus of the animal laboratory that can be terminated by the avoidance response, but most involve the more complex stimulus condition consisting of the request and the following brief period described earlier.

Thanks and You’re Welcome. When a person does something for another person that is a kindness of some sort, it is customary to thank the person. What evokes the thanking response, and what is its reinforcement? It is clearly evoked by the person’s performing the favor or the kindness. Should performing the favor be consid- ered purely an SD, in the presence of which one can say “thanks” and receive the reinforcement consisting of the other person saying “you’re welcome”? In many cases, ordinary courteous remarks may involve a CMO- R component. Consider the following scenario. Person A has his arms full carrying something out of the build-

11This procedure is often incorrectly referred to as extinguishing the avoid- ance response, but a true avoidance extinction procedure requires response occurrence without termination of the warning stimulus.

ing to his car. As he approaches the outer door, Person B opens the door and holds it open while Person A goes out. Person A then usually smiles and says “Thanks”. The CMO-R component can be illustrated by suppos- ing that Person A just walks out without acknowledging the favor. In such circumstances it would not be un- usual for Person B to call out sarcastically, “You’re welcome!” Someone’s doing a favor for someone else is a warning stimulus (a CMO-R) that has systemati- cally preceded some form of disapproval if the favor is not acknowledged in some way.

In applied behavior analysis, the CMO-R is often part of procedures for training or teaching individuals with defective social and verbal repertoires. In language training programs, for example, learners are typically asked questions or given verbal instructions that clearly function as reflexive CMOs, which will be followed by further intense social interaction if they do not respond to them appropriately. Such questions and instructions may well be functioning primarily as reflexive CMOs, rather than SDs related to the possibility of receiving praise or other positive reinforcers. Although it may not be possi- ble to completely eliminate this type of aversiveness, it is important to understand its nature and origin.

Weakening the Effects of Reflexive CMOs

Extinction and two forms of unpairing can weaken the effects of CMO-Rs. Extinction consists of the occurrence of a response without its reinforcement. The reinforce- ment for the response evoked by the CMO-R is the ter- mination of the warning stimulus. When the response occurs repeatedly without terminating the warning stim- ulus, and the ultimate worsening occurs when the relevant time period elapses, the response will be weakened as with any extinction procedure.

Two forms of unpairing will also weaken the reflex- ive CMO relation. One involves the nonoccurrence of the ultimate form of worsening when the warning stimulus is not terminated.11 This type of unpairing weakens the CMO relation by weakening the reinforcement that con- sists of warning-stimulus termination. Warning-stimulus termination is only reinforcing to the extent that the warning-stimulus-off condition is an improvement over the warning-stimulus-on condition. When warning- stimulus-on is not followed by the ultimate worsening, it becomes no worse than warning-stimulus-off, and the re- inforcement for the avoidance response decreases.

The other type of unpairing occurs when the re- sponse continues to terminate the warning stimulus, but

Motivating Operations

402

Motivating Operations

the ultimate worsening occurs anyway when it would have occurred if the warning stimulus had not been ter- minated. In this case the warning-stimulus-off condition becomes just as bad as the warning-stimulus-on condi- tion, and again the reinforcement for the avoidance re- sponse decreases.

In the usual academic demand situation with some developmentally disabled individuals, typical problem behavior (e.g., tantruming, self-injury, aggressive be- havior) is sometimes evoked by the early phases of the de- mand sequence and reinforced by terminating the early phase and not progressing to the later and possibly more demanding phases. Assuming that the ultimate phases of the demand sequence must occur because of the impor- tance of the relevant repertoire being taught, and assum- ing that they cannot be made less aversive, then extinction of the problem behavior is the only procedure that is of practical value. This would consist of continuing the de- mand sequence irrespective of the occurrence of prob- lem behavior. Neither of the unpairing procedures would be effective. The first would not result in any training, and the other would result in the problem behavior oc- curring as soon as the later phases of training began.

But of course, one should not assume that the ulti- mate phases of the demand cannot be made less aversive. Increasing instructional effectiveness will result in less failure, more frequent reinforcement, and other general improvements in the demand situation to the point at which it may function as an opportunity for praise, edi- bles, and so forth, rather than a demand.

Transitive CMO (CMO-T): A Stimulus That Alters the Value of Another Stimulus

CMO-T Description

When an environmental variable is related to the relation between another stimulus and some form of improve- ment, the presence of that variable functions as a transi- tive CMO, or CMO-T, to establish the second condition’s reinforcing effectiveness and to evoke the behavior that has been followed by that reinforcer. All variables that function as UMOs also function as transitive CMOs for the stimuli that are conditioned reinforcers because of their relation to the relevant unconditioned reinforcer. Consider the simple operant chain described earlier: A food-deprived rat pulls a cord that turns on a buzzer sound. In the presence of the buzzer sound the rat emits some other response that causes the delivery of a food pellet. Food deprivation makes food effective as an un- conditioned reinforcer, a relation that requires no learn- ing history. Food deprivation also makes the buzzer sound effective as a conditioned reinforcer, which clearly does

12This scenario was first described in Michael, 1982. At that time CMO-T was called an establishing stimulus, or SE.

require a learning history. Thus, food deprivation is a UMO with respect to the reinforcing effectiveness of food, but a CMO-T with respect to the reinforcing effec- tiveness of the buzzer sound. In the human situation food deprivation establishes not only food as a reinforcer, but also all of the stimuli that have been related to obtaining food—an attentive server in a restaurant, a menu, the utensils with which one transports the food to the mouth, and so forth.

Understanding transitive CMOs that result from UMOs in this way requires no special knowledge beyond what is needed to understand the effects of UMOs. Also, the evocative effect of such transitive CMOs is not easily confused with the evocative effect of an SD. If one can see food deprivation as an MO (rather than an SD) with re- spect to the behavior that has been reinforced with food, then its function as an MO (rather than an SD) with respect to behavior that has been reinforced with the various food- related conditioned reinforcers is an easy extension.

The reinforcing effectiveness of many (probably most) conditioned reinforcers is not only altered by rel- evant UMOs as described previously, but also dependent on other stimulus conditions because of an additional learning history. This notion underlies the fact that con- ditioned reinforcing effectiveness is often said to be de- pendent on a “context.” When the context is not appropriate, the stimuli may be available but are not ac- cessed because they are not effective as reinforcers in that context. A change to an appropriate context will evoke behavior that has been followed by those stimuli, which are now effective as conditioned reinforcers. The occurrence of the behavior is not related to the availabil- ity, but rather to the value, of its consequence. For ex- ample, flashlights are usually available in home settings, but are not accessed until a power failure makes them valuable. In this sense the power failure (the sudden dark- ness) evokes the behavior that has obtained the flashlight in the past (rummaging around in a particular drawer). The motivative nature of this CMO-T relation is not widely recognized, and the evocative variable (the sudden darkness) is usually interpreted as an SD.

Human CMO-T Examples

Consider a workman disassembling a piece of equipment, with his assistant handing him tools as he requests them.12

The workman encounters a slotted screw that must be re- moved and requests a screwdriver. The sight of the screw evoked the request, the reinforcement for which is re- ceiving the tool. Prior to an analysis in terms of the CMO- T, the sight of the screw would have been considered an

403

SD for the request, but such screws have not been differ- entially related to the availability of reinforcement for re- quests for tools. In the typical workman’s history, assistants have generally provided requested tools irre- spective of the stimulus conditions in which the request occurred. The sight of the screw is more accurately in- terpreted as a CMO-T for the request, not as an SD.

The fact that several SDs are involved in this com- plex situation makes the analysis more difficult. The screw is an SD for unscrewing movements (with a screw- driver in hand). The verbal request, although evoked by the sight of the slotted screw as a CMO-T, is dependent on the presence of the assistant as an SD. The offered screwdriver is an SD for reaching movements. The criti- cal issue, however, is the role of the screw in evoking the request, and this is a motivating rather than a discrimi- native relation.

Another common human example involves a stim- ulus related to some form of danger that evokes some relevant protective behavior. A night security guard is patrolling an area and hears a suspicious sound. He pushes a button on his phone that signals another secu- rity guard, who then activates his own phone and asks if help is needed (which reinforces the first guard’s call). The suspicious sound is not an SD in the presence of which the second security guard’s response is more available, but rather a CMO-T in the presence of which it is more valuable. SDs are involved, however. A phone’s ringing is an SD in the presence of which one has activated the phone, said something into the receiver, and been reinforced by hearing a response from another person. Answering phones that are not ringing has typ- ically not been so reinforced. (Note, incidentally, that the effect of the danger signal is not to evoke behavior that produces its own termination as with the CMO-R, but rather behavior that produces some other event, in this case, the sound of the security guard’s colleague offering to help.)

For an animal analog of the CMO-T13 consider a food-deprived monkey in a chamber with a retractable lever and a chain hanging from the ceiling. Pulling the chain causes the lever to come into the chamber for 5 seconds. If a light (on the wall of the chamber) is on, a lever press delivers a food pellet, but if the light is off, the lever press has no effect. The light comes on and goes off on a random time basis, unrelated to the monkey’s behavior. Note that the chain pull will cause the lever to come into the chamber for 5 seconds irrespective of the condition of the light. A well-trained monkey’s chain-

13This animal example was first described in Michael, 1982. The language has been changed to be more in line with current terminology.

pulling behavior would be infrequent when the light is off, but evoked by the light onset. The light onset would be the CMO-T in this situation, like the slotted screw or the suspicious sound in the previous examples.

Until recently most behavior analysts interpreted transitive CMOs as SDs. The distinction hinges on the re- lation between reinforcer availability and the presence or absence of the stimulus. If the reinforcer is more available in the presence than in the absence of the stimulus, the stimulus is an SD; if it is just as available in the absence as in the presence of the stimulus, the stimulus is a CMO- T. Screwdrivers have typically been just as available in the absence as in the presence of screws. The response by the security guard’s colleague has been just as available in the absence as in the presence of a suspicious noise. The retractable lever was just as available to the monkey (for pulling the chain) in the absence as in the presence of the light.

Weakening the Effects of Transitive CMOs

The evocative effect of the CMO-T can be temporarily weakened by weakening the MO related to the ultimate outcome of the sequence of behaviors. Consider the ex- ample of the workman requesting a screwdriver and also the example of the monkey pulling the chain to cause the lever to come into the chamber. Temporary weakening of the CMO relation could be accomplished by elimi- nating the reason for doing the work—the workman is told that the equipment does not have to be disassem- bled, and the monkey is fed a large amount of food prior to being placed in the experimental chamber. Of course, the next time the workman saw a screw that had to be unscrewed, he would again ask his assistant for a screw- driver. When the monkey was again food deprived in the chamber and the light came on, the chain-pulling re- sponse would occur.

More permanent weakening can be accomplished by an extinction procedure and by two kinds of unpairing. To extinguish the request that is evoked by the slotted screw, something in the environment would have to change so that such requests are no longer honored (e.g., assistants now believe that workmen should get their tools them- selves). In the monkey example, the chain pull would no longer cause the lever to come into the chamber. One type of unpairing would be illustrated if screwdrivers no longer worked to unscrew slotted screws if, say, all such screws were now welded. The workman can still obtain a screw- driver by asking, but the value of the screwdriver is ulti- mately lost because such a tool wouldn’t work anymore. In the monkey scenario, the monkey can still cause the lever to come into the chamber by pulling the chain, but pressing the lever does not deliver food. A second kind of

Motivating Operations

404

unpairing would be illustrated if construction practices changed so that slotted screws could now be easily un- screwed by hand as well as with a screwdriver, or if the lever press delivers food when the light is off as well as when it is on.

Importance of the CMO-T for Language Training

It is increasingly being recognized that mand training is an essential aspect of language programs for individuals with severely defective verbal repertoires. For such indi- viduals, manding does not spontaneously arise from tact and receptive language training. The learner has to want something, make an appropriate verbal response, and be reinforced by receiving what was wanted. With this pro- cedure the response comes under the control of the rele- vant MO. UMOs can be taken advantage of to teach mands for unconditioned reinforcers, but this is a rela- tively small repertoire. Use of a CMO-T, however, is a way to make the learner want anything that can be a means to another end. Any stimulus, object, or event can be a basis for a mand simply by arranging an environ- ment in which that stimulus can function as a conditioned reinforcer. Thus, if a pencil mark on a piece of paper is required for an opportunity to play with a favored toy, a mand for a pencil and for a piece of paper can be taught.

Practical Implications of the CMO-T in General

A CMO-T is a stimulus onset that evokes behavior because of its relation to the value of a consequence rather than to the availability of a consequence. This distinction must be relevant in subtle ways to the effective understanding and manipulation of behavioral variables for a variety of prac- tical purposes. Two forms of behavioral control, the SD

and the CMO-T, which are so different in origin, would be expected to differ in other important ways. This issue is an example of terminological refinement, not a discovery of any new empirical relations. The value of this refine- ment, should it have value, will be found in the improved theoretical and practical effectiveness of those whose ver- bal behavior has been affected by it.

General Implications of Motivating Operations for Behavior Analysis Behavior analysis makes extensive use of the three-term contingency relation involving stimulus, response, and con- sequence. However, the reinforcing or punishing effec- tiveness of the consequence in developing control by the stimulus depends on an MO, and the future effectiveness of the stimulus in evoking the response depends on the presence of the same MO in that future condition. In other words, the three-term contingency cannot be fully under- stood, or most effectively used for practical purposes, with- out a thorough understanding of motivating operations.

Summary

Motivating Operations

Definition and Characteristics of Motivating Operations

1. A motivating operation (MO) (a) alters the effectiveness of some stimulus as a reinforcer, the value-altering effect; and (b) alters the current frequency of all behavior that has been reinforced by that stimulus, the behavior-altering effect.

2. The value-altering effect is either (a) an increase in the re- inforcing effectiveness of some stimulus, in which case the MO is an establishing operation (EO); or (b) a decrease in reinforcing effectiveness, in which case the MO is an abolishing operation (AO).

3. The behavior-altering effect is either (a) an increase in the current frequency of behavior that has been reinforced by some stimulus, called an evocative effect; or (b) a decrease in the current frequency of behavior that has been rein- forced by some stimulus, called an abative effect.

4. The alteration in frequency can be (a) the direct evocative or abative effect of the MO on response frequency and/or (b) the indirect effect on the evocative or abative strength of relevant discriminative stimuli (SDs).

5. In addition to frequency, other aspects of behavior such as response magnitude, latency, and relative frequency can be altered by an MO.

6. It is not correct to interpret the behavior-altering effect of an MO as due to the organism’s encountering a more or less effective form of reinforcement; a strong relation ex- ists between MO level and responding when no reinforcers are being received.

7. Behavior-altering versus function-altering effects: MOs and SDs are antecedent variables that have behavior- altering effects. Reinforcers, punishers, or the occurrence of a response without its reinforcer (extinction procedure)

405

or without its punisher (recovery from punishment proce- dure) are consequences that change the organism’s reper- toire so that it behaves differently in the future. SDs and MOs alter the current frequency of behavior, but rein- forcers, punishers, and response occurrence without con- sequence alter the future frequency of behavior.

A Critical Distinction: Motivative versus Discriminative Relations

8. An SD controls a type of behavior because it has been re- lated to the differential availability of an effective rein- forcer for that type of behavior. This means that the relevant consequence has been available in the presence of, and unavailable in the absence of, the stimulus. Most vari- ables that qualify as motivating operations fail to meet this second SD requirement because in the absence of the vari- able, there is no MO for the relevant reinforcer, and thus no reinforcer unavailability.

9. A useful contrast: SDs are related to the differential avail- ability of a currently effective form of reinforcement for a particular type of behavior; MOs are related to the differ- ential reinforcing effectiveness of a particular type of en- vironmental event.

Unconditioned Motivating Operations (UMOs)

10. The main UMOs for humans are those related to depriva- tion and satiation with respect to food, water, oxygen, ac- tivity, and sleep; and those related to sexual reinforcement, comfortable temperature conditions, and painful stimula- tion. For each variable, there are two MOs, one with an es- tablishing operation (EO), and one with an abolishing operation (AO). Also, each variable has an evocative effect and an abative effect. Thus food deprivation is an EO and has evocative effects on relevant behavior, and food ingestion is an AO and has abative effects on relevant behavior.

11. The cognitive interpretation of behavior-altering effects is that the person understands (i.e., can verbally describe) the situation and then behaves appropriately as a result of that understanding. But in fact, reinforcement automati- cally adds the reinforced behavior to the repertoire that will be evoked and abated by the relevant UMO; the per- son does not have to “understand” anything for an MO to have its effects. Two different kinds of ineffectiveness can result from this misinterpretation. There may be insuffi- cient effort to train individuals who have very limited ver- bal repertoires, and inadequate preparation for an increase in problem behavior that preceded reinforcement.

12. The role of stimulus conditions in the generality of train- ing effects is well known, but unless the MO for the rein- forcers that were used in training is also in effect, the trained behavior will not occur in the new conditions.

13. Temporary weakening of EO effects can be accomplished by the relevant AO and abative operations. For example, undesirable behavior based on food deprivation can be abated by food ingestion, but the behavior will come back

when deprivation is again in effect. More permanent weak- ening of behavior-altering effects can be accomplished by an extinction procedure (i.e., letting undesired behavior evoked by the MO occur without reinforcement).

14. A variable that alters the punishing effectiveness of a stim- ulus, object, or event, and alters the frequency of the be- havior that has been so punished, is an MO for punishment. If the value-altering effect does not depend on a learning history, then the variable is a UMO. An increase in pain will function as punishment as long as the current level is not so high that an increase cannot occur. If a stimulus is a punisher because of its relation to a reduced availability of a reinforcer, then the MO for that reinforcer is the MO for the punisher. Thus the MO for food removal as a pun- isher is food deprivation.

15. Social disapproval, time-out from reinforcement, and re- sponse cost are stimulus conditions that usually function as punishment because they are related to a reduction in the availability of some kinds of reinforcers. The MOs for those forms of punishment are the MOs for the reinforcers that are being made less available.

16. Multiple effects: Environmental events that function as UMOs will typically have behavior-altering effects on the current frequency of a type of behavior, and (as conse- quences) function-altering effects with respect to the future frequency of whatever behavior immediately preceded the onset of the event.

17. A behavioral intervention is often chosen because of (a) its MO behavior-altering effect, or (b) its repertoire-altering effect (as a reinforcer or a punisher). But irrespective of the purpose of the intervention, an effect in the opposite di- rection from the targeted one will also occur, and should be planned for.

Conditioned Motivating Operations (CMOs)

18. Motivating variables that alter the reinforcing effective- ness of other stimuli, objects, or events, but only as a re- sult of the organism’s learning history, are called conditioned motivating operations (CMOs). As with the UMOs, they also alter the momentary frequency of all be- havior that has been reinforced (or punished) by those other events.

19. The surrogate CMO (CMO-S) is a stimulus that acquires its MO effectiveness by being paired with another MO, and has the same value-altering and behavior-altering ef- fects as the MO with which it was paired.

20. A stimulus that acquires MO effectiveness by preceding some form of worsening or improvement is called a re- flexive CMO (CMO-R). It is exemplified by the warning stimulus in a typical escape–avoidance procedure, which establishes its own offset as reinforcement and evokes all behavior that has accomplished that offset.

21. In evoking the avoidance response, the CMO-R has usually been interpreted as an SD. The CMO-R fails to qualify as

Motivating Operations

406

Motivating Operations

an SD, however, because in its absence there is no MO for a reinforcer that could be unavailable and thus no rein- forcer unavailability. It clearly qualifies as an MO and, be- cause its MO characteristics depend on its learning history, as a CMO.

22. The CMO-R identifies a negative aspect of many everyday interactions that otherwise would be interpreted as a se- quence of opportunities for positive reinforcement. One example is a request for information, which initiates a brief period during which a response must be made to termi- nate a period of increasing social awkwardness.

23. The CMO-R is often an unrecognized component of pro- cedures used in teaching effective social and verbal be- havior. Learners are asked questions or given instructions, which are followed by further intense social interaction if they do not respond to them appropriately. The question or instruction may be functioning more as a warning stimu- lus, as a CMO-R, than as an SD related to an opportunity for receiving praise or other positive reinforcers.

24. The CMO-R can be weakened by extinction if the response occurs without terminating the warning stimulus (e.g., con- tinuing the demand sequence irrespective of the occur- rence of problem behavior), or by two kinds of unpairing (the ultimate worsening fails to occur or occurs irrespec- tive of the avoidance response).

25. An environmental variable that establishes (or abolishes) the reinforcing effectiveness of another stimulus and evokes (or abates) the behavior that has been reinforced by that other stimulus is a transitive CMO, or CMO-T.

26. Variables that function as UMOs also function as transitive CMOs for the stimuli that are conditioned reinforcers be- cause of their relation to the unconditioned reinforcer. Food deprivation (as a UMO) establishes as a reinforcer not only food but also (as a CMO-T) all of the stimuli that have been related to obtaining food (e.g., the utensils with which one transports food to the mouth).

27. The reinforcing effectiveness of many conditioned rein- forcers is not only altered by relevant UMOs, but also may be dependent on other stimulus conditions because of an additional learning history. Those stimulus conditions, as CMO-Ts, then also evoke behavior that has obtained the conditioned reinforcers. The occurrence of the behavior is not related to the availability, but rather to the value, of its consequence.

28. A model for the CMO-T is a feature of the environment that must be manipulated with a tool—evoking behavior that obtains that tool—for example, requesting it of an- other person. Another human example is a stimulus re- lated to some form of danger evoking the relevant protective behavior.

29. The evocative effect of the CMO-T can be temporarily weakened by weakening the MO related to the ultimate outcome of the sequence of behaviors (e.g., the job related to the request for a tool is no longer necessary). More per- manent weakening can be accomplished by an extinction procedure (e.g., requests for tools are no longer honored) and by two kinds of unpairing (e.g., the tool no longer ac- complishes the task, or the task can be accomplished with- out the tool).

30. The CMO-T is especially valuable in language programs for teaching mands. It is a way to make the learner want anything that can be a means to another end, and then to reinforce an appropriate mand with the object that is wanted.

General Implications of Motivating Operations for Behavior Analysis

31. The three-term contingency of stimulus, response, and consequence is an essential component of behavior analy- sis, but this relation cannot be fully understood, or most ef- fectively used, without a thorough understanding of motivating operations.

407

Stimulus Control

Key Terms antecedent stimulus class arbitrary stimulus class concept formation discriminative stimulus (SD) feature stimulus class

matching-to-sample reflexivity stimulus control stimulus delta (SΔ) stimulus discrimination training

stimulus equivalence stimulus generalization stimulus generalization gradient symmetry transitivity

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes and Concepts

3-2 Define and provide examples of stimulus and stimulus class.

3-7 Define and provide examples of stimulus control.

3-8 Define and provide examples of establishing operations.

3-12 Define and provide examples of generalization and discrimination.

Content Area 9: Behavior Change Procedures

9-7 Use discrimination training procedures.

9-8 Use prompt and prompt fading.

9-21 Use stimulus equivalence procedures.

408

Stimulus Control

Reinforcement of an operant response increases the frequency of future responding and influ- ences the stimuli that immediately precede the

response. The stimuli preceding the response (i.e., an- tecedent stimuli) acquire an evocative effect on the rele- vant behavior. In a typical laboratory demonstration of operant conditioning, a rat is placed in an experimental chamber and given the opportunity to press a lever. Con- tingent on a lever press the rat receives a food pellet. Re- inforcement of the lever press increases the frequency of lever pressing.

Researchers can make this simple demonstration more complex by manipulating other variables. For example, oc- casionally a buzzer might sound, and the rat receives a food pellet only when the lever is pressed in the presence of the buzzer. The buzzer sound preceding the lever press is called a discriminative stimulus (SD, pronounced “ess-dee”). With some experience, the rat will make more lever presses in the presence of the buzzer sound (SD) than in its absence, a condition called stimulus delta (SΔ, pronounced “ess- delta”). Behavior that occurs more often in the presence of an SD than in its absence is under stimulus control. Tech- nically, stimulus control occurs when the rate, latency, du- ration, or amplitude of a response is altered in the presence of an antecedent stimulus (Dinsmoor, 1995a, b). A stimu- lus acquires control only when responses emitted in the presence of that stimulus produce reinforcement more often than responses in the absence of stimulus.

Stimulus control should not be viewed as just an in- teresting procedure for laboratory demonstrations. Stim- ulus control plays a fundamental role in everyday complex behaviors (e.g., language systems, conceptual behavior, problem solving), education, and treatment (Shahan & Chase, 2002; Stromer, 2000). People do not answer the telephone in the absence of a ring. A person driving a car stops the car more often in the presence of a red traffic light than in its absence. People who use both Spanish and English languages will likely use Spanish, not English, to communicate with a Spanish-speaking audience.

Behaviors considered inappropriate in one context are accepted as appropriate when emitted in another context. For example, teachers will accept loud talking as appropriate on the playground, but not in the class- room. Arriving 15 to 20 minutes late is appropriate for a party, but not for a job interview. Some behaviors that parents, teachers, and society in general may call inap- propriate are not behavior problems per se. The problem is emitting behaviors at a time or in a place or circum- stance that is deemed inappropriate by others. This rep- resents a problem of stimulus control, and it is a major concern for applied behavior analysts. This chapter ad- dresses the factors related to the development of stim- ulus control.

Antecedent Stimuli

Stimulus control of an operant response appears similar to the control of respondent behavior by a conditioned stimulus. The SD and the conditioned stimulus are an- tecedent stimuli that evoke the occurrence of behavior. Applied behavior analysts, however, need to distinguish between the function of an SD for operant behavior and a conditioned stimulus for respondent conditioning; this is a crucial distinction for understanding environmental control of operant behavior. In a typical laboratory demonstration of respondent conditioning, the experi- menter presents food to a dog. The food functions as an unconditioned stimulus to elicit the unconditioned re- sponse—salivation. The experimenter then introduces a buzzer sound (a neutral stimulus). The buzzer sound does not elicit salivation. Then after several occurrences of pairing the buzzer sound with food delivery, the buzzer sound becomes a conditioned stimulus that will elicit sali- vation (a conditioned response) in the absence of food (an unconditioned stimulus).

Laboratory experiments on operant and respondent conditioning have demonstrated consistently that ante- cedent stimuli can acquire control over a behavior. A buzzer sounds, and a rat presses a bar. A buzzer sounds, and a dog salivates. Despite the similarities, the bar press is an operant behavior, salivation is a respondent behavior, and the manner in which the SD and the con- ditioned stimulus acquire their controlling functions is very different. An SD acquires its controlling function through association with stimulus changes that occur immediately following behavior. Conversely, a condi- tioned stimulus acquires its controlling function through association with other antecedent stimuli that elicit the behavior (i.e., an unconditioned stimulus or conditioned stimulus).

The environment contains many forms of energy that a person can perceive. Evolutionary adaptation to the en- vironment has provided organisms with anatomical struc- tures (i.e., organ receptors) that detect these forms of energy. For example, the eye detects electromagnetic ra- diation; the ear, air-pressure vibrations; the tongue and nose, chemical energies; skin receptors, mechanical pres- sure and thermal changes (Michael, 1993).

Applied behavior analysts use the physical properties of a stimulus to investigate its effect on behavior. The physical energy, however, must relate to the sensory ca- pabilities of the organism. For instance, ultraviolet radi- ation is physical energy, but ultraviolet radiation does not function as a stimulus for humans because that energy does not result in any relationship with operant behav- ior. Ultraviolet radiation would function as a stimulus for humans in the presence of a special device that detects the radiation. A dog whistle functions as a stimulus for dogs

409

Stimulus Control

but not for humans. Dogs can hear the air pressure vi- brations of the whistle, but humans cannot. Any form of physical energy capable of detection by the organism can function as a discriminative stimulus.

Discriminative and Motivational Functions of Stimuli

Discriminative stimuli and establishing operations share two important similarities: (a) both events occur before the behavior of interest, and (b) both events have evoca- tive functions. To evoke behavior means to occasion be- havior by calling it up or producing it. Distinguishing the nature of the antecedent control is often difficult. Was the behavior evoked by an SD, an establishing operation (EO), or both?

In some situations, an antecedent stimulus change alters the frequency of a response and appears to have an SD effect. For example, in a typical shock–escape pro- cedure an animal is placed in an experimental chamber. Shock is administered until a response removes the shock for a designated period of time. Then the shock is reintroduced until it is again terminated with a response, and so on. An experienced animal removes the shock immediately. In such a situation some would say that the shock serves as an SD. The shock, an antecedent stimu- lus, evokes a response that is negatively reinforced (the shock is removed). In this situation, however, shock does not function as an SD. A response in the presence of an SD must produce more frequent reinforcement than it does in its absence. Even though the animal receives re- inforcement by removing the shock, the absence of shock does not constitute a state of lower frequency reinforce- ment. Before the response can be reinforced, the shock must be on. Shock in this example is functioning as an EO because it changes what functions as reinforcement rather than the availability of reinforcement (Michael, 2000). Often the apparent SD effect does not have a his- tory of effective differential reinforcement correlated with the altered frequency of response. These situations are probably related to motivating operations (MOs) rather than stimulus control.

The following scenario puts Michael’s laboratory ex- ample into an applied context: A teacher requests a re- sponse from a student. The student emits aggressive behaviors immediately following the request. The ag- gressive behaviors remove the request. Later, the teacher again requests a response, and the cycle of aggressive be- haviors and removal of the request continues. As in the laboratory example, some would say that the requests serve as an SD. The request, an antecedent stimulus, evokes aggression that is negatively reinforced (the re- quest is removed). In this applied example, the request

is an EO that evokes the aggressive behaviors rather than an SD. It makes no sense to talk about an SD evoking the aggressive behavior in the absence of the request (McGill, 1999), just as it makes no sense to talk about an SD evok- ing a response in the absence of the shock. The organ- ism does not “want” to escape in the absence of the shock/demand (EO) situation. An antecedent stimulus functions as an SD only when in its presence a specific re- sponse or set of responses produces reinforcement, and the same response does not produce reinforcement in the absence of that stimulus (Michael, 2000).

The laboratory and applied examples can be changed to show the difference between the evocative functions of MOs and stimulus control. The experimen- tal conditions could change so that a buzzer sounds at different periods of time throughout the session, the shock would be removed only when a response is made while the buzzer is sounding, and a response would not produce reinforcement in the absence of the buzzer (i.e., the shock would not be removed). Under these condi- tions the buzzer would function as an SD, and stimulus control would be demonstrated. The applied conditions could change so that two teachers work with the student. One teacher allows the student’s aggressive behavior to remove requests. The other teacher does not remove the requests. The different procedures produce effective neg- ative reinforcement in the presence of one teacher but not in the presence of the other. The teacher who allows the removal of the request would become an SD that evokes aggressive behaviors from the student. In these modified examples, the characteristics of the antecedent control would be different in the presence and absence of the buzzer and the other teacher, and the buzzer and the other teacher would be correlated with an increased frequency of reinforcement.

An understanding of the SD and MO evocative func- tions will improve the technological descriptions of an- tecedent control and the understanding of behavior change (Laraway, Snycerski, Michael, & Poling, 2001). Ultimately, these understandings will produce greater ef- fectiveness in education and treatment.

Stimulus Generalization When an antecedent stimulus has a history of evoking a response that has been reinforced in its presence, there is a general tendency for similar stimuli to also evoke that response. This evocative function occurs with stimuli that share similar physical properties with the controlling an- tecedent stimulus. This tendency is called stimulus gen- eralization. Conversely, stimulus discrimination occurs when different stimuli do not evoke the response. Dif- ferent degrees of stimulus control produce the defining

410

Stimulus Control

characteristics of stimulus generalization and discrimina- tion. Stimulus generalization and discrimination are rel- ative relations. Stimulus generalization reflects a loose degree of stimulus control, whereas discrimination has a relatively tight degree of control. In a simple everyday sit- uation, stimulus generalization can be observed when a young child who has learned to say “daddy” in the pres- ence of her father says “daddy” in the presence of a neigh- bor, a clerk in a store, or Uncle Joe. Further conditioning will sharpen the degree of stimulus control to one specific stimulus, the child’s father.

Stimulus generalization occurs with new stimuli that share similar physical dimensions with the controlling antecedent stimulus. For instance, if the response has a history of producing reinforcement in the presence of a blue stimulus, stimulus generalization is more likely with a lighter or darker color of blue than with a red or yellow stimulus. Also, stimulus generalization is likely when the new stimuli have other elements (e.g., size, shape) in common with the controlling stimulus. A student whose behavior has produced reinforcement for making a re- sponse to a circle is more likely to make the same re- sponse to an oval shape than to a triangular shape.

A stimulus generalization gradient graphically de- picts the degree of stimulus generalization and discrim- ination by showing the extent to which responses reinforced in one stimulus condition are emitted in the presence of untrained stimuli. When the slope of the gra- dient is relatively flat, little stimulus control is evident. However, an increasing slope of the gradient shows more stimulus control.

Behavior analysts have used several procedures to produce stimulus generalization gradients. The classic technique of Guttman and Kalish (1956) provides a rep- resentative example. Their technique is important because many prior researchers had obtained stimulus general- ization gradients by conditioning groups of subjects on the same stimulus value and then testing them individu- ally, each with a different stimulus value. Obviously, this type of technique cannot demonstrate the degree of stim- ulus control for individual subjects. Guttman and Kalish provided a method of acquiring gradients for each subject and laid the foundation for greater understanding of the principles governing stimulus control.

Guttman and Kalish reinforced pigeons on a VI 1-minute schedule for pecking a disk illuminated with a light source appearing yellow-green to humans (i.e., a wavelength of 550 mμ). After the disk peck had stabilized, the pigeons were tested under extinction conditions on the original stimulus and a randomized series of 11 dif- ferent wavelengths never presented during training as the test for stimulus generalization.

Stimulus generalization occurs with responses to a new stimulus after a response has been conditioned in

the presence of another similar stimulus. If responding produces reinforcement during testing for stimulus gen- eralization, it cannot be clear whether any responses to a new stimulus after the first response represent general- ization or if the responses are a function of the rein- forcement schedule. Guttman and Kalish avoided this problem of confounding their results by testing for gen- eralization during extinction.

Lalli, Mace, Livezey, and Kates (1998) reported an excellent applied example of assessing and displaying a stimulus generalization gradient. They used a stimulus generalization gradient in an assessment of the relation between the physical proximity of an adult and the self- injurious behavior (SIB) of a 10-year-old girl with severe mental retardation. The results presented in Figure 1 show that the percentage of total SIB across sessions be- came progressively lower as the distance between the therapist and the girl increased.

Development of Stimulus Control Stimulus Discrimination Training

The conventional procedure for stimulus discrimina- tion training requires one behavior and two antecedent stimulus conditions. Responses are reinforced in the pres- ence of one stimulus condition, the SD, but responses are not reinforced in the presence of the other stimulus, the

100 90 80 70 60 50 40 30 20 10 0P

er ce

n ta

ge o

f T

o ta

l R

es p

o n

se s

A cr

o ss

S es

si o

n s

0 <.5M 1.5 3.0 4.5 6.0 7.5 9.0 Distance (Meters)

Stimulus Generalization Gradient of Self-Injurious Behavior

Figure 1 The percentage of total responses across sessions at a given distance during generalization tests. <0.5, 1.5, 3.0, 4.5, 6.0, 7.5, and 9.0 refer to the distance (in meters) between the therapist and the participant. From “Assessment of Stimulus Generalization Gradients in the Treatment of Self- Injurious Behavior” by J. S. Lalli, F.C. Mace, K. Livezey, and K. Kates (1998). Journal of Applied Behavior Analysis, 31, p. 481. Copyright 1988 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

411

Stimulus Control

SΔ. When a teacher applies this training procedure ap- propriately and consistently, responding in the presence of the SD will come to exceed responding in the presence of the SΔ. Often, over time, the participant will learn not to respond in the presence of the SΔ.

Applied behavior analysts often describe the con- ventional procedures for discrimination training with dif- ferential reinforcement as alternating conditions of reinforcement and extinction, meaning that a response produces reinforcement in the SD condition but not in the SΔ condition. To clarify and stress an important point, however, the SΔ is used not only to show a condition of zero reinforcement (extinction) but also to denote a con- dition that provides a lesser amount or quality of rein- forcement than the SD condition (Michael, 1993).

Maglieri, DeLeon, Rodriguez-Catter, and Sevin (2000) used discrimination training as part of an inter- vention to decrease the food stealing of a 14-year-old girl with Prader-Willi syndrome, a serious medical condition usually correlated with obesity and food stealing. Dur- ing discrimination training, a teacher showed the girl two containers of cookies. One container contained a warn- ing label, the SΔ, and the other container had no warning label, the SD. The teacher told the girl that she could eat cookies only from the container without the warning label. The teacher asked the girl in the presence of the two containers, “Which cookies can you eat?” If the girl answered that she could eat the cookies from the con- tainer without the warning, the teacher let her eat one cookie from the container. This discrimination training procedure decreased stealing food from the containers marked with the warning label.

Concept Formation

The preceding section on discrimination training de- scribes how an antecedent stimulus can acquire control over a response, meaning that the behavior occurs more frequently in the presence of the stimulus than in its ab- sence. A discrimination training procedure might be used to teach a preschool student to name the primary colors. For instance, to teach the color red, the teacher could use a red object such as a red ball as the SD condition and a nonred object such as a yellow ball as the SΔ con- dition. The teacher could position both balls randomly in front of the student, direct the student to name and point to the red ball, and reinforce correct responses, but not incorrect responses. After a few trials the red ball would acquire stimulus control over the student’s response, and the student would reliably differentiate the red ball from the yellow ball. This simple discrimination training, how- ever, may not sufficiently meet the instructional objec- tive of identifying the color red. The teacher may want the student to learn not only to discriminate between red

balls and balls of different color, but also the concept of redness.

Terms such as concept formation or concept ac- quisition imply for many people some hypothetical con- struct of a mental process. Yet acquiring a concept is clearly dependent on responses in the presence of an- tecedent stimuli and the consequences following those responses. Concept formation is a behavioral outcome of stimulus generalization and discrimination (Keller & Schoenfeld, 1950/1995). Concept formation is a com- plex example of stimulus control that requires both stimulus generalization within a class of stimuli and dis- crimination between classes of stimuli. An antecedent stimulus class is a set of stimuli that share a common relationship. All of the stimuli in an antecedent stimu- lus class will evoke the same operant response class, or elicit the same respondent behavior. This evocative or eliciting function is the only common property among the stimuli in the class (Cuvo, 2000). For example, con- sider a stimulus class for the concept red. A red object is called red because of a particular conditioning his- tory. This conditioning history of differential rein- forcement will evoke the response red to light waves of different wavelengths, from light red to dark red. These different shades of the color red that will evoke the re- sponse red share a conditioning history and are included in the same stimulus class. The different shades of red (e.g., light red) that do not evoke the response red are not members of that stimulus class. Therefore the con- cept of redness requires stimulus generalization from the trained stimuli to many other stimuli within the stim- ulus class. If the preschool student described earlier had acquired the concept of redness, he would be able to identify the red ball and, without specific training or re- inforcement, choose a red balloon, a red toy car, a red pencil, and so on.

In addition to stimulus generalization, a concept re- quires discrimination between members and nonmem- bers of the stimulus class. For example, the concept of redness requires discriminating between red and other colors and irrelevant stimulus dimensions such as shape or size. The concept begins with discrimination between the red ball and the yellow ball but results in discrimi- nating a red dress from a blue dress, a red toy car from a white toy car, and a red pencil from a black pencil.

Discrimination training is fundamental to teaching conceptual behavior. Antecedent stimuli representative of a group of stimuli sharing a common relationship (i.e., the stimulus class) and antecedent stimuli from other stimulus classes must be presented. Before a concept can be acquired, the teacher must present exemplars of what the concept is (i.e., the SD condition) and what the con- cept is not (i.e., the SΔ condition). This approach holds true for all conceptual development, even for highly

412

Stimulus Control

abstract concepts (e.g., honesty, patriotism, justice, free- dom, sharing). It is also possible to acquire a concept through vicarious discrimination training and differen- tial reinforcement. A verbal definition of a concept, with examples and nonexamples of the concept, may be suf- ficient for concept formation without additional direct training.

Authors of children’s literature often teach concepts vicariously, such as good and bad, honest and dishonest, courageous and cowardly. For example, consider the story of an owner of a mom-and-pop grocery store who wanted to hire a young person to work in the store. The job in- cluded sweeping the floor, bagging groceries, and keep- ing the shelves neat. The owner wanted an honest person to work for him, so he decided to test all applicants to see whether they were honest. The first young person who applied was given the opportunity to try the job be- fore the owner made the commitment to hire him. But before the applicant came to work, the owner hid a dol- lar bill where he knew the young person would find it. At the end of the test period the owner asked the applicant how he had liked working in the store, whether he wanted the job, and whether anything surprising or unusual had happened to him. The applicant replied that he wanted the job and that nothing surprising had happened to him. The grocer told the first applicant that he wanted to con- sider others who had applied also. The second applicant worked a test period with the same results as the first per- son. He did not get the job. The third young person to work for the grocer was sweeping the floor, found the dollar bill, and took it immediately to the grocer. The third applicant said she turned in the dollar bill in case one of the customers or the grocer had dropped it. The gro- cer asked the applicant whether she liked the job and wanted to work for him. The young person replied that she did. The grocer told the applicant that she had the job because she was an honest person. The grocer also let her keep the dollar bill.

The previous children’s story presents exemplars of honest and dishonest behavior. The honest behavior was rewarded (i.e., the honest person obtained the job), and the dishonest behavior was not rewarded (i.e., the first two applicants did not get the job). This story might vic- ariously teach a certain concept of honesty.

Stimuli comprising a class may function within fea- ture stimulus classes and arbitrary stimulus classes (McIlvane, Dube, Green, & Serna, 1993). Stimuli in a feature stimulus class share common physical forms (e.g., topographical structures) or common relative re- lations (e.g., spatial arrangements). Feature stimulus classes include an infinite number of stimuli and com- prise a large portion of our conceptual behavior. For in- stance, a concept of dog is based on a feature stimulus class. The common physical forms of all dogs will be members of that stimulus class. A young child, through differential reinforcement, will learn to differentiate dogs from horses, cats, cows, and so on. Physical forms pro- vide common relations for many feature stimulus classes, such as stimulus classes that evoke the responses book, table, house, tree, cup, cat, rug, onion, and car. A rela- tional, or a relative relation, exists among stimuli in other feature classes. Examples of these feature stimulus classes based on relative relations are found with con- cepts such as bigger than, hotter than, higher than, on top of, and to the left of.

Stimuli comprising an arbitrary stimulus class evoke the same response, but they do not share a common stimulus feature (i.e., they do not resemble each other in physical form, nor do they share a relational relation- ship). Arbitrary stimulus classes are comprised of a lim- ited number of stimuli. For example, a teacher could form an arbitrary stimulus class using the stimuli 50%, 1/2, di- vided evenly, and .5 (see Figure 2). Following training, each of these stimuli of different physical forms will evoke the same response, a half. Green bean, asparagus, potato, and corn could be developed into an arbitrary stimulus class to evoke the response vegetable. Students learn to associate vowels with the arbitrary class of the letters A, E, I, O, U and sometimes Y.

The development of concepts and complex verbal relations plays an important role in parenting, caregiv- ing, education, and treatment. Applied behavior analysts need to consider different instructional procedures when teaching concepts and complex verbal relations that pro- duce feature stimulus classes and arbitrary stimulus classes. A common instructional procedure used with fea- ture stimulus classes is to differentially reinforce re- sponses to is (SD) and is not (SΔ) examples of the concept.

Antecedent Stimuli Response Reinforcement

50%

1/2

Divided evenly

.5 A half Reinforcement

Figure 2 Antecedent stimuli with different physical forms that evoke the same response, a half. An example of an arbitrary stimulus class.

413

Stimulus Control

Broad generalization is common with feature stimulus classes; depending on the functioning level of the partic- ipants, a few trained examples may be sufficient to de- velop the concept. Stimulus generalization, however, is not a characteristic of arbitrary stimulus classes. Applied behavior analysts have developed arbitrary stimulus classes by using matching-to-sample procedures to cre- ate stimulus equivalence among arbitrary stimuli.

Stimulus Equivalence In a historically influential experiment, Sidman (1971) demonstrated the development of an equivalence stimu- lus class among arbitrary stimuli. A boy with severe re- tardation served as the participant. Before the experiment, the boy could

1. match pictures to their spoken names and

2. name the pictures.

Also before the experiment, the boy could not

3. match written words to the spoken names,

4. match written words to the pictures,

5. match pictures to the written words, or

6. say the written words.

Sidman discovered that after the boy was taught to match written names to spoken names (#3), he could, without additional instruction, match written names to the pictures (#4), match pictures to the written names (#5), and say the written words (#6). In other words, as a result of learning one new stimulus–stimulus relation (#3), the other three stimulus–stimulus relations (#4, #5, and #6) emerged without additional training or rein- forcement. When Sidman linked two or more sets of stim- ulus–stimulus relations, these other noninstructed or unreinforced stimulus–stimulus relations emerged. With- out instruction, the boy’s receptive and expressive lan- guage expanded beyond what existed before the start of the experiment. This is the big prize—something to shoot for in curriculum design and training programs.

Following Sidman’s research (see Sidman, 1994), stimulus equivalence has become a major area of basic and applied research in many areas of complex verbal re- lations, such as reading (Kennedy, Itkonen, & Lindquist, 1994), language arts (Lane & Critchfield, 1998), and mathematics (Lynch & Cuvo, 1995). For example, Rose, De Souza, and Hanna (1996) taught seven nonreading children to read 51 training words. The children matched written words to spoken words, and copied words and named the words. All of the children learned to read the 51 training words, and five of the seven children read

generalization words. Data from Rose and colleagues demonstrate the potential power of stimulus equivalence as a process for teaching reading.

Research on stimulus equivalence has contributed to the understanding of stimulus–stimulus relations among complex human behaviors. Sidman (1971) and others who researched stimulus equivalence during the 1970s (e.g., Sidman & Cresson, 1973; Spradlin, Cotter, & Bax- ley, 1973) gave future applied behavior analysts a pow- erful method for teaching stimulus–stimulus relations (i.e., conditional stimulus control).

Defining Stimulus Equivalence

Equivalence describes the emergence of accurate re- sponding to untrained and nonreinforced stimulus– stimulus relations following the reinforcement of re- sponses to some stimulus–stimulus relations. Behav- ior analysts define stimulus equivalence by testing for reflexivity, symmetry, and transitivity among stimulus–stimulus relations. A positive demonstration of all three behavioral tests (i.e., reflexivity, symme- try, and transitivity) is necessary to meet the definition of an equivalence relation among a set of arbitrary stim- uli. Sidman and Tailby (1982) based this definition on the mathematical statement:

a. If A = B, and

b. B = C, then

c. A = C

Reflexivity occurs when in the absence of training and reinforcement a response will select a stimulus that is matched to itself (e.g., A = A). For example, a partici- pant is shown a picture of a bicycle and three choice pic- tures of a car, an airplane, and a bicycle. Reflexivity, also called generalized identity matching, has occurred if the participant without instruction selects the bicycle from the three choice pictures.

Lane and Critchfield (1998) taught generalized iden- tity matching of printed letters and a spoken word (vowel or consonant) to two adolescent females with moderate mental retardation. The comparison stimuli of A and D, and O and V accompanied the sample spoken word vowel. Generalized identity matching occurred when par- ticipants presented with the sample spoken word vowel selected comparison O from the O and V stimuli and re- jected the nonmatching sample V stimulus; and selected the comparison A from the A and D stimuli and rejected the nonmatching sample D stimulus.

Symmetry occurs with the reversibility of the sam- ple stimulus and comparison stimulus (e.g., if A = B, then B = A). For example, the learner is taught, when presented with the spoken word car (sample stimulus A), to select

414

Stimulus Control

a comparison picture of a car (comparison B). When pre- sented with the picture of a car (sample stimulus B), with- out additional training or reinforcement, the learner selects the comparison spoken word car (comparison A).

Transitivity, the final and critical test for stimulus equivalance, is a derived (i.e., untrained) stimulus– stimulus relation (e.g., A = C, C = A) that emerges as a product of training two other stimulus–stimulus relations (e.g., A = B and B = C). For example, transitivity would be demonstrated if, after training the two stimulus– stimulus relations shown in 1 and 2 below, the relation shown in 3 emerges without additional instruction or re- inforcement:

1. If A (e.g., spoken word bicycle) = B (e.g., the pic- ture of a bicycle) (see Figure 3), and

2. B (the picture of a bicycle) = C (e.g., the written word bicycle) (see Figure 4), then

3. C (the written word bicycle) = A (the spoken name, bicycle) (see Figure 5).

Matching-to-Sample Basic and applied researchers have used the matching-to- sample procedure to develop and test for stimulus equiv- alence. Dinsmoor (1995b) reported that Skinner introduced the experimental procedure called matching-to-sample, and he described Skinner’s procedure in the following way. A pigeon was presented with three horizontal keys to peck. The middle key was illuminated with a color to start the trial. A peck on the illuminated key turned it off and lighted two side keys. One of the side keys was the same color as the sample color illuminated on the middle key. A peck on the side key with the same color as the sample key pro- duced reinforcement. An error response was not reinforced.

The three-term contingency is the basic unit of analy- sis in the development of complex stimulus-control

bicycle Figure 3 Example of an A = B relation (spoken name and picture).

bicycle airplane car

Figure 4 Example of a B = C relation (picture and written word).

415

Stimulus Control

bicycle

car airplane bicycle

Figure 5 Example of a transitive relation, C = A (written word and spoken name), that emerges as a result of training A = B (spoken name and picture) and B = C (picture and written word) relations.

repertoires. Skinner’s matching-to-sample procedure in- cluded the three-term contingency:

SD ------------→ Response ------------→ Reinforcement Side key with Key peck Grain color

This basic contingency, however, is incomplete because of constraints from the environmental context. The con- textual events operating on the three-term contingency become conditional discriminations (Sidman, 1994). The sample stimulus in Skinner’s procedure is the conditional stimulus. The three-term contingency is effective only when it matches the sample stimulus. Other three-term contingencies (nonmatches) are ineffective. Reinforce- ment is conditional on the context of discriminative stim- uli other than the SD; that is, the effectiveness of the three-term contingency comes under contextual control. Conditional discriminations operate at the level of a four- term contingency:

Contextual stimulus --→ SD -----→ Response → Reinforcement

Conditional sample Side key Key peck Grain with color

Key color

SΔ

Color not matching sample

To start a matching-to-sample trial, the participant will make a response (called the observing response) to pre- sent the sample stimulus (i.e., the conditional sample). The comparison stimuli (i.e., the discriminative events) are pre- sented usually after removing the sample stimulus, but not always, and provide for an effective three-term contin- gency and other noneffective three-term contingencies. A certain comparison will match the conditional sample. A response that selects the matched comparison and rejects

the nonmatched comparisons will produce reinforcement. Figure 6 presents an example of the observing response, the conditional sample, the discriminative events, and the comparison match. Responses that select the nonmatch- ing comparison stimuli are not reinforced. During condi- tional discrimination training, the same selection must be correct with one conditional stimulus, but incorrect with one or more other sample stimuli. Most matching-to- sample applications use a correction procedure with error responses. One correction procedure has the learner re- spond to the same sample and comparison stimuli until a correct response has been reinforced. Error correction pro- cedures and the random positioning of the comparison stimuli control for position responding.

Factors Affecting the Development of Stimulus Control Applied behavior analysts establish stimulus control with frequent differential reinforcement of behavior in the presence and in the absence of the SD condition. Effective differential reinforcement requires the consistent use of consequences that function as reinforcers. Additional fac- tors such as preattending skills, stimulus salience, mask- ing, and overshadowing also will affect the development of stimulus control.

Preattending Skills

The development of stimulus control requires certain pre- requisite skills. For academic or social skills the student should engage in orienting behaviors appropriate to the SDs in the instructional setting. Such preattending skills include looking at the instructional materials, looking at

416

Stimulus Control

Observing response

Conditional sample

Discriminative events

Comparison match

SΔ SD SΔ SΔ

gested that salient cues facilitated the efficiency of func- tional analyses by producing faster and clearer outcomes in the presence of the salient cues than in the absence of the salient cues.

Some stimuli have more salience than others de- pending on the sensory capabilities of the individual, the past history of reinforcement, and the context of the en- vironment. For instance, a student may not attend to words written on the blackboard because of poor vision, or to the teacher’s oral directions because of poor hear- ing, or to the curriculum materials because of past failures to learn, or to teacher directions because of focusing at- tention on a toy in the student’s desk.

Masking and Overshadowing

Masking and overshadowing are methods for decreasing the salience of stimuli (Dinsmoor, 1995b). In masking, even though one stimulus has acquired stimulus control over behavior, a competing stimulus can block the evoca- tive function of that stimulus. For example, a student may know the answers to a teacher’s questions, but will not re- spond in the presence of the peer group, which in this ex- ample is competition of different contingencies of reinforcement, not just an antecedent stimulus that makes it harder to “attend to” the relevant SD. With overshad- owing, the presence of one stimulus condition interferes with the acquisition of stimulus control by another stim- ulus. Some stimuli are more salient than others. For ex- ample, looking out of the window to watch the cheerleaders’ practice may distract some students from at- tending to the instructional stimuli presented during the algebra lesson.

Applied behavior analysts need to recognize that masking and overshadowing can hinder the development of stimulus control, and apply procedures that reduce these effects. Examples of reducing the influence of masking and overshadowing include (a) rearranging the physical environment (e.g., lowering the window shade, removing distractions, changing seating assignments), (b) making instructional stimuli appropriately intense (e.g., providing a rapid pace of instruction, many oppor- tunities to respond, an appropriate level of difficulty, and opportunities to set goals), and (c) consistently reinforc- ing behavior in the presence of the instructionally rele- vant stimuli.

Using Prompts to Develop Stimulus Control Prompts are supplementary antecedent stimuli used to occasion a correct response in the presence of an SD that will eventually control the behavior. Applied behavior analysts give response and stimulus prompts before or

the teacher whenever a response is modeled, listening to oral instructions, and sitting quietly for short periods of time. Teachers should use direct behavioral interventions to specifically teach preattending skills to learners who have not developed these skills. Learners must emit be- haviors that orient the sensory receptors to the appropri- ate SD for the development of stimulus control.

Stimulus Salience

The salience of the stimulus influences attention to the stimulus and ultimately the development of stimulus con- trol (Dinsmoor, 1995b). Salience refers to the prominence of the stimulus in the person’s environment. For example, Conners and colleagues (2000) included salient cues (e.g., specific room color, specific therapist) into the context of multielement functional analyses. Their results sug-

Figure 6 An example of the observing response, the conditional sample, the discriminative events, and the comparison match during a matching-to-sample trial.

417

Stimulus Control

during the performance of a behavior. Response prompts operate directly on the response. Stimulus prompts op- erate directly on the antecedent task stimuli to cue a cor- rect response in conjunction with the critical SD.

Response Prompts

The three major forms of response prompts are verbal in- structions, modeling, and physical guidance.

Verbal Instructions

Applied behavior analysts use functionally appropriate verbal instructions as supplementary response prompts. Verbal response prompts occur frequently in almost all training contexts in the forms of vocal verbal instruction (e.g., oral, telling) and nonvocal verbal instruction (e.g., written words, manual signs, pictures)

Teachers often use vocal verbal instructional prompts. Suppose a teacher asked a student to read the sentence, “Plants need soil, air, and water to grow.” The student reads, “Plants need . . . Plants need . . . Plants need . . .” The teacher could use any number of verbal prompts to occasion the next word. She might say, “The next word is soil. Point to soil and say soil.” Or she might use a rhyming word for soil. For another example, Adkins and Mathews (1997) taught in-home caregivers to use vocal verbal response prompts to improve voiding pro- cedures for two adults with urinary incontinence and cog- nitive impairments. The in-home caregiver checked for dryness each hour or every 2 hours between 6:00 A.M. and 9:00 P.M. The caregiver praised dryness, asked the adult to use the toilet, and provided assistance as needed when the adult was dry at the regularly scheduled check. This simple response prompt procedure, which was in- troduced following a baseline condition, produced in one of the adults a mean 22% reduction of grams of urine collected per day in wet diapers during the 2-hour prompted voiding condition, and a mean 69% reduction during the 1-hour prompt condition. The second adult re- ceived only the 1-hour prompted voiding condition, which resulted in a mean 55% reduction of urine col- lected per day.

Krantz and McClannahan (1998) and Sarokoff, Tay- lor, and Poulson (2001) used nonvocal verbal instruc- tional response prompts in the form of embedded scripts to improve the spontaneous social exchanges of children with autism. Examples of the embedded scripts in the children’s photographic activity schedules included look, watch, and let’s eat our snack. In another example of non- vocal verbal instructional response prompts, Wong, Seroka, and Ogisi (2000) developed a checklist with 54 steps to prompt self-assessment of blood glucose level by a diabetic woman with memory impairments. This

participant followed the checklist sequence and checked off each step as she completed it.

Modeling

Applied behavior analysts can demonstrate or model the desired behavior as a response prompt. Modeling can ef- fectively prompt behaviors especially for learners who have already learned some of the component behaviors required for the imitation. Modeling is an easy, practical, and successful way for a coach to show a player an ap- propriate form for shooting a basketball through a hoop when the player already can hold the ball, raise it over his head, and push the ball away from his body. Few teachers would use modeling to teach a child with severe disabilities to tie her shoes if she could not hold the laces in her hands. In addition, attending skills are important. The learner must observe the model to enable the imita- tion of the performance. Finally, modeling as a response prompt should be used only with students who have al- ready developed imitative skills. The use of models to assist in the development of appropriate academic and social behavior has been demonstrated repeatedly.

Physical Guidance

Physical guidance is a response prompt applied most often with young children, learners with severe disabili- ties, and older adults experiencing physical limitations. Using physical guidance, the teacher partially physically guides the student’s movements, or physically guides the student throughout the entire movement of the response.

Hanley, Iwata, Thompson, and Lindberg (2000) re- ported the use of physical guidance to help participants with profound mental retardation manipulate leisure items. Conaghan, Singh, Moe, Landrum, and Ellis (1992) used physical guidance to prompt adults with mental re- tardation and hearing impairments to use manual signs. When a participant made an error in sign production, the teacher physically guided the person’s hands to prompt a correct response. In another example, a personal trainer worked with three older adults with severe disabilities, osteoporosis, and arthritis. The trainer physically guided the participants’ arm movements whenever they did not begin independent pushes with a dumbbell or stopped the pushes before reaching their exercise criterion (K. Cooper & Browder, 1997).

Physical guidance is an effective response prompt, but it is more intrusive than verbal instruction and mod- eling. It requires direct physical involvement between the teacher and the student, making precise assessment of

418

Stimulus Control

student progress difficult. Physical response prompts pro- vide little opportunity for the student to emit the behav- ior without the direct assistance of the teacher. Another possible problem is some learners’ resistance to physical touch. Some learners, however, will require the use of physical guidance.

Stimulus Prompts

Applied behavior analysts have frequently used movement, position, and redundancy of antecedent stimuli as stimu- lus prompts. For example, movement cues can help a learner discriminate between a penny and a dime by point- ing to, tapping, touching, or looking at the coin to be iden- tified. In the coin discrimination task, the teacher could use a position cue and place the correct coin closer to the student. Redundancy cues occur when one or more stim- ulus or response dimensions (e.g., color, size, shape) are paired with the correct choice. For instance, a teacher might use a color mediation procedure of associating a numeral with a color, then link the name of a color to an answer for an arithmetic fact (Van Houten & Rolider, 1990).

Transfer of Stimulus Control Applied behavior analysts should provide response and stimulus prompts as supplementary antecedent stimuli only during the acquisition phase of instruction. With the reliable occurrence of behavior, applied behavior ana- lysts need to transfer stimulus control from the response and stimulus prompts to the naturally existing stimulus. Applied behavior analysts transfer stimulus control by gradually fading stimuli in or out, gradually presenting or removing antecedent stimuli. Eventually, the natural stim- ulus, a partially changed stimulus, or a new stimulus will evoke the response. Fading response prompts and stim- ulus prompts is the procedure used to transfer stimulus control from the prompts to the natural stimulus, and also to minimize the number of error responses occurring in the presence of the natural stimulus.

Terrace’s (1963a, b) influential research on the trans- fer of stimulus control using fading and superimposition of stimuli provides a classic example of transferring stim- ulus control. In these studies Terrace taught pigeons to make red–green and vertical–horizontal discriminations with a minimum of errors. His use of techniques for grad- ually transferring stimulus control was called errorless learning. To teach a red–green discrimination, Terrace presented the SΔ (red light) at the beginning of discrimi- nation training, before the SD (green light) had stimulus control over the pigeon’s responses. The initial introduc- tion of the red light was with low illumination and for

brief time intervals. During successive presentations of the stimuli, Terrace gradually increased the intensity of the red light and the duration of time it was illuminated until it differed from the green light only in hue. With this procedure Terrace taught pigeons to discriminate red from green with only a minimum number of errors (re- sponses to the SΔ).

Terrace further demonstrated that stimulus control acquired with red and green lights could be transferred to vertical and horizontal lines with a minimum number of errors (i.e., responses in the presence of the SΔ). His pro- cedure consisted of first superimposing a white vertical line on the green light (SD) and a white horizontal line on the red light (SΔ). Then the pigeons were given several presentations of the two compound stimuli. Finally, the amplitude of the red and green lights was reduced grad- ually until only the vertical and horizontal lines remained as stimulus conditions. The pigeons showed almost per- fect transfer of stimulus control from the red–green lights to the vertical–horizontal lines. That is, they emitted re- sponses in the presence of the vertical line (SD) and sel- dom responded in the presence of the horizontal line (SΔ).

Following Terrace’s work, other pioneer researchers (e.g., Moore & Goldiamond, 1964) produced landmark studies showing that the transfer of stimulus control with few incorrect responses was possible with human learn- ers, which provided the foundations for developing ef- fective procedures to transfer stimulus control from response prompts to natural stimuli in the applied context.

Transferring Stimulus Control from Response Prompts to Naturally Existing Stimuli

Wolery and Gast (1984) described four procedures for transferring stimulus control from response prompts to natural stimuli. They describe these procedures as most- to-least prompts, graduated guidance, least-to-most prompts, and time delay.

Most-to-Least Prompts

The applied behavior analyst can use most-to-least re- sponse prompts to transfer stimulus control from response prompts to the natural stimulus whenever the participant does not respond to the natural stimulus or makes an in- correct response. To apply most-to-least response prompts, the analyst physically guides the participant through the entire performance sequence, then gradually reduces the amount of physical assistance provided as training progresses from trial to trial and session to ses- sion. Customarily, most-to-least prompting moves from physical guidance to visual prompts to verbal instruc- tions, and finally to the natural stimulus without prompts.

419

Stimulus Control

Graduated Guidance

The applied behavior analyst provides physical guidance as needed, but using graduated guidance she will imme- diately start to fade out the physical prompts to transfer stimulus control. Graduated guidance begins with the ap- plied behavior analyst following the participant’s move- ments closely with her hands, but not touching the participant. The analyst then increases the distance of her hands from the participant by gradually changing the lo- cation of the physical prompt. For example, if the applied behavior analyst used physical guidance for a partici- pant’s hand movement in zippering a coat, she might move the prompt from the hand to the wrist, to the elbow, to the shoulder, and then to no physical contact. Gradu- ated guidance provides the opportunity for an immedi- ate physical prompt as needed.

Least-to-Most Prompts

When transferring stimulus control from response prompts using least-to-most prompts, the applied behav- ior analyst gives the participant an opportunity to per- form the response with the least amount of assistance on each trial. The participant receives greater degrees of as- sistance with each successive trial without a correct re- sponse. The procedure for least-to-most prompting requires the participant to make a correct response within a set time limit (e.g., 3 seconds) from the presentation of the natural SD. If the response does not occur within the specified time, the applied behavior analyst will again present the natural SD and a response prompt of least as- sistance, such as a verbal response prompt. If after the same specified time limit (e.g., another 3 seconds), the participant does not make a correct response, the analyst gives the natural SD and another response prompt, such as a gesture. The participant receives partial or full phys- ical guidance if the lesser prompting does not evoke a correct response. Applied behavior analysts using the least-to-most response prompt procedure present the nat- ural SD and the same time limit during each training trial. For example, Heckaman, Alber, Hooper, and Heward (1998) used instructions, nonspecific verbal prompts, modeling, and physical prompting in a least-to-most 5-second response prompt hierarchy to improve the dis- ruptive behavior of four students with autism.

Time Delay

To produce a transfer of stimulus control, most-to-least prompts, graduated guidance, and least-to-most prompts occur as consequences to gradual changes in the form, position, or intensity of a response evoked by the natural stimulus. Conversely, as an antecedent response prompt,

the time delay procedures use only variations in the time intervals between presentation of the natural stimulus and the presentation of the response prompt. Constant time delay and progressive time delay transfer stimulus con- trol from a prompt to the natural stimulus by delaying the presentation of the prompt following the presentation of the natural stimulus.

The constant time delay procedure first presents sev- eral trials using a 0-second delay—that is, the simulta- neous presentation of the natural stimulus and the response prompt. Usually, but not always (Schuster, Grif- fen, & Wolery, 1992), the trials that follow the simulta- neous prompt condition apply a fixed time delay (e.g., 3 seconds) between the presentation of the natural stimu- lus and the presentation of the response prompt (Cald- well, Wolery, Werts, & Caldwell, 1996).

The progressive time delay procedure, like constant time delay, starts with a 0-second delay between the pre- sentation of the natural stimulus and the response prompt. Usually, a teacher will use several 0-second tri- als before extending the time delay. The number of 0- second trials will depend on the task difficulty and the functioning level of the participant. Following the si- multaneous presentations, the teacher will gradually and systematically extend the time delay, often in 1-second intervals. The time delay can be extended after a spe- cific number of presentations, after each session, after a specific number of sessions, or after meeting a perfor- mance criterion.

Heckaman and colleagues (1998) used the follow- ing progressive time delay procedure: They began with 0-second trials, simultaneously presenting the control- ling response prompt (e.g., physical prompt, model) and the task instruction. The simultaneous presentations con- tinued until the participant met the criterion of nine correct responses. The first time delay was set at 0.5 seconds. After meeting the 0.5-second criterion, the researchers increased the delay by 1-second increments up to 5 sec- onds. Correct responses occurring before the response prompt or within 3 seconds after the prompt received a positive feedback statement (e.g., “That’s right”). Error responses received a negative feedback statement (e.g., “No, that’s not right”) and the controlling prompt. Also, the next trial following an error response moved back to the previous delay level. The controlling prompt was also presented with disruptive behavior.

Transfer of Stimulus Control Using Stimulus Control Shaping

The preceding sections focused on response prompts that do not change the task stimuli or materials. The stimulus control shaping procedures presented here modify the

420

Stimulus Control

task stimuli or materials gradually and systematically to prompt a response. Supplementary stimulus conditions are faded in or faded out to transfer stimulus control from the stimulus prompt to the natural stimulus. Stimulus con- trol shaping can be accomplished with stimulus fading and stimulus shape transformations (McIlvane & Dube, 1992; Sidman & Stoddard, 1967).

Stimulus Fading

Stimulus fading involves highlighting a physical dimen- sion (e.g., color, size, position) of a stimulus to increase the likelihood of a correct response. The highlighted or exaggerated dimension is faded gradually in or out. The following examples of (a) handwriting the uppercase let- ter A, and (b) giving the answer 9 to an arithmetic prob- lem are illustrations of systematically fading out stimuli.

Krantz and McClannahan (1998) faded out scripts (i.e., the words Look and Watch me) embedded in photo- graphic activity schedules. The embedded scripts prompted the social exchanges of children with autism. The words Look and Watch me were printed with 72- point font and bold letters on white 9-cm note cards. Krantz and McClannahan began fading out the words by removing one third of the script card, then another third. Sometimes, during the script fading, portions of letters were still shown on the card such as a part of an o in Look. Finally, the scripts and cards were removed.

The treatment of a feeding disorder reported by Patel, Piazza, Kelly, Ochsner, and Santana (2001) provides an example of fading stimuli in and out. Severe food selec- tivity is a common problem for children with feeding dis- orders. For example, some children find highly textured foods aversive and dangerous because of gagging. Patel and colleagues faded in Carnation Instant Breakfast (CIB) and then milk to water in treating a 6-year-old boy with pervasive developmental feeding disorder. The boy would drink small amounts of water. The researchers began the fading in procedure by adding 20% of the CIB packet to 240 ml of water. Following three sessions of drinking the 20% mixture, more CIB was gradually added to the water, initially by 5% and later 10% increments. The researchers

then faded in milk to the CIB and water mixture after the boy would drink one CIB packet with 240 ml of water. Milk was gradually added to CIB/water mixture in 10% increments as the water was faded out (e.g., 10% milk and 90% water, plus one packet of CIB; then 20% milk and 80% water).

Applied behavior analysts have used the superimpo- sition of stimuli with stimulus fading. In one instance, the transfer of stimulus control occurs when one stimu- lus is faded out; in another application, one stimulus is faded in as the other stimulus is faded out. The research by Terrace (1963a, b) demonstrating the transfer of stim- ulus control from a red–green discrimination to a verti- cal–horizontal discrimination shows the superimposition of two specific classes of stimuli and the fading out of one stimulus class. The lines were superimposed on the colored lights; then the lights were gradually faded out, leaving only the vertical and horizontal lines as the dis- criminative stimuli. Figure 7 provides an applied ex- ample of the superimposition and stimulus fading procedures used by Terrace. The figure shows a series of steps from an arithmetic program to teach 7 � 2 = _____.

The other frequently used procedure fades in the nat- ural stimulus and fades out the stimulus prompt. Figure 8 illustrates this superimposition procedure, in which the prompt is faded out and the natural stimulus 8 + 5 = _____ is faded in.

Stimulus Shape Transformations

The procedure for stimulus shape transformations uses an initial stimulus shape that will prompt a correct re- sponse. That initial shape is then gradually changed to form the natural stimulus, while maintaining correct re- sponding. For example, a program to transform the shape of stimuli to teach number recognition could include the following steps (Johnston, 1973):

The shape of the stimulus prompt must change grad- ually so that the student continues to respond correctly. In teaching word identification, using stimulus shape transformations could include the following steps (John- ston, 1973):

421

Stimulus Control

Figure 7 Illustration of two classes of superimposed stimuli with one class then faded out. From Addition and Subtraction Math Program with Stimulus Shaping and Stimulus Fading by T. Johnson, 1973, unpublished project, Ohio Department of Education. Reprinted by permission.

Figure 9 shows how superimposition of stimuli can be used with stimulus shaping in arithmetic instruc- tion. The + and = signs are superimposed on the stimu- lus shaping program and are gradually faded in.

In summary, a variety of procedures exist for trans- ferring stimulus control from response and stimulus

prompts to natural stimuli. Currently, procedures for transferring stimulus control of response prompts are more practical in teaching situations than stimulus shape transformations because of the greater skill and time re- quired for material preparation with stimulus change procedures.

422

Stimulus Control

Figure 8 Illustration of superimposition and stimulus fading to fade in the natural stimulus and fade out the stimulus prompt. From Addition and Subtraction Math Program with Stimulus Shaping and Stimulus Fading by T. Johnson, 1973, unpublished project, Ohio Department of Education. Reprinted by permission.

423

Stimulus Control

Figure 9 Illustration of superimposition of stimuli and stimulus shaping. (From Addition and Subtraction Math Program with Stimulus Shaping and Stimulus Fading by T. Johnson, 1973, unpublished project, Ohio Department of Education. Reprinted by permission.)

Summary Antecedent Stimuli

1. Reinforcement of an operant response increases the fre- quency of future responding and influences the stimuli that immediately precede the response. The stimuli pre- ceding the response (i.e., antecedent stimuli) will acquire an evocative effect on the relevant behavior.

2. Stimulus control occurs when the rate, latency, duration, or amplitude of a response is altered in the presence of an antecedent stimulus (Dinsmoor, 1995a, b). A stimulus ac- quires control only when responses emitted in the pres- ence of that stimulus produce reinforcement.

3. Discriminative stimuli and motivating operations share two important similarities: (a) Both events occur before the behavior of interest, and (b) both events have evocative functions.

4. Often an evocative effect that appears to be the function of an SD is not the product of a history of differential reinforcement correlated with the altered frequency of response. These sit- uations are probably related to motivational operations (MOs) rather than stimulus control.

Stimulus Generalization

5. When an antecedent stimulus has a history of evoking a behavior that has been reinforced in its presence, there

is a general tendency for other antecedent stimuli to also evoke that behavior. This evocative function occurs with stimuli that share similar physical properties with the controlling antecedent stimulus. This tendency is called stimulus generalization. Conversely, stimulus discrimi- nation occurs when the new stimuli do not evoke the response.

6. Stimulus generalization reflects a loose degree of stimu- lus control, whereas discrimination indicates a relatively tight degree of stimulus control.

7. A stimulus generalization gradient graphically depicts the degree of stimulus generalization and discrimination by showing the extent to which responses reinforced in one stimulus condition are emitted in the presence of un- trained stimuli.

Development of Stimulus Control

8. The conventional procedure for stimulus discrimination training requires one behavior and two antecedent stimu- lus conditions. Responses are reinforced in the presence of one stimulus condition, the SD, but responses are not re- inforced in the presence of the other stimulus, the SΔ.

9. Concept formation is a complex example of stimulus con- trol that requires both stimulus generalization within a class of stimuli and discrimination between stimulus classes.

424

Stimulus Control

10. An antecedent stimulus class is a set of stimuli that share a common relationship. All of the stimuli in the class will evoke the same operant response class, or elicit the same response in the case of respondent behavior.

11. Stimuli comprising a class may function within feature stimulus classes and arbitrary stimulus classes.

Stimulus Equivalence

12. Equivalence describes the emergence of accurate re- sponding to untrained and nonreinforced stimulus– stimulus relations following the reinforcement of re- sponses to some other stimulus–stimulus relations.

13. Behavior analysts define stimulus equivalence by test- ing for reflexivity, symmetry, and transitivity among stimulus–stimulus relations. A positive demonstration of all three tests (i.e., reflexivity, symmetry, and transitivity) is necessary to meet the definition of an equivalence rela- tion among a set of arbitrary stimuli.

14. Basic and applied researchers have used the matching-to- sample procedure to develop and test for stimulus equiv- alence.

Factors Affecting the Development of Stimulus Control

15. Applied behavior analysts establish stimulus control with frequent differential reinforcement of behavior in the pres- ence and in the absence of the SD condition. Effective dif- ferential reinforcement requires the consistent use of consequences that function as reinforcers. Additional fac- tors such as preattending skills, stimulus salience, mask- ing, and overshadowing will also affect the development of stimulus control.

Using Prompts to Develop Stimulus Control

16. Prompts are supplementary antecedent stimuli (e.g., in- structions, modeling, physical guidance) used to occasion a correct response in the presence of an SD that will even- tually control the behavior. Applied behavior analysts pro- vide response and stimulus prompts before or during the performance of a behavior.

17. Response prompts operate directly on the response. Stim- ulus prompts operate directly on antecedent task stimuli to cue a correct response in conjunction with the critical SD.

Transfer of Stimulus Control

18. Applied behavior analysts should provide response and stimulus prompts as supplementary antecedent stimuli only during the acquisition phase of instruction.

19. Fading response prompts and stimulus prompts is the pro- cedure used to transfer stimulus control from the prompts to the natural stimulus, and also to minimize the number of error responses occurring in the presence of the natural stimulus.

20. Procedures for transferring stimulus control from response prompts to natural stimuli include (a) most-to-least prompts, (b) graduated guidance, (c) least-to-most prompts, and (d) time delay.

21. Stimulus fading involves exaggerating a dimension of a stimulus to increase the likelihood of a correct response. The exaggerated dimension is faded gradually in or out.

22. The procedure for stimulus shape transformations uses an initial stimulus shape that will prompt a correct response. That initial shape is then gradually changed to form the natural stimulus, while maintaining correct responding.

425

Imitation

Key Terms imitation

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes and Concepts

3-15 Define and provide examples of echoics and imitation.

426

Imitation

An imitative repertoire promotes the relatively quick acquisition of behaviors such as a young child’s development of social and communica-

tion skills. With an understanding of the imitation process, applied behavior analysts can use imitation as an intervention to evoke new behaviors. Without an imi- tative repertoire, a person has little chance for the agile acquisition of behaviors.

Imitation has received considerable experimental and theoretical attention spanning several decades (e.g., Baer & Sherman, 1964; Carr & Kologinsky, 1983; Garcia & Batista-Wallace, 1977; Garfinkle & Schwartz, 2002; Wol- ery & Schuster, 1997). The experimental literature shows that individuals can acquire and maintain imitative and echoic behavior in the same fashion as they acquire and maintain other operant behaviors.1 That is, (a) reinforce- ment increases the frequency of imitation; (b) when some imitative behaviors receive reinforcement, other imita- tive behaviors occur without specific training and rein- forcement; and (c) some children who do not imitate may be taught to do so. This chapter defines imitation and suggests a protocol for imitation training with learners who do not imitate.

Definition of Imitation Four behavior–environment relations functionally define imitation: (a) Any physical movement may function as a model for imitation. A model is an antecedent stimulus that evokes the imitative behavior. (b) An imitative be- havior must immediately follow the presentation of the model (e.g., within 3 to 5 seconds). (c) The model and the behavior must have formal similarity. (d) The model must be the controlling variable for an imitative behavior (Holth, 2003).

Models

Planned Models

Planned models are prearranged antecedent stimuli that help learners acquire new skills or refine the topography of certain elements of existing skills. A planned model shows the learner exactly what to do. A videotape of a person emitting specific behaviors can serve as a planned model. LaBlanc and colleagues (2003), for example, used video models to teach three children with autism perspective-taking skills. The children learned to imitate a video model touching or pointing to an object such as

1Chapter entitled “Verbal Behavior” defines the echoic operant and func- tionally separates it from imitation. Basically, echoic is the technical term used in the context of vocal verbal behavior, and imitation applies to non- vocal verbal and nonverbal behaviors.

a bowl or box. (The children also echoed vocal-verbal behaviors such as saying “under the bowl,” or saying “one” to identify a box marked 1.)

Unplanned Models

All antecedent stimuli with the capacity to evoke imita- tion are potentially unplanned models. Unplanned mod- els occasion many new forms of behavior because imitating the behavior of others in everyday social in- teractions (e.g., school, work, play) frequently produces new and helpful adaptive behaviors. For example, a young person during her first outing on a city bus may learn how to pay the fare by imitating the fare-paying of other boarders.

Formal Similarity

Formal similarity occurs when the model and the behav- ior physically resemble each other and are in the same sense mode (i.e., they look alike, sound alike) (Michael, 2004). For example, when a student observes a teacher finger-spell the word house (i.e., the model), and then duplicates the finger spelling (i.e., the imitation), the im- itative finger spelling has formal similarity with the model. A baby sitting in a high chair immediately taps the tray with his hand after seeing his mother tap the tray with her hand. The baby’s imitative tap has formal sim- ilarity to the mother’s tap.

Immediacy

The temporal relation of the immediacy between the model and the imitative behavior is an important feature of imitation. However, a form (topography) of an imita- tion may occur at a later time in the context of everyday life situations. For example, the young person on her first bus ride who used imitation to learn how to pay the fare may use her newly learned behavior to pay the bus fare for her return trip when no other boarders have paid fares before her. The parent of the student who imitated the finger spelling of house might ask, “What did you learn at school today?” and in response, the student may pro- duce the finger spelling for house.

When the topography of a previous imitation occurs in the absence of the model (e.g., paying the bus fare, fin- ger spelling), that delayed behavior is not an imitative behavior. The delayed behaviors of paying the bus fare and the finger spelling in the contexts previously pre- sented have similar topographies as the imitated behav- iors but occur as the result of different controlling variables. The relation between discriminative stimuli (e.g., the device for collecting the bus fare) or motivating

427

Imitation

operations (e.g., the parent’s question) and delayed be- haviors are functionally different from the relation be- tween a model and an imitative behavior. Therefore, delayed behaviors using the topography of a prior imita- tive model, by definition, are not imitative.2

A Controlled Relation

We often view imitative behavior as doing the same. Al- though the formal similarity of doing the same is a nec- essary condition for imitation, it is not sufficient. Formal similarity can exist without the model functionally con- trolling the similar behavior. The controlling relation be- tween a model and a similar behavior is the most important property that defines imitation. A controlling relation between the behavior of a model and the behav- ior of the imitator is inferred when a novel model evokes a similar behavior in the absence of a history of rein- forcement for that behavior. An imitative behavior is a new behavior that follows a novel antecedent event (i.e., the model). After the model evokes an imitation, that be- havior comes into contact with contingencies of rein- forcement. These new contingencies of reinforcement then become the controlling variable for the discrimi- nated operant (i.e., MO/SD → R → SR).

Holth (2003) explained the discriminated operant in the context of imitation training this way:

Let us imagine that a dog is trained to sit whenever the owner sits down in a chair, and to turn around in a circle whenever the owner turns around. Does the dog imitate the owner’s behavior? Almost certainly not. The dog could have as easily been taught to sit whenever the owner turns around in a circle and to turn around in a circle when the owner sits down. Thus, what may look like imitation may be nothing more than a series of di- rectly taught discriminated operants. The source of con- trol can be determined only by introducing novel exemplars. In the absence of a demonstration that the dog responds to novel performances by “doing the same,” there is no evidence that a similarity to the owner’s behavior is important in determining the form of the dog’s response. Hence, there is no true demon- stration that the dog imitates the behavior of the owner unless it also responds to new instances of the owner’s behavior by “doing the same.” (p. 157)

Is and Is-Not Examples of Controlled Relations and Imitation

Consider two guitarists performing in a contemporary rock band. One guitarist improvises a short riff (i.e., a novel model). The other guitarist hears the riff and then

2Delayed imitation is, however, a commonly used term in the imitation lit- erature (e.g., Garcia, 1976).

immediately reproduces the riff cord-for-cord and note- for-note (i.e., imitative behavior—a new behavior). This is example meets all of the conditions for imitation; im- mediacy, formal similarity, and the model (i.e., the im- provised riff) produced the controlling relation.

Continuing with the two guitarists in the contempo- rary rock band, suppose one musician says to the other, “I have a great idea for our opening song. Let me play it for you. If you like it, I will teach it to you.” The other gui- tarist likes what she hears. They practice the riff together until the second guitarist learns it. The guitarists add the riff to the opening number. On stage, one guitarist plays the riff, and then the other guitarist immediately repro- duces the riff. This is not an example of imitation be- cause the riff played by the first guitarist was not a novel model and the similar behavior of the second guitarist was a trained similar response with a history of rein- forcement. It is an example of a discriminated operant.

As another is-not example of imitative behavior, let’s now consider two classical musicians playing a fugue using a musical score. A fugue is a form of music based on a short melody that is played at the beginning by one player or one section of an ensemble, and then another player or section repeats the short melody. The perfor- mance of a fugue appears similar to imitation in that one player or section presents a melody, then another player or section immediately repeats the melody, and the two melodies have formal similarity. A fugue is not an ex- ample of imitation, however, because the first introduc- tion of the short melody does not control the similar behavior. The printed notes on the music manuscript pro- vide the controlling variables. Playing a fugue using a musical score is an example of a discriminated operant rather than of imitation.

As a final is-not example, Tim throws a baseball to Bill (a person with some baseball experience). Bill catches the ball, and then immediately throws it back to Tim. Did Bill imitate Tim’s throw? It gives an ap- pearance of imitation. Tim’s throw was an antecedent event (a model); Bill’s throw immediately followed and had formal similarity to the model. This is not an ex- ample of imitation because Bill’s throw to Tim was not a new behavioral instance of throwing under the con- trol of Tim’s throw. Again, the controlling variable was the history of reinforcement that produced a discrimi- nated operant.

Imitation Training Typically developing children acquire many skills by im- itating unplanned models. Parents and other caregivers do not usually have to apply specific interventions to fa- cilitate the development of imitative skills. However,

428

Imitation

some infants and children with developmental disabilities do not imitate. Without an imitative repertoire, these chil- dren will have great difficulty developing beyond the most basic of skills. Still, it is possible to teach imitation to some children who do not imitate.

Applied behavior analysts have validated repeatedly the procedures used by Baer and colleagues as an effec- tive method for teaching imitation to children who do not imitate (e.g., Baer, Peterson, & Sherman, 1967; Baer & Sherman, 1964). For example, three children with severe to profound mental retardation served as participants in one of their studies (Baer, Peterson, & Sherman, 1967). During imitation training, the children were taught to emit simple discriminated (i.e., similar to the model) re- sponses (e.g., raising an arm) when the teacher presented the verbal response cue “do this,” and then provided the model (e.g., raised his arm). Baer and colleagues selected appropriate skill levels for their participants (e.g., gross motor movements, fine motor movements) Also, the teacher initially used physical guidance to prompt the similar response, and then gradually reduced the guid- ance over several trials; shaped the discriminated re- sponse by using bits of food to reinforce closer and closer similarity to the model; and reinforced responses with formal similarity.

The imitation training protocol developed by Baer and colleagues taught some nonimitative learners to im- itate, meaning that a novel model controlled imitative be- haviors in the absence of specific training and reinforcement of those behaviors. One participant imi- tated a novel model only after receiving imitation train- ing for similarity on 130 different models. The second participant showed similar results to those of the first. The third participant learned to imitate with less training than the other participants had. This participant imitated the ninth training model, a novel model without a history of training and reinforcement.

To summarize the results from Baer and colleagues: (a) Children who did not have an imitative repertoire learned to imitate with training that used response cues and prompts, shaping, and reinforcement; (b) when some imitative behaviors produced reinforcement, the partici- pants imitated novel models without reinforcement; and (c) the participants demonstrated an effect sometimes known as learning set (Harlow, 1959), or a learning-to- learn phenomenon. As the participants progressed through imitation training, they required fewer and fewer training trials to learn a new discriminated response with similarity to the models.

The major objective of imitation training is to teach learners to do what the person providing the model does regardless of the behavior modeled. A learner who learns to do what the model does is likely to imitate models that have not been associated with specific training, and

3Generalized imitation is a term used frequently in the imitation literature to label a participant’s unprompted, untrained, nonreinforced responses with formal similarity to the actions of a model. We label such responses to novel models as, simply, imitation.

those imitations are likely to occur in many situations and settings, frequently in the absence of planned rein- forcement.3 Imitation, however, may depend on the pa- rameters of the response class used during training. For example, Young, Krantz, McClannahan, and Poulson (1994) found that children with autism imitated novel models within the vocal, toy-play, and pantomime re- sponse types used for training, but the imitations did not generalize across response types. Even so, imitation pro- vides for the relatively quick acquisition of new, complex behaviors that are characteristic of many human en- deavors. Imitation produces those new behaviors with- out undue reliance on physical help or previous reinforcement considerations.

Building on the experimental methods of Baer and colleagues, Striefel (1974) developed an imitation train- ing program for practitioners. The components of Striefel’s protocol are as follows: (a) assessing, and teach- ing if necessary, prerequisite skills for imitation training, (b) selecting models for training, (c) pretesting, (d) se- quencing models for training, and (e) performing imita- tion training.

Assessing, and Teaching If Necessary, Prerequisite Skills for Imitation Training

Learners cannot imitate if they do not attend to the pre- sentation of the model. Therefore, attending to the model is a prerequisite for imitation training. Striefel (1974) de- fined attending as staying seated during instruction, keep- ing one’s hands in one’s lap, looking at the trainer whenever one’s name is called, and looking at objects identified by the trainer. Also, practitioners often need to decrease problem behaviors that interfere with training (e.g., aggression, screaming, odd hand movements).

Suggested procedures for assessing attending skills include the following:

1. Staying seated. Seat the learner and record the du- ration of time the learner remains seated.

2. Looking at the teacher. Say the learner’s name in a commanding voice and record whether the student makes eye contact.

3. Keeping hands in lap. Prompt the student to put his hands in his lap and record the duration of time the student’s hands remain in that position.

4. Looking at objects. Place several objects on a table and say, “Look at this.” Immediately following the

429

Imitation

command, move a finger from in front of the learner’s eye to one of the objects and record whether the student looked at the object.

Teachers often assess attending skills for a minimum of three sessions. The teacher can begin imitation training if the assessment data show adequate attending skills. The teacher will need to teach these skills before beginning imitation training if attending skills need to be developed.

Selecting Models for Imitation Training

Practitioners may need to select and use about 25 be- haviors as models during initial imitation training. In- cluding gross motor movements (e.g., raising a hand) and fine motor movements (e.g., manual sign language) as models provides learners with opportunities to develop more refined differentiations with their imitative skills.

Practitioners usually use one model at a time during the initial training trials rather than a sequence of move- ments. Practitioners may choose to use more complex models such as sequences of behavior after the learner can imitate one model per occasion successfully. Also, the initial training usually includes models of (a) the movement of body parts (e.g., touching nose, hopping on one foot, bringing hand to mouth) and (b) the manip- ulation of physical objects (e.g., passing a basketball, picking up a glass, zipping a coat).

Pretesting

The learner’s responses to the selected models should be pretested. The pretest may show that the learner will im- itate some models without training. The pretesting pro- cedures advocated by Striefel (1974) are as follows:

1. Prepare the learner’s attending behaviors for the pretest (e.g., seated, hands in lap) and often assume the same ready position as the learner.

2. If you are using an object model, place one object in front of the learner and one object in front of yourself.

3. Say the learner’s name to start the pretest, and when the learner makes eye contact, say, “Do this” (i.e., child’s name, pause, “Do this”).

4. Present the model. For example, if the selected be- havior is to pick up a ball, pick up the ball yourself and hold it for a few seconds.

5. Immediately praise each response that has formal similarity to the model, and deliver the reinforcer (e.g., a hug, an edible) as quickly as possible.

6. Record the learner’s response as correct or incor- rect (or no response), or as an approximation of

the model (e.g., touches the ball, but does not pick it up).

7. Continue pretesting with the remaining models.

Practitioners can use the pretest procedure with all motor and vocal verbal models (e.g., name, pause, “Do this,” “Say ball”). Striefel recommended pretesting for several sessions until all models have been pretested at least three times. If the learner correctly responds during pretesting to a selected model at a set criterion level (e.g., three of three correct), then the practitioner should ad- vance to other models. If the learner does not meet the cri- terion, the practitioner should select that model for imitation training.

Sequencing the Selected Models for Training

Practitioners use the pretest results to arrange the pre- sentation sequence for the selected models, arranging the sequence from the easiest to the most difficult models to imitate. The first models selected for imitation training are those that the learner imitated correctly on some, but not all, of the pretest trials. The models that the learner responded to incorrectly, but approximated the model, are selected next. Finally, the models that the learner failed to perform, or performed incorrectly, are the last to be selected for training.

Conducting Imitation Training

Striefel (1974) suggested using four conditions for imi- tation training: preassessment, training, postassessment, and probing for imitative behaviors. The procedures used in imitation training are the same as those in the pretest with the exception of when and how often the practitioner presents the selected models.

Preassessment

The preassessment is a short pretest given before each training session. Practitioners use the first three models currently selected for training for the preassessment. These three models are presented three times each in ran- dom order during the preassessment. If the learner’s behavior has similarity to the model on all three presen- tations, that model is removed from the training. The pre- assessment procedure allows practitioners to evaluate the learner’s current performance on the models selected for training that session, and to determine the learner’s progress in learning to respond to the model.

430

Imitation

Training

During training, practitioners use repeated presentations of one of the three models used in the preassessment. The model selected first for training is the one most often re- sponded to during the preassessment (i.e., the behavior was similar to the model on some, but not all, of the pre- assessment presentations). If, however, the learner made approximations only, the behavior with the closest simi- larity to the model is selected first for training. Training continues until the learner responds to the model cor- rectly on five consecutive trials.

Imitation training will likely include physical guid- ance to prompt the response if the learner fails to respond. For example, the practitioner may physically guide the learner’s behavior through the complete response. Phys- ical guidance allows the learner to experience the re- sponse and the reinforcer for that specific movement. After physically assisting the complete response, the practitioner will gradually withdraw the physical guid- ance, letting go of the student’s body just before the en- tire movement is completed, and then continue to fade the physical guidance by withdrawing the physical sup- port earlier on each subsequent trial. Eventually, the learner may complete the movement without assistance. When the learner responds to the model without prompt- ing for five consecutive trials, that model is included in the postassessment.

Postassessment

During the postassessment, the practitioner presents, three times each, five previously learned models and five mod- els that are still included in imitation training. The practi- tioner will remove a most recently learned behavior from imitation training following three consecutive postassess- ments in which the learner responds correctly without physical guidance to the model 14 out of the 15 opportu- nities. Physical guidance, however, is appropriate to use during the postassessment. If the learner does not reach this criterion (14 out of 15 postassessment opportunities), Striefel (1974) recommended continuing imitation training with that model. The postassessment procedure allows the practitioner to evaluate how well a learner performs the previously and most recently learned behaviors.

Probes for Imitative Behaviors

Practitioners will use approximately five nontrained, novel models to probe for occurrences of imitation at the end of each imitation training session, or they will intermix the probes with the training trials. The probe procedure uses the same procedures as the preassessment activities, but without using the antecedent verbal response prompt (i.e.,

child’s name, pause, “Do this”) or other forms of response prompts (e.g., physical guidance). Probing for nontrained imitations provides data on the learner’s progress in de- veloping an imitation repertoire—in other words, learn- ing to do what the model does.

Guidelines for Imitation Training Keep Training Sessions Active and Brief

Most practitioners use short training sessions during im- itation training, typically 10 to 15 minutes, but often schedule more than one session per day. Two or three brief sessions can be more effective than one long ses- sion. To maintain quick and active training, practition- ers should allow no more than a few seconds between trials.

Reinforce Both Prompted and Imitative Responses

In the early stages of imitation training, practitioners should reinforce each occurrence of either prompted re- sponses or true imitation. If the learner’s participation re- quires a reinforcer other than praise, it should be presented immediately in small amounts that the learner can consume quickly (e.g., cereal bits, 5 seconds music). Practitioners should reinforce only matching responses or imitative behaviors that occur within 3 to 5 seconds of the model. Learners who consistently emit correct matching responses but not immediately following the model should be reinforced for shorter response laten- cies (e.g., a decreasing contingency from 7 seconds to 6 seconds to 5 seconds to 4 seconds).

Pair Verbal Praise and Attention with Tangible Reinforcers

During imitation training, many learners, particularly children with severe to profound developmental disabil- ities, need tangible consequences such as edibles or liq- uids. As training progresses, practitioners try to use social attention and verbal praise to maintain the matching re- sponses or imitative behaviors. They do this by pairing the delivery of other consequences with social and verbal praise. Social attention (e.g., patting the student’s arm af- fectionately) and/or descriptive verbal praise should im- mediately follow each correct response or approximation, simultaneously with the other consequence. A learner’s willingness to participate in imitation training may in- crease when the practitioner schedules a preferred activ- ity to follow each session.

431

Imitation

If Progress Breaks Down, Back Up and Move Ahead Slowly

There may be identifiable reasons for worsening perfor- mance, such as reinforcer satiation or context distrac- tions; or perhaps the practitioner presented models that were too complex for the learner. Regardless of whether a clear reason exists for the worsening performance, the practitioner should return to an earlier level of success- ful performance. Once successful imitative responses are reestablished, training can advance.

Keep a Record

As with all behavior change programs, applied behavior analysts should directly measure and record the learner’s performance and review the data after each session. With the use of frequent and direct measurement, the practi- tioner can make objective, informed, data-based deci- sions concerning the effects of the training program.

Fade Out Verbal Response Prompts and Physical Guidance

Parents and caregivers in the everyday environments of young children almost always teach imitation skills using verbal response prompts and physical guidance. For ex-

ample, a caregiver may tell a child to “Wave bye-bye,” model waving, and then physically guide the wave of the child. Or, a parent may ask a child, “What does the cow say?” The parent then presents a model (“The cow says moo”), tells the child to say “moo,” and then provides praise and attention if the child says “moo.” This natural instructional process is the same process advocated in this chapter for teaching imitative skills to learners who do not imitate: A verbal response prompt is given (“Do this”), a model is presented, and physical guidance is used when needed. However, imitation training is not complete until all response prompts have been with- drawn. Children need to learn to do what the model does without the supports of response prompting. Therefore, to promote the effective use of imitation, practitioners should fade out the response prompts used during the ac- quisition of trained matching responses.

Ending Imitation Training

Decisions to stop imitation training depend on the learner’s behavior and program goals. For example, the practitioner could stop motor imitation training when the student imitates the first presentations of novel models, or when the student imitates a sequence of behaviors (e.g., washing hands, brushing teeth, finger spelling).

Summary Definition of Imitation

1. Four behavior–environment relations define imitation: (a) Any physical movements may function as a model for imitation. A model is an antecedent stimulus that evokes the imitative behavior. (b) An imitative behavior must be emitted within 3 seconds of the presentation of the model. (c) The model and the behavior must have formal simi- larity. (d) The model must be the controlling variable for an imitative behavior.

Formal Similarity

2. Formal similarity occurs when the model and the behav- ior physically resemble each other and are in the same sense mode (i.e., they look alike, sound alike).

Immediacy

3. The temporal relation of the immediacy between the model and the imitative behavior is an important feature of imi- tation. However, a form (topography) of imitation may occur in the context of everyday life situations, and at any time. When the form of a previous imitation occurs in the absence of the model, that delayed behavior is not an im- itative behavior.

A Controlled Relation

4. We often view imitative behavior as doing the same. Al- though the formal similarity of doing the same is a neces- sary condition for imitation, it is not sufficient. Formal similarity can exist without the model functionally con- trolling the similar behavior.

5. A controlling relation between the behavior of a model and the behavior of the imitator is inferred when a novel model evokes a similar behavior in the absence of a history of reinforcement.

6. An imitative behavior is a new behavior emitted following a novel antecedent event (i.e., the model). After the model evokes the imitation, that behavior comes into contact with contingencies of reinforcement. These new contingencies of reinforcement then provides the controlling variables for the discriminated operant (i.e., MO/SD → R → SR).

Imitation Training for Learners Who Do Not Imitate

7. Applied behavior analysts have validated repeatedly the instructional method used by Baer and colleagues as an effective method for teaching imitation to children who do not imitate.

432

Imitation

8. Building on the experimental methods of Baer and col- leagues, Striefel (1974) developed an imitation training program for practitioners.

9. The components of Striefel’s protocol are as follows: (a) assessing, and teaching if necessary, prerequisite skills for imitation training, (b) selecting models for training, (c) pretesting, (d) sequencing models for training, and (e) conducting imitation training.

Guidelines for Imitation Training

10. Keep training sessions active and brief.

11. If the learner’s participation requires a reinforcer other than praise, it should be presented immediately in small amounts that the learner can consume quickly.

12. Pair verbal praise and attention with tangible reinforcers.

13. If progress breaks down, back up and move ahead slowly.

14. Measure and record the learner’s performance and review the data after each session.

15. Fade out verbal response prompts and physical guidance.

16. Stop motor imitation training when the student consis- tently imitates novel models, or when the student imitates a sequence of behaviors (e.g., washing hands, brushing teeth, finger spelling).

433

Chaining

Key Terms backward chaining backward chaining with leaps ahead behavior chain

behavior chain interruption strategy behavior chain with a limited hold

chaining

forward chaining task analysis total-task chaining

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 8: Selecting Intervention Outcomes and Strategies

8-1 Conduct a task analysis.

Content Area 9: Behavior Change Procedures

9-12 Use chaining.

434

Chaining

This chapter defines a behavior chain, provides a rationale for establishing behavior chains in applied settings, and discusses the importance

of task analysis in behavior chain training. The chapter presents a procedure for constructing and validating a task analysis, along with procedures for assessing indi- vidual mastery levels. Forward chaining, total-task chain- ing, backward chaining, and backward chaining with leap aheads are addressed, followed by guidelines for decid- ing which behavior chain procedure to use in applied set- tings and a description of behavior chains with limited holds. Techniques for breaking inappropriate chains are also addressed. The chapter concludes with an examina- tion of factors affecting the performance of a behavior chain.

Definition of a Behavior Chain A behavior chain is a specific sequence of discrete re- sponses, each associated with a particular stimulus con- dition. Each discrete response and the associated stimulus condition serve as an individual component of the chain. When individual components are linked to- gether, the result is a behavior chain that produces a ter- minal outcome. Each response in a chain produces a stimulus change that simultaneously serves as a condi- tioned reinforcer for the response that produced it and as a discriminative stimulus (SD) for the next response in the chain. Each stimulus that links the two response com- ponents together serves a dual function: It is both a con- ditioned reinforcer and an SD (Reynolds, 1975; Skinner, 1953). Reinforcement produced by the final response in the chain maintains the effectiveness of the stimulus changes that serve as conditioned reinforcers and SDs

for each response in the chain. The no exception to the dual function of the components occurs with the first and last stimulus in the chain. In these instances, the stimulus serves only one function: either as the SD or as the conditioned reinforcer.

Reynolds (1975) provided this laboratory example of a behavioral chain (illustrated in Figure 1):

An experimental example of a chain may begin when a pigeon is presented with a blue key. When the pigeon pecks the key, it changes to red. After the key turns red, the pigeon presses a pedal that turns the key yellow. During yellow, displacing a bar changes the key to green. Finally, during green, pecks are reinforced with the operation of a grain-delivery mechanism and its as- sociated stimuli, in the presence of which the bird ap- proaches the grain magazine and eats. (pp. 59–60)

In Reynolds’ example, the links are arranged in the fol- lowing sequence: blue-key-peck-red; red-peddle-press- yellow; yellow-bar-press-green; green-key-peck-grain magazine; grain magazine-eat-food intake. “Because each stimulus has a dual function, as discriminative stimulus and conditioner reinforcer, the links overlap. In fact, it is this dual function of the stimuli that holds the chain together” (Reynolds, 1975, p. 60).

As Figure 1 shows, this chain includes four re- sponses (R1, R2, R3, and R4), with a specific stimulus con- dition (S1, S2, S3, and S4) associated with each response. The blue light (S1) is an SD to occasion the first response (R1), a key peck that terminates the blue light and pro- duces the onset of the red light (S2). The red light serves as conditioned reinforcement for R1 and as an SD for R2

(a pedal press). The pedal press (R2) terminates the red light and produces the onset of the yellow light (S3), and so on. The last response produces food, thus completing and maintaining the chain.

S1 Blue Light

R1 Key Peck (terminates blue light, activates red light)

S2 Red Light

R2 Pedal Press (terminates red light, activates yellow light)

S3 Yellow Light

R3 Bar Press (terminates yellow light, activates green light)

S4 Green Light

R4 Key Peck (terminates green light, activates food magazine)

(food)

S = Stimulus R = Response SR = Reinforcement

Figure 1 Illustration of a behavior chain consisting of four components. Based on a chain described by Reynolds (1975, pp. 59–60).

435

Chaining

Table 1 Delineation of the Relationship between Each Discriminative Stimulus, Response, and Reinforcement in a Sample Behavior Chain

Discriminative stimulus Response Conditioned reinforcement

S1 “Put on your coat.” R1 Obtain coat from closet Coat in hands

S2 Coat in hands R2 Place one arm in sleeve One arm in sleeve

S3 One arm in sleeve/one arm out R3 Place second arm in sleeve Coat on

S4 Coat on R4 Zip up the coat Teacher praise

Table 1 shows the links of a behavior chain using a classroom example. In preparing a class of preschool students for recess, the teacher might say to one student, “Please put on your coat.” The teacher’s statement would serve as the SD (S1) evoking the first response (R1) in the chain, obtaining the coat from the closet. That response in the presence of the teacher’s statement terminates the teacher’s statement and produces the onset of the coat in the student’s hands (S2). The coat in the student’s hands serves as conditioned reinforcement for obtaining the coat from the closet (R1) and as an SD (S2) for putting one arm through a sleeve (R2). That response, putting one arm through the sleeve in the presence of the coat in both hands, terminates the stimulus condition of the coat in the student’s hands and produces the onset of one arm in a sleeve and one arm out (S3), a stimulus change that serves as conditioned reinforcement for putting one arm through a sleeve and as an SD (S3) for placing the second arm through the other sleeve (R3). That response termi- nates the condition of one arm in a sleeve and one arm out, and produces the SD (S4) of the coat being completely on. The coat being fully on serves as a conditioned rein- forcer for putting the second arm through the other sleeve and as an SD for zipping it (R4). Finally, zipping up the coat completes the chain and produces teacher praise.

A behavior chain has the following three important characteristics: (a) A behavior chain involves the perfor- mance of a specific series of discrete responses; (b) the performance of each behavior in the sequence changes the environment in such a way that it produces condi- tioned reinforcement for the preceding response and serves as an SD for the next response; and (c) the re- sponses within the chain must be performed in a specific sequence, usually in close temporal succession.

Behavior Chains with a Limited Hold

A behavior chain with a limited hold is a sequence of behaviors that must be performed correctly and within a specified time to produce reinforcement. Thus, behavior chains with limited holds are characterized by perfor-

mance that is accurate and proficient. An assembly task on a production line exemplifies a chain with a limited hold. To meet the production requirements of the job, an employee might have to assemble 30 couplers onto 30 shafts within 30 minutes (one per minute). If the coupler is placed on the shaft, the retaining clip applied, and the unit forwarded to the next person in line within the pre- scribed period of time, reinforcement is delivered. In a behavior chain with a limited hold, the person must not only have the prerequisite behaviors in his repertoire, but must also emit those behaviors in close temporal suc- cession to obtain reinforcement.

The behavior analyst must recognize that accuracy and rate are essential dimensions of chains with limited holds. For example, if a person is able to complete a chain in the correct sequence but the speed with which he per- forms one or more responses in the chain is slow, chang- ing the focus to increase the rate of performance is warranted. One way to do this is to set a time criterion for completion of each response within the chain; another is to set a designated time criterion for completion of the total chain.

Rationale for Using Chaining Whereas behavior chain connotes a particular sequence of responses ending in reinforcement, the term chaining refers to various methods for linking specific sequences of stimuli and responses to form new performances. In forward chaining, behaviors are linked together begin- ning with the first behavior in the sequence. In backward chaining, the behaviors are linked together beginning with the last behavior in the sequence. Both of these pro- cedures and their variations are discussed in detail later in this chapter.

There are several reasons for teaching new behav- ior chains to people. First, an important aspect of the ed- ucation of students with development disabilities is to increase independent living skills (e.g., using public util- ities, taking care of personal needs, executing travel skills, socializing appropriately). As these skills develop, the student is more likely to function effectively in least

436

Chaining

restrictive environments or participate in activities with- out adult supervision. Complex behaviors developed with chaining procedures allow individuals to function more on their own.

Second, chaining provides the means by which a series of discrete behaviors can be combined to form a series of responses that occasion the delivery of positive reinforcement. That is, essentially, chaining can be used to add behaviors to an existing behavioral repertoire. For example, a person with a developmental disability might consistently seek assistance from a coworker or job coach while completing an assembly task. A chain- ing procedure could be used to increase the number of tasks that must be performed before reinforcement is delivered. To that end, the teacher might give the indi- vidual a written or pictorial list of the parts that must be assembled to complete the job. When the first portion of the task is finished, the individual crosses off the first word or picture on the list and proceeds to the second task. In behavioral terms, the first word or picture on the list serves as the SD to occasion the response of com- pleting the first task. That response, in the presence of the word or picture on the list, terminates the initial stimulus and produces the next stimulus, the second word or picture on the list. The completion of the sec- ond task serves as conditioned reinforcement for com- pleting the task and produces the onset of the third SD. In this way the chaining procedure enables simple be- haviors to be combined into a longer series of complex responses.

Finally, chaining can be combined with other be- havior change prompting, instructing, and reinforcing procedures to build more complex and adaptive reper- toires (McWilliams, Nietupski, & Hamre-Nietupski, 1990).

Task Analysis Before individual components of a chain can be linked to- gether, the behavior analyst must construct and validate a task analysis of the components of the behavioral se- quence, and assess the mastery level of the learner with respect to each behavior in the task analysis. Task analy- sis involves breaking a complex skill into smaller, teach- able units, the product of which is a series of sequentially ordered steps or tasks.

Constructing and Validating a Task Analysis

The purpose of constructing and validating a task analy- sis is to determine the sequence of behaviors that are nec- essary and sufficient to complete a given task efficiently.

The sequence of behaviors that one person might have to perform may not be identical to what another person needs to achieve the same outcome. A task analysis should be in- dividualized according to the age, skill level, and prior experience of the person in question. Further, some task analyses are comprised of a limited number of major steps, with each major step containing four to five subtasks. Figure 2 illustrates a task analysis for making a bed developed by McWilliams and colleagues (1990).

At least three methods can be used to identify and validate the components of a task analysis. In the first method the behavioral components of the sequence are developed after observing competent individuals perform the desired sequence of behaviors. For example, Test, Spooner, Keul, and Grossi (1990) produced a task analy- sis for using a public telephone by observing two adults perform the task, and then validating that sequence by training a person with developmental disabilities to use the task analysis. Based on the training, the researchers made subsequent modifications to the original sequence of tasks. Table 2 shows the final task analysis con- taining 17 steps.

Some of the steps in the task analysis of using a pub- lic telephone might appear in a different order or be com- bined differently. For instance, Step 3, choose the correct change, could be interchanged with Step 2, find the tele- phone number; or an 18th step could be added, “Place change back in your pocket.” There are no absolute rules for determining the number or sequence of steps. How- ever, practitioners are advised to consider the learner’s physical, sensory, or motoric skill levels when determin- ing the scope and sequence of the steps, and be prepared to adjust if necessary.

A second method of validating a task analysis is to consult with experts or persons skilled in performing the task (Snell & Brown, 2006). For instance, when teach- ing mending skills to young adults with developmental disabilities, consultation with a seamstress or a sec- ondary-level home economics teacher is advisable. Based on such expert evaluation, a task analysis could be constructed that would serve as the basis for assessment and training.

A third method of determining and validating the se- quence of behaviors in a task analysis is to perform the behaviors oneself (Snell & Brown, 2006). For example, a practitioner interested in teaching shoe tying could re- peatedly tie her own shoes, noting the discrete, observable steps necessary to achieve a correctly tied shoe. The ad- vantage of self-performing the task is the opportunity to come into contact with the task demands of the sequence prior to training a learner, getting a clearer idea of the be- haviors to be taught and the associated SDs necessary to occasion each behavior. Self-performing the task repeat- edly would refine the response topography necessary for

437

Chaining

the learner to use the sequence most efficiently. Table 3 shows how an initial 7-step sequence for tying a shoe might be expanded to 14 steps after self-performing the behavior (cf. Bailey & Wolery, 1992).

A systematic trial-and-error procedure can assist the behavior analyst in developing a task analysis. Using a systematic trial-and-error method, an initial task analysis is generated and then refined and revised as it is being tested. With revisions and refinements obtained through field tests, a more functional and appropriate task analy- sis can be achieved. For instance, as mentioned earlier, Test and colleagues (1990) generated their initial task analysis for using a public telephone by watching two adults complete the task. They subsequently carried the process a step further by asking a person with develop- mental disabilities to perform the same task and modi- fied the task analysis accordingly.

Regardless of the method used to put the steps in se- quence, the SDs and corresponding responses must be identified. But being able to perform a response is not sufficient; the individual must be able to discriminate the conditions under which a given response should be per- formed. Listing the discriminative stimuli and associated responses helps the trainer determine whether naturally occurring SDs will evoke different or multiple responses. This topic will be discussed in detail later in the chapter.

Assessing Mastery Level

Mastery level is assessed to determine which compo- nents of the task analysis a person can perform inde- pendently. There are two principal ways to assess a person’s mastery level of task analysis behaviors prior to

Figure 2 Task analysis of bedmaking skills.

Given an unmade bed that included a messed-up bedspread, blanket, pillow, and flat sheet and a fitted sheet, the student will:

Section 1: bed preparation 1. remove pillow from bed 2. pull bedspread to foot of bed 3. pull blanket to foot of bed 4. pull flat sheet to foot of bed 5. smooth wrinkles of fitted sheet

Section 2: flat sheet 6. pull top of flat sheet to headboard 7. straighten right corner of flat sheet at foot of bed 8. repeat #7 for the left corner 9. make sheet even across top end of mattress

10. smooth wrinkles

Section 3: blanket 11. pull top of blanket to headboard 12. straighten right corner of blanket at foot of bed 13. repeat #12 for the left corner 14. make blanket even with sheet across top end of mattress 15. smooth wrinkles

Section 4: bedspread 16. pull top of bedspread to headboard 17. pull right corner of bedspread at foot to floor 18. repeat #17 for left side 19. make bedspread even with floor on both sides 20. smooth wrinkles

Section 5: pillow 21. fold top of spread to within 4 inches of the pillow width 22. place pillow on folded portion 23. cover pillow with the folded portion 24. smooth bedspread wrinkles on and around the pillow

From “Teaching Complex Activities to Students with Moderate Handicaps Through the Forward Chaining of Shorter Total Cycle Response Sequences,” by R. McWilliams, J. Nietupski, & S. Hamre-Nietupski, 1990, Edu- cation and Training in Mental Retardation, 25, p. 296. Copyright 1990 by the Council for Exceptional Children. Reprinted by permission.

438

Chaining

Table 2 Task Analysis and Time Limits for Performing Each Task for Using a Public Telephone

Step Time Limit

1. Locate the telephone in the environment 2 minutes

2. Find the telephone number 1 minute

3. Choose the correct change 30 seconds

4. Pick up receiver using left hand 10 seconds

5. Put receiver to left ear and listen for dial tone 10 seconds

6. Insert first coin 20 seconds

7. Insert second coin 20 seconds

8-14. Dial seven-digit number 10 seconds per

15. Wait for telephone to ring a minimum of five times 25 seconds

16. If someone answers, initiate conversation 5 seconds

17. If telephone is busy, hang up phone and collect money 15 seconds

From “Teaching Adolescents with Severe Disabilities to Use the Public Telephone,” by D. W. Test, F. Spooner, P. K. Keul, & T. A. Grossi, 1990, Behavior Modification, 3rd ed., p. 161. Copyright 1990 by Sage Publications. Reprinted by permission.

Table 3 Initial and Expanded Steps for Teaching Shoe Tying

Shorter sequencea Longer sequenceb

1. Partially tighten shoe laces. 1. Pinch lace.

2. Pull shoe laces tight—vertical pull. 2. Pull lace.

3. Cross shoe laces. 3. Hang lace ends from corresponding sides of shoe.

4. Tighten laces—horizontal pull. 4. Pick up laces in corresponding hands.

5. Tie laces into a knot. 5. Lift laces above shoe.

6. Make a bow. 6. Cross right lace over the left to form a tepee.

7. Tighten bow. 7. Bring left lace toward student.

8. Pull left lace through tepee.

9. Pull laces away from each other.

10. Bend left lace to form a loop.

11. Pinch loop with left hand.

12. Bring right lace over the fingers—around loop.

13. Push right lace through hole.

14. Pull loops away from each other.

Sources: (a) Santa Cruz County Office of Education, Behavioral Characteristics Progression. Palo Alto, California, VORT Corporation, 1973. (b) Smith, D. D., Smith, J. O., & Edgar, E. “Research and Application of Instructional Materials Development.” In N. G. Haring & L. Brown (Eds.), Teaching the Severely Handicapped (Vol. 1). New York: Grune & Stratton, 1976. From Teaching Infants and Preschoolers with Handicaps, p. 47, by D. B. Bailey & M. Wolery, 1984, Columbus, OH: Charles E. Merrill. Used by permission.

439

Chaining

training—the single-opportunity method and the multiple-opportunity method (Snell & Brown, 2006).

Single-Opportunity Method

The single-opportunity method is designed to assess a learner’s ability to perform each behavior in the task analysis in correct sequence. Figure 3 is an example of a form used to record a learner’s performance. Specif- ically, a plus sign (+) or minus sign (–) is marked for each behavior that is correctly or incorrectly emitted.

Assessment in this example began when the teacher said, “Tom, put on your hearing aid.” Tom’s responses to the steps in the task analysis were then recorded. The figure shows Tom’s data for the first four days of as- sessment. On Day 1 Tom opened the hearing aid con- tainer and removed the harness; each of these steps was performed correctly, independently, sequentially, and within a 6-second time limit. However, Tom then at- tempted to put the hearing aid harness over his head (Step 5) without first doing Steps 3 and 4. Because he continued to perform this behavior for more than 10 sec-

onds, the teacher stopped the assessment and scored Steps 3 and 4 and all remaining steps as incorrect. On Day 2 Tom was stopped after Step 1 because he per- formed a subsequent step out of order. On Days 3 and 4 the assessment was discontinued after Step 4 because Tom took more than 6 seconds to perform Step 5. Given a criterion for mastery of 100% accuracy within 6 sec- onds over three consecutive probes, the data indicate that Tom only met the criterion for Step 1 (the three +’s for Step 2 were not recorded consecutively).

Multiple-Opportunity Method

The multiple-opportunity method of task analysis as- sessment evaluates the person’s level of mastery across all the behaviors in the task analysis. If a step is per- formed incorrectly, out of sequence, or the time limit for completing the step is exceeded, the behavior ana- lyst completes that step for the learner and then posi- tions her for the next step. Each step performed correctly is scored as correct response, even if the learner erred on the previous steps.

Figure 3 Task analysis data sheet for single-opportunity assessment of inserting a hearing aid.

Task Analysis Assessment of Inserting a Hearing Aid Instructional cue: “Put on your hearing aid” Teacher: Christine Assessment method: Single opportunity Student: Tom

Date Step Behavior 10/1 10/2 10/3 10/4 1 Open container + + + + 2 Remove harness + � + + 3 Arm 1/Strap 1 � � + + 4 Arm 2/Strap 2 � � + + 5 Harness over head � � � � 6 Fasten harness � � � � 7 Unsnap pocket � � � � 8 Remove aid from container � � � � 9 Insert aid into pocket � � � �

10 Snap pocket � � � � 11 Pick up earmold � � � � 12 Insert earmold into ear � � � � 13 Turn on aid � � � � 14 Set the control � � � � Percentage of Steps Correct 14% 7% 28% 28%

Materials: Hearing aid container, harness, earmold Response latency: 6 seconds Recording key: + (correct) � (incorrect) Criterion: 100% correct performance for 3 consecutive days

From “Teaching Severely Multihandicapped Students to Put on Their Own Hearing Aids,” by D. J. Tucker & G. W. Berry, 1980, Journal of Applied Behavior Analysis, 13, p. 69. Copyright 1980 by the Society for the Ex- perimental Analysis of Behavior, Inc. Adapted by permission.

440

Chaining

Figure 4 Task analysis data sheet for multiple-opportunity assessment of inserting a hearing aid.

Task Analysis Assessment of Inserting a Hearing Aid Instructional cue: “Put on your hearing aid” Teacher: Marge Assessment method: Multiple opportunity Student: Kathy

Date Step Behavior 10/1 10/2 10/3 10/4 1 Open container � + + + 2 Remove harness + � + + 3 Arm 1/Strap 1 + � + + 4 Arm 2/Strap 2 � � + + 5 Harness over head + � + � 6 Fasten harness � + � + 7 Unsnap pocket + � + + 8 Remove aid from container + � � + 9 Insert aid into pocket + + � +

10 Snap pocket + � + � 11 Pick up earmold + � + � 12 Insert earmold into ear � � � + 13 Turn on aid � � � � 14 Set the control � � � � Percentage of Steps Correct 57% 21% 57% 64%

Materials: Hearing aid container, harness, earmold Response latency: 6 seconds Recording key: + (correct) � (incorrect) Criterion: 100% correct performance for 3 consecutive days

From “Teaching Severely Multihandicapped Students to Put on Their Own Hearing Aids,” by D. J. Tucker and G. W. Berry, 1980, Journal of Applied Behavior Analysis, 13, p. 69. Copyright 1980 by the Society for the Ex- perimental Analysis of Behavior, Inc. Adapted by permission.

Figure 4 shows that Kathy, after receiving the instructional cue, did not perform the first step in the sequence (opening the container), so a minus was recorded. The teacher then opened the container and po- sitioned Kathy in front of it. Kathy then removed the harness (Step 2) and placed her arm through the strap (Step 3), so a plus was recorded for each of these steps. Because 6 seconds passed before Step 4 was completed, the teacher performed the step for Kathy, scored a minus, and positioned her to do Step 5. Kathy then per- formed Step 5, and the rest of the assessment contin- ued in this fashion.

The key to using the multiple-opportunity method for a task analysis assessment is to ensure that teaching is not commingled with assessment. That is, if the teacher physically guides or models the step, an accurate assess- ment cannot be obtained. Therefore, it is important that the teacher does not help the student do any step.

Single-opportunity and multiple-opportunity meth- ods can both be effective ways to determine mastery of initial skills in a behavior chain. Of the two, the single-

opportunity method is the more conservative measure be- cause the assessment terminates at the first step at which performance breaks down. It also provides less informa- tion to the teacher once instruction is initiated; but it is probably quicker to conduct, especially if the task analy- sis is long, and it reduces the likelihood of learning tak- ing place during assessment (Snell & Brown, 2006). The multiple-opportunity method takes more time to com- plete, but it provides the behavior analyst with more in- formation. That is, the teacher could learn which steps in the task analysis the learner has already mastered, thus eliminating the need to instruct steps already in the learner’s repertoire.

We have looked at two of the prerequisites for link- ing components of a chain: (a) conducting and validating a task analysis of the components of the behavioral se- quence and (b) assessing the learner’s pre-existing skill with each component in the chain. In the following sec- tion, we will address the third component: teaching the in- dividual to perform each step of the chain in close temporal succession.

441

Chaining

Behavior Chaining Methods After the task analysis has been constructed and validated and the criterion for success and the data collection pro- cedures have been determined, the next step is to decide which chaining procedure to use to teach the new se- quence of behavior. The behavior analyst has four op- tions: forward chaining, total-task chaining, backward chaining, and backward chaining with leap aheads.

Forward Chaining

In forward chaining the behaviors identified in the task analysis are taught in their naturally occurring order. Specifically, reinforcement is delivered when the prede- termined criterion for the first behavior in the sequence is achieved. Thereafter, reinforcement is delivered for cri- terion completion of Steps 1 and 2. Each succeeding step requires the cumulative performance of all previous steps in the correct order.

For example, a child learning to tie her shoes ac- cording to the 14-step task analysis shown in Table 3 would be reinforced when the first step, “Pinch lace,” is performed accurately three consecutive times. Next, re- inforcement would be delivered when that step and the next one, “Pull lace,” are performed to the same crite- rion. Then, “Hang lace ends from corresponding sides of shoe” would be added, and all three steps would have to be performed correctly before reinforcement is delivered. Ultimately, all 14 steps in the task analysis should be per- formed in a similar manner. However, at any given train- ing step a variety of response prompting and other strategies may be employed to occasion the response.

Longer chains of behaviors can be broken down into smaller chains or skill clusters, each of which can be taught in a manner similar to that used with a single-re- sponse unit. When one skill cluster is mastered, it is linked to the next. The final response in the first skill cluster sets the occasion for the first response in the second skill clus- ter. Essentially, in this variation skill clusters become the analogue for units of behaviors, and these clusters are linked.

McWilliams and colleagues (1990) combined skill clusters in forward chaining to teach bedmaking skills to three students with developmental disabilities. Based on a task analysis of bedmaking responses (see Figure 2), five skill clusters were identified, each containing four to five subtasks. Once baseline was collected showing minimal accuracy with completing the entire chain in the presence of the cue, “Show me how to make the bed,” an instructional procedure was chosen that involved teach- ing the clusters one at a time by breaking the complex behaviors into smaller chains (see Figure 5).

Initial training began with the teacher demonstrat- ing the chain of tasks for Sections 1 and 2. Subsequent training involved teaching the prior section, the current section targeted for instruction, and the next section in the sequence. For example, if training was to be conducted on Section 2, Sections 1, 2, and 3 were demonstrated. Next, the students practiced the targeted sequences two to five times. When the sequence was performed cor- rectly, praise was delivered. When an error occurred, a three-part correction procedure involving verbal redi- rection, redirection plus modeling, and/or physical guid- ance was initiated until the trial ended on a correct response. When the first skill cluster was mastered (S1), the second skill cluster (S2) was introduced, followed by the third (S3), the fourth (S4), and so forth.

The results indicated that the forward chaining pro- cedure was effective in teaching the students bedmaking skills. That is, all the students were able to make their beds independently, or with only minimal assistance, when the teacher directed them to do so. Furthermore, all of them were able to make their beds in the general- ization setting (i.e., the home).

The McWilliams and colleagues study (1990) illus- trates two main advantages of forward chaining: (a) It can be used to link smaller chains into larger ones, and (b) it is relatively easy, so teachers are likely to use it in the classroom.

Our discussion thus far has focused on teaching be- havior chains through direct teacher instruction. How- ever, some evidence suggests that chained responses can also be learned through observation (Wolery, Ault, Gast, Doyle, & Griffen, 1991). Griffen, Wolery, and Schuster (1992) used a constant time delay (CTD) procedure to teach students with mental retardation a series of chains tasks food preparation behavior. During CTD with one student where the chained response was taught, two other students watched. Results showed that the two observ- ing students learned at least 85% of the correct steps in the chain even though direct instruction on those steps had not occurred.

Total-Task Chaining

Total-task chaining (sometimes called total-task pre- sentation or whole-task presentation) is a variation of for- ward chaining in which the learner receives training on each step in the task analysis during every session. Trainer assistance is provided with any step the person is unable to perform independently, and the chain is trained until the learner is able to perform all the behaviors in the se- quence to the predetermined criterion. Depending on the complexity of the chain, the learner’s repertoire, and available resources, physical assistance and/or graduated guidance may be incorporated.

442

Chaining

M it

D eb

B et

N u

m b

er o

f C

o rr

ec tl

y P

er fo

rm ed

T as

k S

te p

s 5

s 4

s 3

s 2

s 1

Intervention Base- line

Generalization

Session Probe Trials

Figure 5 Number of correctly performed task steps across baseline, intervention, and generalization probe trials. From “Teaching Complex Activities to Students with Moderate Handicaps Through the Forward Chaining of Shorter Total Cycle Response Sequences,” by R. McWilliams, J. Nietupski, & S. Hamre- Nietupski, 1990, Education and Training in Mental Retardation, 25, p. 296. Copyright 1990 by the Council for Exceptional Children. Reprinted by permission.

Werts, Caldwell, and Wolery (1996) used peer mod- els and total-task chaining to teach skills such as operat- ing an audiotape, sharpening a pencil, and using a calculator to three elementary-level students with dis- abilities who were enrolled in general education class- rooms. Response chains were individualized for each student based on teacher-recommended sequences that the students had to learn. Each session was divided into three parts: (a) The students with disabilities were probed on their ability to perform the total task response chain; (b) a peer model competent with the chain demonstrated the chain in its entirety while simultaneously describing each step; and (c) the student partner was probed again to determine performance of the chain. During the probes before and after peer modeling, the students with dis- abilities were directed to complete the chain. If a student was successful, a correct response was noted, but no feed- back was delivered. If a student was unsuccessful, the student’s view was blocked temporarily while the teacher completed that step in the chain, and then the student was

redirected to complete the remaining steps. Each behav- ior in the response chain was scored.

All three students learned to complete the response chain after peer modeling, and reached criterion on the re- sponse chain (100% correct for 2 out of 3 days) over the course of the study. The results for Charlie, one of the three students in the study, are shown in Figure 6.

Test and colleagues (1990) used total-task chaining with a least-to-most prompting procedure to teach two adolescents with severe mental retardation to use a pub- lic telephone. After identifying and validating a 17-step task analysis (see Table 2), baseline data were ob- tained by giving the students a 3 × 5 index card with their home phone number and two dimes, and directing them to “phone home.” During training, a least-to-most prompting procedure consisting of three levels of prompts (verbal, verbal plus gesture, and verbal plus guidance) was implemented when errors occurred on any of the 17 steps that comprised the task analysis. Each instructional session consisted of two training tri- als followed by a probe to measure the number of in- dependent steps completed. In addition, generalization probes were conducted in two other settings at least once per week.

Results showed that the combination of total-task chaining plus prompts increased the number of steps in the task analysis performed correctly by each student and that the skills generalized to two community-based settings (see Figure 7). Test and colleagues (1990) concluded that total-task chaining, especially as used within the context of a two-per-week training regimen, of- fered benefits to practitioners working with learners in community-based settings.

Backward Chaining

When a backward chaining procedure is used, all the behaviors identified in the task analysis are initially com- pleted by the trainer, except for the final behavior in the chain. When the learner performs the final behavior in the sequence at the predetermined criterion level, rein- forcement is delivered. Next, reinforcement is delivered when the last and the next-to-last behaviors in the se- quence are performed to criterion. Subsequently, rein- forcement is delivered when the last three behaviors are performed to criterion. This sequence proceeds back- ward through the chain until all the steps in the task analysis have been introduced in reverse order and prac- ticed cumulatively.

Pierrel and Sherman (1963) conducted a classic demonstration of backward chaining at Brown Univer- sity with a white rat named Barnabus. Pierrel and Sher- man taught Barnabus to climb a spiral staircase, push down and cross a drawbridge, climb a ladder, pull a toy

443

Chaining

Probe 1 Peer

Modeling Probe 2 Probe 3 Probe 4 Post-Test

100

0A cc

es si

n g

C o

m p

u te

r P

ro gr

am Imitation probe

Initial Probe

100

A d

d in

g w

it h

a C

al cu

la to

Imitation Probe

Initial Probe

Peer Modeling Eq ua

tio ns

Car d Atte

nd in

Cue Ph ys

ica l

Gui da

nc e

No Gui

da nc

Imitation Probe

100

Sh ar

p en

in g

a P

en ci

P er

ce n

ta ge

o f

C o

rr ec

t R

es p

o n

se s

5 10 15 20 25 30 35 40 45 50 55

Sessions

Peer Modeling

Initial Probe

Charlie

Figure 6 Percentage of steps performed correctly by Charlie for the three response chains. Triangles represent percentage of steps correct in initial probes; open circles represent percentage of steps correct in imitation probes. From “Peer Modeling of Response Chains: Observational Learning by Students with Disabilities,” by M. G. Werts, N. K. Caldwell, & M. Wolery, 1996, Journal of Applied Behavior Analysis, 29, p. 60. Copyright 1996 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

car by a chain, enter the car and pedal through a tunnel, climb a flight of stairs, run through an enclosed tube, enter an elevator, raise a miniature replica of the Brown University flag, exit the elevator, and finally press a bar for which he received a pellet of food. Barnabus be- came so famous for his performance of this elaborate sequence that he acquired the reputation of being “the rat with a college education.” This chain of 11 re- sponses was trained by initially conditioning the last response in the sequence (the bar press) in the presence of a buzzer sound, which was established as the SD for the bar press. Then the next-to-last response in the sequence (exiting the elevator) was conditioned in

the presence of the elevator at the bottom of the shaft. Each response in the chain was added in turn so that a discrete stimulus served as the SD for the next response and as the conditioned reinforcer for the preceding response.

To illustrate backward chaining with a classroom ex- ample, let us suppose that a preschool teacher wants to teach a student to tie his shoes. First, the teacher con- ducts a task analysis of shoe tying and puts the compo- nent behaviors in logical sequence.

1. Cross over laces on the shoe.

2. Tie a knot.

444

Chaining

15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0

N u

m b

er C

o rr

ec t

Baseline TT + Prompts

5 10 15 20 25 30 35 40 45

Setting 1 Setting 2 Setting 3

Student 1

15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0

N u

m b

er C

o rr

ec t

5 10 15 20 25 30 35 40 45

Student 2

Sessions

Figure 7 Number of correctly performed steps in the public-telephone task analysis by two students during baseline and total-task chaining (TT) plus response prompts. Data points for settings 2 and 3 show performance on generalization probes. From “Teaching Adolescents with Severe Disabilities to Use the Public Telephone,” by D. W. Test, F. Spooner, P. K. Keul, and T. A. Grossi, 1990, Behavior Modification, p. 165. Copyright 1990 by Sage Publications. Reprinted by permission.

3. Make a loop with the lace on the right side of the shoe, and hold it in the right hand.

4. With the left hand, wrap the other lace around the loop.

5. Use index or middle finger of the left hand to push the left lace through the opening between the laces.

6. Grasp both loops, one in each hand.

7. Draw the loops snug.

The teacher begins by training with the last step in the sequence, Step 7, until the student is able to com- plete it without mistakes for three consecutive trials. After each correct trial at Step 7, reinforcement is delivered. The teacher then introduces the next-to-last step, Step 6, and begins training the student to perform that step in conjunction with the student’s performance of the final

step in the chain. Reinforcement is contingent on the suc- cessful performance of Steps 6 and 7. The teacher then introduces Step 5, making sure that the training step and all previously learned steps (i.e., Steps 5, 6, and 7) are ex- ecuted in the proper sequence prior to reinforcement. The teacher can use supplemental response prompts to evoke correct responding on behavior at any step. How- ever, any supplemental response prompts—verbal, pic- torial, demonstration, or physical—introduced during training must be faded later in the program so that the student’s behavior comes under stimulus control of the naturally occurring SDs. In brief, when using a backward chaining procedure, the system of task analysis is arranged in reverse order and the last step is trained first.

With backward chaining, the first behavior the learner performs independently produces the terminal re- inforcement: The shoe is tied. The next-to-last response produces the onset of a stimulus condition that reinforces that step and serves as an SD for the last behavior, which is now established in the learner’s behavioral repertoire. This reinforcing sequence is repeated for the remainder of the steps.

Hagopian, Farrell, and Amari (1996) combined back- ward chaining with fading to reduce the life-threatening behavior of Josh, a 12-year-old male with autism and mental retardation. Because of medical complications as- sociated with a variety of gastrointestinal factors, fre- quent vomiting, and constipation during the 6 months prior to the study, Josh refused to ingest any food or liq- uids by mouth. Indeed, when he did ingest food by mouth previously, he expelled it.

Over the 70-day course of Josh’s treatment program, data were collected on liquid acceptance, expulsions, swallows, and avoidance of the target behavior: drinking water from a cup. After baseline was collected on Josh’s ability to swallow 10 cc of water, a swallow condition without water was implemented. Basically, an empty sy- ringe was placed and depressed in his mouth, and he was directed to swallow. Next, the syringe was filled with a small volume of water, and reinforcement delivered when Josh swallowed the water. In subsequent phases, the vol- ume of water in the syringe was gradually increased from 0.2 cc to 0.5 cc to 1 cc, and ultimately to 3 cc. To earn re- inforcement in subsequent conditions, Josh had to emit the target chain using at first 3 cc of water from a cup to ultimately 30 cc of water. At the end of the study, a suc- cessful generalization probe was implemented with 90 cc of a mixture of water and juice. The results showed that the chaining procedure improved Josh’s ability to emit the target chain (see Figure 8). Hagopian and col- leagues explained their procedure this way:

We began by targeting a preexisting response (swallow- ing), which was the third and final response in the chain of behaviors that constitute drinking from a cup. Next,

445

Chaining

100

P er

ce n

ta ge

o f

T ri

al s

51 10 15 20 25 30 35 40 45 50 55 60 65 70 Sessions

Success

Avoidant Behavior

10 c

c cu

p , b

as el

in e

Sw al

lo w

, n o

w at

Em p

ty s

yr in

Sy ri

n ge

d ip

p ed

in w

at er

0. 2

cc s

yr in

10 c

c cu

p , p

ro b

0. 5

cc s

yr in

1 cc

s yr

in ge

3 cc

s yr

in ge

3 cc

c u

10 c

c cu

p , p

ro b

e 3

cc c

u p

10 c

c cu

p , p

ro b

e 30

c c

cu p

90 c

c cu

p ge

n er

al iz

at io

break

Figure 8 Percentage of trials with successful drinking and trials with avoidance behaviors. From “Treating Total Liquid Refusal with Backward Chaining and Fading,” by L. P. Hagopian, D. A. Farrell, & A. Amari, 1996, Journal of Applied Behavior Analysis, 29, p. 575. Copyright 1996 by the Society for the Experimental Analysis of Behavior, Inc. Used with permission.

reinforcement was delivered for the last two responses in the chain (accepting and swallowing). Finally, rein- forcement was delivered only when all three responses in the chain occurred. (p. 575)

A primary advantage of backward chaining is that on every instructional the learner comes into contact with the terminal reinforcer for the chain. As a direct outcome of reinforcement, the stimulus that is present at the time of reinforcement increases its discriminative properties. Further, the repeated reinforcement of all behaviors in the chain increases the discriminative capability of all the stimuli associated with these behaviors and with the re- inforcer. The main disadvantage of backward chaining is that the potential passive participation of the learner in earlier steps in the chain may limit the total number of re- sponses made during any given training session.

Backward Chaining with Leap Aheads

Spooner, Spooner, and Ulicny (1986) reported the use of a variation of backward chaining they called back- ward chaining with leap aheads. Backward chaining with leap aheads follows essentially the same proce- dures as backward chaining, except that not every step in the task analysis is trained. Some steps are simply probed. The purpose of the leap-aheads modification is to decrease total training time needed to learn the chain. With conventional backward chaining, the stepwise rep- etition of the behaviors in the sequence may slow down the learning process, especially when the learner has thoroughly mastered some of the steps in the chain. For instance, in the previous shoe-tying illustration, the child

might perform Step 7, the last behavior in the sequence, and then leap ahead to Step 4 because Steps 5 and 6 are in her repertoire. It is important to remember, however, that the learner must still perform Steps 5 and 6 cor- rectly and in sequence with the other steps to receive reinforcement.

Forward, Total-Task, or Backward Chaining: Which to Use?

Forward chaining, total-task chaining, and backward chaining have each been shown to be effective with a wide range of self-care, vocational, and independent- living behaviors. Which chaining procedure should be the method of first choice? Research conducted to date does not suggest a clear answer. After examining evi- dence reported between 1980 and 2001, Kazdin (2001) concluded: “Direct comparisons have not established that one [forward chaining, backward chaining, or total- task chaining] is consistently more effective than the other” (p. 49).

Although overwhelming conclusive data do not favor one chaining method over another, anecdotal evidence and logical analysis suggest that total-task chaining may be appropriate when the student (a) can perform many of the tasks in the chain, but needs to learn them in se- quence; (b) has an imitative repertoire; (c) has moderate to severe disabilities (Test et al., 1990); and/or (d) when the task sequence or cycle is not very long or complex (Miltenberger, 2001).

The uncertainty about which approach to follow may be minimized by conducting a personalized task analysis for the learner, systematically applying the

446

Chaining

single- or multiple-opportunity method to determine the starting point for instruction, relying on empirically sound data-based studies in the literature, and collect- ing evaluation data on the approach to determine its ef- ficacy for that person.

Interrupting and Breaking Behavior Chains Our discussion thus far has focused on procedures for building behavior chains. As we have seen, chaining has been used successfully to increase and improve the repertoires of a wide range of people and across a vari- ety of tasks. Still, simply knowing how to link behavior to behavior is not always sufficient for practitioners. In some situations, knowing how a behavior chain works can be used to interrupt the execution of an existing chain (e.g., making toast) to advance a different re- sponse class of skills (e.g., a speech skill). Further, be- cause some chains are inappropriate (e.g., excessive consumption of food), knowing how to break an inap- propriate behavior chain can achieve positive results (e.g., the person eats an appropriate amount of food and stops eating at that point).

Behavior Chain Interruption Strategy

The behavior chain interruption strategy (BCIS) relies on the participant’s skill to perform the critical elements of a chain independently, but the chain is interrupted at a pre- determined step so that another behavior can be emitted. Initially developed to increase speech and oral responding (Goetz, Gee, & Sailor, 1985; Hunt & Goetz, 1988), the BCIS has been extended into pictorial communication sys- tems (Roberts-Pennell & Sigafoos, 1999), manual signing (Romer, Cullinan, & Schoenberg, 1994), and switch acti- vation (Gee, Graham, Goetz, Oshima, & Yoshioka, 1991).

The BCIS works as follows. First, an assessment is conducted to determine if the person can independently complete a chain of three or more components. Figure 9 shows an example of how a toast-making chain, di- vided into five steps, was assessed across degrees of dis- tress and attempts to complete when specific steps were blocked or interrupted by an observer. Degree of distress was ranked on a 3-point scale, and attempts were regis- tered using a dichotomous scale (i.e., yes or no). The as- sessment yielded a mean distress ranking of 2.3 and a total percentage of attempts (66%).

With respect to using BCIS in applied settings to in- crease behavior, a chain is selected for training based on

Figure 9 Sample score sheet used to evaluate chains.

Chain Pretesting

Student’s Name: Chuck Chain: making toast

Date: 5/2 *interrupt at start of step

Attempts to Sequence of steps Degree of distress complete Comments

1. Pull bread from bag* 1 2 3 yes no (low) (high)

2. Put in toaster 1 2 3 yes no

3. Push down* 1 2 3 yes no tries to touch knob

4. Take toast out 1 2 3 yes no

5. Put on plate* 1 2 3 yes no very upset; self-abusive

Total 2/3 = 66%

From “Using a Chain Interruption Strategy to Teach Communication Skills to Students with Severe Disabili- ties,” by L. Goetz, K. Gee, & W. Sailor, Journal of the Association of Persons with Severe Handicaps, 13 (1), p. 24. Copyright by the Association for Persons with Severe Handicaps. Used by permission.

x distress: 7 3

� 2.3

447

Chaining

Figure 10 Key components and features of the BCIS.

• Instruction begins in the middle of the chain sequence, not the start, differing from forward and backward chaining procedures. Because the instruction begins in the middle of the sequence, interrupting the sequence might function as a transitive conditioned motivating operation or negative reinforcer, the removal of which increases behavior.

• Procedurally, the BCIS is predicated on an assessment that verifies that the person is able to complete the chain independently, but experiences moderate distress when the chain is interrupted in the middle of the sequence.

• Verbal prompts are used at the point of interruption (e.g., “What do you want?”), but a full range of response prompts—modeling and physical guidance—has also been employed.

• Interruption training occurs in the natural setting (e.g., a water basin for washing hair; the microwave oven for making cookies).

• Maintenance, generalization, and social validity data, although not overwhelmingly evident in every study, are sufficiently robust to suggest that the BCIS be included with other interventions (e.g., mand-model, time delay, and incidental teaching).

it being moderately distressful for the individual when she is interrupted while performing the chain, but not so distressful that the person’s behavior becomes episodic or self-injurious. After collecting baseline on a target be- havior (e.g., vocalizations), the person is directed to start the chain (e.g., “Make toast”). At a predetermined step in the chain—say, Step 3 (push down)—the individual’s ability to complete the chain is restricted. For example, the practitioner might momentarily and passively block access to the toaster. The person would then be prompted, “What do you want?” A vocalization would be required for the chain to be completed. That is, the person would have to make a response such as saying, “Push down.”

Although the exact behavioral mechanisms respon- sible for improved performance as a result of using BCIS have not been pinpointed, field-based research efforts, including those examining generalized outcomes (Grun- sell & Carter, 2002), have established it as an efficacious approach, especially for persons with severe disabilities. For example, Carter and Grunsell’s (2001) review of the literature on the efficacy of BCIS shows that it is positive and beneficial. In their view,

The BCIS may be seen as empirically supported and complementary to other naturalistic techniques. . . . A small, but growing body of research literature has demonstrated that individuals with severe disabilities have acquired requesting with the BCIS. In addition, re- search demonstrates that implementation of the BCIS may increase the rate of requesting. (p. 48)

The assumption underlying BCIS assessment is that “persistence in task completion and emotional response to the interruption would serve as operational definitions of high motivation for task completion” (Goetz et al., 1985, p. 23). It could also be claimed that the interruption,

because it momentarily detains the learner from obtain- ing the reinforcer for completing the task, and serves as a blocked establishing operation, a condition that exists when a reinforcer cannot be obtained without some other additional action or behavior occurring. Further, the role that negative reinforcement plays in BCIS and the sys- tematic environmental change that occurs at the point of the interruption have not been fully analyzed to deter- mine their relative contributions.

According to Carter and Grunell (2001), the BCIS consists of several components and features that make it useful as a behavior change tactic in natural settings. Figure 10 provides a review of these key features.

Breaking an Inappropriate Chain

An inappropriate behavior chain (e.g., nail biting, smok- ing, excessive food consumption) can be broken by de- termining the initial SD and substituting an SD for an alternative behavior, or by extending the chain and build- ing in time delays. Given that the first SD in a chain evokes for the first response, which in turn terminates that SD and produces the second SD, and so on through- out the chain. If the first SD appears less frequently, the entire chain occurs less often. Martin and Pear (2003) suggested that a behavior chain that reinforces excessive food consumption, for example, might be broken by in- troducing links that require the individual to place the eating utensil on the table between bites or introducing a 3- to 5-second time delay before the next bite can be initiated. In their view, “In the undesirable chain, the person gets ready to consume the next mouthful before even finishing the present one. A more desirable chain separates these components and introduces brief delays” (p. 143).

448

Chaining

Figure 11 Illustration of an original behavior chain (upper panel) and its revision (lower panel) to break an inappropriate response sequence.

With respect to separating chain components, let us consider the case of a restaurant trainee with mod- erate mental retardation who has been taught to bus ta- bles. Upon completion of the training program, the trainee was able to perform each of the necessary be- haviors in the table-bussing chain accurately and pro- ficiently (see Figure 11, upper panel). However, on the job site, the new employee started to emit an inap- propriate sequence of behaviors. Specifically, in the presence of the dirty dishes on the empty table, the in- dividual scraped excess food scraps onto the table itself, instead of placing the dirty items on the food cart. In other words, the initial SD for the table-bussing chain, the empty table with the dirty plates, was setting the occasion for a response (dish scraping) that should have occurred later in the chain. To break this inappropriate chain, the behavior analyst should consider several pos- sible sources of difficulty, including (a) reexamining the SDs and responses, (b) determining whether similar SDs cue different responses, (c) analyzing the job set- ting to identify relevant and irrelevant SDs, (d) deter- mining whether SDs in the job setting differ from training SDs, and (e) identifying the presence of novel stimuli in the environment.

Reexamine the SDs and Responses

The purpose of reexamining the list of SDs and re- sponses in the task analysis is to determine whether the original chain of associated responses they evoke is ar- bitrary, based primarily on expert opinion, time and mo- tion studies, and practical efficiency. In our example, the trainer wants the presence of the dirty dishes on the

empty table to evoke the response of placing those dishes in the bus cart. Consequently, a rearranged chain of SDs and responses was trained (see Figure 11, lower panel).

Determine Whether Similar SDs Cue Different Responses

Figure 11 (upper panel) shows two similar SDs— dirty dishes on the empty table and dirty dishes on the cart—which might have contributed to the scraping of food onto the table. In other words, R2 (scrape dishes) might have come under the control of S1 (dirty dishes on table). Figure 11 (lower panel) shows how the be- havior analyst corrected the sequence by rearranging the SDs and their associated responses. Scraping the dishes is now the fifth response in the chain and occurs at the kitchen sink, an area away from the restaurant ta- bles and diners. Thus, any potential confusion would be reduced or eliminated.

Analyze the Natural Setting to Identify Relevant and Irrelevant SDs

The training program should be designed so that the learner is trained to discriminate the relevant (i.e., critical) compo- nents of a stimulus from the irrelevant variations. Figure 11 (lower panel) shows at least two relevant character- istics of S1: an empty table and the presence of dirty dishes on that table. An irrelevant stimulus might be the location of the table in the restaurant, the number of place settings on the table, or the arrangement of the place settings. A rel- evant characteristic of S5, presence of the kitchen sink,

449

Chaining

would be water faucets, the sink configuration, or dirty dishes. Finally, irrelevant stimuli might be the size of the kitchen sink and the type or style of faucets.

Determine Whether SDs in the Natural Setting Differ from Training SDs

It is possible that some variations of the SDs cannot be taught during training phases. For this reason many au- thors recommend conducting the final training sessions in the natural setting where the behavior chain is expected to be performed. This allows any differences that exist to be recognized by the trainer, and subsequent discrim- ination training to be further refined on site.

Identify the Presence of Novel Stimuli in the Environment

The presence of a novel stimulus unexpected in the orig- inal training situation can also prompt the occurrence of an inappropriate chain. In the restaurant example, the presence of a crowd of customers might set the occasion for the chain to be performed out of sequence. Likewise, distracting stimuli (e.g., customers coming and going, tips left on the table) might set the occasion for an inap- propriate chain. Also, a coworker might unwittingly give contradictory instructions to the trainee. In any of these situations, the novel stimuli should be identified and the learner taught to discriminate them along with other SDs in the environment.

Factors Affecting the Performance of a Behavior Chain Several factors affect the performance of a behavior chain. The following sections outline these factors and provide recommendations for addressing them.

Completeness of the Task Analysis

The more complete and accurate the task analysis is, the more likely a person will be to progress through the se- quence efficaciously. If the elements making up the chain are not sequenced appropriately, or if the corresponding SDs are not identified for each response, learning the chain will be more difficult.

The behavior analyst should remember two key points when attempting to develop an accurate task analy- sis. First, planning must occur before training. The time

devoted to constructing and validating the task analysis is well spent. Second, after the task analysis is con- structed, training should begin with the expectation that adjustments in the task analysis, or the use of more in- trusive prompts may be needed at various steps in the task analysis. For example, McWilliams and colleagues (1990) noted that one of their students required exten- sive trials and more intrusive prompts with some steps in the task analysis before improved performance was obtained.

Length or Complexity of the Chain

Longer or more complex behavior chains take more time to learn than shorter or less complex behavior chains. Likewise, the behavior analyst might expect training time to be longer if two or more chains are linked.

Schedule of Reinforcement

When a reinforcer is presented subsequent to the per- formance of a behavior in a chain, it affects each of the responses making up the chain. However, the effect on each response is not identical. For example, in back- ward chaining, responses performed at the end of the chain are strengthened faster than responses made ear- lier in the chain because they are reinforced more fre- quently. The behavior analyst is advised to remember two points: (a) A chain can be maintained if an appro- priate schedule of reinforcement is used, and (b) the number of responses in a chain may need to be consid- ered when defining the schedule of reinforcement.

Stimulus Variation

Bellamy, Horner, and Inman (1979) provided an excel- lent pictorial representation of how stimulus variation affects the performance of a chain. The upper photo- graph in Figure 12 shows a cam switch bearing be- fore and after its placement on a cam switch axle. The lower photograph shows the four different types of bear- ings that can be used in the assembly process. The re- sponse of placing the bearing on the axle must be under the control of the presence of the bearing (SD); how- ever, variation among bearings requires that the response be under the control of several SDs, each sharing a critical feature. In the illustration each bearing has a 1.12-cm hole in the center and one or more hex nut slots on a face of the bearing. Any bearing with these stimu- lus features should evoke the response of placing the bearing on the axle, even if other irrelevant dimensions

450

Chaining

Figure 12 Upper panel: A cam switch bearing before and after placement on a cam switch axle. Lower panel: Four different types of bearings used in cam switch assembly. From Vocational Habilitation of Severely Retarded Adults, pp. 40 & 42, by G. T. Bellamy, R. H. Horner, and D. P. Inman, 1979, Austin, TX: PRO-ED. Copyright 1979 by PRO-ED. Reprinted by permission.

are present (e.g., color, material composition, weight). Stimuli without these features should not occasion the response.

If possible, the behavior analyst should introduce all possible variations of the SD that the learner will en- counter. Regardless of the behavior chain, presentation of stimulus variations increases the probability that the correct response will occur in their presence. This in- cludes, for example, in an assembly task, various canis- ters, and shafts; with dressing skills, different fasteners, zippers, and buttons; and with tooth brushing, assorted tubes and pumps.

Response Variation

Often when stimulus variations occur, response varia- tion also must occur to produce the same effect. Again,

Bellamy and colleagues (1979) provided an illustration with the cam shaft assembly. In the upper-left photo- graph of Figure 13, the bearing has been placed on the cam shaft, and the retaining clip is being positioned for placement with a pair of nose pliers. The upper- right photograph shows the clip in position. The lower- left photograph shows a different bearing configuration (i.e., the SD is different) requiring a different response. Instead of the retaining clip being lifted over the bear- ing cap with pliers, it must be pushed over the cap with a wrench-type tool. The response of lifting or pushing has changed, as has the response of selecting the ap- propriate tool. Thus, the behavior analyst should be aware that when stimulus variation is introduced, train- ing or retraining of responses within the chain might also be required.

451

Chaining

Figure 13 Two ways that retaining rings are applied to affix bearings to cam axles. From Vocational Habilitation of Severely Retarded Adults, p. 44, by G. T. Bellamy, R. H. Horner, and D. P. Inman, 1979, Austin, TX: PRO-ED. Copyright 1979 by PRO-ED. Reprinted by permission.

Summary Definition of a Behavior Chain

1. A behavior chain is a specific sequence of discrete re- sponses, each associated with a particular stimulus con- dition. Each discrete response and the associated stimulus condition serve as an individual component of the chain. When individual components are linked together, the re- sult is a behavior chain that produces a terminal outcome.

2. Each stimulus that links two sequential responses in a chain serves dual functions: It is a conditioned reinforcer for the response that produced it and an SD for the next re- sponse in the chain.

3. In a behavior chain with a limited hold, a sequence of be- haviors must be performed correctly and within a specific time for reinforcement to be delivered. Proficient re- sponding is a distinguishing features of chains with lim- ited holds.

Rationale for Using Chaining

4. Three reasons a behavior analyst should be skilled in build- ing behavior chains: (a) chains can be used to improve in- dependent-living skills; (b) chains can provide the means by which other behaviors are combined into more com- plex sequences; and (c) chains can be combined with other procedures to build behavioral repertoires in generalized settings.

5. Chaining refers to the way these specific sequences of stimuli and responses are linked to form new perfor- mances. In forward chaining, behaviors are linked together beginning with the first behavior in the sequence. In back- ward chaining, the behaviors are linked together begin- ning with the last behavior in the sequence.

Task Analysis

6. Task analysis involves breaking a complex skill into smaller, teachable units, the product of which is a series of sequentially ordered steps or tasks.

7. The purpose of constructing and validating a task analy- sis is to determine the sequence of stepwise critical be- haviors that comprise the complete task and that would lead to it being performed efficiently. Task analyses can be constructed by observing a competent person perform the task, consulting experts, and self-performing the sequence.

8. The purpose of assessing mastery level is to determine which components of the task analysis can already be per- formed independently. Assessment can be conducted with the single-opportunity or multiple-opportunity method.

Behavior Chaining Methods

9. In forward chaining, the behaviors identified in the task analysis are taught in their natural sequence. Specifically, reinforcement is delivered when the predetermined criterion for the first behavior in the sequence is achieved. There- after, reinforcement is delivered for criterion completion of Steps 1 and 2. With each successive step, reinforcement is delivered contingent on the correct performance of all steps trained thus far.

10. Total-task chaining is a variation of forward chaining in which the learner receives training on each step in the task analysis during every session. Trainer assistance is provided using response prompts with any step that the individual is not able to perform. The chain is trained until the learner performs all the behaviors in the sequence to criterion.

452

Chaining

11. In backward chaining, all the steps identified in the task analysis are completed by the trainer, except the last one. When the final step in the sequence is performed to crite- rion, reinforcement is delivered. Next, reinforcement is delivered when the next-to-last step and the last step are performed. Subsequently, the individual must perform the last three steps before reinforcement is delivered, and so on. The primary advantage of backward chaining is that the learner comes into contact with the contingencies of rein- forcement immediately, and the functional relationship be- gins to develop.

12. Backward chaining with leap aheads follows essentially the same procedures as in backward chaining, except that not every step in the task analysis is trained. The leap- aheads modification provides for probing or assessing un- trained behaviors in the sequence. Its purpose is to speed up training of the behavior chain.

13. The decision to use forward chaining, total-task chaining, or backward chaining should be based on the results of a task analysis assessment, empirically sound data-based studies, and a functional evaluation, taking into consider- ation the cognitive, physical, and motoric abilities and needs of the individual.

Interrupting and Breaking Behavior Chains

14. The behavior chain interruption strategy (BCIS) is an in- tervention that relies on the participant’s skill in perform- ing the critical elements of a chain independently, but the chain is interrupted at a predetermined step so that another behavior can be emitted.

15. An inappropriate chain can be broken if the initial SD that sets the occasion for the first behavior in the chain is rec- ognized and an alternative SD is substituted. Alternative SDs can be determined by reexamining the list of SDs and responses in the task analysis, determining whether simi- lar SDs cue different responses, analyzing the natural set- ting to identify relevant and irrelevant SDs, determining whether the SDs in the natural setting differ from training SDs, and/or identifying the presence of novel SDs in the environment.

Factors Affecting the Performance of a Behavior Chain

16. Factors affecting the performance of a behavior chain in- clude: (a) completeness of the task analysis, (b) length or complexity of the chain, (c) schedule of reinforcement, (d) stimulus variation, and (e) response variation.

453

clicker training differential reinforcement

response differentiation shaping

successive approximation

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 9: Behavior Change Procedures

9-6 Use differential reinforcement. 9-11 Use shaping.

Shaping

Key Terms

454

Shaping is the process of systematically and differentially reinforcing successive approxi- mations to a terminal behavior. Shaping is used

in many everyday situations to help learners acquire new behaviors. For example, language therapists use shaping when they develop speech with a client by first reinforcing lip movements, then sound production, and finally word and sentence expression. Teachers working with students with severe disabilities shape social in- teractions when they differentially reinforce eye con- tact, one-word greetings, and conversational speech. A basketball coach shapes the foul-shooting behavior of her players when she differentially reinforces accurate shooting from positions a few feet in front of the basket to positions closer to the 15-foot regulation foul line. Further, trainers use shaping to teach desirable behav- ior to animals for function (e.g., loading horses into a trailer without injury to the horse or the attendant), or for appeal or utility duty (e.g., teaching porpoises to execute show routines).

Depending on the complexity of a given behavior and the learner’s prerequisite skills, shaping may require many successive approximations before the terminal be- havior is achieved. Achievement of the terminal behav- ior with respect to time, trials, or direction is seldom predictable, immediate, or linear. If the learner emits a closer approximation to the terminal behavior, and the practitioner fails to detect and reinforce it, achievement of the terminal behavior will be delayed. However, if a systematic approach is used—that is, if each instance of closer approximations to the terminal behavior is detected and reinforced—progress can usually be attained more quickly. Although shaping can be time-consuming, it rep- resents an important approach to teaching new behav- iors, especially those behaviors that cannot easily be learned by instructions, incidental experience or expo- sure, imitation, physical cues, or verbal prompts.

This chapter defines shaping, presents examples of how to shape behavior across and within different re- sponse topographies, and suggests ways to improve the efficiency of shaping. Clicker training is illustrated as one method that trainers use to shape new behaviors in animals. Next, guidelines for implementing shaping are presented. The chapter concludes with a look at future applications of shaping.

Definition of Shaping In Science and Human Behavior, Skinner (1953) pre- sented the concept of shaping with an analogy:

Operant conditioning shapes behavior as a sculptor shapes a lump of clay. … The final product seems to

have a special unity or integrity of design, but we cannot find a point at which this suddenly appears. In the same sense, an operant is not something which appears full grown in the behavior of the organism. It is the result of a continuous shaping process. (p. 91)

Through careful and skillful manipulation of the orig- inal, undifferentiated lump of clay, the artisan keeps some aspects of the clay in its primary location, cuts other pieces away, and reforms and molds still other sections so that the form is slowly transfigured into the final sculpted design. Similarly, the skillful practitioner can shape novel forms of behavior from responses that initially bear little resemblance to the final product. A practitioner using be- havioral shaping differentially reinforces successive ap- proximations toward a terminal behavior. The end product of shaping—a terminal behavior—can be claimed when the topography, frequency, latency, duration, or ampli- tude/magnitude of the target behavior reaches a prede- termined criterion level. The two key procedural components of shaping, differential reinforcement and successive approximations, are described next.

Differential Reinforcement When a heavy ball is thrown beyond a certain mark, when a horizontal bar is cleared in vaulting or in jump- ing, when a ball is batted over the fence (and when, as a result, a record is broken or a match or game won), dif- ferential reinforcement is at work.

– B. F. Skinner (1953, p. 97)

Differential reinforcement is a procedure in which re- inforcement is provided for responses that share a predetermined dimension or quality, and in which re- inforcement is withheld for responses that do not demonstrate that quality. For example, differential re- inforcement is used by a parent who complies with his child’s requests for an item at the dinner table when those requests include polite words such as “please” or “may I,” and who does not pass the requested item fol- lowing requests in which polite words are missing. Dif- ferential reinforcement has two effects: responses similar to those that have been reinforced occur with greater frequency, and responses resembling the unre- inforced members are emitted less frequently (i.e., they undergo extinction).

When differential reinforcement is applied consis- tently within a response class, its dual effects result in a new response class composed primarily of responses sharing the characteristics of the previously reinforced subclass. This emergence of a new response class is called response differentiation. Response differentiation as

Shaping

455

Shaping

Figure 1 Approximations in differentially reinforcing the wearing of eyeglasses. Shaded portion includes behaviors no longer reinforced. From How to Use Shaping by M. Panyan, 1980, p. 4, Austin, TX: PRO-ED. Copyright 1980 by PRO-ED. Reprinted by permission.

a result of the parent’s differential reinforcement at the dinner table would be evident if all of the child’s requests contained polite words.

Successive Approximations

The practitioner using shaping differentially reinforces responses that resemble the terminal behavior in some way. The shaping process begins with reinforcement of responses in the learner’s current repertoire that share an important topographical feature with the terminal be- havior or are a prerequisite behavior for the final termi- nal behavior. When the initially reinforced responses become more frequent, the practitioner shifts the crite- rion for reinforcement to responses that are a closer ap- proximation of the terminal behavior. The gradually changing criterion for reinforcement during shaping results in a succession of new response classes, or suc- cessive approximations, each one closer in form to the terminal behavior than the response class it replaces. Skinner (1953) discussed the critical nature of succes- sive approximations as follows:

The original probability of the response in its final form is very low; in some cases it may even be zero. In this way we build complicated operants that would never ap- pear in the repertoire of the organism otherwise. By re- inforcing a series of successive approximations, we bring a rare response to a very high probability in a short time. This is an effective procedure because it recog- nizes and utilizes the continuous nature of a complex act. (p. 92)

Figure 1 illustrates the progression of successive approximations used by Wolf, Risley, and Mees (1964) in

shaping the wearing of eyeglasses in a preschooler who was in danger of losing his eyesight if he did not wear corrective glasses regularly. Touching glasses was the first behavior that was reinforced. When touching glasses was established, picking up glasses was reinforced, and touching glasses was placed on extinction (see shaded portion of figure). Next, putting glasses up to his face was reinforced, and the two previously reinforced be- haviors were placed on extinction. Training continued until the terminal behavior, placing the glasses on his face, was emitted, and all previous behaviors were placed on extinction.

Shaping Different Dimensions of Performance

Behavior can be shaped in terms of topography, fre- quency, latency, duration, and amplitude/magnitude (see Table 1). Differential reinforcement could also be used to teach a child to talk within a conversational decibel range. Let us suppose a behavior analyst is working with a student who usually talks at such a low volume (e.g., below 45 decibels) that his teacher and peers have diffi- culty hearing him. Successive approximations to 65 deci- bels (dB)—the amplitude of normal conversational speech—might be 45, 55, and ultimately 65 dB. Differ- ential reinforcement of speaking at a minimum volume of 45 dB would place responses below that amplitude on extinction. When the student is speaking consistently at or above 45 dB, the criterion would be raised to 55 db. Likewise, when 55 dB and finally 65 dB are achieved, previous lower amplitude levels are not reinforced (i.e., they are placed on extinction).

Fleece and colleagues (1981) used shaping to in- crease the voice volume of two children enrolled in a private preschool for students with physical and devel- opmental disabilities. Baseline data were collected on voice volume in the regular classroom. Voice volume was measured on a 0- to 20-point scale with 0 indicat- ing that the child’s voice level was usually inaudible, 10 indicating normal voice volume, and 20 indicating screaming. The shaping procedure consisted of having the children recite a nursery rhyme in the presence of a voice-activated relay device, whereby voice volume ac- tivated a light display. The intensity of the light corre- sponded to increased levels of voice volume: Higher voice volume produced brighter light, and lower voice volume produced a dimmer light. The teacher shaped voice volume by increasing the sensitivity threshold of the relay device. That is, whereas a low voice was suffi- cient to activate the dim light in the beginning stages of training, a much higher volume was needed to produce the same effect in later stages. Each raising of the volume

putting glasses up to his face

touching glasses

picking up glasses

touching glasses

picking up glasses

touching glasses

456

Shaping

Table 1 Examples of Performance Improvements that Could Be Shaped in Various Dimensions of Behavior

Dimension Example

Topography (form of • Refining motor movements associated with a golf swing, throwing motion, or vaulting behavior. the behavior) • Improving cursive or manuscript letter formation during handwriting exercises.

Frequency (number of • Increasing the number of problems completed during each minute of a math seatwork responses per unit of time) assignment.

• Increasing the number of correctly spelled and appropriately used words written per minute.

Latency (time between the • Decreasing compliance time between a parental directive to “clean your room” and the onset of onset of the antecedent room-cleaning behavior. stimulus and the occurrence • Increasing the delay between the onset of an aggressive remark and retaliation by a student with of the behavior) severe emotional disabilities.

Duration (total elapsed time • Increasing the length of time that a student stays on task. for occurrence of the • Increasing the number of minutes of engaged study behavior. behavior)

Amplitude/Magnitude • Increasing the projected voice volume of a speaker from 45 dB to 65 dB. (response strength • Increasing the height of the high jump bar for students enrolled in a physical education class. or force)

level required to activate the light represented a succes- sive approximation to the terminal volume.

Analysis of the children’s performance using a mul- tiple-baseline across students design indicated that the

voice volume increased in the classroom setting as a func- tion of treatment (see Figure 2). Also, the children’s voice volume remained high after a 4-month period. Fi- nally, according to anecdotal reports by school staff, the

Figure 2 Voice-volume levels per session in the classroom setting. From “Elevation of Voice Volume in Young Developmentally Delayed Children via an Operant Shaping Procedure” by L. Fleece, A. Gross, T. O’Brien, J. Kistner, E. Rothblum, and R. Drabman, 1981, Journal of Applied Behavioral Analysis, 14, p. 354. Copyright by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Usually screaming

Normal

Usually inaudible

Usually screaming

Normal

V o

ic e

V o

lu m

Sessions

Usually inaudible

Follow-up

1 month follow-up 4 month follow-up

457

Shaping

children’s higher voice volume generalized to other set- tings beyond the classroom.

From a practical standpoint, measures of increased decibels could be obtained with a voice-sensitive tape recorder or an audiometric recording device that emits a signal (light or sound) only when a specific threshold level is achieved (Fleece et al., 1981). Productions below the criterion level would not activate the recording device.

Shaping across and within Response Topographies

Shaping behavior across different response topogra- phies means that select members of a response class are differentially reinforced, whereas members of other response classes are not reinforced. As stated previ- ously, lip movements, speech sounds, one-word utter- ances, and phrase or sentence productions represent the different topographies of the layer response class of speaking behaviors; they are the prerequisite behaviors of speaking. When shaping behavior across different response topographies, the practitioner gradually in- creases the criterion of performance before delivering reinforcement.

Isaacs, Thomas, and Goldiamond (1960) reported a classic study showing how behaviors can be shaped across and within response topographies. They success- fully shaped the vocal behavior of Andrew, a man diag- nosed with catatonic schizophrenia, who had not spoken for 19 years despite many efforts to encourage speech production. Essentially, the shaping procedure was ini- tiated when an astute psychologist noticed that Andrew’s usually passive expression changed slightly when a pack- age of chewing gum inadvertently dropped on the floor. Realizing that gum might be an effective reinforcer for building behaviors in the response class of talking, the psychologist selected speech production as the terminal behavior.

The next step in the shaping process was to select an initial behavior to reinforce. Lip movement was chosen because the psychologist noted that slight lip movements had occurred in the presence of the pack of gum, and more important, lip movement was in the response class of speech. As soon as lip movement was established by differential reinforcement, the psychologist waited for the next approximation of the terminal behavior. During this phase, lip movement alone was no longer reinforced; only lip movements with sound produced the reinforce- ment. When Andrew began making guttural sounds, vo- calizations were differentially reinforced. Then the

guttural sound itself was shaped (differential reinforce- ment within a response topography) until Andrew said the word gum. After the 6th week of shaping, the psy- chologist asked Andrew to say gum, to which Andrew responded, “Gum, please.” During that session and after- wards, Andrew went on to converse with the psychologist and others at the institution about his identity and back- ground. In this powerful demonstration of shaping, after selection of the terminal behavior and the initial starting point, each member of the response class was shaped by differential reinforcement of successive approximations to the terminal behavior.

Shaping a behavior within a response topography means that the form of the behavior remains constant, but differential reinforcement is applied to another mea- surable dimension of the behavior. To illustrate, let us suppose that in a college-level physical education class, the teacher is instructing the students on water safety. Specifically, she is teaching them how to throw a life pre- server a given distance to a person struggling in the water. Since the important skill in this activity is to throw the life preserver near the person, the physical education teacher might shape accurate tossing by reinforcing successive approximations to a toss of a given distance. In other words, each toss that is near the person (e.g., within 2 meters) will be praised, whereas tosses outside that range will not be. As students become more accurate, the area can be reduced so that the terminal behavior is a toss within arm’s length of the person. In this case, the mag- nitude of the behavior is being shaped; the form of the toss remains the same.

Another example of shaping within a response topog- raphy is a parent who attempts to increase the duration of her child’s piano practice. The criterion for success—the terminal behavior—in this particular program might be to have the child practice for 30 minutes three times per week (e.g., Monday, Wednesday, and Friday). To ac- complish her objective, the parent could reinforce pro- gressively longer periods of practice, perhaps beginning with a few minutes one night per week. Next, the parent might reinforce increasing durations of practice: 10, 12, 15, 20, 25, and ultimately 30 minutes of practice only on Mondays. No contingency would be in effect for the other two days. As soon as an intermediate criterion level is reached (e.g., 20 minutes), reinforcement is no longer de- livered for less than 20 minutes of practice unless per- formance stalls at a higher level and progress is impeded.

During the next phase of shaping, the process is re- peated for Wednesday nights. Now the child must meet the criterion on both days before reinforcement is deliv- ered. Finally, the sequence is repeated for all three nights. It is important to remember that the behavior being shaped in this example is not piano playing. The child can already

458

Shaping

play the piano; the topography of that response class has been learned. What is being shaped by differential rein- forcement is a dimension of behavior within the response class, namely, the duration of piano practice.

Positive Aspects of Shaping

Shaping teaches new behaviors. Because shaping is im- plemented systematically and gradually, the terminal be- havior is always in sight. Also, shaping uses a positive approach to teach new behavior. Reinforcement is deliv- ered consistently upon the occurrence of successive approximations to the terminal behavior, whereas nonapproximations are placed on extinction. Punishment or other aversive procedures are typically not involved in a shaping program. Finally, shaping can be combined with other established behavior change or behavior-build- ing procedures (e.g., chaining). For example, suppose that a behavior analyst designed a seven-step task analy- sis to teach a child to tie his shoes. However, the child was unable to complete Step 5 in the task analysis. Shap- ing can be used in isolation to teach a closer approxima- tion to that step. Once Step 5 was learned through shaping, chaining would continue with the rest of the steps in the task analysis.

Limitations of Shaping

At least five limitations of shaping can be identified. Prac- titioners should be aware of these limitations and be pre- pared to deal with them as they arise. First, shaping new behavior can be time-consuming because many approx- imations may be necessary before the terminal behavior is achieved (Cipani & Spooner, 1994).

Second, progress toward the terminal behavior is not always linear. That is, the learner does not always proceed from one approximation to the next in a con- tinuous, logical sequence. Progress may be erratic. If the behavior is too erratic (i.e., not resembling a closer approximation to the terminal behavior), an approxi- mation may need to be further reduced, allowing for more reinforcement and progress. The skill of the prac- titioner in noting and reinforcing the next smallest ap- proximation to the terminal behavior is critical to the success of shaping. If the practitioner fails to reinforce responses at the next approximation—because of ne- glect, inexperience, or preoccupation with other tasks— the occurrence of similar responses may be few and far between. If reinforcement for performance at a given approximation continues longer than necessary, progress toward the terminal behavior will be impeded. 1The researcher provided a safety net so that the rat was not injured.

Third, shaping requires the practitioner to consis- tently monitor the learner to detect subtle indications that the next approximation that is closer to the terminal be- havior has been performed. Many practitioners—for ex- ample, teachers in busy or demanding classrooms—are not able to monitor behavior closely to note small changes. Consequently, shaping may be conducted inap- propriately or at least inefficiently.

Fourth, shaping can be misapplied. Consider the child trying to obtain her father’s attention by emitting low-volume demands (e.g., “Dad, I want some ice cream”). The father does not attend to the initial calls. As the child’s attempts become increasingly unsuccess- ful, she may become more determined to gain her fa- ther’s attention. In doing so, the frequency and amplitude of the calls may increase (e.g., “DAD, I want ice cream!”). After listening to the crescendo of vocal de- mands, the father eventually provides the ice cream. The next time, the child states what she wants in an even louder voice before getting her father’s attention. In this scenario, the father has differentially reinforced an ever- rising level of attention-getting behavior and shaped higher levels of call-outs for ice cream. Using this ex- ample as a backdrop, Skinner (1953) pointed out: “Dif- ferential reinforcement applied by a preoccupied or negligent parent is very close to the procedure we should adopt if we were given the task of conditioning a child to be annoying” (p. 98).

Finally, harmful behavior can be shaped. For exam- ple, Rasey and Iversen (1993) showed that differential re- inforcement could be used to shape a rat’s behavior to the point that the rat fell from the edge of a platform. By dif- ferentially reinforcing the rat for sticking its nose farther over the edge for food, the rat eventually fell off the ledge.1 It is not difficult to speculate that adolescent games such as Dare and Double Dare, which have evolved and become popularized in thrill- and fear-seeking television shows, capitalize on persons receiving differential rein- forcement for increasingly higher levels of risk-taking that can lead to dangerous—and sometimes tragic— behavior.

Shaping versus Stimulus Fading

Shaping and fading both change behavior gradually, al- beit in much different ways. In shaping, the antecedent stimulus stays the same, while the response progressively becomes more differentiated. In stimulus fading, the

459

Shaping

opposite occurs: The antecedent stimulus changes grad- ually, while the response stays essentially the same.

Increasing the Efficiency of Shaping In addition to showing how behaviors are shaped across and within response topographies, the Isaacs and col- leagues (1960) study illustrates another aspect of shaping— efficiency. During the early stages of the program, the psy- chologist waited for the next approximation of the behav- ior to appear before delivering the reinforcer. Waiting can be time-consuming and wasteful, so Isaacs and colleagues improved efficiency by using a verbal response prompt, “Say gum,” after the sixth training session. Presumably, if the psychologist had not used a verbal prompt, several ad- ditional sessions would have been necessary before a suc- cessful outcome was achieved.

Shaping can be enhanced in three ways. First, a dis- criminative stimulus (SD) may be combined with shap- ing. For example, when attempting to shape hand shaking as a greeting skill for an adult with develop- mental disabilities, the teacher might say, “Frank, hold out your arm.” Scott, Scott, and Goldwater (1997) used a vocal prompt (“Reach!”) as a university-level pole- vaulter ran a path to plant his pole in the vault box. The prompt was designed to focus the vaulter’s attention on extending his arms prior to take-off in the vault. Kazdin (2001) suggested that priming a response using any va- riety of prompting mechanisms can be helpful, espe- cially if the individual’s repertoire is weak and the likelihood of distinguishable successive approximations is low. “Even if the responses are not in the client’s reper- toire of responses, the priming procedure can initiate early response components and facilitate shaping” (p. 277). The second way to enhance shaping is through physical guidance. In the Frank example cited earlier, the teacher might manually assist Frank in holding out his arm. Using the third shaping-enhancing technique, the teacher might use an imitative prompt to demonstrate arm extension (e.g., “Frank, hold out your arm like this”). Any contrived prompt that is used to speed shap- ing should later be faded.

Clicker Training Pryor (1999) described clicker training as a science- based system for shaping behavior using positive rein- forcement. The clicker is a handheld device that produces a click sound when a metal tab is pressed. Reinforcement

is paired with the sound of the clicker so that the sound becomes a conditioned reinforcer. Initially used to shape dolphin behavior without physical guidance (Pryor & Norris, 1991), clicker training was later used with other animals (e.g., cats and horses) and ultimately humans to shape complex behaviors such as airplane pilot skills (Pryor, 2005).

Clicker trainers focus on building behavior, not stopping behavior. Instead of yelling at the dog for jumping up, you click it for sitting. Instead of kicking the horse to make it go, you click it for walking. Then, click by click, you “shape” longer sits, or more walking, until you have the final results you want. Once the behavior is learned, you keep it going with praise and approval and save the clicker and treats for the next new thing you want to train. (Pryor, 2003, p. 1)

Figure 3 provides 15 tips for getting started with clicker training (Pryor, 2003).

Guidelines for Implementing Shaping The practitioner should consider many factors before de- ciding to use shaping. First, the nature of the terminal be- havior to be learned and the resources available should be assessed. For example, a fifth-grade teacher might be in- terested in increasing the number of math problems per- formed by a student with learning disabilities. Perhaps the student currently completes 5 problems per math pe- riod, with a range of 0 to 10 problems. If the student is able to finish and check her performance independently at the end of the period, shaping could be implemented in which the number of completed math problems would be differentially reinforced. Reinforcement might be pre- sented for 5, 7, 9, and then 11 or more problems com- pleted correctly per period. In this case, solving the math problem is within a specific response topography, and the student can monitor her own performance.

In addition, because shaping requires multiple ap- proximations and linear progression cannot be predicted, the practitioner is advised to estimate the total amount of time required to achieve the terminal behavior. Such es- timates can be determined by asking colleagues how many sessions it took them to shape similar behaviors, or by differentially reinforcing a few behaviors that ap- proximate the terminal behavior and then extrapolating from that experience the total amount of time that might be needed to shape the terminal behavior. These two pro- cedures are likely to yield only rough estimates because any number of unforeseen factors can accelerate or

460

Shaping

Figure 3 Pryor’s 15 tips for getting started with the clicker.

1. Push and release the springy end of the clicker, making a two-toned click. Then treat. Keep the treats small. Use a delicious treat at first: for a dog or cat, little cubes of roast chicken, not a lump of kibble.

2. Click DURING the desired behavior, not after it is completed. The timing of the click is crucial. Don’t be dismayed if your pet stops the behavior when it hears the click. The click ends the behavior. Give the treat after that; the timing of the treat is not important.

3. Click when your dog or other pet does something you like. Begin with something easy that the pet is likely to do on its own (ideas: sit, come toward you, touch your hand with its nose, lift a foot, touch and follow a target object such as a pencil or a spoon).

4. Click once (in-out). If you want to express special enthusiasm, increase the number of treats, not the number of clicks.

5. Keep practice sessions short. Much more is learned in three sessions of five minutes each than in an hour of boring repetition.You can get dramatic results, and teach your pet many new things, by fitting a few clicks a day here and there in your normal routine.

6. Fix bad behavior by clicking good behavior. Click the puppy for relieving itself in the proper spot. Click for paws on the ground, not on the visitors. Instead of scolding for making noise, click for silence. Cure leash-pulling by clicking and treating those moments when the leash happens to go slack.

7. Click for voluntary (or accidental) movements toward your goal.You may coax or lure the animal into a movement or position, but don’t push, pull, or hold it. Let the animal discover how to do the behavior on its own. If you need a leash for safety’s sake, loop it over your shoulder or tie it to your belt.

8. Don’t wait for the “whole picture” or the perfect behavior. Click and treat for small movements in the right direction.You want the dog to sit, and it starts to crouch in back: click.You want it to come when called, and it takes a few steps your way: click.

9. Keep raising your goal. As soon as you have a good response—when a dog, for example, is voluntarily lying down, coming toward you, or sitting repeatedly—start asking for more. Wait a few beats, until the dog stays down a little longer, comes a little further, sits a little faster. Then click. This is called “shaping” a behavior.

10. When your animal has learned to do something for clicks, it will begin showing you the behavior spontaneously, trying to get you to click. Now is the time to begin offering a cue, such as a word or a hand signal. Start clicking for that behavior if it happens during or after the cue. Start ignoring that behavior when the cue wasn’t given.

11. Don’t order the animal around; clicker training is not command-based. If your pet does not respond to a cue, it is not disobeying; it just hasn’t learned the cue completely. Find more ways to cue it and click it for the desired behavior. Try working in a quieter, less distracting place for a while. If you have more than one pet, separate them for training, and let them take turns.

12. Carry a clicker and “catch” cute behaviors like cocking the head, chasing the tail, or holding up one foot.You can click for many different behaviors, whenever you happen to notice them, without confusing your pet.

13. If you get mad, put the clicker away. Don’t mix scoldings, leash-jerking, and correction training with clicker training; you will lose the animal’s confidence in the clicker and perhaps in you.

14. If you are not making progress with a particular behavior, you are probably clicking too late. Accurate timing is important. Get someone else to watch you, and perhaps to click for you, a few times.

15. Above all, have fun. Clicker-training is a wonderful way to enrich your relationship with any learner.

From Click to Win! by Karen Pryor, 2002. Sunshine Books, Waltham, MA.

461

Shaping

decelerate progress. If it appears that more time is needed than can be arranged, the practitioner might consider an- other strategy. Some behaviors seem to preclude the use of shaping as a behavior-building technique. For instance, if a high school English teacher is interested in increas- ing the public-speaking repertoires of his students, prompting, modeling, or peer tutoring might be more ef- ficient than shaping. That is, telling or showing the stu- dents how to use gestures, inflection, or metaphors would be much faster than attempting to shape each of these distinct response classes alone.

After the decision has been made to use shaping, Pryor’s Ten Laws of Shaping can assist the practitioner in implementing this process in applied settings (see Figure 4).

Select the Terminal Behavior

Practitioners often work with learners who have to change multiple behaviors. Consequently, they must identify the highest priority behavior quickly. The ultimate criterion in this decision is the individual’s expected independence after the behavior change; that is, the likelihood of her at- taining additional reinforcers from the environment (Snell & Brown, 2006). For example, if a student frequently roams the classroom poking other students, taking their papers, and verbally harassing them, a shaping procedure might best begin with a behavior that is incompatible with roaming the room because of the utility it would have for

the individual and the other students. In addition, if in- seat behavior is developed, the staff with whom the stu- dent interacts will likely notice and reinforce it. In this case, shaping should differentially reinforce longer dura- tions of in-seat behavior.

It is also important to define the terminal behavior precisely. For example, a behavior analyst might want to shape the appropriate sitting behavior of an individual with severe retardation. To do so, sitting might be defined as being located upright in the chair, facing the front of the room with buttocks against the bottom of the chair and back against the chair rest during a 15-minute morn- ing activity. Using this definition, the analyst can deter- mine when the behavior is achieved, as well as what does not constitute the behavior—an important discrimination if shaping is to be conducted efficiently (e.g., the student might be half in and half out of the seat, or the student might be in the seat but turned toward the back of the room).

Determine the Criterion for Success

After identifying the terminal behavior, the practitioner should specify the criterion for success. Here, the prac- titioner decides how accurate, fast, intense, or durable the behavior must be before it can be considered shaped. Several measures can be applied to establish a criterion of success. Some of the more common include frequency,

Figure 4 Pryor’s ten laws of shaping.

1. Raise criteria in increments small enough so that the subject always has a realistic chance of reinforcement.

2. Train one aspect of any particular behavior at a time. Don’t try to shape for two criteria simultaneously.

3. During shaping, put the current level of response on a variable ratio schedule of reinforcement before adding or raising the criteria.

4. When introducing a new criterion, or aspect of the behavioral skill, temporarily relax the old ones.

5. Stay ahead of your subject: Plan your shaping program completely so that if the subject makes sudden progress, you are aware of what to reinforce next.

6. Don’t change trainers in midstream.You can have several trainers per trainee, but stick to one shaper per behavior.

7. If one shaping procedure is not eliciting progress, find another. There are as many ways to get behavior as there are trainers to think them up.

8. Don’t interrupt a training session gratuitously; that constitutes a punishment. 9. If behavior deteriorates, “Go back to kindergarten.” Quickly review the whole shaping

process with a series of easily earned reinforcers. 10. End each session on a high note, if possible, but in any case quit while you’re ahead.

Reprinted with the permission of Simon & Schuster Adult Publishing Group from Don’t Shoot the Dog! The New Art of Teaching and Training (revised edition) by K. Pryor, 1999, pp. 38–39. Copyright © 1984 by Karen Pryor.

462

Shaping

magnitude, and duration. Depending on the terminal behavior, any or all of these dimensions can be used to as- sess achievement. For instance, a special education teacher might shape frequency performance in solving mathematics calculations problems that would proceed from the student’s current level of performance (say, half a problem per minute) to a higher rate (e.g., four problems per minute).

Norms for success can be determined by measuring the behavior in a similar peer group or consulting estab- lished norms in the literature. For instance, norms and guidelines for reading rate by grade level (Kame´enui, 2002), physical education skills for children by age group (President’s Council on Physical Fitness, 2005), and homework guidelines per grade level (Heron, Hippler, & Tincani, 2003) can be referenced to validate progress to- ward the terminal behavior.

In the earlier illustration, the criterion for success could be the student’s sitting in her seat appropriately 90% of the time during the morning activity for 5 con- secutive days. In this example, two criteria are speci- fied: the percentage of acceptable sitting behavior per session (i.e., 90%) and the number of days that the cri- terion must be met to establish the behavior (i.e., 5 con- secutive days).

Analyze the Response Class

The purpose of analyzing the response class is to at- tempt to identify the approximations that might be emit- ted in the shaping sequence. When approximation behaviors are known—and anticipated—the practitioner is in a better position to observe and reinforce an ap- proximation when it is emitted. However, the practi- tioner must recognize the projected approximations are guesstimates of behaviors that might be emitted. In ac- tuality, the learner may emit behaviors that are merely approximations of these projections. The practitioner must be able to make the professional judgment that the previously emitted behavior either is or is not a closer approximation to the terminal behavior than the behav- iors that occurred and were reinforced in the past. Ac- cording to Galbicka (1994):

The successful shaper must carefully ascertain charac- teristics of an individual’s present response repertoire, explicitly define characteristics the final behavior will have at the end of training, and plot a course between re- inforcement and extinction that will bring the right re- sponses along at the right time, fostering the final behavioral sequence while never losing responding altogether. (p. 739)

The relevant approximations across or within a re- sponse class can be analyzed in several ways. First, prac- titioners can consult experts in the field to determine their views on the proper sequence of approximations for a given behavior (Snell & Brown, 2006). For example, teachers who have taught three-digit multiplication for several years can be consulted to learn their experiences with prerequisite behaviors for performing this type of calculation task. Second, as stated earlier, normative data from published studies may be used to provide an esti- mate of the approximations involved. Third, a videotape can be used to analyze the component behaviors. The videotape would help the practitioner to see motion that might be undetectable in an initial live performance, but become detectable as the practitioner becomes skilled with observing it. Finally, the practitioner can perform the target behavior himself, carefully noting the discrete behavioral components as he produces them.

The ultimate determination of the approximations, the approximation order, the duration of time that rein- forcement should be delivered at a given approximation, and the criteria for skipping or repeating approximations rests with the skill and the judgment of the practitioner. Ultimately, the learner’s performance should dictate when approximation size should be increased, maintained, or decreased, making consistent and vigilant monitoring by the practitioner essential.

Identify the First Behavior to Reinforce

Two criteria are suggested for identifying the initial be- havior for reinforcement:(a) The behavior should already occur at some minimum frequency, and (b) the behavior should be a member of the targeted response class. The first condition reduces the need to wait for the occurrence of the initial behavior. Waiting for a behavior to be emit- ted can be therapeutically counterproductive and is usu- ally unnecessary. The second criterion sets the occasion for reinforcing an existing behavioral component that has one dimension in common with the terminal behavior. For example, if the terminal behavior is expressive speech, as was the case with Andrew, lip movement would be a good first choice.

Eliminate Interfering or Extraneous Stimuli

Eliminating sources of distraction during shaping en- hances the effectiveness of the process. For example, if

463

Shaping

Figure 5 Percentage of correct arm extensions at seven different heights. The numbers contained within each criterion condition refer to the height of the photoelectric beam. From “A Performance Improvement Program for an International-Level Track and Field Athlete” by D. Scott, L. M. Scott, & B. Goldwater, 1997, Journal of Applied Behavior Analysis, 30 (3), p. 575. Copyright by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

a parent is interested in shaping one dimension of her daughter’s dressing behavior at the local gymnasium (e.g., the rate of dressing) and decides to begin the shap- ing procedure in a room where cartoons are shown on the television, the shaping program may not be success- ful because the cartoons will be competing for the daugh- ter’s attention. It would be more efficient to choose a time when and a location where sources of distraction can be reduced or eliminated.

Proceed in Gradual Stages

The importance of proceeding toward the terminal goal in gradual stages cannot be overemphasized. The practi- tioner should anticipate changes in the rate of progress and be prepared to go from approximation to approxi- mation as the learner’s behavior dictates. Each new oc- currence of a successive approximation to the terminal behavior must be detected and reinforced. If not, shaping will be, at worst, unsuccessful or, at best, haphazard, re- quiring much more time. Furthermore, the occurrence of a behavior at a given approximation does not mean that the next response that approximates the terminal behav- ior will be immediately or accurately produced. Figure 5 shows the results of a study by Scott and colleagues (1997), who found that the percentage of correct arm ex- tensions by the pole-vaulter decreased initially as the

height of the bar was raised. Whereas a 90% correct arm extension was noted at the initial height, the vaulter’s cor- rect performance typically dropped to approximately 70% each time the bar was raised. After successive at- tempts at the new height, the 90% correct criterion was reestablished.

The practitioner must also be aware that many tri- als may be required at a given approximation before the subject can advance to the next approximation. Figure 5 illustrates this point as well. On the other hand, only a few trials may be required. The practitioner must watch carefully and be prepared to reinforce many tri- als at a given approximation or to move rapidly toward the terminal objective.

Limit the Number of Approximations at Each Level

Just as it is important to proceed gradually from approx- imation to approximation, it is equally important to en- sure that progress is not impeded by offering too many trials at a given approximation. This may cause the be- havior to become firmly established, and because that ap- proximation has to be extinguished before progress can begin again (Catania, 1998), the more frequently rein- forcement is delivered at a given approximation, the longer the process can take. In general, if the learner is

100

Pe rc

en ta

ge o

f C

or re

ct A

rm E

x te

n si

o n

1 21 41 61 81 101 121 141 161 181 201

Sessions

225 230 235 240 245 250 252

515 520 525 527 530 535Jumped Maximum Height

537

464

Shaping

progressing steadily, reinforcement is probably being de- livered at the correct pace. If too many mistakes are being made or behavior ceases altogether, the criterion for re- inforcement has probably being raised too quickly. Fi- nally, if the performance stabilizes at a certain level, shaping is probably going too slowly. Practitioners may experience all three of these conditions over the course of successive approximations that ultimately lead to the be- havior being shaped. Therefore, they must be vigilant and adjust the procedure, if necessary.

Continue to Reinforce When the Terminal Behavior Is Achieved

When the terminal behavior is demonstrated and rein- forced, it is necessary to continue to reinforce it. Other- wise the behavior will be lost, and performance will return to a lower level. Reinforcement must continue until the criterion for success is achieved and a maintenance schedule of reinforcement is established.

Future Applications of Shaping New applications of shaping are being reported in the lit- erature that extend its utility, promote its efficiency, and ultimately benefit the individual with respect to shorter training time and/or skill development (Shimoff & Cata- nia, 1995). At least three future applications of shaping can be considered: using percentile schedules, using com- puters to teach shaping skills, and combining shaping procedures with robotics engineering.

Percentile Schedules

Galbicka (1994) argued against the often-heard notion that shaping is more art than science and that practition- ers can learn to use shaping only by hard-earned direct experience. Building on more than two decades of labo- ratory research on percentile reinforcement schedules, Galbicka’s conceptualization of shaping specifies the cri- teria for responses and reinforcement in mathematical terms, thereby standardizing the shaping process regard- less of the person doing the shaping. He stated:

Percentile schedules disassemble the process of shaping into its constituent components, translate those compo- nents into simple, mathematical statements, and then use these equations, with parameters specified by the experi- menter or trainer, to determine what presently consti- 2They also developed simulations for reinforcement schedules.

tutes a criterional response and should therefore be rein- forced. (p. 740)

Galbicka (1994) conceded that for percentile sched- ules to be used effectively in applied settings, behaviors must be (a) measured constantly and (b) rank-ordered using a system of comparisons to previous responses. If a re- sponse exceeds a criterion value, reinforcement is deliv- ered. If not, reinforcement is withheld. Galbicka projected that understanding how percentile schedules work may

increase our understanding of the complex social and non-social dynamics that shape behavior . . . allow[ing] unprecedented control over experimentally relevant stim- uli in operant conditioning and differential procedures and provide a seemingly endless horizon against which to cast our sights for extensions and applications. (p. 759)

Using Computers to Teach Shaping

Several options are currently available to teach shaping skills using computers or a combination of specialized software with computer-based applications. For exam- ple, Sniffy, the Virtual Rat, is a computer-based, digitized animation of a white rat within a Skinner box. Students can practice basic shaping on their own personal com- puters using this virtual simulation.

Acknowledging the practical limitations of teaching behavior analysis principles (e.g., shaping) interactively to a large student body, Shimoff and Catania (1995) de- veloped and refined computer simulations to teach the concepts of shaping.2 In their simulation, The Shaping Game, four levels of difficulty (easy, medium, hard, and very hard) were programmed over a series of refinements. Beginning with easier tasks and advancing to the very hard level, students are taught to detect successive ap- proximations to bar-pressing behaviors that should be re- inforced, and those that should not. In still later versions, students are expected to shape the movement of the ani- mated rat from one side of a Skinner box to another. Shimoff and Catania provided a case for computer sim- ulations when they stated, “Computers can be important and effective tools for establishing sophisticated behav- ior that appears to be contingency shaped even in classes that do not provide real (unsimulated) laboratory experi- ence” (pp. 315–316).

Martin and Pear (2003) contended that computers might be used to shape some dimensions of behavior (e.g., topography) more quickly than humans because a

465

Shaping

computer can be calibrated and programmed with spe- cific decision rules for delivering reinforcement. They further suggested that in medical application cases (e.g., amputees, stroke victims), in which minute muscle move- ments may go undetected by a human observer, a micro- processor could detect shapeable responses. In their view, “Computers . . . are both accurate and fast and may there- fore be useful in answering fundamental questions con- cerning which shaping procedures are most effective . . . computers may be able to shape at least some kinds of be- havior as effectively as humans” (p. 133).

Combining Shaping with Robotics Engineering

Some researchers have begun to explore how shaping can be applied to robot training (e.g., Dorigo & Colom- betti, 1998; Saksida, Raymond, & Touretzky, 1997). Es- sentially, shaping is being considered as one method of

programming robots to progress from a start state through a series of intermediate and end-goal states to achieve more complicated sets of commands. According to Sav- age (1998):

Given that shaping is a significant determinant of how a wide variety of organisms adapt to the changing circum- stances of the real world, then shaping may, indeed, have potential as a model for real-world robotic interac- tions. However, the success of this strategy depends on a clear understanding of the principles of biological shap- ing on the part of the roboticists, and the effective ro- botic implementation of these principles. (p. 321)

Although Dorigo and Colombetti (1998) and Sak- sida and colleagues (1997) concede the difficulty in ap- plying behavioral principles of successive approximations to robots, this line of inquiry, combined with emerging knowledge on artificial intelligence, offers exciting prospects for the future.

Summary Definition of Shaping

1. Shaping is the differential reinforcement of successive ap- proximations to a desired behavior. In shaping, differential reinforcement is applied to produce a series of slightly dif- ferent response classes, with each successive response class becoming a closer approximation to the terminal be- havior than the previous one.

2. A terminal behavior—the end product of shaping—can be claimed when the topography, frequency, latency, dura- tion, or amplitude/magnitude of the target behavior reaches a predetermined criterion level.

3. Differential reinforcement is a procedure in which rein- forcement is provided for responses that share a predeter- mined dimension or quality, and in which reinforcement is withheld for responses that do not demonstrate that quality.

4. Differential reinforcement has two effects: responses simi- lar to those that have been reinforced occur with greater fre- quency, and responses resembling the unreinforced members are emitted less frequently (i.e., undergo extinction).

5. The dual effects of differential reinforcement result in re- sponse differentiation, the emergence of a new response class composed primarily of responses sharing the char- acteristics of the previously reinforced subclass.

6. The gradually changing criterion for reinforcement dur- ing shaping results in a succession of new response classes, or successive approximations, each one closer in form to the terminal behavior than the response class it replaces.

Shaping across and within Response Topographies

7. Shaping behavior across different response topographies means that select members of a response class are differ- entially reinforced, while members of other response classes are not reinforced.

8. When shaping a behavior within a response topography, the form of the behavior remains constant, but differen- tial reinforcement is applied to a dimension of the behav- ior (e.g., frequency, duration).

9. Shaping entails several limitations that practitioners should consider before applying it.

10. In shaping, the antecedent stimulus stays the same, while the response progressively becomes more differentiated. In stimulus fading, the response remains the same, while the antecedent stimulus changes gradually.

Increasing the Efficiency of Shaping

11. The efficiency of shaping may be improved in several ways, including using a discriminative stimulus, a vocal prompt, physical guidance, an imitative prompt, or prim- ing. Any prompt that is introduced is later faded.

Clicker Training

12. Clicker training is a science-based system for shaping be- havior using positive reinforcement.

13. The clicker, a handheld device that produces a click sound when a metal tab is pressed, provides the signal that when a behavior is being performed in the presence of the click, reinforcement follows.

466

Shaping

Guidelines for Implementing Shaping

14. Before deciding to use shaping, the nature of the behaviors to be learned and the resources available must be assessed.

15. After the decision has been made to use a shaping proce- dure, the practitioner performs the following steps: select the terminal behavior, decide the criterion for success, an- alyze the response class, identify the first behavior to re- inforce, eliminate interfering or extraneous stimuli,

proceed in gradual stages, limit the number of approxi- mations at each level, and continue reinforcement when the terminal behavior is achieved.

Future Applications of Shaping

16. Future applications of shaping include applying percentile schedules, integrating computer technology to teach be- havior analysis using shaping procedures, and combining shaping procedures with emerging robotics technology.

467

Extinction

Key Terms escape extinction extinction (operant)

extinction burst resistance to extinction

sensory extinction spontaneous recovery

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes and Concepts

3-11 Define and provide examples of extinction.

Content Area 9: Behavior Change Procedures

9-4 Use extinction:

(a) Identify possible reinforcers maintaining behavior and use extinction.

(b) State and plan for the possible unwanted effects of the use of extinction.

468

This chapter describes how to reduce the fre- quency of a previously reinforced behavior by withholding reinforcement, a principle known

as extinction. Extinction as a procedure provides zero probability of reinforcement. It is also a behavioral process of a diminishing rate of response. Practitioners have applied this principle of behavior effectively in a wide variety of settings, such as homes, schools, and in- stitutions, and with diverse problem behaviors ranging from severe self-destruction to mild disruptions. How- ever, the effectiveness of extinction in an applied setting depends primarily on the identification of reinforcing consequences and consistent application of the proce- dure. Extinction does not require the application of aver- sive stimuli to decrease behavior; it does not provide verbal or physical models of punishers directed toward others. Extinction simply requires the withholding of re- inforcers. This chapter defines extinction and describes procedures for applying extinction. Additionally, it dis- cusses extinction behavior and resistance to extinction. Although extinction appears to be a simple process, its application in applied settings can be difficult.

Definition of Extinction Extinction as a procedure occurs when reinforcement of a previously reinforced behavior is discontinued; as a re- sult, the frequency of that behavior decreases in the fu- ture.1 To restate this principle: Keller and Schoenfeld (1950/1995) defined extinction this way; “Conditioned operants are extinguished by severing the relation be- tween the act and the effect. . . . The principle of Type R [operant] extinction may be put as follows: The strength of a conditioned operant may be diminished by with- holding its reinforcement” (pp. 70–71). Similarly, Skin- ner (1953) wrote, “When reinforcement is no longer forthcoming, a response becomes less and less frequent in what is called ‘operant extinction’” (p. 69).

DeShawna, a third-grade student, frequently inter- rupted her teacher when he instructed other students. It appeared that DeShawna tried to direct the teacher’s at- tention from the other students to herself. The teacher al- ways responded to DeShawna’s interruptions by answering her questions, asking her to return to her seat, telling her not to interrupt him while he worked with the other students, or explaining that she needed to wait for her turn. DeShawna usually agreed that she would not in- terrupt her teacher, but her annoying interruptions did not

1The term extinction is also used with respondent conditioned reflexes. Presenting a conditioned stimulus (CS) again and again without the un- conditioned stimulus until the CS no longer elicits the conditioned response is called respondent extinction.

stop. The teacher knew that DeShawna interrupted his in- struction to get his attention. Since telling DeShawna not to interrupt him at certain times did not work, the teacher decided to try ignoring her interruptions. DeShawna stopped interrupting her teacher after 4 days during which her interruptions produced no teacher attention.

In this example, student interruptions (i.e., the prob- lem behavior) were maintained by teacher attention (i.e., the reinforcer). The teacher applied the extinction pro- cedure by ignoring DeShawna’s interruptions, and the procedure produced a decrease in the frequency of the interruptions. Note that the extinction procedure does not prevent occurrences of a problem behavior (e.g., the in- terruptions). Rather, the environment is changed so that the problem behavior no longer produces reinforcement (e.g., teacher attention).

Procedural and Functional Forms of Extinction

The use of functional behavior assessments have enabled applied behavior analysts to distinguish clearly between the procedural variations of extinction (ignoring) and the functional variations of extinction (withholding main- taining reinforcers). This clarification between proce- dural and functional variations of extinction has led to more effective treatments (Lerman & Iwata, 1996a).

Historically, some applied behavior analysts have emphasized the procedural form of extinction (e.g., just ignore the problem behavior and it will go away) rather than the functional form (i.e., withholding the main- taining reinforcers). Applications of the procedural form of extinction are often ineffective. Functional behavior assessments enabled applied behavior analysts to dis- tinguish clearly between the procedural (ignoring) and the functional forms of extinction (withholding main- taining reinforcers), and as an outcome, may have in- creased research in using basic extinction procedures in applied settings. When the extinction procedure is matched to the behavioral function, the intervention is usually effective.

Extinction: Misuses of a Technical Term

With the possible exception of negative reinforcement, extinction is perhaps the most misunderstood and mis- used technical term in applied behavior analysis. Extinction is a technical term that applied behavior ana- lysts should use only to identify the procedure of with- holding reinforcers that maintain behavior. Four common misuses of the technical term are presented in the fol- lowing sections.

Extinction

469

Using Extinction to Refer to Any Decrease in Behavior

Some use the term extinction when referring to any de- crease in responding, regardless of what produced the behavior change. For example, if a person’s responding decreased as a result of punishment such as time-out from reinforcement or a physical restriction of the behavior, a statement such as “the behavior is extinguishing” is mis- leading and incorrect in a technical sense. Labeling any reduction in behavior that reaches a zero rate of occur- rence as extinction is another common misuse of this technical term.

Confusing Forgetting and Extinction

A misuse of the term extinction occurs when the speaker confuses extinction with forgetting. In forgetting, a be- havior is weakened by the passage of time during which the person does not have an opportunity to emit the be- havior. In extinction, behavior is weakened because it does not produce reinforcement.

Confusing Response Blocking and Sensory Extinction

Applied behavior analysts have used goggles, gloves, hel- mets, and wrist weights to block the occurrence of re- sponses maintained by automatic reinforcement, rather than masking the sensory stimulation. The applications of response blocking to reduce problem behaviors appear similar to sensory extinction. Response blocking, how- ever, is not an extinction procedure. With all extinction procedures, including sensory consequences, the persons can emit the problem behavior, but that behavior will not produce reinforcement. By contrast, response blocking prevents the occurrence of the target behavior (Lerman & Iwata, 1996).

Confusing Noncontingent Reinforcement and Extinction

Two definitions of extinction now appear in the applied behavior analysis literature. Each definition uses a dif- ferent procedure to diminish behavior. We use one defi- nition for extinction throughout this chapter (i.e., Keller & Schoenfeld, 1950/1995; Skinner 1953). It is the defi- nition most associated with the principle of behavior called extinction. This definition of extinction has served the experimental analysis of behavior and applied be- havior analysis well for many years.

The procedure for the second definition does not withhold the reinforcers that maintain the problem be- havior. It presents those reinforcers noncontingently

2Extinction linked to behaviors maintained by positive reinforcement is often called attention extinction in the applied behavior analysis literature, especially in the functional behavior assessment (FBA) literature in which reinforcement by social attention is one of the conditions or hypotheses explored by FBA.

(NCR), meaning that the applied behavior analyst deliv- ers stimuli with known reinforcing properties to an individual on a fixed-time or variable-time schedule com- pletely independent of responding.

NCR provides an important and effective interven- tion for diminishing problem behaviors, but NCR oper- ates on behavior in a different way than does the principle of extinction. Extinction diminishes behavior by chang- ing consequence stimuli; NCR diminishes behavior by changing motivating operations. Although both proce- dures produce a decrease in behavior, the behavioral ef- fects result from different controlling variables. Using an established technical term to describe effects produced by two different procedures is confusing.

Extinction Procedures Specifically, procedures for extinction take three distinct forms that are linked to behavior maintained by positive reinforcement,2 negative reinforcement, and automatic reinforcement.

Extinction of Behavior Maintained by Positive Reinforcement

Behaviors maintained by positive reinforcement are placed on extinction when those behaviors do not pro- duce the reinforcer. Williams (1959) in a classic study described the effects of the removal of positive rein- forcement (extinction) on the tyrant-like behavior of a 21-month-old boy. The child had been seriously ill for the first 18 months of his life but had completely recov- ered when the study began. The boy demanded special at- tention from his parents, especially at bedtime, and responded with temper tantrums (e.g., screaming, fussing, crying) when his parents did not provide the attention. A parent spent 1⁄2 to 2 hours each bedtime waiting in the bedroom until the boy went to sleep.

Although Williams did not speculate about how the tantrums developed, a likely explanation is not difficult to imagine. Because the child had been seriously ill for much of his first 18 months, crying had been his signal that he was experiencing discomfort or pain or needed help. In effect, his crying may have been reinforced by his parents’ attention. Crying became a high-frequency be- havior during the extended illness and continued after the child’s health improved. The parents probably realized

Extinction

470

D u

ra ti

o n

o f

C ry

in g

in M

in u

te s

Times Child Put to Bed

Figure 1 Two extinction series showing duration of crying as a function of being put to bed. From “The Elimination of Tantrum Behavior by Extinction Procedures” by C. D. Williams, 1959, Journal of Abnormal and Social Psychology, 59, p. 269. Copyright 1959 by the American Psychological Association. Reprinted by permission of the publisher and author.

eventually that their child cried at bedtime to gain their attention and tried to ignore the crying. The crying in- creased in intensity and emotion when they did not stay in the room. As the days and weeks went by, the child’s demand for absolute attention at bedtime grew worse. The parents probably decided again not to stay with the child at bedtime; but the intensity of the crying increased, and some new tantrum behaviors occurred. The parents returned to the room and, by giving in to the tantrums, taught their child to behave like a tyrant.

After 3 months of the tyrant-like behavior, the par- ents decided that they must do something about their child’s tantrums. It appeared that parental attention main- tained the tantrums; therefore, they planned to apply the principle of extinction. The parents, in a leisurely and re- laxed manner, put the child to bed, left the bedroom, and closed the door. The parents recorded the duration of screaming and crying from the moment the door was closed.

Figure 1 shows the duration of the tantrums during extinction. The child had a tantrum for 45 minutes the first time he was put to bed without his parents staying in the room. The tantrums gradually decreased until the 10th session, when the child “no longer whimpered, fussed, or cried when the parents left the room. Rather, he smiled as they left. The parents reported that their child made happy sounds until he dropped off to sleep” (Williams, 1959, p. 269).

No tantrums occurred for approximately 1 week, but they resumed after his aunt put him to bed. When the

child started to tantrum, the aunt returned to the bedroom and stayed with the child until he went to sleep. Tantrums then returned to the previous high level and had to be de- creased a second time.

Figure 1 also shows the duration of tantrums for the 10 days following the aunt’s intervention. The data curve is similar to that of the first removal of parent attention. The duration of the tantrums was somewhat greater dur- ing the second removal, but reached zero by the ninth session. Williams reported that the child had no further bedtime tantrums during 2 years of follow-up.

Extinction of Behavior Maintained by Negative Reinforcement

Behaviors maintained by negative reinforcement are placed on extinction (a.k.a., escape extinction) when those behaviors do not produce a removal of the aversive stimulus, meaning that the person cannot escape from the aversive situation. Anderson and Long (2002) and Daw- son and colleagues (2003) provided excellent examples of using escape extinction as a behavioral intervention.

Anderson and Long (2002) provided treatment to re- duce the problem behaviors of Drew, an 8-year-old boy with autism and moderate to severe retardation. Drew’s problem behaviors consisted of self-injurious behavior (SIB), aggression, and disruptions. Anderson and Long completed a functional behavior assessment, and then hypothesized that escape from task situations maintained the problem behaviors. Based on this hypothesis, the speech therapist used escape extinction to diminish the problem behaviors that occurred while Drew worked on match-to-sample and receptive language tasks. These tasks evoked the highest rates of problem behavior. The speech therapist provided instructional prompts during the tasks. When Drew emitted problem behaviors fol- lowing the instructional prompt, the speech therapist physically guided him to task completion. Escape ex- tinction produced a significant reduction in problem be- haviors during the match-to-sample and receptive language tasks. Figure 2 shows the number of problem behaviors emitted per minute during baseline (i.e., es- cape) and escape extinction conditions.

Dawson and colleagues (2003) reported the use of escape extinction to diminish food refusal by Mary, a 3-year-old who was admitted to a day program for the treatment of total food refusal. Her medical history in- cluded gastroesophageal reflex, delayed gastric empty- ing, and gastrostomy tube dependence among other medical issues. Her food refusal included head turning when presented with a bite of food, putting her hand on the spoon or therapist’s hand or arm, and using her hands or bib to cover her face.

Extinction

471

0P ro

b le

m B

eh av

io r

p er

M in

u te

2 4 6 8 10 12 14 16 18 20 22 24 26 Sessions

Baseline Escape Extinction Baseline Escape Extinction

Drew

Figure 2 The number of problem behaviors per minute by Drew during baseline and escape extinc- tion conditions. From “Use of a Structured Descriptive Assessment Methodology to Identify Variables Affecting Problem Behavior” by C. M. Anderson & E. S. Long, 2002, Journal of Applied Behavior Analysis, 35 (2), p. 152. Copyright 2002 by Society for the Experimental Analysis of Behavior, Inc. Adapted by permission.

The escape extinction procedure followed 12 ses- sions of food refusal. If Mary refused the bite of food, the therapist held the spoon to Mary’s mouth until she took the bite. When Mary expelled the bite, the food was re-presented until swallowed. With this procedure, Mary’s refusals did not produce an escape from the food. The therapist terminated the session after Mary had swallowed 12 bites of food. Mary accepted no food during the 12 baseline sessions; but after two sessions of escape extinction, Mary’s food acceptance increased to 100% compliance.

Extinction of Behavior Maintained by Automatic Reinforcement

Behaviors maintained by automatic reinforcement are placed on extinction (a.k.a. sensory extinction;) by masking or removing the sensory consequence (Rincover, 1978). Some behaviors produce natural sensory conse- quences that maintain the behavior. Rincover (1981) de- scribed a naturally occurring sensory consequence as a stimulus that “sounds good, looks good, tastes good, smells good, feels good to the touch, or the movement itself is good” (p. 1).

Extinction linked to automatic reinforcement is not a recommended treatment option for problem behaviors, even self-stimulatory behaviors that are maintained by social consequences or negative reinforcement. Auto- matic reinforcement, however, can maintain self-injurious behaviors (SIB) and persistent, nonpurposeful, repetitive self-stimulatory behaviors (e.g., flipping fingers, head rocking, toe walking, hair pulling, fondling body parts). For example, Kennedy and Sousa (1995) reported that a 19-year-old man with profound disabilities had poked his eyes for 12 years, which resulted in visual impairment in both eyes. Eye poking was thought to serve a sensory stimulation function because it occurred most frequently when he was alone. When Kennedy and Sousa used gog- gles to mask contact with his eyes, eye-poking behavior decreased markedly.

Social consequences frequently maintain aggression, but not always. Automatic reinforcement can maintain

aggression also, as it can with SIB and self-stimulatory behaviors (Thompson, Fisher, Piazza, & Kuhn, 1998). Deaver, Miltenberger, and Stricker (2001) used extinc- tion to decrease hair twirling. Hair twirling is a frequent precursor to hair pulling, a serious self-injury. Tina, at 2 years and 5 months of age, received treatment for hair twirling and pulling. A functional analysis documented that Tina did not twirl or pull hair to obtain attention, and that hair twirling most often occurred when Tina was alone at bedtime. The sensory extinction procedure con- sisted of Tina wearing thin cotton mittens on both hands at naptime during day care and during bedtime at home. Figure 3 shows that sensory extinction diminished hair twirling to near-zero levels at home and at day care.

Rincover, Cook, Peoples, and Packard (1979) and Rincover (1981) provided the following examples of ap- plying extinction with automatic reinforcement:

1. A child persisted in flipping a light switch on and off. The visual sensory consequence was removed by disconnecting the switch.

2. A child persistently scratched his body until it bled. The tactile (touch) sensory consequence was removed by putting a thin rubber glove on his hand so that he could not feel his skin. Later, the glove was faded by gradually cutting off portions of it.

3. A child would throw up, and then eat the vomit. The gustatory (taste) extinction procedure con- sisted of adding lima beans to the vomit. The child did not like lima beans; therefore, the vomit did not taste as good, and the positive sensory conse- quence was masked.

4. A child received kinesthetic stimulation (i.e., stim- ulation of muscles, tendons, and joints) by holding his arms out to his side and incessantly flapping his fingers, wrists, and arms. The extinction proce- dure consisted of taping a small vibratory mecha- nism on the back of his hand to mask the kinesthetic stimulation.

5. A child produced auditory stimulation by persis- tently twirling an object, such as a plate, on a table.

Extinction

472

100

P er

ce n

ta ge

o f

Se ss

io n

T im

e Baseline Mittens Baseline Mittens

Bedroom

Day care

100

Baseline Mittens Baseline Mittens

Sessions

Figure 3 The percentage of session time of hair twirling during baseline and treatment conditions in the home and day care settings. From “Functional Analysis and Treatment of Hair Twirling in a Young Child” by C. M. Deaver, R. G. Miltenberger, and J. M. Stricker, 2001, Journal of Applied Behavior Analysis, 34 p. 537. Copyright 2001 by Society for Experimental Analysis of Behavior, Inc. Used by permission.

Carpeting the surface of the table he used for spin- ning objects masked the auditory stimulation from his plate spinning.

Extinction Effects When a previously reinforced behavior is emitted but is not followed by the usual reinforcing consequences, the occurrence of that behavior should either gradually de- crease to its prereinforcement level or stop entirely. Be- haviors undergoing extinction are usually associated with predictable characteristics in rate and topography of re- sponse. These extinction effects have strong generality across species, response classes, and settings (Lerman & Iwata, 1995, 1996a; Spradlin, 1996). Applied behavior analysts, however, have not devoted much research ef- fort to the basic extinction procedure beyond using ex- tinction as a component in treatment packages for problem behaviors (Lerman & Iwata, 1996a). Conse- quently, extinction effects have not been documented clearly in applied settings.

Lerman and Iwata (1996) cautioned applied behav- ior analysts that these clearly documented extinction ef- fects may have limited generality in applied behavior analysis. Applied researchers and practitioners should

view all of the following comments on the extinction ef- fects tentatively when they relate to behavioral interven- tions or applied research rather than basic research. Applied behavior analysts almost always apply extinc- tion as an element of a treatment package, thereby con- founding the understanding of extinction behavior in applied settings.

Gradual Decrease in Frequency and Amplitude

Extinction produces a gradual reduction in behavior. However, when reinforcement is removed abruptly, nu- merous unreinforced responses can follow. This gradual decrease in response frequency will tend to be sporadic with a gradual increase in pauses between responses (Keller & Schoenfeld, 1950/1995). The extinction pro- cedure is often difficult for teachers and parents to apply because of the initial increase in frequency and magnitude and the gradual decrease in behavior. For example, par- ents may be unwilling to ignore tantrum behavior for a sufficient amount of time because tantrums are so aver- sive to parents. Rolider and Van Houten (1984) presented a tactic for this practical problem. They suggested teach- ing parents to ignore gradually increasing durations of

Extinction

473

R es

p o

n se

Before Extinction (Reinforcement)

During Extinction

Extinction Burst

Spontaneous Recovery

Time Figure 4 Illustration of an extinction burst and spontaneous recovery.

bedtime crying. They used baseline data to assess how long the parents could ignore bedtime crying comfort- ably before attending to their child. Then, the parents gradually increased their duration of ignoring. Every 2 days they waited an additional 5 minutes before attend- ing to the child until a sufficient total duration of time was achieved.

Extinction Burst

A general effect of the extinction procedure is an imme- diate increase in the frequency of the response after the removal of the positive, negative, or automatic rein- forcement. The behavioral literature uses the term extinction burst to identify this initial increase in re- sponse frequency. Figure 4 presents an illustration of an extinction burst. Operationally, Lerman, Iwata, and Wal- lace (1999) defined extinction burst as “an increase in re- sponding during any of the first three treatment sessions above that observed during all of the last five baseline sessions or all of baseline” (p. 3). The extinction burst is well documented in basic research, but not well docu- mented in applied research (Lerman & Iwata, 1995, 1996a). When reported, the bursts have occurred for only a few sessions without notable problems.

Goh and Iwata (1994) provided data showing an ex- tinction burst. Steve was a 40-year-old man with pro- found mental retardation. He had been referred for evaluation for self-injury (head banging, head hitting). A functional analysis showed that self-injury was reinforced by escape from instructions. Goh and Iwata used extinc- tion for the treatment of Steve’s self-injury. The top panel in Figure 5 shows extinction bursts occurring with the onset of each of the two extinction phases.

Even though applied researchers have rarely re- ported on extinction bursts, they do occur in applied set- tings (e.g., Richman, Wacker, Asmus, Casey, & Anderson, 1999; Vollmer et al., 1998). Problem behav- iors can worsen during extinction before they show im- provement. For example, teachers should anticipate an initial increase in disruption during extinction. There- after, problem behaviors should begin to decrease and should return to their prereinforcement level. Applied behavior analysts need to anticipate extinction bursts and be prepared to consistently withhold the reinforcing con- sequence. Extinction bursts usually suggest that the re- inforcer(s) maintaining the problem behavior was successfully identified, indicating that there is a good chance of an effective intervention.

Initial Increase in Response Amplitude

In addition to the extinction burst, which is an increase in the frequency of the response, an initial increase in the amplitude or force of the response also may occur during extinction. Parents who begin to ignore a child’s bedtime tantrums might experience an increase in the loudness (i.e., amplitude) of screaming and the force of kicking before the tantrums begin to diminish.

Spontaneous Recovery

During extinction a behavior will typically continue a de- creasing trend until it reaches a prereinforced level or ul- timately ceases. However, a phenomenon commonly associated with the extinction process is the reappear- ance of the behavior after it has diminished to its prere- inforcement level or stopped entirely. Basic researchers

Extinction

474

SI B

Baseline ExtinctionBLExtinction

Sessions

A gg

re ss

io n

R es

p o

n se

s p

er M

in u

20 40 60 80

Figure 5 Number of responses per minute of self- injurious behavior (SIB) (top panel) and aggression (bottom panel) for Steve during baseline and extinction. From “Behavioral Persistence and Variability during Extinction of Self-Injury Maintained by Escape” by H-L. Goh and B. A. Iwata, 1994, Journal of Applied Behavior Analysis, 27, p. 174. Copyright 1994 by Society for the Experimental Analysis of Behavior, Inc. Used by permission.

commonly report this extinction effect and call it spont- aneous recovery. With spontaneous recovery, the be- havior that diminished during the extinction process recurs even though the behavior does not produce rein- forcement. Spontaneous recovery is short-lived and lim- ited if the extinction procedure remains in effect (see Figure 4).

Applied behavior analysts have not researched the characteristics and prevalence of spontaneous recovery (Lerman & Iwata, 1996a). Therapists and teachers need to know about spontaneous recovery, however, or they might conclude erroneously that the extinction procedure is no longer effective.

Variables Affecting Resistance to Extinction Behavior analysts refer to continued responding during the extinction procedure as resistance to extinction. Be- havior that continues to occur during extinction is said to have greater resistance to extinction than behavior that diminishes quickly. Resistance to extinction is a relative concept. Three methods are commonly used to measure resistance to extinction. Reynolds (1968) described two measurement procedures: the rate of decline in response frequency, and the total number of responses emitted before responding either attains some final low level or ceases. Lerman, Iwata, Shore, and Kahng (1996) reported measuring resistance to extinction as the duration of time required for a behavior to reach a predetermined criterion.

Continuous and Intermittent Reinforcement

Three tentative statements describe resistance to extinc- tion as it relates to continuous and intermittent rein- forcement. (a) Intermittent reinforcement may produce behavior with greater resistance to extinction than the re- sistance produced by continuous reinforcement. For ex- ample, behavior maintained on a continuous schedule of reinforcement may diminish more quickly than behavior maintained on intermittent schedules of reinforcement (Keller & Schoenfeld, 1950/1995). (b) Some intermittent schedules may produce more resistant than others (Ferster & Skinner, 1957). The two variable schedules (VR, VI) may have more resistance to extinction than the fixed schedules (FR, FI). (c) To a degree, the thinner the inter- mittent schedule of reinforcement is, the greater the re- sistance to extinction will be.

An undergraduate student in an introductory class in applied behavior analysis gave an interesting example of the effects of intermittent reinforcement on persistence. About 2 months after a lecture on resistance to extinc- tion, the student shared this experience:

You know, intermittent reinforcement really does affect persistence. This guy I know was always calling me for a date. I didn’t like him and didn’t enjoy being with him. Most of the time when he asked for a date I would not go out with him. Occasionally, because of his persis- tence, I would give in and accept a date. After your lec- ture I realized that I was maintaining his calling on an

Extinction

475

intermittent schedule of reinforcement, which could ac- count for his persistence. I decided to change this situa- tion and put him on a continuous schedule of reinforcement. Every time he called for a date, I agreed to go out with him. We spent about four evenings a week together for 3 weeks. Then without giving any reason, I abruptly stopped accepting dates with him. Since then he’s called only three times, and I have not heard from him in 4 weeks.

Establishing Operation

All stimuli that function as reinforcers require a mini- mum level of an establishing operation (i.e., motivation must be present). The strength of the establishing opera- tion (EO) above the minimum level will influence resis- tance to extinction. Basic research has shown that “[R]esistance to extinction is greater when extinction is carried out under high motivation than under low” (Keller & Schoenfeld, 1950/1995, p. 75). We assume that the functional relation between an establishing operation and resistance to extinction exists in applied settings as well as laboratory settings. For example, the fellow might per- sist in calling the woman for a date if he had made a siz- able bet with his buddies that the woman would date him again.

Number, Magnitude, and Quality of Reinforcement

The number of times a behavior has produced reinforce- ment may influence resistance to extinction. A behavior with a long history of reinforcement may have more re- sistance to extinction than a behavior with a shorter his- tory of reinforcement. If bedtime tantrums have produced reinforcement for 1 year, they might have more resistance to extinction than tantrums that have produced rein- forcement for only 1 week.

Lerman, Kelley, Vorndran, Duhn, and LaRue (2002) reported that applied behavior analysts have not estab- lished what effects the magnitude of a reinforcer has on the resistance to extinction. However, the magnitude and quality of a reinforcer will likely influence resistance to extinction. A reinforcer of greater magnitude and quality might produce more resistance to extinction than a rein- forcer with less magnitude and quality.

Number of Previous Extinction Trials

Successive applications of conditioning and extinction may influence the resistance to extinction. Sometimes problem behaviors diminish during extinction, and then are accidentally strengthened with reinforcement. When

this happens, applied behavior analysts can reapply the extinction procedure. Typically, behavior may diminish quickly with fewer total responses during a reapplication of extinction. This effect is additive when the participant can discriminate the onset of extinction. With each suc- cessive application of extinction, decreases in behavior become increasingly rapid until only a single response occurs following the withdrawal of reinforcement.

Response Effort

Even though limited, applied researchers have produced some data on response effort and its effect on resistance to extinction (Lerman & Iwata, 1996a). The effort re- quired for a response apparently influences its resistance to extinction. A response requiring greater effort dimin- ishes more quickly during extinction than a response re- quiring less effort.

Using Extinction Effectively Numerous guidelines for using extinction effectively have been published, with most authors providing similar rec- ommendations. Presented below are 10 guidelines to fol- low before and during the application of extinction. They are: withholding all reinforcers maintaining the problem behavior, withholding reinforcement consistently, com- bining extinction with other procedures, using instruc- tions, planning for extinction-produced aggression, increasing the number of extinction trials, including sig- nificant others in extinction, guarding against uninten- tional extinction, maintaining extinction-decreased behavior, and circumstances under which extinction should not be used.

Withholding All Reinforcers Maintaining the Problem Behavior

A first step in using extinction effectively is to identify and withhold all possible sources of reinforcement that maintain the target behavior. The effectiveness of ex- tinction is dependent on the correct identification of the consequences that maintain the problem behavior (Iwata, Pace, Cowdery, & Miltenberger, 1994). Functional be- havior assessments have improved greatly the applica- tion and effectiveness of using extinction in applied settings (Lerman, Iwata, & Wallace, 1999; Richman, Wacker, Asmus, Casey, & Anderson, 1999).

Applied behavior analysts collect data on antecedent and consequent stimuli that are temporally related to the

Extinction

476

problem behavior and provide answers to questions such as the following:

1. Does the problem behavior occur more frequently when something happens in the environment to oc- casion the behavior (e.g., a demand or request)?

2. Is the frequency of the problem behavior unrelated to antecedent stimuli and social consequences?

3. Does the problem behavior occur more frequently when it produces attention from other persons?

If the answer to the first question is yes, the problem be- havior may be maintained with negative reinforcement. If the answer to the second question is yes, the applied be- havior analyst will need to consider withholding tactile, auditory, visual, gustatory, olfactory, and kinesthetic con- sequences alone or in combination. If the answer to the third question is yes, the behavior may be maintained by positive reinforcement in the form of social attention.

The consequence that maintains the problem behav- ior appears obvious in some applied settings. In the Williams (1959) study, for instance, parental attention appeared to be the only source of reinforcement main- taining the tyrant-like behavior at bedtime. However, be- haviors are frequently maintained by multiple sources of reinforcement. The class clown’s behavior might be main- tained by the teacher’s reaction to the disruptive behav- ior, by the attention received from peers, or a combination of both. Johnny might cry when his parent brings him to preschool to escape from preschool, to keep his parents with him, to occasion his teacher’s concern and attention, or to achieve some combination of all three. When mul- tiple sources of reinforcement maintain problem behavior, identifying and withholding one source of reinforcement may have minimal or no affect on behavior. If teacher and peer attention maintain clownish classroom behavior, then withholding only teacher attention may produce lit- tle change in the problem behavior. The teacher must withhold her attention and also teach the other students to ignore the clownish behavior to effectively apply the extinction procedure.

Withholding Reinforcement Consistently

When the reinforcing consequences have been identified, teachers must withhold them consistently. All behavior change procedures require consistent application, but con- sistency is essential for extinction. Teachers, parents, ther- apists, and applied behavior analysts often report that consistency is the single most difficult aspect in using extinction. The error of not withholding reinforcement consistently negates the effectiveness of the extinction procedure, and this point cannot be overemphasized.

Combining Extinction with Other Procedures

Extinction is an effective singular intervention. However, teachers and therapists, applied researchers, and writers of textbooks have rarely recommended it as such (Vollmer et al., 1998). We also recommend that applied behavior analysts always consider combining extinction with other treatments, especially the reinforcement of al- ternative behaviors. Two reasons support this recom- mendation. First, the effectiveness of extinction may increase when it is combined with other procedures, es- pecially positive reinforcement. By combining extinction with differential reinforcement of appropriate behaviors, for example, the applied behavior analyst alters the en- vironment by reinforcing appropriate alternative behav- iors and placing problem behaviors on extinction. During intervention, the extinction procedure should not reduce the overall amount of positive consequences received by a participant. Second, differential reinforcement and an- tecedent procedures hold promise for reducing extinc- tion effects such as bursting and aggression (Lerman, Iwata, & Wallace, 1999).

Rehfeldt and Chambers (2003) combined extinction with differential reinforcement for the treatment of ver- bal perseverance maintained by positive reinforcement in the form of social attention. The participant was an adult with autism. Rehfeldt and Chambers presented so- cial attention contingent on appropriate and nonpresev- eration speech, and applied attention extinction by not responding to the participant’s inappropriate verbal be- havior. This combined treatment produced an increase in appropriate verbal responses and a decrease in persever- ance. These data suggest that contingencies of rein- forcement might maintain the unusual speech of some individuals with autism.

Using Instructions

The contingencies of reinforcement affect the future fre- quency of behavior automatically. It is not necessary for people to know, to describe, or even to perceive that con- tingencies are affecting their behavior. However, behav- iors sometime diminish more quickly during extinction when teachers describe the extinction procedure to stu- dents. For example, teachers frequently provide small- group instruction while other students do independent seatwork. When seatwork students ask questions, they interrupt instruction. Many teachers correct this problem by placing question asking on extinction. They simply ignore student questions until after the end of small-group instruction. This tactic is often effective. The extinction procedure, however, tends to be more effective when

Extinction

477

Extinction

teachers tell students that they will ignore all their ques- tions until after the end of small-group instruction.

Planning for Extinction-Produced Aggression

Behaviors that occurred infrequently in the past will sometime become prominent during extinction by re- placing the problem behaviors. Frequently, these side effect replacement behaviors are aggressive (Lerman et al., 1999). Skinner (1953) interpreted the change in response topographies (e.g., side effects) as emotional behaviors, including aggression that sometimes ac- companies extinction.

Goh and Iwata (1994) provided a convincing demon- stration showing how aggression-induced behaviors oc- curred when a target behavior was placed on extinction. Steve was a 40-year-old man with profound mental re- tardation. He had been referred for evaluation for self- injury (head banging, head hitting). A functional analysis showed that self-injury was reinforced by escape from instructions. Goh and Iwata used extinction for the treat- ment of Steve’s self-injury. Steve rarely slapped or kicked other persons (i.e., aggression) during the two baseline phases, but aggression increased with the onset of the two extinction phases (see Figure 5, bottom panel). Ag- gression essentially stopped by the end of each extinc- tion phase when self-injury was stable and low, even though Steve’s aggression remained untreated during the baseline and extinction conditions.

Applied behavior analysts need to plan on manag- ing aggressive behaviors when they occur as side effects of extinction. It is critical that extinction-produced ag- gression not produce reinforcement. Frequently, parents, teachers, and therapists react to the aggression with at- tention that may function as reinforcement of the extinction-produced aggression. For instance, a teacher decides to ignore DeShawna’s questions while providing small-group instruction. When DeShawna interrupts the teacher with a question, the teacher does not respond. DeShawna then starts disrupting other seatwork students. To quiet DeShawna down, the teacher responds, “Oh, all right, DeShawna, what did you want to know?” In effect, the teacher’s behavior may have reinforced DeShawna interruptions during small-group instruction and the ac- companying inappropriate disruptions of the other seat- work students.

Many times extinction-produced aggression takes the form of verbal abuse. Often, teachers and parents do not need to react to it. If extinction-produced aggression produces reinforcement, individuals simply use other in- appropriate behavior such as verbal abuse to produce re- inforcement. Teachers and parents cannot and should not

ignore some forms of aggression and self-injurious be- havior. Teachers and parents need to know (a) that they can ignore some aggression, (b) when they need to in- tervene on the aggression, and (c) what they will do for the intervention.

Increasing the Number of Extinction Trials

An extinction trial occurs each time the behavior does not produce reinforcement. Whenever possible, applied behavior analysts should increase the number of extinc- tion trials for the problem behaviors. Increasing extinc- tion trials improves the efficiency of extinction by accelerating the extinction process. Applied behavior an- alysts can increase extinction trials when applied settings will accommodate frequent occurrences of the problem behavior. For example, Billy’s parents employed an ex- tinction procedure to reduce his tantrums. His parents no- ticed that Billy had tantrums most often when he did not get his way about staying up late, eating a snack, and going outside. For the purpose of the program, they de- cided to set up several additional situations each day in which Billy did not get his way. Billy was then more likely to emit the inappropriate behavior at a higher rate, thereby giving his parents more occasions to ignore it. As a result, his tantrums decreased in a shorter period of time than they would have if left at their usual rate.

Including Significant Others in Extinction

It is important that other persons in the environment not re- inforce undesirable behavior. A teacher, for example, needs to share extinction plans with other people who might help in the classroom—parent volunteers, grandparents, music teachers, speech therapists, industrial arts specialists—to avoid their reinforcing inappropriate behaviors. All people in contact with the learner must apply the same extinction procedure for effective treatment.

Guarding against Unintentional Extinction

Desirable behaviors are often unintentionally placed on extinction. A beginning teacher confronted with one stu- dent who is on task and many other students who are not will probably direct most of his attention to the majority and will provide little or no attention to the student who is working. It is common practice to give the most atten- tion to problems (the squeaky wheel gets the grease) and to ignore situations that are going smoothly. However, behaviors must continue to be reinforced if they are to

478

Extinction

be maintained. All teachers must give attention to stu- dents who are on-task.

Maintaining Extinction-Decreased Behavior

Applied behavior analysts leave the extinction procedure in effect permanently for maintaining the extinction- diminished behavior. A permanent application of escape extinction and attention extinction is a preferred proce- dure. Applied behavior analysts can use a permanent ap- plication also with some sensory extinction procedures. For example, the carpeting on a table used for spinning objects could remain in place indefinitely. Some appli- cations of sensory extinction appear inappropriate and inconvenient if kept permanently in effect. For example, requiring Tina to permanently wear thin cotton mittens on both hands at naptime during day care and during bedtime at home appears inappropriate and inconvenient (see Figure 3). In such a case, the applied behavior ana- lysts can maintain treatment gains by gradually fading out the sensory extinction procedure—for example, cutting off 1 inch of the glove palm every 3 to 4 days until the glove palm is removed. The analyst may then remove the fingers and thumb one at a time to gradually remove the mittens.

When Not to Use Extinction

Imitation

Extinction can be inappropriate if the behaviors placed on extinction are likely to be imitated by others. Some be- haviors can be tolerated if only one person emits them, but become intolerable if several persons emit them.

Extreme Behaviors

With few exceptions, most applications of extinction as a singular intervention have focused on important but rel- atively minor behavior problems (e.g., disruptive class- room behavior, tantrums, excessive noise, mild forms of aggression). However, some behaviors are so harmful to self or others or so destructive to property that they must be controlled with the most rapid and humane procedure available. Extinction as a singular intervention is not rec- ommended in such situations.

The use of extinction as a singular intervention to decrease severe aggression toward self, others, or prop- erty raises ethical concerns. Addressing the issue of ethics, Pinkston, Reese, LeBlanc, and Baer (1973) ana- lyzed the effects of an extinction technique that did not allow a person to harm himself or his victim. In their ap- proach the aggressor was ignored, but the victim was sheltered from attack. Pinkston and colleagues demon- strated the effectiveness of a safe extinction technique with an extremely aggressive preschool boy, whose ag- gressive behaviors included choking, biting, pinching, hitting, and kicking classmates. During the baseline con- dition the teachers responded to the child’s aggression as they had in the past. “Typically, this took the form of ver- bal admonitions or reproofs such as: ‘Cain, we do not do that here,’ or ‘Cain, you can’t play here until you are ready to be a good boy’” (p. 118). The teachers did not attend to the boy’s aggressive behaviors during extinction. When he attacked a peer, the teachers immediately attended to the peer. The victim was consoled and provided with an opportunity to play with a toy. In addition, the teachers at- tended to the boy’s positive behaviors. This extinction procedure was effective in greatly reducing aggression. The application of extinction requires sound, mature, hu- mane, and ethical professional judgment.

Summary Definition of Extinction

1. Extinction as a procedure provides zero probability of re- inforcement. It is also a behavioral process of a diminish- ing rate of response.

2. Functional behavior assessments have enabled applied behavior analysts to distinguish clearly between the pro- cedural variations of extinction (ignoring) and the func- tional variations of extinction (withholding maintaining reinforcers).

3. Extinction is a technical term that applied behavior analysts should use only to identify the procedure of withholding reinforcers that maintain behavior.

Extinction Procedures

4. Procedures for extinction take three distinct forms that are linked to behavior maintained by positive reinforcement, negative reinforcement, and automatic reinforcement.

5. Behaviors maintained by positive reinforcement are placed on extinction when those behaviors do not pro- duce the reinforcer.

6. Behaviors maintained by negative reinforcement are placed on extinction (a.k.a., escape extinction) when those behaviors do not produce a removal of the aversive stim- ulus, meaning that the individual cannot escape from the aversive situation.

479

7. Behaviors maintained by automatic reinforcement are placed on extinction (a.k.a., sensory extinction) by mask- ing or removing the sensory consequence.

Extinction Effects

8. Behaviors undergoing extinction are usually associated with predictable characteristics in rate and topography of response.

9. Extinction produces a gradual reduction in behavior.

10. A general effect of the extinction procedure is an imme- diate increase in the frequency of the response after the removal of the positive, negative, or automatic reinforce- ment. The behavioral literature uses the term extinction burst to identify this initial increase in response frequency.

11. With spontaneous recovery, the behavior that diminished during the extinction process recurs even though the be- havior does not produce reinforcement.

Variables Affecting Resistance to Extinction

12. Behavior that continues to occur during extinction is said to have greater resistance to extinction than behavior that diminishes quickly. Resistance to extinction is a relative concept.

13. Intermittent schedules of reinforcement may produce be- havior with greater resistance to extinction than the resis- tance produced by a continuous reinforcement schedule.

14. Variables schedules of reinforcement (e.g., VR, VI) may have more resistance to extinction than fixed schedules (e.g., FR, FI).

15. To a degree, the thinner the intermittent schedule of rein- forcement is, the greater the resistance to extinction will be.

16. Resistance to extinction is likely to increase with the strength of the establishing operation (EO) for the rein- forcer being withheld.

17. The number, magnitude, and quality of the reinforcer may affect resistance to extinction.

18. Successive applications of conditioning and extinction may influence the resistance to extinction.

19. The effort required for a response apparently influences its resistance to extinction.

Using Extinction Effectively

20. A first step in using extinction effectively is to withhold all reinforcers that are maintaining the problem behavior.

21. When the reinforcing consequences have been identified, teachers must withhold them consistently.

22. Applied behavior analysts should always consider com- bining extinction with other procedures.

23. Behaviors often decrease more quickly during extinction when the person is informed of the procedure being applied.

24. When using extinction, applied behavior analysts should plan for extinction-produced aggression.

25. Increasing the number of extinction trials improves the ef- ficiency of extinction by accelerating the extinction process.

26. All people in contact with the learner must apply the same extinction procedure for effective treatment.

27. Extinction should not be used for behaviors that are likely to be imitated by others or for behaviors that are harmful to self or others.

Extinction

480

differential reinforcement of alternative behavior (DRA)

differential reinforcement of incompatible behavior (DRI)

differential reinforcement of low rates (DRL)

differential reinforcement of other behavior (DRO)

fixed-interval DRO (FI-DRO) fixed-momentary DRO (FM-DRO) full-session DRL interval DRL

spaced-responding DRL variable-interval DRO (VI-DRO) variable-momentary DRO

(VM-DRO)

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 9: Behavior Change Procedures

9-6 Use differential reinforcement.

Differential Reinforcement

Key Terms

481

Differential Reinforcement

Practitioners can choose from among a wide range of effective procedures for decreasing or eliminating problem behaviors. Although often

effective in reducing targeted problem behaviors, inter- ventions based primarily on extinction or punishment may produce unwanted side effects. Maladaptive emo- tional behavior and a higher-than-usual rate of respond- ing are commonly observed when a behavior with a long and consistent history of reinforcement is placed on ex- tinction. Punishment may evoke escape, avoidance, ag- gression, and other forms of undesirable countercontrol. In addition to unwanted side effects, another limitation of extinction and punishment as primary methods for re- ducing problem behaviors is that neither approach strengthens or teaches adaptive behaviors with which in- dividuals can attain the reinforcers previously achieved by the undesirable behaviors. Finally, beyond the possibil- ity of unwanted side effects and their lack of educative value, interventions that rely on extinction alone or pun- ishment in any of its varied forms raise important ethical and legal concerns (Repp & Singh, 1990).

Because of all these concerns, applied behavior an- alysts have developed effective reinforcement-based pro- cedures for reducing problem behaviors (e.g., Kazdin, 1980; Lerman & Vorndran, 2002; Singh & Katz, 1985). These positive reductive procedures are based on dif- ferential reinforcement to diminish or eliminate prob- lem behaviors.

Basic Description of Differential Reinforcement All applications of differential reinforcement entail rein- forcing one response class and withholding reinforce- ment for another response class.1 When used as a reductive procedure for problem behavior, differential re- inforcement consists of two components: (a) providing re- inforcement contingent on either the occurrence of a behavior other than the problem behavior or the problem behavior occurring at a reduced rate, and (b) withholding reinforcement as much as possible for the problem be- havior. Although implementing differential reinforcement involves extinction, as Cowdery, Iwata, and Pace (1990) pointed out,

differential reinforcement does not involve extended in- terruptions of ongoing activities (e.g., time out), contin- gent removal of positive reinforcers (e.g., response cost),

1Differential reinforcement is not only for reducing problem behaviors. As described elsewhere in this text, differential reinforcement is a defining feature of shaping new behaviors. Various forms of differential reinforce- ment contingencies are also used as experimental control procedures (see Thompson & Iwata, 2005).

or presentation of aversive stimuli (e.g., punishment). These characteristics make differential reinforcement the least intrusive of all behavior interventions and prob- ably account for its widespread popularity. (p. 497)

Differential reinforcement in its various forms is one of the most effective, widely known, and commonly used techniques to reduce problem behavior. The four most researched variations of differential reinforcement for de- creasing inappropriate behavior are differential rein- forcement of incompatible behavior (DRI), differential reinforcement of alternative behavior (DRA), differen- tial reinforcement of other behavior (DRO), and differ- ential reinforcement of low rates (DRL). This chapter defines, gives examples of applications of, and suggests guidelines for the effective use of each of these differen- tial reinforcement procedures for decreasing problem behavior.

Differential Reinforcement of Incompatible Behavior and Differential Reinforcement of Alternative Behavior DRI and DRA have the dual effects of weakening the problem behavior while simultaneously strengthening acceptable behaviors that are either incompatible with or an alternative to the targeted problem behaviors. When properly implemented as a treatment for problem be- havior, differential reinforcement of an incompatible or alternative behavior can be conceptualized as a sched- ule of reinforcement in which two concurrent operants— the inappropriate behavior targeted for reduction and the appropriate behavior selected—receive reinforcement at different rates (Fisher & Mazur, 1997). Because the dif- ferential reinforcement schedule favors the appropriate behavior, the client allocates more responding to the ap- propriate behavior and less responding to the problem behavior (which is placed on extinction) (Vollmer, Roane, Ringdahl, & Marcus, 1999). For example, when Friman (1990) reinforced the in-seat behavior of a preschool student with hyperactivity, out-of-seat behav- ior decreased markedly.

With the proper selection of behaviors, these two in- terventions may promote educational, social, and personal skill development. With DRI/DRA, the practitioner con- trols the development of appropriate behaviors and con- currently measures both the problem behavior and the desired replacement behavior. Teachers, therapists, and parents have a long history of using DRI/DRA inter- ventions in education, treatment, and everyday social

482

Differential Reinforcement

interactions. Practitioners usually find DRI/DRA the easi- est of the four differential reinforcement procedures to apply.

Differential Reinforcement of Incompatible Behavior

A practitioner applying differential reinforcement of incompatible behavior (DRI) reinforces a behavior that cannot occur simultaneously with the problem behavior and withholds reinforcement following instances of the problem behavior. The behavior that gets reinforced (e.g., a student seated at his seat) and the problem behavior that is placed on extinction (e.g., student is out of seat) are mutually exclusive response classes whose different topographies make it impossible to emit both behaviors at the same time.

Dixon, Benedict, and Larson (2001) used DRI to treat the inappropriate verbal behavior of Fernando, a 25- year-old man with moderate mental retardation and a psy- chotic disorder. Inappropriate vocalizations included utterances not related to the context, sexually inappro- priate remarks, illogical placement of words within a sen- tence, and “psychotic” statements (e.g., “There’s a purple moose on my head named Chuckles,” p. 362). The re- searchers defined appropriate verbal behavior as any vocal utterance that did not meet the defining character- istics of inappropriate vocalization. A functional analysis revealed that the inappropriate verbal statements were maintained by social attention.

The researchers implemented DRI by ignoring Fer- nando’s inappropriate verbal behavior and attending to his appropriate statements with 10 seconds of comments that were relevant to Fernando’s statements. For example, if Fernando said something about an activity he liked to do, the experimenter told him it was interesting and hoped he could do it again soon. During a comparison baseline condition, inappropriate behavior resulted in attention and appropriate verbal behavior was ignored. The DRI intervention effectively reduced Fernando’s inappropriate verbal behavior and increased appropriate utterances (Figure 1).

Differential Reinforcement of Alternative Behavior

Differential reinforcement of alternative behavior (DRA) and DRI are similar procedures. A practitioner using differential reinforcement of alternative behavior (DRA) reinforces occurrences of a behavior that pro- vides a desirable alternative to the problem behavior but is not necessarily incompatible with it. Behavior ana- lysts can use an alternative behavior to occupy the time that the problem behavior would ordinarily use. The al-

2Any behavior selected for reinforcement with DRI provides the learner with an “alternative” to the problem behavior, but not all behaviors se- lected for reinforcement in a DRA are incompatible with the problem be- havior. Sometimes the difference between DRI and DRA is a fine one and provokes legitimate debate. For example, the interventions used in studies by Dixon and colleagues (2001) and Wilder, Masuda, O’Conner, and Baham (2001) were described as applications of DRA. However, because appropriate verbal behavior was defined in each study as vocal utterances that did not meet the defining characteristics of inappropriate vocaliza- tions, the response classes they reinforced were incompatible with the prob- lem behavior. If the “alternative” behavior selected for reinforcement cannot occur simultaneously with the problem behavior, we believe the procedure is properly labeled DRI.

ternative behavior and the problem behavior, however, are not topographically incompatible. For example, a classroom teacher could assign two students who fre- quently argue with each other to work on a class project together, thus reinforcing cooperative behaviors associ- ated with project development. Working together on a class project is not incompatible with arguing; the two re- sponse classes could occur together. However, the two students might argue less when engaged in cooperative behaviors.2

Roane, Lerman, and Vorndran (2001) selected plac- ing a plastic block in a bucket as an alternative behavior to screaming (i.e., a brief vocal sound above conversa- tional levels). Lerman, Kelley, Vorndran, Kuhn, and LaRue (2002) shaped and maintained touching a com- munication card as an alternative behavior to disruptions such as throwing task materials and aggression.

When escape from a task or demand situation is used as the reinforcer in a differential reinforcement procedure for reducing inappropriate behavior, the in- tervention is sometimes called differential negative reinforcement of alternative (or incompatible) behavior (DNRA or DNRI) (Vollmer & Iwata, 1992). DRI/DRA interventions can use negative reinforcers also (Mar- cus & Vollmer, 1995; Piazza, Moses, & Fisher, 1996). Applying DNRA/DNRI to reduce the occurrence of a problem behavior maintained by escape from a task or demand context consists of providing negative rein- forcement of the alternative behavior in the form of brief periods of escape from the task and escape ex- tinction for the problem behavior. Positive reinforce- ment for the alternative/incompatible behavior is often provided as well.

Lalli, Casey, and Kates (1995) allowed participants to escape from a task for 30 seconds contingent on using an alternative response (e.g., giving the therapist a card with “BREAK” printed on it, or saying “no”) as an al- ternative to their aberrant behavior. The researchers also provided praise for using the alternative responses. As training progressed, the participants’ escape from a task was contingent on using the trained verbal responses and completing a gradually increasing number of required

483

Differential Reinforcement

steps of the task. The DNRA intervention increased the use of the trained verbal response and decreased the prob- lem behavior.3

Guidelines for Using DRI/DRA

Practitioners often use DRI and DRA as interventions for a wide range of problem behaviors. Many re- searchers and practitioners have discovered that the fol- lowing guidelines improve the effectiveness of DRI and DRA.

Select Incompatible/Alternative Behavior

Ideally, the behavior selected to be incompatible with or alternative to the inappropriate behavior (a) already ex- ists in the learner’s current repertoire; (b) requires equal, or preferably less, effort than the problem behavior; (c) is being emitted at a rate prior to the DRI/DRA inter- vention that will provide sufficient opportunities for re- inforcement; and (d) is likely to be reinforced in the learner’s natural environment after intervention is termi- nated. Behaviors meeting these criteria will increase the initial effectiveness of DRI/DRA and facilitate the main- tenance and generalization of behavior changes after in- tervention is terminated.

When possible, the practitioner should differentially reinforce incompatible/alternative behaviors that will lead to or increase the opportunities for the learner to acquire new or other useful skills. When no behaviors meeting those criteria can be identified, the practitioner can se- lect an alternative behavior that can be easily taught or consider using a DRO procedure. Figure 2 shows some examples of incompatible or alternative behaviors (Webber & Scheuermann, 1991).

3DRA is a primary component of functional communication training (FCT), an intervention in which an alternative behavior is selected to serve the same communicative function as the problem behavior (e.g., saying “Break please” results in the same reinforcement as aggression or tantrumming did previously). FCT is described in detail in Chapter entitled “Antecedent Interventions.”

Select Reinforcers That Are Powerful and Can Be Delivered Consistently

Perhaps the greatest threat to the effectiveness of any re- inforcement-based intervention is the use of stimulus changes following behaviors that the practitioner only assumes function as reinforcement. Providing as conse- quences for the incompatible/alternative behavior stimu- lus changes identified by stimulus preference assessment and reinforcer assessment or functional behavior assess- ments will increase the effectiveness of DRI/DRA. In ad- dition, reinforcers should be selected that are relevant to the establishing operations that naturally exist in the treat- ment setting or that can be created (e.g., by deprivation prior to treatment sessions).

The same consequence that is maintaining the prob- lem behavior prior to intervention is often the most ef- fective reinforcer for the alternative or incompatible behavior (Dixon et al., 2001; Wilder et al., 2001). After discovering that the problem behaviors emitted by dif- ferent students in a classroom were maintained by dif- ferent reinforcers, Durand, Crimmins, Caufield, and Taylor (1989) effectively used individually selected re- inforcers to differentially reinforce appropriate alternative behaviors for each student, and the inappropriate behav- iors decreased in frequency.

The effectiveness of any intervention involving dif- ferential reinforcement depends on the practitioner’s abil- ity to deliver and withhold consistently stimulus changes that currently function as reinforcement. Therefore, in addition to their potential effectiveness as reinforcers, the stimulus changes that the practitioner selects must be ones he can deliver immediately and consistently when the alternative/incompatible behavior occurs and with- hold following instances of the problem behavior.

The magnitude of the reinforcer used in a differen- tial reinforcement intervention is probably less impor- tant than its consistent delivery and control. Lerman, Kelley, Vorndran, Kuhn, and LaRue (2002) provided pos- itive reinforcement (access to toys) and negative rein- forcement (escape from demands) in magnitudes ranging from 20 seconds to 300 seconds. They found that rein- forcement magnitude only affected responding minimally

N u

m b

er o

f V

er b

al U

tt er

an ce

1 3 5 7 9 11 13 15 17 19 21 23 25 Sessions

Appropriate

Inappropriate

DRI Baseline DRI Baseline DRI

Figure 1 Number of appropriate and inappropriate verbal utterances by an adult male during DRI and baseline conditions. Adapted from “Functional Analysis and Treatment of Inappropriate Verbal Behavior” by M. R. Dixon, H. Benedict, & T. Larson (2001), Journal of Applied Behavior Analysis, 34, p. 362. Copyright 2001 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

484

Differential Reinforcement

during treatment and had little effect on maintenance after treatment was discontinued.

Reinforcing Incompatible/Alternative Behavior Immediately and Consistently

Initially, a practitioner should use a continuous (CRF) schedule of reinforcement for the incompatible or alterna- tive behavior, and then transition to an intermittent sched- ule. The reinforcer should be presented consistently and immediately following each occurrence of the incompati- ble or alternative behavior. After firmly establishing the in- compatible or alternative behavior, the practitioner should gradually thin the reinforcement schedule.

Withhold Reinforcement for the Problem Behavior

The effectiveness of differential reinforcement as an in- tervention for problem behavior depends on the incom- patible or alternative behavior yielding a higher rate of reinforcement than the problem behavior. Maximizing the

difference between rates of reinforcement obtained by the two response classes entails withholding all reinforcement for the problem behavior (i.e., extinction schedule).

Ideally, the alternative behavior would always be re- inforced (at least initially) and the problem behavior would never be reinforced. As Vollmer and colleagues (1999) noted, “Perfect implementation of differential re- inforcement entails providing reinforcers as immediately as possible after an appropriate behavior occurs (e.g., within 5 s). Treatment effects may degrade as the delay to reinforcement increases, especially if inappropriate be- havior is occasionally reinforced” (p. 21). However, prac- titioners must often implement DRI/DRA procedures in less than optimal conditions, in which some occurrences of the alternative/incompatible behavior are not followed by a reinforcer and the problem behavior sometimes in- advertently produces reinforcement.

Results from a study by Vollmer and colleagues (1999) suggest that even when such treatment “mistakes” occur, differential reinforcement may still be effective. These researchers compared a “full implementation” of DRA, in which 100% of instances of alternative behavior

Figure 2 Examples of behaviors selected as incompatible/alternative behaviors when implementing DRI/DRA for common classroom behavior problems.

Problem Behavior Positive Incompatible Alternative

Talking back Positive response such as “Yes, sir” or “OK” or “I understand”; or acceptable questions such as “May I ask you a question about that?” or “May I tell you my side?”

Cursing Acceptable exclamations such as “darn,” “shucks.”

Being off-task Any on-task behavior: looking at book, writing, looking at teacher, etc.

Being out of seat Sitting in seat (bottom on chair, with body in upright position).

Noncompliance Following directions within _____ seconds (time limit will depend upon age of student); following directions by second time direction is given.

Talking out Raising hand and waiting quietly to be called on.

Turning in messy papers No marks other than answers.

Hitting, pinching, kicking, Using verbal expression of anger; pounding fist into hand; pushing/shoving sitting or standing next to other students without touching them.

Tardiness Being in seat when bell rings (or by desired time).

Self-injurious or self- Sitting with hands on desk or in lap; hands not touching any stimulatory behaviors part of body; head up and not touching anything (desk,

shoulder, etc.).

Inappropriate use Holding/using materials appropriately (e.g., writing only on of materials appropriate paper, etc.).

Adapted from “Accentuate the Positive . . . Eliminate the Negative” by J. Webber and B. Scheuermann, 1991, Teaching Exceptional Children, 24(1), p. 15. Copyright 1991 by Council for Exceptional Children. Used by permission.

485

Differential Reinforcement

6.0

4.0

2.0

R es

p o

n se

s p

er M

in u

Inapp.

Appropriate Rachel

10 20 30 40 50 60

0/100 100/0 0/100 25/75 100/0 50/50 75/25

100

50.0

P er

ce n

ta ge

o f

R es

p o

n se

s A

ll o

ca te

Inapp.

App.

Rachel

10 20 30 40 50 60

0/100 100/0 0/100 25/75 100/0 50/50 75/25

Sessions

Figure 3 Number of appropriate and inappropriate responses per minute (upper panel) and allocation of appro- priate and inappropriate behavior (lower panel) by 17-year-old student with disabilities during full (100/0) and partial levels of implementation of DRA. Adapted from “Evaluating Treatment Challenges with Differential Reinforcement of Alternative Behavior” by T. R. Vollmer, H. S. Roane, J. E. Ringdahl, and B. A. Marcus, 1999, Journal of Applied Behavior Analysis, 32, p. 17. Copyright 1999 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

were reinforced (CRF) and 0% of instances of aberrant behavior were reinforced (extinction), with various lev- els of “partial implementation.” For example, with a 25/75 implementation schedule, only one in every four in- stances of appropriate behavior was reinforced, whereas a reinforcer followed three of every four instances of in- appropriate behavior. As expected, full implementation of differential reinforcement produced the greatest effects with inappropriate behavior being “virtually replaced by appropriate behavior, and lower levels of implementa- tion eventually reducing treatment efficacy if the sched- ule of reinforcement favored inappropriate behavior” (p. 20). Figure 3 shows the results for Rachel, a 17-year- old girl with profound mental retardation who engaged in self-injurious behavior (e.g., head hitting and hand biting) and aggression toward others (e.g., scratching, hitting, hair pulling).

An important and perhaps surprising finding of the Vollmer and colleagues (1999) study was that partial im- plementation was effective under certain conditions. Dur- ing partial implementation, the participants’ behavior showed a “disproportional tendency toward appropriate behavior” (p. 20) if they had prior exposure to full im- plementation. The researchers concluded that this finding suggests the possibility of intentionally thinning imple- mentation levels prior to extending differential rein- forcement interventions to settings in which maintaining treatment fidelity will be difficult. Recognizing that treat-

ment effects might erode over time, they recommended that full implementation booster sessions be conducted periodically to reestablish the predominance of appro- priate behavior.

Combine DRI/DRA with Other Procedures

Considering that the DRI/DRA interventions do not specifically provide consequences for the problem be- havior, a practitioner will seldom apply DRI/DRA (or DRL) as a single intervention if the problem behavior is destructive, dangerous to the learner or to others, or if it interferes with health and safety. In these situations, the practitioner might combine DRI/DRA with other reduc- tive procedures (e.g., response blocking, time-out, stim- ulus fading) to produce a more potent intervention (e.g., Patel, Piazza, Kelly, Oschsher, & Santana, 2001).

For example, Ringdahl and colleagues (2002) com- pared the effects of DRA with and without instructional fading on the frequency of problem behaviors of an 8- year-old girl, Kristina, who had been diagnosed with autism and functioned in the moderate range of mental re- tardation. Kristina’s problem behaviors during academic tasks (destruction: throwing, tearing, or breaking work materials; aggression: hitting, kicking, biting; and self- injurious behavior: biting herself) had become so severe that she had been admitted to a hospital day treatment program for help. Because a previous functional analysis

486

Differential Reinforcement

P ro

b le

m B

eh av

io r

p er

M in

u te

0.8

0.6

0.4

0.2

In st

ru ct

io n

s p

er M

in u

Problem Behavior—DRA only Problem Behavior—DRA + Fading Instruction Rate—DRA only Instruction Rate—DRA + Fading

5 10 15 20 25 30 35 Sessions

Figure 4 Problem behavior by an 8-year-old student with autism and developmental disabil- ities (left y axis) and instructions given per minute (right y axis) under DRA only and DRA imple- mented with instructional fading. From “Differential Reinforcement with and without Instructional Fading” by J. E. Ringdahl, K. Kitsukawa, M. S. Andelman, N. Call, L. C. Winborn, A. Barretto, & G. K. Reed (2002), Journal of Applied Behavior Analysis, 35, p. 293. Copyright 2002 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

revealed that Kristina’s problem behavior was maintained by escape from instructions to complete tasks, the DRA procedure in both conditions consisted of giving Kristina a 1-minute break from the task and access to leisure ma- terials contingent on independent completion of the task (e.g., counting items, matching cards with the same col- ors or shapes) without exhibiting problem behavior. In the DRA without instructional fading condition, Kristina was given an instruction to complete a work task at a rate of approximately one instruction every other minute. In the DRA with instructional fading, Kristina was given no instructions during the first three sessions, then one instruction every 15 minutes. The rate of instructions was gradually increased until it equaled the rate in the DRA without the fading condition. Problem behavior was high at the outset of the DRA without the instructional fading condition but decreased across sessions to a mean of 0.2 responses per minute over the final three sessions (see Figure 4). During DRA with instructional fading, problem behavior was low from the outset and occurred during only two sessions (Sessions 9 and 14). Over the final three sessions of DRA with instructional fading, Kristina exhibited no problem behavior and the instruc- tion rate was equal to the DRA without instructional fad- ing (0.5 instructions per minute).

Differential Reinforcement of Other Behavior A practitioner using differential reinforcement of other behavior (DRO) delivers a reinforcer whenever the problem behavior has not occurred during or at specific times (see momentary DRO later in chapter). As de-

4Because instances of target problem behavior in DRO and DRL (to be discussed later in this chapter) often postpone reinforcement, some be- havior analysts have suggested that DRO and DRL be conceptualized as negative punishment procedures rather than reinforcement techniques. For example, Van Houten and Rolider (1990) consider DRO and DRL varia- tions of a procedure they called contingent reinforcement postponement (CRP).

scribed by Reynolds (1961), DRO provides “reinforce- ment for not responding” (p. 59). Because reinforcement is contingent on the absence or omission of target be- havior, DRO is sometimes called differential reinforce- ment of zero responding or omission training (e.g., Weiher & Harman, 1975).4

The delivery of reinforcement with DRO is deter- mined by a combination of how the omission requirement is implemented and scheduled. The omission requirement can make reinforcement contingent on the problem be- havior not occurring either (a) throughout an entire in- terval of time (interval DRO), or (b) at specific moments of time (momentary DRO). With an interval DRO, rein- forcement is delivered if no occurrences of the problem behavior were observed throughout the entire interval. Any instance of the target behavior resets the interval, thereby postponing reinforcement. With a momentary DRO procedure, reinforcement is contingent on the ab- sence of the problem behavior at specific points in time.

The researcher or practitioner can determine whether the omission requirement has been met (i.e., at the ends of intervals or at specific moments in time) according to a fixed or variable schedule. The combinations of these two factors—an interval or momentary omission requirement implemented on either a fixed or a variable schedule— yield the four basic DRO arrangements shown in Figure 5 (Lindberg, Iwata, Kahng, & DeLeon, 1999).

487

Differential Reinforcement

Fixed-Interval DRO (FI-DRO)

Fixed-Momentary DRO (FM-DRO)

Variable-Interval DRO (VI-DRO)

Variable-Momentary DRO (VM-DRO)

Interval

M om

entaryFixed

Variable

Omission Requirement

Sc h

ed u

Figure 5 Four basic variations of DRO that can be created by altering the schedule (fixed or variable) and omission requirement (interval or momentary). Adapted from “DRO Contingencies: An Analysis of Variable- Momentary Schedules” by J. S. Lindberg, B. A. Iwata, S. W. Kahng, & I. G. DeLeon, 1999. Journal of Applied Behavior Analysis, 32, p. 125. Copyright 1999 by the Society for the Experimental Analysis of Behavior, Inc. Used by permission.

Interval DRO

The following sections introduce procedures for applying the fixed-interval DRO, variable-interval DRO, fixed- momentary DRO, and variable-momentary DRO sched- ules. Most researchers and practitioners use the fixed-interval DRO as an intervention for problem be- haviors. However, a growing number of researchers and practitioners are exploring applications of variable- interval DRO and momentary DRO schedules.

Fixed-Interval DRO (FI-DRO)

Most applications of interval DRO apply the omission re- quirement at the end of successive time intervals of equal duration. To apply a fixed-interval DRO (FI-DRO) pro- cedure, a practitioner (a) establishes an interval of time; (b) delivers reinforcement at the end of that interval if the problem did not occur during the interval; and (c) upon any occurrence of the problem behavior, immediately re- sets the timer to begin a new interval.5 For example, Allen, Gottselig, and Boylan (1982) applied an FI-DRO proce- dure as a group contingency to decrease the disruptive classroom behaviors of third-grade students.6 A kitchen timer was set to 5-minute intervals and continued to run as long as no disruptive behaviors occurred. If any student engaged in disruptive behavior at any time during the 5- minute interval, the timer was reset and a new 5-minute interval began. When the timer signaled the end of a 5- minute interval without disruptive behavior, the class was awarded 1 minute of free time, which they accumulated and used at the end of the class period.

The DRO interval and treatment session length can be increased gradually as the person’s behavior improves. Cowdery, Iwata, and Pace (1990) used FI-DRO to help

5A DRO contingency can also be applied using data from permanent prod- uct measurement. For example, Alberto and Troutman (2006) described a procedure in which a teacher provided reinforcement each time a student submitted a paper that contained no doodles. If the papers were basically of the same length or type, the procedure would be a variation of the fixed- interval DRO. 6Group contingencies are described in Chapter entitled “Contigency Con- tracting, Token Economy, and Group Contingencies.”

Jerry, a 9-year-old who had never attended school and spent most of his time hospitalized because he scratched and rubbed his skin so often and hard that he had open le- sions all over his body. During treatment, the experi- menter left the room and watched Jerry through an observation window. Cowdery and colleagues initially made praise and token reinforcement contingent on 2- minute “scratch-free” intervals, which were determined by baseline assessment. If Jerry scratched during the in- terval, the experimenter entered the room and told him she regretted he had not earned a penny and asked him to try again. Jerry’s SIB immediately decreased to zero when DRO was first implemented and remained low as the DRO interval and session length were gradually in- creased (from three 2-minute intervals to three 4-minute intervals). Following a brief return to baseline conditions in which Jerry’s scratching quickly increased, DRO was reinstated with five intervals per session and session length was increased in 1-minute increments. Jerry could now earn a total of 10 tokens per session, a token for each interval in which he did not scratch plus a 5-token bonus if he refrained from scratching for all five intervals. Jerry’s SIB gradually reduced to zero as session length and DRO intervals were extended (see Figure 6). Ses- sion 61 consisted of a single 15-minute DRO interval, which was followed by 25- and 30-minute intervals.

Variable-Interval DRO (VI-DRO)

When reinforcement is delivered contingent on the ab- sence of the targeted problem behavior during intervals of varying and unpredictable durations, a variable- interval DRO (VI-DRO) schedule is in effect. For ex- ample, on a VI-DRO 10-second schedule, reinforcement would be delivered contingent on the omission of the be- havior throughout intervals of varying duration that av- erage 10 seconds (e.g., a random sequence of 2-second, 5-second, 8-second, 15-second, and 20-second intervals).

Chiang, Iwata, and Dorsey (1979) used a VI-DRO intervention to decrease the inappropriate behaviors dur- ing bus rides to and from school of a 10-year-old student

488

Differential Reinforcement

100

P er

ce n

ta ge

I n

te rv

al s

o f

SI B

SIB—Demand SIB—Attention SIB—Play SIB—Alone with Toys SIB—Alone Session Length

Baseline DRO

BL DRO 30

Sessions 20 40 60

Se ss

io n

L en

gt h

( M

in u

te s)

Figure 6 Percentage intervals of SIB by Jerry during baseline and DRO and duration of DRO sessions (right y axis). Other data points during first baseline phase show results of functional analysis confirming that Jerry’s SIB occurred most often when he was left alone. From Cowdery, G., Iwata, B. A., & Pace, G. M. (1990). Effects and Side Effects of DRO as Treatment for Self-injurious Behavior. Journal of Applied Behavior Analysis, 23, 497–506. Copyright 1990 by the Society for the Exper- imental Analysis of Behavior, Inc. Used by permission.

with developmental disabilities. The boy’s disruptive be- haviors included aggression (e.g., slapping, poking, hit- ting, kicking), being out of seat, stereotypic behaviors, and inappropriate vocalizations (e.g., screaming, yelling). Instead of basing the DRO intervals on elapsed time, the researchers used an interesting procedure they described as a “distance-based” DRO schedule. The driver divided the bus route into sections designated by landmarks such as stop signs and traffic lights and mounted a four-digit hand counter on the dashboard of the bus. The counter was within reach of the driver and within view of the stu- dent. At each predetermined landmark the bus driver praised the student’s behavior and added one point to the hand counter if no disruptive behaviors had been emitted during the DRO interval. Upon arrival at home or school the driver recorded the number of points earned on a card and gave the card to the boy’s foster father or teacher. The student exchanged the points earned for rewards (e.g., ac- cess to toys, home and school privileges, small snacks).

Chiang and colleagues used a two-tier multiple base- line across settings design to evaluate the VI-DRO inter- vention. During baseline, disruptive behavior ranged from 20 to 100% with an average of 66.2% for the afternoon rides and 0 to 92% with an average of 48.5% in the morn- ings. When the VI-DRO procedure was implemented on the afternoon rides only, disruptive behavior decreased immediately to an average of 5.1% (range of 0 to 40%) on the afternoon rides but remained unchanged during the morning rides. When the intervention was applied also in the mornings, all disruptive behavior was eliminated.

Progar and Colleagues (2001) implemented VI-DRO as one component of a conjunctive schedule of rein- forcement to reduce the aggressive behaviors of Milty, a 14-year-old boy with autism. Reinforcement is delivered on a conjunctive schedule when the requirements of both schedule components had been met; in this case a fixed- ratio 3 schedule for task completion and a variable- interval 148-second schedule for the absence of aggression. Milty received an edible reinforcer each time he completed three components of a task (e.g., making his bed, vacuuming, straightening objects in his room) in the absence of aggression for the duration of the variable interval. An occurrence of aggression prior to completing the FR 3-second requirement reset both com- ponents of the conjunctive schedule. This intervention produced a substantial reduction of aggressive behavior.

Momentary DRO

Fixed-momentary DRO (FM-DRO) and variable- momentary DRO (VM-DRO) schedules use the same pro- cedures as interval DRO (FI-DRO, VI-DRO) except that reinforcement is contingent on the absence of the problem behavior only when each interval ends, rather than through- out the entire interval as with the whole-interval DRO.

Lindberg, Iwata, Kahng, and DeLeon (1999) used a VM-DRO intervention to help Bridget, a 50-year-old woman with severe developmental disabilities who often banged her head and hit herself on the head and body. A functional analysis indicated that Bridget’s self-injurious

489

Differential Reinforcement

R es

p o

n se

s p

er M

in u

te (

SI B

)

10 20 30 40 50 60 70 Sessions

Baseline BL

VM-DRO

Bridget

300 s22 s

11 s Figure 7 Responses per minute of self- injurious behavior (SIB) during baseline and treatment conditions. From “DRO Contingencies: An Analysis of Variable-momentary Schedules” by J. S. Lindberg, B. A. Iwata, S. W. Kahng, & I. G. DeLeon, 1999. Journal of Applied Behavior Analysis, 32, p. 131. Copyright 1999 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

behaviors (SIB) were maintained by social-positive re- inforcement. During the VM-DRO intervention, her SIB was placed on extinction, and Bridget received 3 to 5 sec- onds of attention from the therapist if she was not bang- ing her head or hitting herself when each interval ended. Bridget did not have to refrain from SIB throughout the interval, only at the end of the interval. As shown in Figure 7, when a VM-DRO 15-second schedule was implemented for five sessions following baseline, Brid- get’s SIB decreased abruptly from baseline levels to al- most zero. Following a return to baseline in which Bridget’s SIB increased, the researchers returned to a VM-DRO treatment condition of 11 seconds, which was later thinned to 22 seconds, and then to the maximum target interval of 300 seconds.

Guidelines for Using DRO

Interval DRO has been used more widely than momen- tary DRO. Some researchers have found that interval DRO is more effective than momentary DRO for sup- pressing problem behavior, and that momentary DRO might be most useful in maintaining reduced levels of problem behavior produced by interval DRO (Barton, Brulle, & Repp, 1986; Repp, Barton, & Brulle, 1983). Lindberg and colleagues (1999) noted two potential ad- vantages of VM-DRO schedules over FI-DRO schedules. First, the VM-DRO schedule appears more practical be- cause the practitioner does not need to monitor the par- ticipant’s behavior at all times. Second, data obtained by these researchers showed that participants obtained higher overall rates of reinforcement with VM-DRO than they did with FI-DRO.

In addition to the importance of selecting potent re- inforcers, we recommend the following guidelines for the effective use of DRO.

Recognize the Limitations of DRO

Although DRO is often positive and highly effective for reducing problem behaviors, it is not without shortcom- ings. With interval DRO, reinforcement is delivered con- tingent only on the absence of problem behavior during the interval, even though another inappropriate behavior might have occurred during that time. For instance, let us assume that a 20-second interval DRO is to be imple- mented to reduce the facial tics of an adolescent with Tourette’s syndrome. Reinforcement will be delivered at the end of each 20-second interval containing no facial tics. However, if the adolescent engages in cursing be- havior at any time during the interval or at the end of the interval, reinforcement will still be delivered. It is possi- ble that reinforcement delivered contingent on the ab- sence of facial tics will occur in close temporal proximity to cursing, thereby inadvertently strengthening another inappropriate behavior. In such cases, the length of the DRO interval should be shortened and/or the definition of the problem behavior expanded to include the other un- desirable behaviors (e.g., reinforcement contingent on the absence of facial tics and cursing).

With momentary DRO, reinforcement is delivered contingent on the problem behavior not occurring at the end of each interval, even though the inappropriate be- havior might have occurred throughout most of an inter- val. Continuing the previous example, on an FM-DRO 20-second schedule, reinforcement will be delivered at

490

Differential Reinforcement

each successive 20th second if facial tics are not occur- ring at those precise moments, even if tics occurred for 50%, 75%, or even 95% of the interval. In such circum- stances, the practitioner should use interval DRO and re- duce the length of the DRO interval.

DRO may not always be successful, and practition- ers must be alert to their data to make informed deci- sions. For instance, Linscheid, Iwata, Ricketts, Williams, and Griffin (1990) found that the systematic application of neither DRO nor DRI eliminated the chronic self- injurious behavior of five people with developmental disabilities. A more intrusive punishment procedure eventually had to be instituted to eliminate the behavior.

Set Initial DRO Intervals That Assure Frequent Reinforcement

Practitioners should establish an initial DRO time inter- val that ensures that the learner’s current level of behav- ior will contact reinforcement when the DRO contingency is applied. For example, Cowdery and col- leagues (1990) initially made praise and token rein- forcement contingent on 2-minute “scratch-free” intervals because that was “the longest we had previously seen Jerry refrain from scratching while left alone” (p. 501). Beginning with an interval that is equal to or slightly less than the mean baseline interresponse time (IRT) usually will make an effective initial DRO interval. To calculate a mean IRT, the practitioner divides the total duration of all baseline measurements by the total num- ber of responses recorded during baseline. For example, if a total of 90 responses were recorded during 30 min- utes of baseline measurement, the mean IRT would be 20 seconds (i.e., 1,800 seconds ÷ 90 responses). Using these baseline data as a guide, the practitioner could set the initial DRO interval at 20 seconds or less.

Do Not Inadvertently Reinforce Other Undesirable Behaviors

Reinforcement on a “pure” DRO schedule is contingent on a very general response class that is defined only by the absence of the targeted problem behavior. Conse- quently, a “pure” DRO is typically used as a treatment of choice for serious behavior problems that occur at very high rates by people whose current repertoires provide few, if any, other behaviors that might function as alter- native or incompatible behaviors and for whom just about anything else they might do is less of a problem than the target behavior.

Because DRO does not require that any certain be- haviors be emitted for reinforcement, whatever the per- son is doing when reinforcement is delivered is likely to

occur more often in the future. Therefore, practitioners using DRO must be careful not to strengthen inadver- tently other undesirable behaviors. When using DRO, the practitioner should deliver reinforcement at the intervals or moments in time specified by the schedule contingent on the absence of the problem behavior and the absence of any other significant inappropriate behaviors.

Gradually Increase the DRO Interval

The practitioner can increase the DRO interval after the initial DRO interval effectively controls the problem be- havior (i.e., thinning the reinforcement schedule). The DRO schedule should be thinned by increasing the in- terval through a series of initially small and gradually in- creasing increments.

Poling and Ryan (1982) suggested three procedures for increasing the duration of the DRO interval.

1. Increase the DRO interval by a constant duration of time. For example, a practitioner could, at each opportunity for an interval increase, increase the DRO interval by 15 seconds.

2. Increase intervals proportionately. For example, a practitioner, at each opportunity to change the in- terval, could increase the DRO interval by 10%.

3. Change the DRO interval each session based on the learner’s performance. For example, a practi- tioner could set the DRO interval for each session equal to the mean IRT from the preceding session.

If problem behavior worsens when a larger DRO in- terval is introduced, the practitioner can decrease the du- ration of the interval to a level that again controls the problem behavior. The DRO interval can then be ex- tended with smaller, more gradual durations after previ- ously obtained reductions in problem behavior have been reestablished.

Extend the Application of DRO to Other Settings and Times of Day

When the frequency of the problem behavior is reduced substantially in the treatment setting, the DRO interven- tion can be introduced during other activities and times in the person’s natural environment. Having teachers, par- ents, or other caregivers begin to deliver reinforcement on the DRO schedule will help extend the effects. For ex- ample, after Jerry’s SIB was under good control during treatment sessions (see Figure 6), Cowdery and Col- leagues (1990) began to implement the DRO interven- tion at other times and places during the day. At first,

491

Differential Reinforcement

various staff members would praise Jerry and give him to- kens for refraining from scratching during single 30- minute DRO intervals during leisure activities, instructional sessions, or free time. Then they began added additional 30-minute DRO intervals until the DRO contingency was extended to all of Jerry’s waking hours. Jerry’s SIB decreased sufficiently to allow him to be dis- charged to his home for the first time in two years. Jerry’s parents were taught how to implement the DRO proce- dure at home.

Combine DRO with Other Procedures

DRO can be used as a single intervention. However, as with DRI/DRA, including DRO in a treatment package with other behavior-reduction procedures often yields more efficient and effective behavior change. De Zu- bicaray and Clair (1998) used a combination of DRO, DRI, and a restitution (punishment) procedure to reduce the physical and verbal abuse and physical aggression of a 46-year-old woman with mental retardation living in a residential facility for people with chronic psychiatric disorders. Further, the results of three experiments by Rolider and Van Houten (1984) showed that an inter- vention consisting of DRO plus reprimands was more ef- fective than DRO alone in decreasing a variety of problem behaviors by children (e.g., physical abuse by a 4-year- old girl toward her baby sister, thumb sucking by a 12- year-old boy, and bedtime tantrums).

DRO can also be added as a supplement to an inter- vention that has produced insufficient results. McCord, Iwata, Galensky, Ellingson, and Thomson (2001) added DRO to a stimulus fading intervention that was “having limited success” in decreasing the problem behaviors of Sarah, a 41-year-old woman with severe mental retarda- tion and visual impairment. Her problem behaviors in- cluded severe self-injurious behavior, property destruction, and aggression evoked by noise. Gradually increasing the noise volume by 2 dB when DRO was in effect, “a therapist delivered a preferred edible item (half a cheese puff) to Sarah following each 6-s interval in which problem behavior was not observed. If problem behavior occurred, the food was withheld and the DRO interval was reset” (p. 455). Each time Sarah completed three consecutive 1-minute sessions with no occurrences of the problem behavior, the researchers increased the DRO interval by 2 seconds. This schedule thinning con- tinued until Sarah could complete the session with no oc- currences of the problem behavior. The therapist then delivered the edible item only at the end of the session. The addition of DRO to the intervention produced an im- mediate decrease in Sarah’s problem behavior, which re- mained at or near zero in the treatment setting for more than 40 sessions. Probes conducted in noisy conditions

after treatment ended showed that the treatment effects had maintained and generalized to Sarah’s home.

Differential Reinforcement of Low Rates of Responding In laboratory research, Ferster and Skinner (1957) found that delivering a reinforcer following a response that had been preceded by increasingly longer intervals of time without a response reduced the overall rate of responding. When reinforcement is applied in this manner as an in- tervention to reduce the occurrences of a target behav- ior, the procedure is called differential reinforcement of low rates (DRL). Because reinforcement with a DRL schedule is delivered following an occurrence of the tar- get behavior, as contrasted with DRO in which rein- forcement is contingent on the complete absence of the behavior, behavior analysts use DRL to decrease the rate of a behavior that occurs too frequently, but not to elim- inate the behavior entirely. For example, a practitioner may identify a certain behavior as a problem behavior not because of the form of that behavior, but because the behavior occurs too often. For example, a student rais- ing his hand or coming to the teacher’s desk to ask for help during independent seatwork is a good thing, but becomes a problem behavior if it occurs excessively.

Some learners may emit a behavior less frequently when told to do the behavior less often, or when given rules stating an appropriate rate of response. When in- struction alone does not diminish occurrences of the prob- lem behavior, the practitioner may need to use a consequence-based intervention. DRL offers practitioners an intervention for diminishing problem behaviors that is stronger than instruction alone, but still falls short of more restrictive consequences (e.g., punishment). Deitz (1977) named and described three DRL procedures: full-session DRL, interval DRL, and spaced-responding DRL.

Full-Session DRL

In a full-session DRL procedure, reinforcement is deliv- ered at the end of an instructional or treatment session if during the entire session the target behavior occurred at a number equal to or below a predetermined criterion. If the number of responses exceeds the specified limit dur- ing the session, reinforcement is withheld. For example, a teacher applying a full-session DRL with a criterion limit of four disruptive acts per class period would provide reinforcement at the end of the class period contingent on four or fewer disruptive acts. Full-session DRL is an effective, efficient, and easy-to-apply intervention for problem behaviors in education and treatment settings.

492

Differential Reinforcement

R es

p o

n se

s p

er M

in u

Sessions

Figure 8 Rate of talk-outs during baseline and treatment phases for a student with devel- opmental disabilities. From “Decreasing Classroom Misbehavior Through the Use of DRL Schedules of Reinforcement” by S. M. Deitz and A. C. Repp, 1973, Journal of Applied Behavior Analysis, 6, p. 458. Copyright 1973 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Deitz and Repp (1973) demonstrated the efficacy and manageability of full-session DRL for diminishing class- room misbehavior. They decreased the talk-outs of an 11-year-old boy with developmental disabilities. During 10 days of baseline, the student averaged 5.7 talk-outs per 50-minute session. With the introduction of full-ses- sion DRL, the student was allowed 5 minutes for play at the end of the day contingent on 3 or fewer talk-outs dur- ing a 50-minute session. Talk-outs during the DRL con- dition decreased to an average of 0.93 talk-outs per 50 minutes. A return to baseline slightly increased the talk- outs to an average of 1.5 per session (see Figure 8)

Interval DRL

To apply an interval DRL schedule of reinforcement, the practitioner divides a total session into a series of equal intervals of time and provides reinforcement at the end of each interval in which the number of occurrences of the problem behavior during that interval is equal to or below a criterion limit. The practitioner removes the op- portunity for reinforcement and begins a new interval if the learner exceeds the criterion number of responses dur- ing the time interval. For example, if a DRL criterion limit is four problem behaviors per hour, the practitioner delivers a reinforcer at the end of each 15 minutes con- tingent on the occurrence of no more than one problem behavior. If the participant emits a second problem be- havior during the interval, the practitioner immediately begins a new 15-minute interval. Beginning a new inter- val postpones the opportunity for reinforcement.

7Because reinforcement in a spaced-responding DRL immediately follows an instance of the target behavior, it is the applied variation of DRL that most closely resembles the DRL schedules of reinforcement described by Ferster and Skinner (1957).

Deitz (1977) initially defined the interval DRL schedule of reinforcement using a criterion of one re- sponse or no responses per interval. However, Deitz and Repp (1983) reported programming an interval DRL schedule with a criterion above one per interval. “A re- sponse which occurs an average of 14 times per hour could be limited to three occurrences per 15-minute in- terval. The response limit of three could then be lowered to two responses per interval, and then to one response per interval” (p. 37).

Deitz and colleagues (1978) used an interval DRL schedule of reinforcement to reduce the disruptive be- haviors of a student with learning disabilities. The 7-year- old student had several difficult classroom misbehaviors (e.g., running, shoving, pushing, hitting, throwing ob- jects). The student received a sheet of paper ruled into 15 blocks, with each block representing a 2-minute in- terval. A star was placed in a block each time the student completed 2 minutes with 1 or no misbehaviors. Each star permitted the student to spend 1 minute on the play- ground with the teacher. If the student emitted 2 misbe- haviors during the interval, the teacher immediately started a new 2-minute interval.

Effective application of interval DRL (as well as the spaced-responding DRL procedure described in the fol- lowing section) requires continuous monitoring of prob- lem behavior, careful timing, and frequent reinforcement. Without the help of an assistant, many practitioners may have difficulty applying the interval DRL procedure in group settings. Interval and spaced-responding DRL pro- cedures are quite reasonable for one-on-one instruction or when competent assistance is available.

Spaced-Responding DRL

Using a spaced-responding DRL schedule of reinforce- ment, the practitioner delivers a reinforcer following an occurrence of a response that is separated by at least a minimum amount of time from the previous response.7

Interresponse time (IRT) is the technical term for the du- ration of time between two responses. IRT and rate of re- sponding are directly correlated: the longer the IRT, the lower the overall rate of responding; shorter IRTs corre- late with higher response rates. When reinforcement is contingent on increasingly longer IRTs, response rate will decrease.

Favell, McGimsey, and Jones (1980) used a spaced- responding DRL schedule of reinforcement and response prompts to decrease the rapid eating of four persons with

493

Differential Reinforcement

profound developmental disabilities. At the start of treat- ment, reinforcement was contingent on short indepen- dent pauses (IRTs) between bites of food, and then gradually required longer and longer pauses between bites. The researchers also manually prompted a separa- tion between bites of food and faded out those response prompts when a minimum of 5-second pauses occurred independently between approximately 75% of all bites. Finally, Favell and colleagues gradually thinned food re- inforcement and praise. The frequency of eating was de- creased from a pretreatment baseline average of 10 to 12 bites per 30 seconds to 3 to 4 bites per 30 seconds dur- ing the spaced-responding DRL condition.

Most behavior-reduction procedures that manipulate consequences have the potential to reduce behavior to zero occurrences. Spaced-responding DRL is unlikely to do that. For that reason, spaced-responding DRL becomes an important intervention for diminishing behavior: A spaced-responding DRL contingency tells learners that their behavior is acceptable, but they should do it less often. For example, a teacher could use a spaced- responding DRL schedule to diminish a student’s prob- lem behavior of asking too many questions. The questions occurred so frequently that they interfered with the class- room learning and teaching. To intervene, the teacher could respond to the student’s question if the student had not asked a question during a preceding minimum of 5 minutes. This spaced-responding DRL intervention could decrease, yet not eliminate, asking questions.

Singh, Dawson, and Manning (1981) used a spaced- responding DRL intervention to reduce the stereotypic responding (e.g., repetitive body movements, rocking) of three teenage girls with profound mental retardation. Dur- ing the first phase of spaced-responding DRL interven- tion, the therapist praised each girl whenever she emitted a stereotypic response at least 12 seconds after the last re- sponse. One of the experimenters timed the intervals and used a system of automated lights to signal the therapist when reinforcement was available. After the DRL 12- second IRT resulted in an abrupt decrease in stereotypic behavior by all three girls (see Figure 9), Singh and colleagues systematically increased the IRT criterion to 30, 60, and then 180 seconds. The spaced-responding DRL procedure not only produced substantial reductions in stereotypic responding in all three subjects but also had the concomitant effect of increasing appropriate be- havior (e.g., smiling, talking, playing).

Guidelines for Using DRL

Several factors influence the effectiveness of the three DRL schedules in diminishing problem behaviors. The following guidelines address those factors.

Recognize the Limitations of DRL

If a practitioner needs to reduce an inappropriate behav- ior quickly, DRL would not be the method of first choice. DRL is slow. Reducing the inappropriate behavior to ap- propriate levels may take more time than the practitioner can afford. Further, DRL would not be advisable for use with self-injurious, violent, or potentially dangerous be- haviors. Finally, from a practical standpoint, using DRL means that the practitioner must focus on the inappro- priate behavior. If a teacher, for example, is not cautious, he may inadvertently give too much attention to the in- appropriate behavior, inadvertently reinforcing it.

Choose the Most Appropriate DRL Procedure

Full-session, interval, and spaced-responding DRL schedules provide different levels of reinforcement for learners. Of the three DRL procedures, only spaced- responding DRL delivers reinforcement immediately fol- lowing the occurrence of specific response; a response must be emitted following a minimum IRT before rein- forcement. Practitioners use spaced-responding DRL to reduce the occurrences of a behavior while maintaining those behaviors at lower rates.

With full-session and interval DRL, a response does not need to occur for the participant to receive the rein- forcer. Practitioners may apply full-session or interval DRL when it is acceptable that the rate of the problem be- havior reaches zero or as an initial step toward the goal of eliminating the behavior.

Spaced-responding and interval DRL usually pro- duce reinforcement at a higher rate than full-session DRL does. Arranging frequent contact with the reinforcement contingency is especially appropriate, and most often necessary, for learners with severe problem behaviors.

Use Baseline Data to Guide the Selection of the Initial Response or IRT Limits

Practitioners can use the mean number of responses emit- ted during baseline sessions, or slightly lower than that average, as the initial full-session DRL criterion. For ex- ample, 8, 13, 10, 7, and 12 responses recorded per session over five baseline sessions produce a mean of 10 re- sponses. Therefore, 8 to 10 responses per session makes an appropriate initial full-session DRL criterion.

Interval DRL and spaced-responding DRL initial time criteria can be set at the baseline mean or slightly lower. For example, 1 response per 15 minutes makes an acceptable initial interval DRL criterion as calculated from a baseline mean of 4 responses per 60-minute ses- sion. With the same baseline data (i.e., 4 responses per 60- minute session), it appears reasonable to use 15 minutes

494

Differential Reinforcement

P er

ce n

t o

f In

te rv

al s

Sessions

Figure 9 Effects of spaced-responding DRL on stereotypic responding of three teenage girls with profound mental retardation. From “Effects of Spaced Responding DRL on the Stereotyped Behavior of Profoundly Retarded Persons” by N. N. Singh, M. J. Dawson, and P. Manning, 1981. Journal of Applied Behavior Analysis, 38, p. 524. Copyright 1981 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

as the initial IRT criterion for a spaced-responding DRL schedule. That is, a response will produce reinforcement only if it is separated from the prior response by a mini- mum of 15 minutes.

Gradually Thin the DRL Schedule

Practitioners should gradually thin the DRL schedule to achieve the desired final rate of responding. Practitioners commonly use three procedures for thinning the initial DRL time criterion.

1. With full-session DRL, the practitioner can set a new DRL criterion using the participant’s current DRL performances. Another option is to set a new DRL criterion at slightly less than the mean num-

ber of responses emitted during recent DRL sessions.

2. With interval DRL, the practitioner can gradually decrease the number of responses per interval if the current criterion is more than one response per interval; or, gradually increase the duration of the criterion interval if the current criterion is one re- sponse per interval.

3. With spaced-responding DRL, the practitioner can adjust the IRT criterion based on the mean IRT of recent sessions, or slightly less than that average. For example, Wright and Vollmer (2002) set the IRT for the DRL component of a treatment pack- age that successfully reduced the rapid eating of a 17-year-old girl at the mean IRT of the previous five sessions. The researchers did not exceed an

495

Differential Reinforcement

IRT of 15 seconds because it was not necessary to reduce the girl’s eating below a rate of four bites per minute.

Practitioners who successfully thin DRL schedules of reinforcement make only gradual, but systematic, changes in time and response criteria associated with full, inter- val, and space-responding DRL variations.

Two possible decision rules for thinning the DRL schedule are:

Rule 1: Practitioners may want to change the DRL criterion whenever the learner meets or exceeds the criterion during three consecutive sessions.

Rule 2: Practitioners may want to change the DRL criterion whenever the learner receives reinforce- ment for at least 90% of the opportunities during three consecutive sessions.

Provide Feedback to the Learner

The effectiveness of a DRL procedure can be enhanced by feedback to help the learner monitor her rate of respond- ing. Full-session, interval, and spaced-responding DRL procedures provide different levels of feedback for par- ticipants. The most accurate feedback comes with spaced- responding DRL because reinforcement follows each response that meets the IRT criterion. When a response occurs that does not meet the IRT criterion, reinforce- ment is withheld, the time interval is immediately reset,

and a new interval begins. This process provides learners with immediate feedback on responding per interval.

Interval DRL also provides a high level of feedback, although it is less than that of spaced-responding DRL. Interval DRL provides the learner with two types of feed- back. The first problem behavior does not provide feed- back. The second response, however, resets the time interval, providing a consequence for the problem behav- ior. Reinforcement occurs at the end of the interval when one or no problem behaviors occurred during the inter- val. These two types of feedback improve the effectiveness of the interval DRL intervention (Deitz et al., 1978).

Applied behavior analysts can arrange full-session DRL with or without feedback. The usual arrangement does not provide feedback concerning moment-to- moment accumulation of responses. Deitz (1977) stated that with full-session DRL, learners would respond only to the DRL criterion: If learners exceeded the criterion they lose the opportunity for reinforcement. Once the learner loses the opportunity for reinforcement, he could then emit high rates of misbehavior without consequence. When the schedule is arranged without moment-to- moment feedback, learners usually stay well below the DRL limit. Full-session DRL, without moment-to- moment feedback, may not be as effective for learners with severe problem behaviors as spaced-responding and interval DRL. The effectiveness of full-session DRL re- lies heavily on an initial verbal description of the con- tingencies of reinforcement (Deitz, 1977).

Summary Basic Description of Differential Reinforcement

1. When used as a reductive procedure for problem behav- ior, differential reinforcement consists of two components: (a) providing reinforcement contingent on either the oc- currence of a behavior other than the problem behavior or the problem behavior occurring at a reduced rate, and (b) withholding reinforcement as much as possible for the problem behavior.

Differential Reinforcement of Incompatible Behavior and Differential Reinforcement of Alternative Behavior

2. Differential reinforcement of an incompatible or alterna- tive behavior can be conceptualized as a schedule of rein- forcement in which two concurrent operants—the inappropriate behavior targeted for reduction and the ap- propriate behavior selected—receive reinforcement at dif- ferent rates.

3. DRI and DRA have the dual effect of weakening the prob- lem behavior while simultaneously strengthening accept- able behaviors that are either incompatible with or an alternative to the targeted problem behaviors.

4. With DRI, reinforcement is delivered for a behavior topographically incompatible with the target problem behavior and withheld following instances of the prob- lem behavior.

5. With DRA, reinforcement is delivered for a behavior that serves as a desirable alternative to the target prob- lem behavior and withheld following instances of the problem behavior.

6. When using DRI/DRA, practitioners should do the following:

• Select incompatible or alternative behaviors that are pre- sent in the learner’s repertoire, require equal or less ef- fort than the problem behavior, are being emitted prior to intervention with sufficient frequency to provide op- portunities for reinforcement, and are likely to produce reinforcement when the intervention ends.

• Select potent reinforcers that can be delivered when the alternative/incompatible behavior occurs and withheld following instances of the problem behavior. The same consequence that has been maintaining the problem

496

Differential Reinforcement

behavior prior to intervention is often the most effec- tive reinforcer for DRI/DRA.

• Reinforce the alternative/incompatible behavior on a continuous reinforcement schedule initially and then gradually thin the schedule of reinforcement.

• Maximize the difference between rates of reinforcement for the alternative/incompatible behavior and the prob- lem behavior by putting the problem behavior on an ex- tinction schedule.

• Combine DRI/DRA with other reductive procedures to produce a more potent intervention.

Differential Reinforcement of Other Behavior

7. With DRO, reinforcement is contingent on the absence of the problem behavior during or at specific times (i.e., mo- mentary DRO).

8. On an interval DRO schedule, reinforcement is delivered at the end of specific intervals of time if the problem did not occur during the interval.

9. On a momentary DRO schedule, reinforcement is deliv- ered at specific moments in time if the problem is not oc- curring at those times.

10. Availability of reinforcement with interval and momen- tary DRO procedures can occur on fixed- or variable- time schedules.

11. When using DRO, practitioners should do the following:

• Establish an initial DRO time interval that ensures that the learner’s current level of behavior will produce frequent reinforcement when the DRO contingency is applied.

• Be careful not to reinforce inadvertently other inappro- priate behaviors.

• Deliver reinforcement at the intervals or moments in time specified by the DRO schedule contingent on the absence of the problem behavior and the absence of any other significant inappropriate behaviors.

• Gradually increase the DRO interval based on decreases in the problem behavior.

• Extend DRO in other settings and times of day after the problem behavior is substantially reduced in the treat- ment setting.

• Combine DRO with other reductive procedures.

Differential Reinforcement of Low Rates of Responding

12. DRL schedules produce low, consistent rates of responding.

13. Reinforcement on a full-session DRL schedule is deliv- ered when responding during an entire instructional or treatment session is equal to or below a criterion limit.

14. On an interval DRL schedule, the total session is divided into equal intervals and reinforcement is provided at the end of each interval in which the number of responses dur- ing the interval is equal to or below a criterion limit.

15. Reinforcement on a spaced-responding DRL schedule follows each occurrence of the target behavior that is sep- arated from the previous response by a minimum inter- response time (IRT).

16. When using DRL, practitioners should do the following:

• Not use DRL if a problem behavior needs to be re- duced quickly.

• Not use DRL with self-injurious or other violent behaviors.

• Select the most appropriate DRL schedules: full-ses- sion or interval DRL when it is acceptable that the rate of the problem behavior reaches zero or as an initial step toward the goal of eliminating the behavior; spaced- responding DRL for reducing the rate of a behavior to be maintained in the learner’s repertoire.

• Use baseline data to guide the selection of the initial re- sponse or IRT limits.

• Gradually thin the DRL schedule to achieve the desired final rate of responding.

• Provide feedback to help the learner monitor the rate of responding.

497

antecedent intervention behavioral momentum fixed-time schedule (FT) functional communication

training (FCT)

high-probability (high-p) request sequence

noncontingent reinforcement (NCR) variable-time schedule (VT)

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 9: Behavior Change Procedures

9-1 Use antecedent-based interventions, such as: contextual or ecological variables, establishing operations, and discriminative stimuli.

9-5 Use response-independent (time-based) schedules of reinforcement.

9-23 Use behavioral momentum.

9-26 Use language acquisition/communication training procedures.

Antecedent Interventions

Key Terms

498

Applied behavior analysts traditionally em- phasized the three-term contingency: how con- sequences affect behavior, and how differential

consequences produce stimulus discrimination (SD) and stimulus control. Rarely did applied behavior analysts address how an antecedent event itself affected behav- ior. However, this situation changed following the pub- lication of two greatly influential articles: one on establishing operations (Michael, 1982) and the other on functional analysis (Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/1994). The convergence of motivat- ing operations (MOs) and functional behavior assess- ments allowed applied behavior analysts to conceptually align applied research on the effects of antecedent con- ditions other than stimulus control (e.g., SDs) to basic principles of behavior. This convergence advanced our understanding of how antecedents affect behavior, a his- torically underemphasized area of applied behavior analysis.

Practitioners have long used antecedents effectively to develop desirable behaviors, to diminish problem be- haviors, and to design environments that select adaptive behaviors in social, academic, leisure, and work en- vironments. Arguably, teachers may use antecedent interventions more frequently than consequence arrange- ments (e.g., reinforcement, punishment, extinction) to change behavior. The long-established practice of ar- ranging antecedent conditions for behavior change adds support to an increasing experimental database of desir- able behavioral outcomes that are functionally related to changing antecedent conditions (e.g., Wilder & Carr, 1998). Table 1 shows common behavior challenges

and the antecedent interventions that practitioners have used to address them.

Defining Antecedent Interventions Conceptual Understanding of Antecedent Interventions

Some textbooks and journal articles classify all antecedent- based behavior change strategies under a single term such as antecedent procedures, antecedent control, antecedent manipulations, or antecedent interventions. Although eco- nomical, using the same term to identify interventions based on stimulus control (SDs) and interventions involv- ing motivating operations (MOs) can lead to confusion about, or failure to recognize, the different functions of these antecedent events. SDs evoke behavior because they have been correlated with increased availability of rein- forcement. The evocative function of MOs, however, is in- dependent of the differential availability of effective reinforcement. MOs (e.g., an establishing operation) in- crease the current frequency of certain types of behavior even when an effective reinforcer is not available. Applied behavior analysts should be cognizant of the different fac- tors underlying the evocative functions of SDs and MOs.

In addition to improved conceptual clarity and con- sistency, understanding the different reasons for the evocative functions of SDs and MOs has important ap- plied implications. Antecedent treatments involving stim- ulus control must include manipulating consequent

Table 1 Examples of Behavior Challenges and Antecedent Interventions That Might Address Them

Challenging behavior situation Antecedent intervention

Disruptive or noncompliant behaviors associated with doing Provide a choice: “Do you want to do your homework first, or homework or taking a bath. take a bath first?”

Poor socialization and communication skills among persons Switch the seating arrangement to a family-style configuration. with developmental disabilities eating at a cafeteria-style- arranged dinner table.

A student misbehaves by disrupting other students. Move the misbehaving student’s desk closer to the teacher and away from the peers bothered by the student’s misbehavior.

Infant during the creeping and crawling stage can encounter Attach gates to stairways, plug electrical outlets, put locks on many potentially dangerous events. cabinet doors, remove table lamps.

Student causes disruptions when requested to complete a Present five sets of five problems each. math worksheet with 25 problems.

Some students misbehave when entering the resource room. The teacher pins an index card for each student on the bulletin Long transition times are clearly correlated with the mis- board. The card contained a personalized question. When behavior and lead to many difficult problem behaviors. entering the resource room, students take their cards from the

bulletin board, go to their desks, and write a response to the question on the card.

Antecedent Interventions

499

events, changing the differential availability of reinforce- ment in the presence and absence of the SD. Behavior change strategies based on motivating operations must change antecedent events. Understanding these differences may improve the development of more effective and effi- cient behavior change efforts involving antecedent events.

Classifying Functions of Antecedent Stimuli

Contingency Dependent

A contingency-dependent antecedent event is dependent on the consequences of behavior for developing evocative and abative effects. All stimulus control functions are contingency dependent. For example, in the presence of 2 + 2 = ?, a student responds 4, not because of the stim- ulus 2 + 2 = ?, but because of a past reinforcement his- tory for saying 4, including perhaps a history of non-reinforcement for making any other response except 4. This chapter uses the term antecedent control to iden- tify contingency-dependent antecedent events.

Contingency Independent

A contingency-independent antecedent event is not de- pendent on the consequences of behavior for developing evocative and abative effects. The antecedent event itself affects behavior–consequence relations. The effects of motivating operations are contingency independent. For example, sleep deprivation can influence the occurrences of problem behaviors in the absence of a history of pair-

ing sleep deprivation with reinforcement or punishment of those behaviors. This chapter uses the term antecedent intervention to identify behavior change tactics based on contingency-independent antecedent events.1

Antecedent Intervention

Abolishing Operations

Applied behavior analysts have used several antecedent interventions, singularly or in treatment packages, to de- crease the effectiveness of reinforcers that maintain prob- lem behaviors (i.e., abolishing operations). Table 2 provides examples of antecedent interventions that used abolishing operations to decrease the effectiveness of re- inforcers maintaining the problem behaviors and a cor- responding reduction of those behaviors.

Temporary Effects

Smith and Iwata (1997) reminded us that the effects of MOs are temporary. Antecedent interventions by them- selves will not produce permanent improvements in be- havior. However, while using an antecedent intervention to diminish a problem behavior, a teacher or therapist can simultaneously apply procedures such as extinction to re- duce the problem behavior and differential reinforcement of alternative behaviors to compete with the problem be- havior. Because of the temporary effects of MOs, an- tecedent interventions most often serve as one component

Table 2 Examples of Antecedent Interventions Using Abolishing Operations

Abolishing operation Example

Provide corrective prompts as an antecedent event. Antecedent corrective academic prompts reduced destructive behavior to zero (Ebanks & Fisher, 2003).

Provide presession exposure to stimuli that function as A father–son playtime preceding compliance sessions reinforcers. improved son’s compliance to father’s requests (Ducharme &

Rushford, 2001).

Provide free access to leisure activities. Manipulation of leisure items effectively competed with SIB maintained by automatic reinforcement (Lindberg et al., 2003).

Reduce noise levels. Reducing noise levels decreased stereotypical covering ears with hands (Tang et al., 2002).

Change levels of social proximity. Low levels of distant-proximity reduced aggressive behaviors (Oliver et al., 2001).

Offer choices. Escape-maintained problem behaviors were decreased when students had opportunities to choose among tasks (Romaniuk et al., 2002).

Increasing response effort. Increasing response effort for pica produced reductions in pica (Piazza et al., 2002).

Antecedent Interventions

1Because Chapter entitled “Stimulus Control” addressed contingency- dependent antecedent events (i.e., stimulus control), this chapter focuses on contingency-independent antecedent events (i.e., MOs).

500

2This chapter uses the phrase presenting stimuli with known reinforcing properties to describe the delivery of noncontinent reinforcers. Interpret- ing the NCR procedure as presenting a reinforcer, or presenting a rein- forcer noncontingent, on a fixed-time or variable-time schedule is technically inconsistent with the functional definition of reinforcement (Poling & Normand, 1999). Reinforcement requires a response–reinforcer relation. We use the term NCR in this chapter to describe the time-based procedures for reducing problem behaviors because the discipline of applied behavior analysis has continued its use, and the term NCR serves a useful descriptive purpose.

of a treatment package (e.g., combining an antecedent in- tervention with extinction, differential reinforcement of alternative behavior, or other procedures). These kinds of treatments may produce maintaining effects.

Three antecedent interventions with established experimental results are noncontingent reinforcement, high-probability request sequence, and functional com- munication training. The remainder of this chapter will elaborate on these three antecedent interventions and pre- sent definitions and guidelines for their effective use.

Noncontingent Reinforcement Noncontingent reinforcement (NCR) is an antecedent intervention in which stimuli with known reinforcing properties are delivered on a fixed-time (FT) or variable- time (VT) schedule independent of the learner’s behav- ior (Vollmer, Iwata, Zarcone, Smith, & Mazaleski, 1993).2

Noncontingent reinforcement may effectively diminish problem behaviors because the reinforcers that maintain

the problem behavior are available freely and frequently. This enriched environment with positive stimuli may function as an abolishing operation (AO), reducing the motivation to engage in the problem behavior.

NCR uses three distinct procedures that identify and deliver stimuli with known reinforcing properties: (a) positive reinforcement (i.e., social mediation), (b) neg- ative reinforcement (i.e., escape), and (c) automatic reinforcement (i.e., without social mediation). NCR pro- vides an important and effective intervention for prob- lem behaviors. It is a common treatment for persons with developmental disabilities.

NCR with Positive Reinforcement

Kahng, Iwata, Thompson, and Hanley (2000) provided an excellent example of applying NCR with positive re- inforcement. A functional analysis showed that social- positive reinforcement maintained self-injurious behavior (SIB) or aggression in three adults with developmental disabilities. During baseline, each occurrence of SIB or aggression produced attention for two of the adults, and a small bit of food for the third adult. During the initial sessions of NCR, the adults received attention or small bits of food on an initial fixed-time (FT) schedule (e.g., 5 seconds). Later, the schedule was thinned to a terminal criterion of 300 seconds. Figure 1 presents the base- line and NCR performances of the three adults and shows that the NCR procedure effectively decreased occurrences of SIB and aggression.

Baseline Noncontingent Reinforcement

Julia

Susan

Matt

10 20 30 40 50 Sessions

Extinction

R es

p o

n se

s p

er M

in u

Figure 1 Number of SIB or aggressive responses per minute during baseline and NCR by three adults with developmental disabilities. From “A Comparison of Procedures for Programming Noncontingent Reinforcement Schedules” by S. W. Kahng, B. A. Iwata, I. G. DeLeon, & M. D. Wallace 2000, Journal of Applied Behavior Analysis, 33, p. 426. Used by permission. Copyright 2000 by the Society for the Experimental Analysis of Behavior. Reproduced by permission.

Antecedent Interventions

501

NCR with Negative Reinforcement

Kodak, Miltenberger, and Romaniuk (2003) analyzed the effects of NCR escape on the instructional task com- pliance and problem behaviors of Andy and John, 4-year-old boys with autism. Andy’s task was to point to cards with teacher-specified pictures, words, or letters on them. John used a marker to trace each letter of writ- ten words. Problem behaviors included resisting prompts, throwing materials, and hitting. During base- line, the therapist gave an instruction for task engage- ment, and contingent on problem behavior following the instruction, removed the task materials and turned away from the child for 10 seconds. During the NCR escape condition, the therapist used an initial FT 10- second schedule for escape, meaning that the student had a break from the instructional requests every 10 sec- onds of the session. The initial 10-second FT schedule was thinned each time the boy achieved a criterion for two consecutive sessions: 10 seconds to 20 seconds, to 30 seconds, to 1 minute, to 1.5 minutes, and finally to a terminal criterion of 2 minutes. The NCR escape pro- cedure increased compliance and decreased problem behaviors.

NCR with Automatic Reinforcement

Lindberg, Iwata, Roscoe, Worsdell, and Hanley (2003) used NCR as a treatment to decrease the self-injurious behavior (SIB) of two women with profound mental

retardation. A functional analysis documented that au- tomatic reinforcement maintained their SIB. The NCR procedure provided Julie and Laura with free access to a variety of home-based, highly preferred leisure items (e.g., beads, string) that they could manipulate through- out the day. Figure 2 shows that NCR object ma- nipulation of preferred leisure items effectively diminished SIB, and the effects were maintained up to a year later. This experiment is important because it showed that NCR object manipulation could compete with automatic reinforcement to reduce the occurrence of SIB.

Using NCR Effectively

Enhancing Effectiveness

The following procedural recommendations identify three key elements for enhancing the effectiveness of NCR. (a) The amount and quality of stimuli with known reinforcing properties influence the effective- ness of NCR. (b) Most treatments include extinction with NCR interventions. (c) Reinforcer preferences can change during intervention. That is, the NCR stimuli may not continue competing with the reinforcers that maintain the problem behavior. DeLeon, Anders, Rodriguez-Catter, and Neider (2000) recommended periodically using a variety of available stimuli with the NCR intervention to reduce problems of changing preferences.

100

% o

f In

te rv

al s

Baseline Noncontingent Reinforcement

Julie

Laura

Object Manipulation

3 months

0SI B

( R

es p

o n

se s

p er

M in

u te

)

10 20 30 40 50 60 70 12 months

Sessions

100

O b

je ct

M an

ip u

la ti

o n

( %

I n

te rv

al s)

SIB

Figure 2 Levels of SIB and object manip- ulation exhibited by Julie and Laura during observations at home while NCR was imple- mented daily. From “Treatment Efficacy of Noncontingent Reinforcement during Brief and Extended Application” by J. S. Lindberg, B. A. Iwata, E. M. Roscoe, A. S. Worsdell, & G. P. Hanley, 2003, Journal of Applied Behavior Analysis, 36, p. 14. Copyright 2003 by the Society for Experimental Analysis of Behavior. Reproduced by permission.

Antecedent Interventions

502

Functional Behavior Assessment

The effectiveness of using NCR is dependent on the cor- rect identification of the positive, negative, or automatic reinforcers maintaining the problem behavior. Advances in functional behavior assessments have greatly improved the effectiveness of NCR by facilitating the identifica- tion of maintaining contingences of reinforcement (Iwata et al., 1982/1994).3

Emphasizing NCR

Applied behavior analysts can enhance the effective- ness of an NCR intervention by presenting a greater amount of stimuli with known reinforcing properties than the rate of reinforcement in the non-NCR condi- tion. For example, Ringdahl, Vollmer, Borrero, and Con- nell (2001) found that NCR was ineffective when the baseline condition and the NCR condition contained a similar amount of reinforcer delivery. NCR was effec- tive, however, when the NCR schedule was denser (i.e., continuous reinforcement) than the baseline schedule. Applied behavior analysts can use the rates of rein- forcement during baseline to establish an initial NCR schedule to ensure a discrepancy between baseline and NCR conditions.

Ringdahl and colleagues (2001) suggested three procedures for emphasizing reinforcement during the NCR intervention: (a) increase the delivery of stimuli with known reinforcing properties, (b) use an obviously different schedule of reinforcement at treatment onset (e.g., continuous reinforcement), and (c) combine dif- ferential reinforcement of other behavior (DRO) with the NCR treatment package. DRO will decrease the ad- ventitious reinforcement of the problem behavior from the time-based NCR schedule.

Time-Based NCR Schedules

Most applications of NCR use a fixed-time schedule (FT) for the delivery of stimuli with known reinforcing properties. The interval of time for the presentation of these stimuli remains the same from delivery to delivery. When applied behavior analysts program the NCR time interval to vary from delivery to delivery, it is called a variable-time schedule (VT). For example, an NCR VT schedule of 10 seconds means that, on the average, stim- uli with known reinforcing properties are presented every 10 seconds. This VT schedule could use time intervals

such as 5, 7, 10, 12 and 15 seconds, arranged to occur in random sequence. Even though most applications of the NCR procedures have used FT schedules, VT schedules can be effective also (Carr, Kellum, & Chong, 2001).

Setting the initial NCR time schedule is an impor- tant aspect of the NCR procedure. The initial schedule can have an have an impact on the effectiveness of the intervention (Kahng, Iwata, DeLeon, & Wallace, 2000). Applied behavior analysts consistently recommend an initial dense FT or VT schedule (e.g., Van Camp, Ler- man, Kelley, Contrucci, & Vorndran, 2000). The therapist can set a dense time value (e.g., 4 seconds) arbitrarily. Usually, however, it is more effective to set the initial time value based on the number of occurrences of the problem behavior, which will ensure frequent contact with the NCR stimuli.

The following procedure can be used to determine an initial NCR schedule: Divide the total duration of all baseline sessions by the total number of occurrences of the problem behavior recorded during baseline, and set the initial interval at or slightly below the quotient. For example, if the participant emitted 300 aggressive acts during 5 days of the baseline, and each baseline session was 10 minutes in duration, then 3,000 seconds divided by 300 responses produces a quotient of 10 seconds. Ac- cordingly, these baseline data suggest an initial FT in- terval of 7 to 10 seconds.

Thinning Time-Based Schedules

Applied behavior analysts use a dense FT or VT sched- ule to begin the NCR procedure. They thin the schedule by adding small time increments to the NCR interval. However, thinning a time-based schedule is best begun only after the initial NCR interval has produced a reduc- tion in the problem behavior.

Applied behavior analysts have used three proce- dures to thin NCR schedules: (a) constant time increase, (b) proportional time increase, and (c) session-to-session time increase or decrease (Hanley, Iwata, & Thompson, 2001; Van Camp et al., 2000).

Constant Time Increase. A therapist can increase the FT or VT schedule intervals by using a constant duration of time, and decrease the amount of time that the learner has access to the NCR stimuli by a con- stant amount of time. For example, a therapist can in- crease the schedule interval by 7 seconds at each opportunity, and each time decrease access to the stimuli by 3 seconds.

Proportional Time Increase. A therapist can in- crease the FT or VT schedule intervals proportionately,

Antecedent Interventions

3Chapter entitled “Functional Behavior Assessment” provides a detailed description of functional behavior assessment.

503

Table 3 Possible Advantages and Disadvantages of NCR

Advantages

NCR is easier to apply than other positive reductive techniques, which require monitoring student behavior for the contingent delivery of the reinforcer (Kahng, Iwata, DeLeon, & Wallace, 2000).

NCR helps create a positive learning environment, which is always desirable during treatment.

A package treatment that includes NCR with extinction may reduce extinction-induced response bursts (Van Camp et al., 2000).

Chance pairings of appropriate behavior and NCR delivery of stimuli with known reinforcing properties could strengthen and maintain those desirable behaviors (Roscoe, Iwata, & Goh, 1998).

Disadvantages

Free access to NCR stimuli may reduce motivation to engage in adaptive behavior.

Chance pairings of problem behavior and NCR delivery of stimuli with known reinforcing properties could strengthen the problem behavior (Van Camp et al., 2000).

NCR escape (i.e., negative reinforcement) can disrupt the instructional process.

meaning that each time, the schedule interval is increased by the same proportion of time. For example, each time interval is increased by 5% (e.g., 60 seconds = initial FT schedule; first schedule increase = 90 seconds [5% of 60 = 30]; second increase = 135 seconds [5% of 90 = 45]).

Session-to-Session Time Increase or Decrease. A therapist can use the learner’s performance to change the schedule interval on a session-to-session basis. For example, at the end of a session, the therapist estab- lishes a new NCR time interval for the next session by dividing the number of problem behaviors that occurred in that session by the duration of the session and using that quotient as the next session’s FT interval.

A therapist will decrease the interval if the problem behavior starts to worsen during schedule thinning. The duration of the NCR interval can be increased again after control of the problem behavior has been reestablished, but in more gradual increments.

Setting Terminal Criteria

Applied behavior analysts usually select an arbitrary ter- minal criterion for NCR schedule thinning. Kahng and colleagues (2000) reported that the terminal criterion of a 5-minute FT schedule has been used commonly in ap- plied research, and that it seems to be a practical and ef- fective criterion. In addition, they reported that research has not established an advantage for a terminal criterion of a 5-minute FT schedule over denser schedules (e.g., 3 minutes) or thinner schedules (e.g., 10 minutes).

Considerations for Using NCR

NCR makes an effective intervention. It has advantages in addition to effectiveness, and some disadvantages. Table 3 lists the advantages and disadvantages of using NCR.

High-Probability Request Sequence When using a high-probability (high-p) request se- quence, the teacher presents a series of easy-to-follow requests for which the participant has a history of com- pliance (i.e., high-p requests); when the learner complies with several such high-p requests in sequence, the teacher immediately gives the target request (i.e., low-p). The be- havioral effects of the high-probability request sequence suggests the abative effects of an abolishing operation (AO) by (a) reducing the value of reinforcement for non- compliance to the low-probability requests (i.e., reduc- ing the value of escape from requests), and (b) reducing the aggression and self-injury often associated with low- p requests.

To apply the high-p request sequence, the teacher or therapist selects two to five short tasks with which the learner has a history of compliance. These short tasks provide the responses for the high-p requests. The teacher or therapist presents the high-p request sequence imme- diately before requesting the target task, the low-p re- quest. Sprague and Horner (1990) used the following dressing anecdote to explain the high-p procedure:

Typical Instruction

Teacher: “Put on the shirt.” (low-p request)

Student: Avoids the hard task by throwing a tantrum

Typical Instruction with High-p Request Sequence

Teacher: “Give me five.” (high-p request)

Student: Slaps teacher’s outstretched hand

Teacher: “All right, nice job! Now, take this ball and put it in your pocket.” (high-p request)

Student: Puts the ball in pocket

Antecedent Interventions

504

Teacher: “Great! That is right! Now, put on the shirt.” (low-p request)

Student: Puts on shirt with assistance

Student compliance provides opportunities for the de- velopment of many important behaviors. Noncompliance, however, is a prevalent problem with persons with devel- opmental disabilities and behavior disorders. The high-p request sequence provides a nonaversive procedure for improving compliance by diminishing escape-maintained problem behaviors. The high-p request sequence may de- crease excessive slowness in responding to requests and in the time used for completing tasks (Mace et al., 1988).

Engelmann and Colvin (1983) provided one of the first formal descriptions of the high-p request sequence in their compliance training procedure for managing se- vere behavior problems. They used the term hard task to identify the procedure of giving three to five easy re- quests immediately before requesting compliance with a difficult task. Applied behavior analysts have used sev- eral labels to identify this intervention, including interspersed requests (Horner, Day, Sprague, O’Brien,

& Heathfield, 1991), pretask requests (Singer, Singer, & Horner, 1987), and behavioral momentum (Mace, & Belfiore, 1990). Currently, most applied behavior ana- lysts identify this antecedent intervention with the label high-p request sequence.

Killu, Sainato, Davis, Ospelt, and Paul (1998) eval- uated the effects of the high-p request sequence on com- pliant responding to low-p requests and occurrences of problem behaviors by three preschool children with de- velopmental delays. They used a compliance criterion of 80% or higher for selecting high-p requests for two chil- dren, and a 60% criterion for the third child. Compliance of less than 40% was used to select the low-p requests.

The request sequence began with the experimenter or a trainer presenting three to five high-p requests. When a child complied to at least three consecutive high-p re- quests, a low-p request was immediately presented. Praise was provided immediately following each compliant re- sponse. Figure 3 shows the children’s performances before, during, and after the high-p sequence. The sequence delivered by two different trainers increased compliant responding to the low-p requests of the three

10 9 8 7 6 5 4 3 2 1 0

Baseline High-Probability Request Sequence Maintenance Follow-Up

10 9 8 7 6 5 4 3 2 1 0

John

BarryExperimenter Secondary Trainer

10 9 8 7 6 5 4 3 2 1 0

N u

m b

er o

f R

es p

o n

se s

William

5 10 15 20 25 30 35 40 45 50 55 3 4 5 6 Sessions Weeks

Figure 3 Number of compliant responses to low-probability requests delivered by the investigator and second trainer across sessions and conditions. Participants were given 10 low-p requests each session. Dashed lines indicate student absences. From “Effects of High-Probability Request Sequences on Preschoolers’ Compliance and Disruptive Behavior” by K. Killu, D. M. Sainato, C. A. Davis, H. Ospelt, & J. N. Paul, 1998, Journal of Behavioral Education, 8, p. 358. Used by permission.

Antecedent Interventions

505

Figure 4 Considerations for using the high-p request sequence.

1. Do not use the high-p request sequence just after an occurrence of the problem behavior. The student might learn that responding to a low-p request with the problem behavior will produce a series of easier requests.

2. Present the high-p request sequence at the beginning and throughout the instructional period to reduce the possibility of problem behaviors producing reinforcement (Horner et al., 1991).

3. Teachers might knowingly or unknowingly let instruction drift from low-p requests to only high-p requests, and select easy tasks to avoid student escape-motivated behavior, a possible outcome from escape-motivated aggression and self-injury associated with low-p requests (Horner et al., 1991).

children. Compliant responding maintained across time and settings.

Using the High-p Request Sequence Effectively

Selecting from Current Repertoire

The tasks selected for the high-p request sequence should be in the learner’s current repertoire, occur with regular- ity of compliance, and have a very short duration of oc- currence. Ardoin, Martens, and Wolfe (1999) selected high-p requests by (a) creating a list of requests that cor- responded to student compliance, (b) presenting each request on the list for five separate sessions, and (c) se- lecting as high-p requests only those tasks that the student complied with 100% of the time.

Mace (1996) reported that the effectiveness of the high-p sequence increases, apparently, as the number of high-p requests increases. A high-p request sequence with five requests may be more effective than a sequence with two requests; but the increase in effectiveness may have a trade-off in efficiency. For example, if the same, or nearly the same, effectiveness can be obtained with two or three high-p requests as can be obtained with five or six, a teacher might select the shorter sequence because it is more efficient. When participants consistently com- ply with the low-p requests, the trainer should gradually reduce the number of high-p requests.

Presenting Requests Rapidly

The high-p requests should be presented in rapid suc- cession, with short interrequest intervals. The first low-p request should immediately follow the reinforcer for high-p compliance (Davis & Reichle, 1996).

Acknowledging Compliance

The learner’s compliance should be acknowledged im- mediately. Notice how the teacher in the previous dress- ing example acknowledged and praised the student’s

compliance before presenting the next request (“All right, nice job!”).

Using Potent Reinforcers

Individuals may emit aggression and self-injurious be- haviors to escape from the demands of the low-p requests. Mace and Belfiore (1990) cautioned that social praise may not increase compliance if motivation for escape be- havior is high. Therefore, high-quality positive stimuli immediately following compliance will increase the ef- fectiveness of the high-p intervention (Mace, 1996). Figure 4 lists consideration for the application of the high-p request sequence.

Functional Communication Training Functional communication training (FCT) establishes an appropriate communicative behavior to compete with problem behaviors evoked by an establishing operation (EO). Rather than changing EOs, functional communi- cation training develops alternative behaviors that are sensitive to the EOs. This is in contrast to the non- contingent reinforcement and high-probability request sequence interventions that alter the effects of EOs.

Functional communication training is an application of differential reinforcement of alternative behavior (DRA) because the intervention develops an alternative communicative response as an antecedent to diminish the problem behavior (Fisher, Kuhn, & Thompson, 1998). The alternative communicative response produces the re- inforcer that has maintained the problem behavior, mak- ing the communicative response functionally equivalent to the problem behavior (Durand & Carr, 1992). The al- ternative communicative responses can take many forms, such as vocalizations, signs, communication boards, word or picture cards, vocal output systems, or gestures (Brown et al., 2000; Shirley, Iwata, Kahng, Mazaleski, & Ler- man, 1997).

Antecedent Interventions

506

100

Baseline Functional Communication Training

Matt at Store

5 10 15 20 25 30 35 40

Challenging Behavior

Trained Response

100

Allison at Mall

5 10 15 20 25 30 35 40 100

Mike at Movies

5 10 15 20 25 30 35 40 100

Ron at Store

5 10 15 20 25 30 35 40 100

David at Library

5 10 15 20 25 30 35 40

Sessions

P er

ce n

ta ge

o f

In te

rv al

Figure 5 Percentage of intervals of challenging behavior by each of the five participants in baseline and FCT in community settings. The hatched bars show the percentage of intervals of unprompted communication by each student. From “Functional Communication Training Using Assistive Devices: Recruiting Natural Communities of Reinforcement” by V. M. Durand, 1999, Journal of Applied Behavior Analysis, 32, p. 260. Copyright 1999 by the Society for Experimental Analysis of Behavior. Reproduced by permission.

Carr and Durand (1985) defined functional com- munication training as a two-step process: (a) complet- ing a functional behavior assessment to identify the stimuli with known reinforcing properties that maintain the problem behavior, and (b) using those stimuli as re- inforcers to develop an alternative behavior to replace the problem behavior. FCT provides an effective treat- ment for many problem behaviors maintained by social attention.

FCT-based interventions typically involve several behavior change tactics in addition to teaching the al- ternative communicative response. For example, applied behavior analysts often use a combination of response prompting, time-out, physical restraint, response block- ing, redirection, and extinction with the problem behavior.

Durand (1999) used FCT in school and community settings to reduce the problem behaviors of five students with severe disabilities. Durand first completed functional behavior assessments to identify the objects and activities maintaining the problem behaviors. Following the func- tional assessments, the students learned to use a com- munication device that produced digitized speech with which they could request the objects and activities iden- tified during the functional behavior assessments. The five students reduced the occurrences of their problem behaviors in a school and in a community setting in which they used digitized speech to communicate with persons in the community. Figure 5 presents data on the per- centage of intervals of problem behaviors occurring in community settings for each student. These data are so- cially significant because they show the importance of teaching skills that recruit reinforcement in natural set- tings, thereby promoting generalization and maintenance of intervention effects.

Effective Use of FCT

Dense Schedule of Reinforcement

The alternative communicative response should produce the reinforcers that maintain the problem behavior on a continuous schedule of reinforcement during the early stages of communication training.

Decreased Use of Verbal Prompts

While teaching the alternative communicative response, verbal prompts such as look or watch me, are used often. After the communicative response is established firmly, the trainer should gradually reduce the verbal prompts and if possible eliminate them altogether to remove any prompt dependence associated with the intervention (Mil- tenberger, Fuqua, & Woods, 1998).

Behavior Reduction Procedures

The effectiveness of FCT is likely to be enhanced if the treatment package includes extinction for the problem behavior and other behavior-reduction procedures such as time-out (Shirley et al., 1997).

Schedule Thinning

Thinning the reinforcement schedule for a firmly estab- lished communicative response is an important part of the FCT treatment package. The time-based procedures described earlier for thinning NCR schedules—constant time increase, proportional time increase, and session- to-session time increase or decrease—are not appropriate for schedule thinning with the alternative communica- tive response. They are incompatible with the methods

Antecedent Interventions

507

Table 4 Possible Advantages and Disadvantages of Functional Communication Training

Advantages

Excellent chance for generalization and maintenance of the alternative communicative response because the communicative response often functions to recruit reinforcement from significant others (Fisher et al., 1998).

May have high social validity. Participants report preferences for FCT over other procedures to diminish behavior (Hanley, Piazza, Fisher, Contrucci, & Maglieri, 1997).

When the alternative communicative behavior and the problem behavior have the same schedule of reinforcement (e.g., FR 1), an FCT intervention may be effective without using extinction (Worsdell, Iwata, Hanley, Thompson, & Kahng, 2000).

Disadvantages

The FCT treatment package usually includes extinction. May produce undesirable effects.

The extinction procedure is very difficult to use consistently, allowing for intermittent reinforcement of problem behaviors.

Participants may emit inappropriately high rates of the alternative communicative response to recruit reinforcement (Fisher et al., 1998).

Recruitment of reinforcement can occur at inconvenient or impossible times for the caregiver (Fisher et al., 1998).

The fact that FCT leaves intact the environment that evoked the problem behavior, may limit its overall effectiveness (McGill, 1999).

used to differentially reinforce the alternative commu- nicative behavior because the FCT intervention does not alter the EO that evokes the problem behavior. The al- ternative communicative behavior must remain sensitive to the evocative function of the EO to compete with the problem behavior. For example, consider a child with de- velopmental disabilities who has a history of engaging in self-stimulatory behavior when presented with diffi- cult tasks. A therapist teaches the child to ask for assis- tance with the difficult tasks (i.e., the alternative communicative behavior), which reduces the self-stimu- latory behavior. After firmly establishing asking for as- sistance, therapists and caregivers would not want to decrease giving assistance with the tasks when asked (thinning the reinforcement schedule). A decrease in giv- ing assistance has the potential to produce the recovery of the self-stimulatory behavior by breaking the alterna- tive communicative behavior-reinforcer contingency.

Hanley and colleagues (2001) recommended a pro- cedure for schedule thinning that used a dense fixed- interval schedule of reinforcement (e.g., FI 2 seconds, FI 3 seconds) during the initial teaching of the alternative commutation response. Once the communicative re- sponse was established, they suggested gradually thin- ning the FI schedule. This procedure, in contrast to time-based procedures, maintained the contingency be- tween responding and reinforcement. They cautioned that thinning the FI schedule during FCT interventions could produce undesirable high rates of the alternative com- municative response that could disrupt the home or class- room settings. Hanley and colleagues further suggested the use of picture cues and external “clocks” to announce when reinforcement is available as a possible way to con- trol the undesirably high rate of the communicative re- sponse. Table 4 summarizes the advantages and disadvantages of using FCT.

Summary Defining Antecedent Interventions

1. The term antecedent refers to the temporal relation of stimuli or events coming before an occurrence of behavior.

2. The convergence of research on motivating operations (MOs) and functional behavior assessments has allowed applied behavior analysts to conceptually align applied research on the effects of antecedent conditions other than stimulus control (e.g., SDs) to basic principles of behavior.

3. Functions of antecedent stimuli or events can be classi- fied as contingency dependent (stimulus control) or con- tingency independent (MO).

4. Conceptually, this chapter uses the term antecedent inter- vention to identify behavior-change tactics based on contingency-independent antecedent stimuli.

5. Applied behavior analysts have used several antecedent interventions, singularly or in treatment packages, to de- crease the effectiveness of reinforcers that maintain prob- lem behaviors (i.e., abolishing operations).

Antecedent Interventions

508

Antecedent Intervention

6. Because of the temporary effects of MOs, antecedent in- terventions most often serve as one component of a multi- component treatment package (i.e., in which antecedent interventions are paired with extinction, differential rein- forcement, or other procedures).

Noncontingent Reinforcement

7. Noncontingent reinforcement (NCR) consists of present- ing of stimuli with known reinforcing properties on a fixed-time (FT) or variable-time (VT) schedule indepen- dent of the participant’s behavior.

8. NCR uses three distinct procedures that identify and de- liver stimuli with known reinforcing properties: (a) posi- tive reinforcement (i.e., social mediation), (b) negative reinforcement (i.e., escape), and (c) automatic reinforce- ment (i.e., without social mediation).

9. An NCR-enriched environment may function as an abol- ishing operation (AO), reducing the motivation to engage in the problem behavior.

High-Probability Request Sequence

10. The behavioral effects of the high-probability (high-p) re- quest sequence suggests the abative effects of an AO by (a) reducing the potency of reinforcement for noncompli- ance to the low-probability requests (i.e., reducing the value of escape from requests), and (b) reducing the aggression and self-injury often associated with low-p requests.

11. To apply the high-p request sequence, the teacher or ther- apist selects two to five short tasks with which the learner has a history of compliance. These short tasks provide the responses for the high-p requests. The teacher or therapist presents the high-p request sequence immediately before requesting the target task, the low-p request.

12. Applied behavior analysts have used several labels to iden- tify this intervention, including interspersed requests, pre- task requests, and behavioral momentum.

Functional Communication Training

13. Functional communication training (FCT) is a form of dif- ferential reinforcement of alternative behavior (DRA) because the intervention develops an alternative communi- cative response as an antecedent to diminish the problem behavior.

14. Functional communication training is an antecedent in- tervention package for establishing an appropriate com- municative behavior to compete with problem behaviors evoked by an establishing operation (EO). Rather than al- tering the value of EOs as does NCR and the high-p request sequence, functional communication training develops alternative behaviors that are sensitive to the EOs that maintain the problem behavior.

15. The alternative communicative response produces the re- inforcer that has maintained the problem behavior, mak- ing the communicative response functionally equivalent to the problem behavior.

Antecedent Interventions

509

Functional Behavior Assessment

Key Terms

conditional probability contingency reversal descriptive functional behavior

assessment

functional analysis functional behavior assessment

(FBA) functionally equivalent

indirect functional assessment

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 4: Behavioral Assessment

4-1 State the primary characteristics of and rationale for conducting a descriptive assessment.

4-2 Gather descriptive data.

(a) Select various methods.

(b) Use various methods.

4-3 Organize and interpret descriptive data.

(a) Select various methods.

(b) Use various methods.

4-4 State the primary characteristics of and rationale for conducting a functional analysis as a form of assessment.

4-5 Conduct functional analyses.

(a) Select various methods.

(b) Use various methods.

4-6 Organize and interpret functional analysis data.

(a) Select various methods.

(b) Use various methods.

This chapter was written by Nancy A. Neef and Stephanie M. Peterson. From Chapter 24 of Applied Behavior Analysis, Second Edition. John O. Cooper, Timothy E. Heron, William L. Heward. Copyright © 2007 by Pearson Education, Inc. All rights reserved.

510

Functional Behavior Assessment

When it is time to wash hands before lunch, one child turns the handles of the faucet and places her hands under the running water, but another

child screams and tantrums. Why? Consistent with the scientific precept of determinism those behaviors are lawfully related to other events in the environment. Functional behavior assessment (FBA) enables hy- potheses about the relations among specific types of en- vironmental events and behaviors. Specifically, FBA is designed to obtain information about the purposes (func- tions) a behavior serves for a person. This chapter describes the basis for FBA, its role in the intervention and prevention of behavior difficulties, and alternative approaches to functional assessment.

Functions of Behavior Evidence from decades of research indicates that both desirable and undesirable behaviors, whether washing hands or screaming and tantrumming, are learned and maintained through interaction with the social and phys- ical environment. These behavior–environment interac- tions are described as positive or negative reinforcement contingencies. Behaviors can be strengthened by either “getting something” or “getting out of something.”

FBA is used to identify the type and source of rein- forcement for challenging behaviors as the basis for in- tervention efforts designed to decrease the occurrence of those behaviors. FBA can be thought of as a reinforcer as- sessment of sorts. It identifies the reinforcers currently maintaining problem behavior. Those reinforcers might be positive or negative social reinforcers provided by someone who interacts with the person, or automatic re- inforcers produced directly by the behavior itself. The idea behind FBA is that if these reinforcement contin- gencies can be identified and interventions can be de- signed to decrease problem behavior and increase adaptive behavior by altering these contingencies. FBA fosters proactive, positive interventions for problem be- havior. Although reinforcement contingencies are dis- cussed in other chapters, a brief review of their role in FBA is warranted.

Positive Reinforcement

Social Positive Reinforcement (Attention)

Problem behavior often results in immediate attention from others, such as head turns; surprised facial expressions; reprimands; attempts to soothe, counsel, or distract; and so on. These reactions can serve to positively reinforce problem behavior (even if inadvertently), and the problem behavior is then more likely to occur in similar circum-

stances. Problem behavior maintained by positive rein- forcement in the form of reactions from others can often occur in situations in which attention is otherwise infre- quent, whether because the person does not have a reper- toire to gain attention in desirable ways, or because others in the environment are typically otherwise occupied.

Tangible Reinforcement

Many behaviors result in access to reinforcing materials or other stimuli. Just as pressing a button on the television remote changes the channel to a desired television show, problem behaviors can produce reinforcing outcomes. A child may cry and tantrum until a favorite television show is turned on; stealing another child’s candy produces ac- cess to the item taken. Problem behaviors may develop when they consistently produce a desired item or event. This often occurs because providing the item temporar- ily stops the problem behavior (e.g., tantrum), although it can have the inadvertent effect of making the problem behavior more probable in the future under similar circumstances.

Automatic Positive Reinforcement

Some behaviors do not depend on the action of others to provide an outcome; some behaviors directly produce their own reinforcement. For example, thumb sucking might be reinforced by physical stimulation of either the hand or the mouth. A behavior is assumed to be main- tained by automatic reinforcement only after social rein- forcers have been ruled out (e.g., when the behavior occurs even when the individual is alone).

Negative Reinforcement

Social Negative Reinforcement (Escape)

Many behaviors are learned as a result of their effective- ness in terminating or postponing aversive events. Hang- ing up the phone terminates interactions with a telemarketer; completing a task or chore terminates re- quests from others to complete it or the demands associ- ated with the task itself. Problem behaviors can be maintained in the same way. Behaviors such as aggres- sion, self-injurious behavior (SIB), and bizarre speech may terminate or avoid unwanted interactions with oth- ers. For example, noncompliance postpones engagement in a nonpreferred activity, and disruptive classroom be- havior often results in the student being sent out of the classroom, thereby allowing escape from instructional tasks or teacher demands. All of these behaviors can be strengthened by negative reinforcement to the extent that

511

Functional Behavior Assessment

they serve to allow the individual to escape or avoid dif- ficult or unpleasant tasks, activities, or interactions.

Automatic Negative Reinforcement

Aversive stimulation, such as a physically painful or un- comfortable condition, is a motivating operation that makes its termination reinforcing. Behaviors that directly terminate aversive stimulation are therefore maintained by negative reinforcement that is an automatic outcome of the response. Automatic negative reinforcement can account for behaviors that are either appropriate or harm- ful. For example, putting calamine lotion on a poison ivy rash can be negatively reinforced by alleviation of itching, but intense or prolonged scratching that breaks the skin can be negatively reinforced in the same manner. Some forms of SIB may serve to distract from other sources of pain, which may account for their correlation with specific medical conditions (e.g., DeLissovoy, 1963).

Function versus Topography

Several points can be made from the previous discussion of the sources of reinforcement for behavior. It is impor- tant to recognize that environmental influences do not make distinctions between desirable and undesirable topographies of behavior; the same reinforcement con- tingencies that account for desirable behavior can also account for undesirable behavior. For example, the child who washes and dries her hands before lunch has prob- ably received praise for doing so. A child who frequently engages in tantrums in the same situation may have re- ceived attention (in the form of reprimands). Both forms of attention have the potential to reinforce the respective behaviors.

Likewise, the same topography of behavior can serve different functions for different individuals. For exam- ple, tantrums may be maintained by positive reinforce- ment in the form of attention for one child, and by negative reinforcement in the form of escape for another child (e.g., Kennedy, Meyer, Knowles, & Shukla, 2000).

Because different behaviors that look quite different can serve the same function, and behavior of the same form can serve different functions under different condi- tions, the topography, or form, of a behavior often re- veals little useful information about the conditions that account for it. Identifying the conditions that account for a behavior (its function), on the other hand, suggests what conditions need to be altered to change the behavior. As- sessment of the function of a behavior can therefore yield useful information with respect to intervention strategies that are likely to be effective.

Role of Functional Behavior Assessment in Intervention and Prevention FBA and Intervention

If the cause-and-effect relation between environmental events and a behavior can be determined, that relation can be altered, thereby diminishing subsequent occur- rences of a problem behavior. FBA interventions can con- sist of at least three strategic approaches: altering antecedent variables, altering consequent variables, and teaching alternative behaviors.

Altering Antecedent Variables

FBA can identify antecedents that might be altered so the problem behavior is less likely to occur. Altering the an- tecedents for problem behavior can change and/or elim- inate either (a) the motivating operation for problem behavior or (b) the discriminative stimuli that trigger problem behavior. For example, the motivating operation for tantrums when a child is asked to wash her hands be- fore lunch could be modified by changing the character- istics associated with lunch so that the avoidance of particular events is no longer reinforcing (e.g., initially re- ducing table-setting demands, altering seating arrange- ments to minimize taunts from a sibling or peer, reducing snacks before lunch and offering more preferred foods during lunch). Alternatively, if the FBA shows that run- ning water is the discriminative stimulus that triggers problem behavior when a child is asked to wash her hands, the child might be given waterless antibacterial hand gel instead. In this case, the discriminative stimulus for problem behavior has been removed, thereby de- creasing problem behavior.

Altering Consequence Variables

FBA can also identify a source of reinforcement to be eliminated for the problem behavior. For example, an FBA that indicates tantrums are maintained by social neg- ative reinforcement (avoidance or escape) suggests a va- riety of treatment options, which, by altering that relation, are likely to be effective, such as the following:

1. The problem behavior can be placed on extinction by ensuring that the reinforcer (e.g., avoidance of lunch) is no longer delivered following problem behavior (tantrums).

2. The schedule might be modified so that hand washing follows (and thereby provides escape from) an event that is less preferred.

512

Functional Behavior Assessment

Teaching Alternative Behaviors

FBA can also identify the source of reinforcement to be provided for appropriate replacement behaviors. Alter- native appropriate behaviors that serve the same function (i.e., produce the same reinforcer) as tantrums could be taught. For example, the student might be taught to touch a card communicating “later” after washing his hands to produce a delay in being seated at the lunch table.

FBA and Default Technologies

Interventions based on an FBA are more likely to be ef- fective than those selected arbitrarily (e.g., Ervin et al., 2001; Iwata et al., 1994b). Understanding why a behav- ior occurs (its function) often suggests how it can be changed for the better. On the other hand, premature ef- forts to treat problem behavior before seeking an under- standing of the purposes it serves for a person can be inefficient, ineffective, and even harmful. For example, suppose that a time-out procedure is implemented in an attempt to attenuate the problem of the child who con- sistently tantrums when she is asked to wash her hands before lunch. The child is removed from the hand-wash- ing activity to a chair in the corner of the room. It may be, however, that the events that typically follow washing hands (those associated with lunch time, such as the de- mands of setting the table or interactions with others) are aversive for the child; tantrums have effectively served to allow the child to avoid those events. In this case, the intervention would be ineffective because it has done nothing to alter the relation between tantrums and the consequence of postponing the aversive events associ- ated with lunch. In fact, the intervention may exacerbate the problem if it produces a desired outcome for the child. If stopping the hand-washing activity and having the child sit on a chair as “time-out” for tantrumming enables the child to avoid the aversive lunchtime events—or to es- cape them altogether—tantrums may be more likely under similar circumstances in the future. When the time- out intervention proves unsuccessful, other interventions might be attempted. Without understanding the function that the problem behavior serves, however, the effective- ness of those interventions cannot be predicted.

At best, a trial-and-error process of evaluating arbi- trarily selected interventions can be lengthy and ineffi- cient. At worst, such an approach may cause the problem behavior to become more frequent or severe. As a result, caregivers might resort to increasingly intrusive, coer- cive, or punishment-based interventions, which are often referred to as default technologies.

FBA can decrease reliance on default technologies and contribute to more effective interventions in several ways. When FBAs are conducted, reinforcement-based

interventions are more likely to be implemented than are interventions that include a punishment component (Pe- lios, Morren, Tesch, & Axelrod, 1999). In addition, the ef- fects of interventions based on FBAs are likely to be more durable than those that do not take the function of prob- lem behavior into account. If contrived contingencies are superimposed on unknown contingencies that are main- taining the behavior, their continuation is often neces- sary to maintain improvements in the behavior. If those superimposed contingencies are discontinued, the be- havior will continue to be influenced by the unchanged operative contingencies.

FBA and Prevention

By furthering understanding of the conditions under which certain behaviors occur, FBA can also contribute to the prevention of difficulties. Although problem behavior may be suppressed without regard to its function by using punishment procedures, additional behaviors not subject to the punishment contingencies may emerge because the motivating operations for problem behavior remain. For example, contingent loss of privileges might eliminate tantrums that occur whenever a child is asked to wash her hands, but it will not eliminate avoidance as a reinforcer or the conditions that establish it as a reinforcer. Thus, other behaviors that result in avoidance may develop, such as aggression, property destruction, or running away. These unintended effects are less likely to occur with in- terventions that address (rather than override or compete with) the reinforcing functions of problem behavior.

On a broader scale, the accumulation of FBA data may further assist in prevention efforts by identifying the conditions that pose risks for the future development of problem behaviors. Preventive efforts can then focus on those conditions. For example, based on data from 152 analyses of the reinforcing functions of self-injurious be- havior (SIB), Iwata and colleagues (1994b) found that escape from task demands or other aversive stimuli ac- counted for the behavior in the largest proportion of cases. The authors speculated that this outcome might have been an unintended result of a move toward providing more aggressive treatment. For example, if a child tantrums when she is required to wash her hands, the teacher might assume that she does not know how to wash her hands. The teacher might decide to replace playtime with a pe- riod of intensive instruction on hygiene. Rather than de- creasing problem behavior, such interventions may exacerbate it. The data reported by Iwata and colleagues (1994b) suggest that preventive efforts should be directed toward modifying instructional environments (such as providing more frequent reinforcement for desirable be- havior, opportunities for breaks, or means to request and

513

Functional Behavior Assessment

obtain help with difficult tasks) so that they are less likely to serve as sources of aversive stimulation (motivating operations) for escape.

Overview of FBA Methods FBA methods can be classified into three types: (a) func- tional (experimental) analysis, (b) descriptive assess- ment, and (c) indirect assessment. The methods can be ordered on a continuum with respect to considerations such as ease of use and the type and precision of infor- mation they yield. Selecting the method or combination of methods that will best suit a particular situation re- quires consideration of each method’s advantages and limitations. We discuss functional analysis first because descriptive and indirect methods of functional assess- ment developed as an outgrowth of functional analysis. As noted later, functional analysis is the only FBA method that allows practitioners to confirm hypotheses regarding functional relations between problem behavior and environmental events.

Functional (Experimental) Analysis

Basic Procedure

In a functional analysis, antecedents and consequences representing those in the person’s natural environment are arranged so that their separate effects on problem behav- ior can be observed and measured. This type of assess- ment is often referred to as an analog because antecedents and consequences similar to those occurring in the nat- ural routines are presented in a systematic manner, but the analysis is not conducted in the context of naturally oc-

curring routines. Analog conditions are often used because they allow the behavior analyst to better control the envi- ronmental variables that may be related to the problem behavior than can be accomplished in naturally occurring situations. Analogs refer to the arrangement of variables rather than the setting in which assessment occurs. Re- search has found that functional analyses conducted in natural environments (e.g., classroom settings) often yield the same (and, in some cases, clearer) results compared to those conducted in simulated settings (Noell, Van- Dertteyden, Gatti, & Whitmarsh, 2001).

Functional analyses typically are comprised of four conditions: three test conditions—contingent attention, contingent escape, and alone—and a control condition, in which problem behavior is expected to be low because re- inforcement is freely available and no demands are placed on the individual (see Table 1). Each test condition contains a motivating operation (MO) and a potential source of reinforcement for problem behavior. The con- ditions are presented systematically one at a time and in an alternating sequence to identify which conditions pre- dictably result in problem behavior. Occurrences of prob- lem behavior are recorded during each session. Sessions are repeated to determine the extent to which problem behavior consistently occurs more often under one or more conditions relative to another.

Interpreting Functional Analyses

The function that problem behavior serves for a person can be determined by visually inspecting a graph of the results of the analysis to identify the condition(s) under which high rates of the behavior occurred. A graph for each potential behavioral function is shown in Figure 1. Problem behavior is expected to be low in the play condition because no motivating operations for problem

Table 1 Motivating Operations and Reinforcement Contingencies for Typical Control and Test Conditions of a Functional Analysis

Condition Antecedent conditions (motivating operation) Consequences for problem behavior

Play Preferred activities continuously available, social Problem behavior is ignored or neutrally redirected. (control) attention provided, and no demands are placed on the

person.

Contingent Attention is diverted or withheld from the person. Attention in the form of mild reprimands or soothing attention statements (e.g., “Don’t do that.You’ll hurt someone.”).

Contingent Task demands are delivered continuously using a three- Break from the task provided by removing task materials escape step prompting procedure (e.g., [1] “You need to fold the and stopping prompts to complete the task.

towel.” [2] Model folding the towel. [3] Provide hand- over-hand assistance to fold the towel.)

Alone Low level of environmental stimulation (i.e., therapist, Problem behavior is ignored or neutrally redirected. task materials, and play materials are absent).

514

Functional Behavior Assessment

N u

m b

er o

f P

ro b

le m

B eh

av io

rs p

er M

in u

2 4 6 8 10 12

Contingent Attention

Alone

Contingent Escape

Free Play

Sessions

Positive Reinforcement (Attention) Function

N u

m b

er o

f P

ro b

le m

B eh

av io

rs p

er M

in u

2 4 6 8 10 12

Sessions

Negative Reinforcement (Escape) Function

Contingent Escape

AloneContingent Attention Free Play

N u

m b

er o

f P

ro b

le m

B eh

av io

rs p

er M

in u

2 4 6 8 10 12

Sessions

Automatic Function

N u

m b

er o

f P

ro b

le m

B eh

av io

rs p

er M

in u

2 4 6 8 10 12

Sessions

Undifferentiated Pattern (Inconclusive Results,

Possible Automatic Function)

Alone

Free Play

Contingent Attention

Contingent Escape

Free Play

AloneContingent Attention

Figure 1 Data patterns typical of each behavioral function during a functional analysis.

behavior are present. Elevated problem behavior in the contingent attention condition suggests that problem be- havior is maintained by social positive reinforcement (see graph in top left of Figure 1). Elevated problem behavior in the contingent escape condition suggests that problem behavior is maintained by negative rein- forcement (graph in top right of Figure 1). Elevated problem behavior in the alone condition suggests that problem behavior is maintained by automatic rein- forcement (graph in bottom left of Figure 1). Further analysis is needed to determine if the source of the au- tomatic reinforcement is positive or negative. Problem behavior may be maintained by multiple sources of re- inforcement. For example, if problem behavior is ele- vated in the contingent attention and contingent escape conditions, it is most likely maintained by both positive and negative reinforcement.

If problem behavior occurs frequently in all condi- tions (including the play condition), or is variable across conditions, responding is considered undifferentiated (see graph in bottom right of Figure 1). Such results are

inconclusive, but can also occur with behavior that is maintained by automatic reinforcement.

Functional analysis has been replicated and extended in hundreds of studies, thereby demonstrating its gener- ality as an approach to the assessment and treatment of a wide range of behavior difficulties. (See the 1994 spe- cial issue of the Journal of Applied Behavior Analysis for a sample of such applications.)

Advantages of Functional Analysis

The primary advantage of functional analysis is its abil- ity to yield a clear demonstration of the variable(s) that relate to the occurrence of a problem behavior. In fact, functional (experimental) analyses serve as the standard of scientific evidence by which other assessment alter- natives are evaluated, and represent the method most often used in research on the assessment and treatment of problem behavior (Arndorfer & Miltenberger, 1993). Be- cause functional analyses allow valid conclusions con- cerning the variables that maintain problem behavior,

515

Functional Behavior Assessment

they have enabled the development of effective reinforcement-based treatments and less reliance on pun- ishment procedures (Ervin et al., 2001; Iwata et al., 1994b; Pelios et al., 1999).

Limitations of Functional Analysis

A risk of functional analysis, however, is that the assess- ment process may temporarily strengthen or increase the undesirable behavior to unacceptable levels, or possibly result in the behavior acquiring new functions. Second, although little is known about the acceptability of func- tional analysis procedures to practitioners (Ervin et al., 2001), the deliberate arrangement of conditions that set the occasion for, or potentially reinforce, problem be- havior can be counterintuitive to persons who do not un- derstand its purpose (or that such conditions are analogs for what occurs in the natural routine). Third, some be- haviors (e.g., those that, albeit serious, occur infrequently) may not be amenable to functional analyses. Fourth, func- tional analyses that are conducted in contrived settings might not detect the variable that accounts for the occur- rence of the problem behavior in the natural environment, particularly if the behavior is controlled by idiosyncratic variables that are not represented in the functional analy- sis conditions (e.g., Noell et al., 2001). Finally, the time, effort, and professional expertise required to conduct and interpret functional analyses have been frequently cited as obstacles to its widespread use in practice (e.g., Spreat & Connelly, 1996; 2001, Volume 2 issue of School Psy- chology Review).

Of course, untreated or ineffectively treated problem behaviors also consume a great deal of time and effort (without a constructive long-term outcome), and imple- mentation of effective treatments (based on an under- standing of the variables that maintain them) is likely to require skills similar to those involved in conducting functional analyses. These concerns have led to research on ways to enhance the practical use of functional analy- ses, such as methods of practitioner training (Iwata et al., 2000; Pindiprolu, Peterson, Rule, & Lignugaris/Kraft, 2003), abbreviated assessments (Northup et al., 1991), and the development of alternative methods of FBA de- scribed in the following sections.

Descriptive Functional Behavior Assessment

As with functional analyses, descriptive functional be- havior assessment encompasses direct observation of behavior; unlike functional analyses, however, observa- tions are made under naturally occurring conditions. Thus, descriptive assessments involve observation of the

problem behavior in relation to events that are not arranged in a systematic manner. Descriptive assess- ments have roots in the early stages of applied behavior analysis; Bijou, Peterson, and Ault (1968) initially de- scribed a method for objectively defining, observing, and coding behavior and contiguous environmental events. This method has been used subsequently to iden- tify events that may be correlated with the target behav- ior. Events that are shown to have a high degree of correlation with the target behavior may suggest hy- potheses about behavioral function. We describe three variations of descriptive analysis: ABC (antecedent- behavior-consequence) continuous recording, ABC nar- rative recording, and scatterplots.

ABC Continuous Recording

With ABC continuous recording, an observer records oc- currences of the targeted problem behaviors and selected environmental events in the natural routine during a pe- riod of time. Codes for recording specific antecedents, problem behaviors, and consequences can be developed based on information obtained from a functional assess- ment interview or ABC narrative recording (described later). For example, following an interview and observa- tions using narrative recording, Lalli, Browder, Mace, and Brown (1993) developed stimulus and response codes to record the occurrence or nonoccurrence of antecedent (e.g., one-to-one instruction, group instruction) and sub- sequent events (attention, tangible reinforcement, escape) for problem behavior during classroom activities.

With ABC continuous recording, the occurrence of a specified event is marked on the data sheet (using par- tial interval, momentary time sampling, or frequency recording) (see Figure 2). The targeted environmental events (antecedents and consequences) are recorded whenever they occur, regardless of whether problem be- havior occurred with it. Recording data in this manner may reveal events that occur in close temporal relations with the target behavior. For example, descriptive data may show that tantrums (behavior) often occur when a student is given an instruction to wash her hands (ante- cedent); the data may also show that tantrums are typi- cally followed by the removal of task demands. A possible hypothesis in this situation is that disruptions are motivated by academic demands and are maintained by escape from those demands (negative reinforcement).

Advantages of ABC Continuous Recording. De- scriptive assessments based on continuous recording use precise measures (similar to functional analyses), and in some cases the correlations may reflect causal

516

Functional Behavior Assessment

Figure 2 Sample data collection form for ABC continuous recording.

ABC Recording Form Observer: R. Van Norman Time begin: 9:30 A.M. Time end: 10:15 A.M. Date: January 25, 2006

Antecedent Behavior Consequence

❑ Task prompt /instruction ❑ Social attention ❑x Attention diverted ❑x Reprimand ❑ Social interaction ❑x Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑ Attention diverted

❑x Task prompt /instruction ❑ Social attention ❑ Attention diverted ❑ Reprimand ❑ Social interaction ❑x Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑x Task removed ❑ Alone (no attention/no activities) ❑ Attention diverted

❑ Task prompt /instruction ❑ Social attention ❑x Attention diverted ❑ Reprimand ❑ Social interaction ❑x Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑x Attention diverted

❑x Task prompt /instruction ❑ Social attention ❑ Attention diverted ❑ Reprimand ❑ Social interaction ❑ Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑ Attention diverted

❑ Task prompt /instruction ❑ Social attention ❑x Attention diverted ❑ Reprimand ❑ Social interaction ❑ Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑ Attention diverted

(continued )

517

Functional Behavior Assessment

Figure 2 (continued)

Antecedent Behavior Consequence

❑ Task prompt /instruction ❑ Social attention ❑ Attention diverted ❑x Reprimand ❑x Social interaction ❑x Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑ Attention diverted

❑ Task prompt /instruction ❑ Social attention ❑ Attention diverted ❑ Reprimand ❑x Social interaction ❑x Tantrum ❑ Task demand ❑ Engaged in preferred activity ❑ Aggression ❑ Access to preferred item ❑ Preferred activity removed ❑ Task removed ❑ Alone (no attention/no activities) ❑x Attention diverted

Source Recording form developed by Renée Van Norman. Used by permission.

relations (e.g., Sasso et al., 1992). Because the assess- ments are conducted in the context in which the prob- lem behaviors occur, they are likely to provide useful information for designing a subsequent functional analysis if that proves necessary. In addition, they do not require disruption to the person’s routine.

Limitations of ABC Continuous Recording. Although descriptive analyses of this type may show a correlation between particular events and the problem behavior, such correlations can be difficult to detect in many situations. This is especially likely if the influen- tial antecedents and consequences do not reliably pre- cede and follow the behavior. In such cases, it may be necessary to analyze descriptive data by calculating conditional probabilities. A conditional probability is the likelihood that a target problem behavior will occur in a given circumstance. Given the example provided in Figure 2, the conditional probability of tantrums is computed by calculating (a) the proportion of occur- rences of tantrums that were preceded by the an- tecedent of an instruction and (b) the proportion of occurrences of tantrums for which task removal was

the consequence. In this example, 9 instances of tan- trums were recorded, and 6 of those occurrences were followed by task removal. Thus, the conditional proba- bility that tantrums occur in the presence of task de- mands and are followed by escape is .66. The closer the conditional probability is to 1.0, the stronger the hy- pothesis is that escape is a variable maintaining prob- lem behavior.

However, conditional probabilities can be mislead- ing. If a behavior is maintained by intermittent rein- forcement, the behavior might occur often even though it is not followed consistently by a particular conse- quence. For example, the teacher might send the student to time-out only when tantrums are so frequent or se- vere that they are intolerable. In this case, only a small proportion of tantrums would be followed by that con- sequence, and the conditional probability would be low. One possibility, therefore, is that a functional relation that exists (e.g., tantrums negatively reinforced by es- cape) will not be detected. Furthermore, the child’s cur- rent behavior intervention plan might require three repetitions of the instruction and an attempt to provide physical assistance before time-out is implemented. In

518

Functional Behavior Assessment

that situation, the conditional probability of tantrums being followed by attention would be high. A descriptive analysis might, therefore, suggest a functional relation (e.g., tantrums positively reinforced by attention) that does not exist. Perhaps for these reasons, studies that have used conditional probability calculations to exam- ine the extent to which descriptive methods lead to the same hypotheses as functional analyses have generally found low agreement (e.g., Lerman & Iwata, 1993; Noell et al., 2001).

ABC Narrative Recording

ABC narrative recording is a form of descriptive assess- ment that differs from continuous recording in that (a) data are collected only when behaviors of interest are ob- served, and (b) the recording is open-ended (any events that immediately precede and follow the target behavior are noted) (see Figure 3). Because data are recorded only when the target behavior occurs, narrative recording may be less time-consuming than continuous recording. On the other hand, narrative recording has several disadvan- tages in addition to those described earlier.

Limitations of Narrative Recording. Because nar- rative recording data are seldom reported in published research, their utility in identifying behavioral function has not been established. However, ABC narrative recording might identify functional relations that do not exist because antecedent and consequent events are recorded only in relation to the target behavior; it would not be evident from ABC data if particular events occurred just as often in the absence of the target behavior. For example, ABC data might erroneously in- dicate a correlation between peer attention and disrup- tion even though peer attention also occurred frequently when the student was not disruptive.

Another potential limitation of ABC narrative record- ing concerns its accuracy. Unless observers receive ade- quate training, they may report inferred states or subjective impressions (e.g., “felt embarrassed,” “was frustrated”) instead of describing observable events in objective terms. In addition, given the likelihood that a number of environmental events occur in close temporal proximity to one another, discriminating the events that occasion a behavior can be difficult. ABC narrative recording may be best suited as a means of gathering pre- liminary information to inform continuous recording or functional analyses.

Scatterplots

Scatterplot recording is a procedure for recording the extent to which a target behavior occurs more often at

particular times than others (Symons, McDonald, & Wehby, 1998; Touchette, MacDonald, & Langer, 1985). Specifically, scatterplots involve dividing the day into blocks of time (e.g., a series of 30-minute segments). For every time segment, an observer uses different sym- bols on an observation form to indicate whether the tar- get problem behavior occurred a lot, some, or not at all. After data have been collected over a series of days, they are analyzed for patterns (specific time periods that are typically associated with problem behavior). If a re- curring response pattern is identified, the temporal dis- tributions in the behavior can be examined for a relation to particular environmental events. For example, a time period in which the behavior occurs often might be cor- related with increased demands, low attention, certain activities, or the presence of a particular person. If so, changes can be made on that basis.

Advantages of Scatterplots. The primary advan- tage of scatterplots is that they identify time periods during which the problem behavior occurs. Such infor- mation can be useful in pinpointing periods of the day when more focused ABC assessments might be con- ducted to obtain additional information regarding the function of the problem behavior.

Limitations of Scatterplots. Although scatterplots are often used in practice, little is known about their utility. It is unclear whether temporal patterns are rou- tinely evident (Kahng et al., 1998). Another problem is that obtaining accurate data with scatterplots may be difficult (Kahng et al., 1998). In addition, the subjective nature of the ratings of how often the behavior occurs (e.g., “a lot” versus “some”) can contribute to difficul- ties with interpretation (standards for these values might differ across teachers or raters).

Indirect Functional Behavior Assessment

Indirect functional assessment methods use struc- tured interviews, checklists, rating scales, or question- naires to obtain information from persons who are familiar with the person exhibiting the problem be- havior (e.g., teachers, parents, caregivers, and/or the individual him- or herself) to identify possible condi- tions or events in the natural environment that cor- relate with the problem behavior. Such procedures are referred to as “indirect” because they do not involve direct observation of the behavior, but rather solicit information based on others’ recollections of the behavior.

519

Functional Behavior Assessment

Behavioral Interviews

Interviews are used routinely in assessment. The goal of a behavioral interview is to obtain clear and objective infor- mation about the problem behaviors, antecedents, and con- sequences. This might include clarifying descriptions of the behavior (consequences); when (times), where (settings, activities, events), with whom, and how often it occurs; what typically precedes the behavior (antecedents); what the child and others typically do immediately following the behavior (consequences); and what steps have previously been taken to address the problem, and with what result. Similar information might be solicited about desirable be- havior (or the conditions under which undesirable behav- ior does not occur) to identify patterns or conditions that predict appropriate versus problem behavior. Information can also be obtained about the child’s apparent preferences (e.g., favorite items or activities), skills, and means of com- municating. A skillful interviewer poses questions in a way that evokes specific, complete, and factual responses about events, with minimal interpretation or inferences.

Lists of interview questions have been published, and they provide a consistent, structured format for obtaining information through either an interview or questionnaire format. For example, the Functional Assessment Inter- view (O’Neill et al., 1997) has 11 sections, which include description of the form (topography) of the behavior, gen- eral factors that might affect the behavior (medications, staffing patterns, daily schedule), antecedents and out- comes of the behavior, functional behavior repertoires, communication skills, potential reinforcers, and treat- ment history.

A form of the Functional Assessment Interview for students who can serve as their own informants is also available (Kern, Dunlap, Clarke, & Childs, 1995; O’Neill et al., 1997). Questions include the behavior(s) that cause trouble for the students at school, a descrip- tion of the student’s class schedule and its relation to problem behavior, rating of intensity of behaviors on a scale of 1 to 6 across class periods and times of day, as- pects of the situation related to the behavior (e.g., dif- ficult, boring, or unclear material; peer teasing; teacher reprimands), other events that might affect the behavior (e.g., lack of sleep, conflicts), consequences (what oc- curs when the individual engages in the behavior), pos- sible behavior alternatives, and potential strategies for a support plan.

Two other questionnaires are the Behavioral Diag- nosis and Treatment Information Form (Bailey & Pyles, 1989) and the Stimulus Control Checklist (Rolider & Van Houten, 1993), which also address questions about the conditions under which the behavior occurs or doesn’t occur, and how often. In addition, they include questions about physiological factors that might affect the behavior.

Behavior Rating Scales

Behavior rating scales designed for functional assessment ask informants to estimate the extent to which behavior occurs under specified conditions, using a Likert scale (e.g., never, seldom, usually, always). Hypotheses about the func- tion of a behavior are based on the scores associated with each condition. Those conditions that are assigned the highest cumulative or average rating are hypothesized to be related to the problem behavior. For example, if an infor- mant states that problem behavior always occurs when de- mands are placed on a child, a negative reinforcement hypothesis might be made. Features of several behavior rating scales are summarized in Table 2.

Advantages of Indirect FBA

Some indirect assessment methods can provide a useful source of information in guiding subsequent, more ob- jective assessments, and contribute to the development of hypotheses about variables that might occasion or maintain the behaviors of concern. Because indirect forms of FBA do not require direct observation of prob- lem behavior, many people view them as convenient.

Limitations of Indirect FBA

A major limitation of indirect FBA is that informants may not have accurate and unbiased recall of behavior and the conditions under which it occurred, or be able to report such recollections in a way that conforms to the require- ments of the question. Perhaps for these reasons, little re- search exists to support the reliability of the information obtained from indirect assessment methods. The Motiva- tion Assessment Scale (MAS) is one of the few behavior rating scales that has been evaluated for its technical ad- equacy. Several studies have evaluated the interrater agree- ment of the MAS, and almost all of them have found it to be low (Arndorfer, Miltenberger, Woster, Rotvedt, & Gaffaney, 1994; Barton-Arwood, Wehby, Gunter, & Lane, 2003; Conroy, Fox, Bucklin, & Good, 1996; Crawford, Brockel, Schauss, & Miltenberger, 1992; Newton & Sturmey, 1991; Sigafoos, Kerr, & Roberts, 1994; Zarcone, Rodgers, & Iwata, 1991). Barton-Arwood and colleagues (2003) also evaluated the technical adequacy of the Prob- lem Behavior Questionnaire (PBQ). They reported vari- able intrarater agreement and questionable stability of the identified behavioral function with both the PBQ and MAS as applied to students with emotional or behavioral disorders. Because of the lack of empirical data to sup- port their validity and interrater agreement, indirect meth- ods of FBA are not recommended as the principal means of identifying the functions of behaviors. These scales can, however, provide information useful for forming ini- tial hypotheses, which can later be tested.

520

Functional Behavior Assessment

Table 2 Overview of Behavior Rating Scales Used to Assess Possible Functions of Problem Behavior

Example item and possible Behavior rating scale Functions assessed Format and number of items function

Motivation Assessment Sensory reinforcement, 16 questions (4 for each of 4 Does the behavior seem to Scale (MAS) (Durand & escape, attention, and functions), 7-point scale from occur in response to your Crimmins, 1992) tangible reinforcement always to never talking to other persons in the

room? (attention)

Motivation Analysis Rating Sensory reinforcement, 6 statements (2 for each of 3 The behavior stops occurring Scale (MARS) (Wieseler, escape, and attention functions), 4-point scale from shortly after you stop working Hanson, Chamberlain, & always to never or making demands on this Thompson, 1985) person. (escape)

Problem Behavior Peer attention, teacher Questions, 7-point range When the problem behavior Questionnaire (PBQ) attention, escape/avoid peer occurs, do peers verbally (Lewis, Scott, & Sugai, 1994) attention, escape/avoid respond to or laugh at the

teacher attention, and student? (peer attention) assessment of setting events

Functional Analysis Screening Social reinforcement (attention, Yes or no as to whether state- When the behavior occurs, do Tool (FAST) (Iwata & DeLeon, preferred items), social re- ments are descriptive you usually try to calm the 1996) inforcement (escape), auto- person down or distract the

matic reinforcement by person with preferred sensory stimulation, auto- activities (leisure items, matic reinforcement by pain snacks, etc.)? (social attenuation reinforcement, attention,

preferred items)

Questions About Behavioral Attention, escape, nonsocial, Statements, 4-point range Participant engages in the Function (QABF) (Paclawskyj, physical, tangible behavior to try to get a Matson, Rush, Smalls, & reaction from you. (attention) Vollmer, 2000)

Conducting a Functional Behavior Assessment Given the strengths and limitations of the different FBA procedures, FBA can best be viewed as a four-step process as follows:

1. Gather information via indirect and descriptive assessment.

2. Interpret information from indirect and descriptive assessment and formulate hypotheses about the purpose of problem behavior.

3. Test hypotheses using functional analysis.

4. Develop intervention options based on the function of problem behavior.

Gathering Information

It is often helpful to begin the FBA process by con- ducting functional assessment interviews with the per- son’s teacher, parent, caregiver, and/or others who work closely with the person. The interview can be helpful in preparing the evaluator to conduct direct observa- tions by identifying and defining the target problem be-

haviors, identifying and defining potential antecedents and consequences that may be observed, and gaining an overall picture of the problem behavior as well as the strengths of the person. The interview can also help determine if other assessments are warranted before a more extensive FBA is conducted. For example, if the interview reveals that the person has chronic ear infec- tions that are currently untreated, a medical evaluation should be conducted before further behavioral assess- ment takes place.

In many cases conducting an interview with the per- son who has problem behavior can be helpful if he has the language skills to understand and respond to interview questions. Sometimes the person has useful insights re- garding why he displays problem behavior in specific contexts (Kern et al., 1995).

At this point, conducting direct observations of the problem behavior within the natural routine is useful. Such observations help confirm or disconfirm the information obtained through the interviews. If it is not clear when the problem behavior occurs most often, a scatterplot analy- sis may be useful to determine when further behavioral observations should be conducted. When the problematic time periods have been determined, the behavior analyst

521

Functional Behavior Assessment

can conduct ABC assessments. Information obtained from the interviews is helpful in guiding the ABC assessment because the evaluator should already have clear defini- tions of the target behavior(s), antecedents, and conse- quences. However, the behavior analyst must also be watchful for additional, unexpected antecedents and con- sequences that might present themselves in the natural en- vironment. Teachers or caregivers sometimes overlook or are unaware of specific stimuli triggering or following problem behavior.

Interpreting Information and Formulating Hypotheses

Results from indirect assessments should be analyzed for patterns of behavior and environmental events so that hypotheses regarding the function of the problem be- havior can be made. If problem behavior occurs most frequently when low levels of attention are available and problem behavior frequently produces attention, a hy- pothesis that attention maintains the problem behavior is appropriate. If the problem behavior occurs most fre- quently in high-demand situations and often produces a reprieve from the task (e.g., through time-out, suspen- sion, or another form of task delay), then a hypothesis that escape maintains the problem behavior is appropri- ate. If problem behavior occurs in an unpredictable pat- tern or at high rates across the school day, a hypothesis that the behavior is maintained by automatic reinforce- ment may be appropriate. In reviewing assessment re- sults and considering possible hypotheses, behavior analysts should remember that behaviors may serve mul- tiple functions and different topographies of problem be- havior may serve different functions.

Hypothesis statements should be written in ABC for- mat. Specifically, the hypothesis statement should state the antecedent(s) hypothesized to trigger the problem be- havior, the topography of problem behavior, and the maintaining consequence. For example:

Hypothesized function Antecedent Behavior Consequence

Escape from When Tonisha she screams termination of hand washing is prompted and tantrums, hand and/or lunch. to wash her which is washing

hands in followed by . . . and lunch preparation by being sent for lunch, . . . to time-out.

Writing hypothesis statements in this manner is use- ful because it requires the behavior analyst to focus on po- tential avenues for intervention: modifying the antecedent and/or modifying the reinforcement contingencies (which may involve teaching a new behavior and altering what behaviors are reinforced or placed on extinction).

Testing Hypotheses

After hypotheses have been developed, a functional analy- sis can be conducted to test them. The functional analy- sis should always contain a control condition that serves to promote the lowest frequency of problem behavior. For most individuals, this is the play condition, which con- sists of (a) continuous availability of preferred toys and/or activities, (b) no demands, and (c) continuously available attention. Then, conditions are selected to test specific hypotheses. For example, if the primary hypothesis is that the problem behavior is maintained by escape, then a con- tingent escape condition should be implemented. Other test conditions may not need to be implemented.

Being selective about the test conditions imple- mented will help keep the functional analysis as brief as possible; however, no conclusions can be made regarding additional functions of problem behavior if additional test conditions are not implemented. For example, if play and contingent escape are the only conditions tested, and problem behavior occurs most frequently in the contin- gent escape condition and seldom or never occurs in the play condition, the conclusion that the problem behavior is maintained by escape is supported. However, because a contingent attention condition was not implemented, one could not rule out the possibility that the problem behavior is also maintained by attention.

One way to test all possible hypotheses within a short period of time is to use a brief functional analysis proce- dure (e.g., Boyagian, DuPaul, Wartel, Handler, Eckert, & McGoey, 2001; Cooper et al., 1992; Derby et al., 1992; Kahng & Iwata, 1999; Northup et al., 1991; Wacker et al., 1990). This technique involves implementing one session each of the control condition and each of the test condi- tions. If an increase in problem behavior is observed in one of the test conditions, a contingency reversal is im- plemented to confirm the hypothesis rather than conduct- ing many repetitions of all of the conditions. For example, after conducting one of each of the control and test con- ditions, assume that Tonisha’s tantrums were elevated in the contingent escape condition (see Figure 3). To con- firm this hypothesis, the behavior analyst could reinstate the contingent escape condition, followed by a contin- gency reversal in which problem behavior no longer pro- duces escape. Rather, another behavior (often a request) is substituted for problem behavior. For example, in Ton- isha’s case, she could be prompted to form the manual sign “break,” and contingent on signing “break,” a task break is provided. The contingency reversal is then fol- lowed by reinstatement of the contingent escape condi- tion, in which the sign “break” is placed on extinction and the tantrums again produces escape. This process could be repeated for any and all conditions in which the problem behavior was elevated during the initial analysis.

522

Functional Behavior Assessment

N u

m b

er o

f R

es p

o n

se s

p er

M in

u te

1 2 3 4 5 6 7

Free Play Alone

Contingent Attention

Tantrums

Sign

"Break"

Contingent Escape

Escape for

Signing Contingent

Escape

Sessions

Contingency Reversal

Figure 3 Hypothetical data from brief functional analysis of Tonisha’s tantrums. The closed data points represent tantrums, and the open data points represent manual signs for breaks. The first four sessions represent the brief functional analysis, while Sessions 5 through 7 represent the contingency reversal.

Kahng and Iwata (1999) found that brief functional analysis identified the function of problem behavior ac- curately (i.e., it yielded results similar to extended func- tional analyses) in 66% of the cases. Brief functional analyses are useful when time is very limited. However, given that brief functional analyses do not match the re- sults of extended functional analyses at least a third of the time, it is probably wise to conduct an extended functional analysis whenever circumstances permit. In addition, when brief functional analyses are inconclu- sive, it is recommended that a more extended functional analysis be conducted (Vollmer, Marcus, Ringdahl, & Roane, 1995).

Developing Interventions

When an FBA has been completed, an intervention that matches the function of problem behavior can be devel- oped. Interventions can take many forms. Although FBA does not identify which interventions will be effective in treating problem behavior, it does identify antecedents that may trigger problem behavior, potential behavioral deficits that should be remedied, and reinforcement con- tingencies that can be altered, as described earlier in this chapter. However, FBA does identify powerful reinforcers that can be used as part of the intervention package. The intervention should be functionally equivalent to prob- lem behavior. That is, if problem behavior serves an es- cape function, then the intervention should provide escape (e.g., in the form of breaks from task demands) for a more appropriate response or involve altering task demands in a fashion that makes escape less reinforcing.

One effective way to design interventions is to re- view confirmed hypotheses to determine how the ABC contingency can be altered to promote more positive be- havior. For example, consider the hypothesis developed for Tonisha:

Hypothesized function Antecedent Behavior Consequence

Escape from When Tonisha she screams termination of hand wash- is prompted and tantrums, hand washing ing and/or to wash her which is and lunch by lunch. hands in followed being sent to

preparation by . . . time-out. for lunch, . . .

The antecedent could be altered by changing the time of day when Tonisha is asked to wash her hands (so that it does not precede lunch, thereby decreasing the moti- vation for escape-motivated tantrums).

Hypothesized function Antecedent Behavior Consequence

Escape from When Tonisha N/A (Problem N/A (The hand washing is prompted behavior is consequence and/or lunch. to wash her avoided.) is irrelevant

hands in because preparation problem for lunch, . . . behavior did

not occur.) Tonisha is prompted to wash her hands before recess.

523

Functional Behavior Assessment

The behavior could be altered by teaching Tonisha a new behavior (e.g., signing “break”) that results in the same outcome (escape from lunch).

Hypothesized function Antecedent Behavior Consequence

Escape from When Tonisha she screams termination of hand washing is prompted and tantrums hand washing and/or lunch. to wash her Tonisha is and lunch.

hands in prompted to preparation sign “break,” for lunch, . . . which is

followed by . . .

Or, the consequences could be altered. For example, the reinforcer for problem behavior could be withheld so that problem behavior is extinguished.

Hypothesized Hypothesized function function Antecedent Behavior (consequence)

Escape from When Tonisha she screams termination hand washing is prompted and tantrums, of hand and/or lunch. to wash her which is washing and

hands in followed lunch by being preparation by . . . sent to for lunch, . . . timeout.

continued presentation of hand-washing and lunch activities.

An intervention can also consist of several different com- ponents. For example, Tonisha could be taught a re- placement behavior (signing “break”), which results in breaks from lunch, while tantrums are simultaneously placed on escape extinction.

FBA can also help identify interventions that are likely to be ineffective or that may worsen the problem behavior. Interventions involving time-out, in- or out-of- school suspension, or planned ignoring are contraindi- cated for problem behaviors maintained by escape. Interventions involving reprimands, discussion, or coun- seling are contraindicated for problem behaviors main- tained by attention.

A final word about intervention: When an interven- tion has been developed, FBA is not “done.” Assessment is an ongoing practice that continues when intervention is implemented. It is important for continued monitoring of intervention effectiveness. The functions of behavior are not static. Rather, they are dynamic and change over time. Intervention may lose its effectiveness over time because the function of problem behavior may change over time (Lerman, Iwata, Smith, Zarcone, & Vollmer, 1994). In such cases, additional functional analyses may need to be conducted to revise the intervention.

Case Examples Illustrating the FBA Process FBA is a highly idiosyncratic process. It is unusual for any two FBAs to be exactly the same because each per- son presents with a unique set of skills and behaviors, as well as a unique history of reinforcement. FBA requires a thorough understanding of behavioral principles to par- cel out the relevant information from interviews and ABC assessments, to form relevant hypotheses, and to test those hypotheses. Beyond these skills, a solid under- standing of behavioral interventions (e.g., differential re- inforcement procedures, schedules of reinforcement, and tactics for promoting maintenance and generalization) is needed to match effective treatments to the function of challenging behavior. This can seem like a daunting process. In an attempt to demonstrate the application of FBA across the idiosyncratic differences in people, we present four case examples.

Brian—Multiple Functions of Problem Behavior

Gathering Information

Brian was 13 years old and diagnosed with pervasive de- velopmental delay, oppositional defiant disorder, and attention-deficit/hyperactivity disorder. He had moder- ate delays in cognitive and adaptive skills. Brian dis- played several problem behaviors, including aggression, property destruction, and tantrums. Brian’s aggression had resulted in several of his teachers having bruises, and his property destruction and tantrums frequently disrupted the daily activities of the classroom.

A Functional Assessment Interview (O’Neill et al., 1997) was conducted with Brian’s teacher, Ms. Baker, who reported that Brian’s problem behavior occurred most frequently when he was asked to perform a task that required any kind of physical labor (e.g., shredding pa- pers) and occurred least during leisure activities. How- ever, Ms. Baker reported that Brian often engaged in problem behavior when he was asked to leave a preferred activity. She noted that Brian used complex speech (sen- tences), although he often used verbal threats (e.g., curse words) and/or aggression, property destruction, and tantrums to communicate his wants and needs.

Because Brian was verbal, a Student-Assisted Func- tional Assessment Interview (Kern et al., 1995) was also conducted. In this interview, Brian reported that he found his math work too difficult but that writing and using a calculator were too easy. He reported that he sometimes received help from his teachers when he asked for it, that sometimes teachers and staff noticed when he was doing a good job, and that he sometimes received rewards for

524

Functional Behavior Assessment

Table 3 Results of ABC Assessments for Brian’s Aggression, Property Destruction, and Tantrums

Antecedent Behavior Consequence

Adult attention diverted to another Yelled at teacher, “That’s not fair! Why do Told to “calm down” student; denied access to Nintendo by you hate me?!” teacher (i.e., told no when he asked if he could play it)

Teacher attending to another student Hit sofa, attempted to leave classroom Given choice of activity and verbal warning to stay in classroom

Teacher attention diverted to another Yelled, “Stop!” at another student Reprimand from teacher: “Don’t worry, student Brian. I will take care of it.”

Story time, teacher attending to other Laughed loudly Reprimand from teacher: “Stop it!” students

Story time, teacher listening to other Interrupted other students while they Reprimand from teacher: “You need to students were talking: “Hey, it’s my turn. I know listen.”

what happens next!”

doing good work. Brian indicated that his work periods were always too long, especially those that consisted of shredding papers. Brian reported that he had the fewest problems in school when he was allowed to answer the phone (his classroom job), when he was completing math problems, and when he was playing with his Gameboy. He stated that he felt he had the most problems at school when he was outside playing with the other students be- cause they often teased him, called him names, and cursed at him.

An ABC assessment was conducted on two separate occasions. The results of the ABC assessment are shown in Table 3.

Interpreting Information and Formulating Hypotheses

Based on the interviews and ABC assessments, the func- tion of Brian’s problem behavior was unclear. It was hy- pothesized that some of Brian’s problem behaviors were

maintained by access to adult attention and preferred items. This hypothesis was a result of the ABC assess- ment, which indicated that many of Brian’s problem be- haviors occurred when adult attention was low or when access to preferred items was restricted. Brian’s problem behavior often resulted in access to adult attention or pre- ferred activities. It was also hypothesized that Brian’s problem behavior was maintained by escape because his teacher reported that Brian frequently engaged in problem behavior in the presence of task demands and because Brian reported that some of his work was too hard and work periods lasted too long. Thus, a functional analysis was conducted to test these hypotheses. The hypothesis statements for Brian are summarized in Table 4.

Testing hypotheses

Next, a functional analysis was completed for Brian. The functional analysis consisted of the same conditions as described previously, with two exceptions. First, an alone condition was not conducted because there was no reason

Table 4 Hypothesis Statements for Brian

Hypothesized function Antecedent Behavior Consequence

Gain attention from adults When adult or peer attention he engages in a variety of attention from adults and and peers is diverted from Brian, . . . problem behaviors, which peers.

result in . . .

Gain access to preferred toys When Brian’s access to pre- he engages in a variety of gaining access to preferred and activities ferred toys and activities is problem behaviors, which toys and activities.

restricted, . . . result in . . .

Escape from difficult and/or When Brian is required to he engages in a variety of the tasks being removed. nonpreferred tasks perform difficult or undesirable problem behaviors, which

tasks, . . . result in . . .

525

Functional Behavior Assessment

100

Fr eq

u en

cy o

f In

ap p

ro p

ri at

e B

eh av

io r

2 4 6 8 10 12 14 16 18 20 22 24

Sessions

Contingent Attention

Contingent Escape

Free Play

Contingent Tangible

Figure 4 Results of Brian’s functional analysis. Inappropriate behavior consists of aggression, property destruction, and tantrums. Functional analysis conducted by Renée Van Norman and Amanda Flaute.

to believe that Brian’s problem behavior served an auto- matic function. Second, a contingent tangible condition was added because there was reason to believe Brian en- gaged in problem behavior to gain access to preferred tangibles and activities. This condition was just like the play condition (i.e., Brian had access to adult attention and preferred toys at the beginning of the session), except that intermittently throughout the session, he was told it was time to give his toy to the teacher and to play with something else (which was less desirable). If Brian com- plied with the request to give the toy to the teacher, he was given a less preferred toy. If he engaged in problem be- havior, he was allowed to continue playing with his pre- ferred toys for a brief period of time.

The results of the functional analysis are shown in Figure 4. Notice that problem behavior never occurred in the play condition, but it occurred at high rates in all three of the test conditions (contingent attention, escape, and tangible). These results indicated that Brian’s prob- lem behavior was maintained by escape, attention, and access to preferred items. During the play condition, how- ever, when continuous attention and preferred items were available and no demands were placed on Brian, prob- lem behavior never occurred.

Developing an Intervention

Based on the results of the functional analysis, a multi- component intervention was implemented. The inter- vention components changed at different points in time depending on the context. These components, as they re- late to the functions of problem behavior, are summa- rized in Table 5. For example, when Brian was engaged in a work task, it was recommended that he be given frequent opportunities to request breaks. In addi- tion, the time-out intervention that the teacher had been using was discontinued in the work context. During

leisure times, when Brian had previously been expected to play alone, the classroom schedule was rearranged so that Brian could play and interact with peers. Brian was also taught to request toys appropriately while playing with peers. In addition, several interventions aimed at in- creasing teacher attention for appropriate behavior were implemented. Brian was taught how to request teacher attention appropriately, and teachers began to respond to these requests rather than ignore them (as they had pre- viously done). In addition, a self-monitoring plan was es- tablished, in which Brian was taught to monitor his own behavior and match his self-recordings to his teachers’. Accurate self-recording resulted in teacher praise and ac- cess to preferred activities with the teacher. Brian’s teach- ers also implemented their own plan to increase attention and praise to Brian every 5 minutes as long as he was not engaged in problem behavior during independent work.

Kaitlyn—Attention Function for Problem Behavior

Gathering Information

Kaitlyn was 12 years old and diagnosed with attention- deficit/hyperactivity disorder. She also displayed some fine and gross motor deficits. Kaitlyn attended classes in both her general education sixth-grade classroom and a special education classroom. She frequently displayed off-task behavior, which consisted of being out of her seat, touching others (e.g., playing “footsie” under the desk with peers), making noises, and talking out of turn. A Functional Assessment Interview (O’Neill et al., 1997) was conducted with Kaitlyn’s teacher, who stated that, typically, Kaitlyn asked questions constantly when she was given a difficult task. The teacher also stated that Kaitlyn often became confused when her routines were changed, and that she therefore needed a lot of assistance.

526

Functional Behavior Assessment

Table 5 Summary of Brian’s Intervention Components

Intervention Options for Attention Function

Intervention Antecedent Behavior Consequence

Teach a new behavior When adult or peer attention he will raise his hand and say, and adults and peers will (social attention) is diverted from Brian, . . . “Excuse me . . .” provide attention to Brian.

Teach a new behavior When adult or peer attention he will self-monitor his appro- and the teachers will provide is diverted from Brian, . . . priate independent work and him with one-on-one time if

match teacher recordings . . . he meets a specific criterion.

Change the antecedent During independent work to increase the probability which will increase adult times, adults will provide at- that Brian will appropriately opportunities to praise and tention to Brian every work independently, . . . attend to appropriate 5 minutes . . . behavior.

Change the antecedent Allow Brian to play with peers to increase the probability which will increase adult during leisure times . . . that Brian will play appropri- opportunities to praise

ately, . . . appropriate behavior and for peers to respond positively.

Intervention Options for Tangible Function

Intervention Antecedent Behavior Consequence

Teach a new behavior When Brian’s access to pre- he will say, “Can I have that and the teacher will provide ferred toys and activities is back, please?” . . . access to preferred toys and restricted, . . . activities.

Intervention Options for Escape Function

Intervention Antecedent Behavior Consequence

Teach a new behavior When Brian is required to he will say, “May I take a and the teacher will allow perform a difficult or undesir- break now?” . . . Brian to take a break from the able task, . . . task.

Change the reinforcement When Brian is required to and he engages in a variety of he will be required to continue contingency perform difficult or undesirable problem behaviors, . . . working on the task and the

tasks . . . time-out intervention will be discontinued.

Because there was only one teacher in the classroom and 25 other students, Kaitlyn received little additional at- tention, and her teacher hypothesized that she engaged in off-task behavior to gain attention.

The results of an ABC assessment showed that Kait- lyn often received little adult attention in the classroom, except when she displayed off-task behavior. Although off-task behavior occurred when attention was diverted, Kaitlyn was often engaged in demanding tasks simulta- neously. Thus, it was unclear which antecedent—the low

attention or the high-demand activity—was related to her problem behavior.

Interpreting Information and Formulating Hypotheses

Based on the information obtained from interviews and ABC assessments, it was hypothesized that Kaitlyn’s off- task behavior served an attention function. The hypothe- ses developed for Kaitlyn are summarized in Table 6.

Table 6 Hypotheses Regarding the Function of Kaitlyn’s Off-Task Behavior

Hypothesized function Antecedent Behavior Consequence

Primary hypothesis—Gain When teacher attention is she engages in off-task be- teacher attention attention from adults diverted from Kaitlyn, . . . havior, which results in . . . (reprimands, task prompts).

Alternative hypothesis— When Kaitlyn is required to she engages in off-task be- those tasks being removed. Escape from difficult academic work on academic tasks, . . . havior, which results in . . . tasks

527

Functional Behavior Assessment

100

P er

ce n

ta ge

o f

10 -S

ec o

n d

I n

te rv

al s

w it

h O

ff -T

as k

B eh

av io

2 4 6 8 10 12 14 16 18 20 Sessions

CA/Easy

CA/Difficult

Free Play

CA/FP

Escape

Figure 5 Results of the functional analysis for Kaitlyn’s off-task behavior. FP = free play; CA/FP = contingent attention during free play activities; CA/Easy = contingent attention during easy academic activities; CA/Difficult = contingent attention during difficult academic activities. Functional analysis conducted by Jessica Frieder, Jill Grunkemeyer, and Jill Hollway.

However, because Kaitlyn was typically observed in aca- demic activities that were paired with low attention sit- uations, it was unclear what role task difficulty played in her off-task behavior. It was unclear whether she en- gaged in problem behavior to gain attention or because the academic tasks were too difficult. Would Kaitlyn en- gage in similar levels of problem behavior when play materials were present as compared to easy or difficult academic materials while attention was diverted? It was important to design a functional analysis that would ad- dress these questions.

Testing Hypotheses

Kaitlyn’s functional analysis demonstrates how func- tional analysis conditions can be constructed to test many kinds of hypotheses. Her functional analysis consisted of the standard play and escape conditions described pre- viously. However, several different contingent attention conditions were implemented to determine whether de- mands interacted with low attention conditions to evoke problem behavior.

Three contingent attention conditions were con- ducted: contingent attention during free play activities (CA/FP), contingent attention during easy academic ac- tivities (CA/Easy), and contingent attention during diffi- cult academic activities (CA/Difficult). In all three conditions, attention was diverted from Kaitlyn until she engaged in off-task behavior. Contingent on off-task be- havior, a teacher approached her and delivered a mild reprimand (e.g., “Kaitlyn, what are you doing? You’re supposed to be doing this activity. I need you to get back to your activity now”.). The conditions differed in the type of activity Kaitlyn was asked to work on. During CA/FP,

1The MotivAider is a device that can be set to vibrate at specific intervals of time. The MotivAider can be obtained at www.habitchange.com.

Kaitlyn was allowed to play with a game of her choice. During CA/Easy, Kaitlyn was required to solve single- digit addition problems. During CA/Difficult, Kaitlyn was required to solve multidigit subtraction problems that required regrouping.

The results of the functional analysis are shown in Figure 5. Little off-task behavior occurred in the free play and escape conditions. Thus, the hypothesis that Kaitlyn engaged in off-task behavior to escape tasks was not supported. Increased off-task behavior was observed in all three of the contingent attention conditions. These data suggested that Kaitlyn’s off-task behavior served an attention function regardless of the type of activity in which she was involved.

Developing an Intervention

Kaitlyn was verbal and often requested attention appro- priately. Therefore, Kaitlyn was taught to monitor her on- and off-task behavior when she was required to work or play independently. Initially, Kaitlyn was taught to mon- itor her behavior every 10 seconds. (This was the longest she was able to stay on task without prompting during baseline.) Both she and a classroom assistant wore a vi- brating timer (a MotivAider)1 that would not distract the other students. When the timer vibrated, Kaitlyn marked on a self-monitoring sheet if she was on or off task. She then looked up at the assistant. If Kaitlyn was on task, the assistant gave Kaitlyn a thumbs-up signal and a smile.

528

Functional Behavior Assessment

Occasionally, the assistant also approached Kaitlyn, gave her a pat on the back, and said, “Way to stay on task!” If she was not on task, the assistant looked away from her. At the end of 10 minutes, the assistant and Kaitlyn com- pared marks. If Kaitlyn was on task on 50 of 60 checks and she matched the teacher on 57 of 60 checks, she was allowed to engage in a special activity with the assistant for 5 minutes. Over time, the interval length and dura- tion of the observation period were increased until Kait- lyn worked appropriately for the entire period.

DeShawn—Automatic Function of Problem Behavior

Gathering Information

DeShawn was 10 years old and diagnosed with autism. He had severe developmental delays and was blind. He took Resperidol for behavioral control. DeShawn fre- quently threw objects across the room, dropped objects and task materials off the table, and knocked objects from tables with a sweeping motion of his arm, which inter- fered with work on academic tasks. A Functional As- sessment Interview (O’Neill et al., 1997) yielded little useful information because the teacher stated that DeShawn’s throwing, dropping, and sweeping were un- predictable. She could not identify any antecedents that typically preceded the problem behavior. The intervention in place at the time was to limit DeShawn’s access or proximity to objects that he might throw. ABC observa- tions conducted within the natural routine also produced little useful information because the teachers prevented

and blocked all occurrences of the problem behavior. Thus, very few occurrences of the problem behavior were observed. However, it was noted that DeShawn was rarely actively involved in lessons. For example, when the teacher read a book to the class, DeShawn was unable to participate because he could not see the pictures. At other times, DeShawn was engaged in individual work tasks, but these tasks did not appear to be meaningful or ap- propriate for his skill level.

Interpreting Information and Formulating Hypotheses

It was difficult to formulate a hypothesis based on the limited information from the teacher and direct obser- vations. Because DeShawn did not seem actively in- volved or interested in the classroom activities, it was hypothesized that throwing, dropping, and sweeping were automatically reinforcing. However, other hy- potheses, such as attention, escape, or multiple functions, could not be ruled out.

Testing Hypotheses

The functional analysis consisted of play, contingent at- tention, and contingent escape conditions. An alone con- dition was not conducted because there was not a room that allowed DeShawn to be observed and supervised covertly. The results of the functional analysis are shown in Figure 6. Throwing, dropping, and sweeping occurred at high and variable rates across all three conditions,

2 4 6 8 10

Sessions

N u

m b

er o

f T

h ro

w s,

D ro

p s,

a n

d S

w ee

p s

Contingent Attention

Free Play

Contingent Escape

Figure 6 Results of DeShawn’s func- tional analysis. Functional analysis conducted by Susan M. Silvestri, Laura Lacy Rismiller, and Jennie E. Valk.

529

Functional Behavior Assessment

C h

ip s

T u

b e

P u

tt y

T w

ir lin

g to

K o

o sh

b al

Sl in

0N u

m b

er o

f T

im es

E ac

h S

ti m

u lu

s W

as S

el ec

te d

Stimulus

M ea

n D

u ra

ti o

n o

f It

em I

n te

ra ct

io n

( se

c) b

ef o

re T

h ro

w n

Figure 7 Results of DeShawn’s prefer- ence assessment. The bars indicate the number of times each stimulus was selected. The line graph represents the mean number of seconds DeShawn played with each item stimulus before throwing it. Preference assessment conducted by Susan M. Silvestri, Laura Lacy Rismiller, and Jennie E. Valk.

demonstrating an undifferentiated pattern. These results were inconclusive, but suggested an automatic reinforce- ment function for throwing, dropping, and sweeping.

Further analysis might have identified more precisely the source of reinforcement (e.g., sweeping motion, sound of objects falling, a clear table surface). Instead, we sought to identify alternative reinforcers that might com- pete with automatically maintained problem behavior.

We conducted a forced-choice preference assessment (Fisher et al., 1992) to identify highly preferred stimuli. The results of this assessment are shown in Figure 7. DeShawn most frequently chose potato chips. Data were also collected on the number of seconds DeShawn played with each item before he threw it. (He was allowed to have access to each item for 30 seconds, so a data point at 30 seconds indicates that he played with the item the entire time and never threw it.) No data were collected on how long DeShawn “played” with potato chips because he did not play with them; he consumed them. However, it should be noted that DeShawn never threw the potato chips, nor did he throw any other objects in the area when he had access to potato chips. He immediately put them in his mouth and ate them. These two pieces of informa- tion indicated that potato chips were highly preferred and might function as a reinforcer that could compete with throwing, dropping, and sweeping. Although the tube that made noise was chosen second most frequently, De- Shawn typically threw it after playing with it for only 12 seconds.

The next assessment evaluated whether potato chips served as a reinforcer for pressing a microswitch that said “Chip, please” and whether the use of potato chips would

compete with throwing (based on the microswitch as- sessment procedures of Wacker, Berg, Wiggins, Mul- doon, & Cavanaugh, 1985). First, baseline data were collected on the number of appropriate switch presses (pressing the switch once to activate the voice); the num- ber of inappropriate switch presses (repeatedly pressing or banging the switch); the number of attempts to throw the switch; and the number of throws, drops, and sweeps of other objects. During baseline, DeShawn was provided with access to the microswitch; however, neither chips nor the Slinky was delivered for pressing the switch. Next, the voice on the switch said “Chip, please,” and a potato chip was provided contingent on appropriate switch pressing. In the next condition, the voice on the switch said “Slinky, please,” and the Slinky (a nonpre- ferred toy) was provided contingent on appropriate switch pressing. Finally, the voice on the switch said “Chip, please” again, and a potato chip was provided contingent on appropriate switch pressing to form an ABCB rever- sal design.

The results of the microswitch assessment are shown in Figure 8. Appropriate microswitch presses in- creased when potato chips were provided contingently. Interestingly, no throwing, dropping, or sweeping oc- curred when the potato chip was available. Some inap- propriate switch presses also occurred, but appropriate switch presses were more frequent. When microswitch pressing resulted in a nonpreferred toy, appropriate switch pressing decreased and throwing, dropping, and sweep- ing increased. This suggested that potato chips competed effectively with throwing, dropping, and sweeping.

530

Functional Behavior Assessment

Figure 8 Results of DeShawn’s microswitch assessment. During Baseline, touching the switch did not result in any tangible stimuli. During Switch = Chip, touching the switch resulted in a bite of potato chip. During Switch = Slinky, touching the switch resulted in access to the Slinky for 30 seconds. Reinforcer assessment conducted by Renée Van Norman, Amanda Flaute, Susan M. Silvestri, Laura Lacy Rismiller, and Jennie E. Valk.

Developing an Intervention

Based on these assessments, an intervention was devel- oped that involved giving DeShawn small bites of potato chips for participating in classroom activities appropri- ately. In addition, classroom activities and routines were modified to increase DeShawn’s participation, and his curriculum was modified to include instruction on more functional activities.

Lorraine—Multiple Topographies That Serve Multiple Functions

Lorraine was 32 years old and functioned in the moder- ate range of mental retardation. She had a diagnosis of Down syndrome and bipolar disorder with psychotic symptoms, for which Zoloft and Resperidol were pre- scribed. She also took Tegretol for seizure control. She was verbal, but her verbal skills were low and her artic- ulation was poor. She communicated through some signs, a simple communication device, gestures, and some words.

Lorraine had resided in a group home for 9 years and attended a sheltered workshop during the day. Lorraine displayed noncompliance, aggression, and SIB in both settings, but the FBA focused on her problem behavior in the group home, where it was more severe and fre- quent. Noncompliance consisted of Lorraine putting her head down on the table, pulling away from people, or leaving the room when requests were made of her; ag- gression consisted of kicking others, throwing objects at others, biting others, and squeezing others’ arms very

hard; SIB consisted of biting her arm, pulling her hair, or pinching her skin.

Gathering Information

Interviews were conducted with Lorraine, her parents, and workshop and group home staff. Lorraine’s parents noted that some of her behavior problems had increased when changes in her medication had been made. Work- shop staff noted that Lorraine was more likely to have problem behavior at work if many people were around her. Workshop staff had also noted that noncompliance had increased shortly after a dosage change in medication 2 months previously. The group home staff noted that they were most concerned about Lorraine’s leaving the group home when she was asked to perform daily chores. Lorraine would often leave the group home and not return until the police had picked her up. Many neighbors had complained because Lorraine would sit on their porches for hours until the police came and removed her.

An ABC assessment was conducted at the workshop and group home to determine whether environmental variables differed across the two settings (e.g., the man- ner in which tasks were presented, the overall level of at- tention). At the workshop, Lorraine was engaged in a jewelry assembly task (one she reportedly enjoyed), and she worked well for 2 1/2 hours. She appeared to work better when others paid attention to her and often became off task when she was ignored; however, no problem be- havior was observed at work. At the group home, ag- gression was observed when staff ignored Lorraine. No

Fr eq

u en

cy o

f B

eh av

io r

2 4 6 8 10 12 14 16 18 20 22 24 Sessions

Baseline (Switch = Nothing) Switch = Potato Chip

Switch = Slinky Switch = Potato Chip

Appropriate Switch Presses

Throwing, Dropping, Sweeping

531

Functional Behavior Assessment

100

Noncompliance

2 4 6 8 10 12

Contingent Escape

Free Play Contingent Attention

100

Aggression

2 4 6 8 10 12

Contingent Escape

Free Play Contingent Attention

100

SIB

2 4 6 8 10 12

Contingent Escape

Free Play

Contingent Attention

Sessions

P er

ce n

ta ge

o f

10 -S

ec o

n d

I n

te rv

al s

w it

h P

ro b

le m

B eh

av io

Figure 9 Results of Lorraine’s func- tional analysis. Functional analysis conducted by Corrine M. Murphy and Tabitha Kirby.

other problem behavior occurred. No demands were placed on Lorraine in the group home during the ABC observation. Group home staff rarely placed any demands on Lorraine in an attempt to avoid her problem behavior.

Interpreting Information and Formulating Hypotheses

Some of Lorraine’s problem behaviors seemed to be re- lated to a dosage change in her medication. Because Lor- raine’s physician judged her medication to be at therapeutic levels, a decision was made to analyze the environmental events related to her problem behavior. Observations of problem behavior during the ABC as- sessment were limited because workshop staff placed minimal demands on Lorraine to avoid problem behav- iors. However, Lorraine’s noncompliance reportedly oc- curred when demands were placed on her. Therefore, it was hypothesized that these problem behaviors were maintained by escape from task demands. Aggression oc- curred during the ABC assessment when Lorraine was ignored. Although SIB was not observed during the ABC

assessment, group home staff reported that Lorraine often engaged in SIB during the same situations that evoked aggression. Therefore, it was hypothesized that both ag- gression and SIB were maintained by attention.

Testing Hypotheses

The functional analysis consisted of free play, contingent attention, and contingent escape conditions (see Figure 9). Because the problem behaviors may have served different functions, each problem behavior was coded and graphed separately. Noncompliance occurred most frequently during the contingent escape condition and rarely occurred during the free play or contingent atten- tion conditions. SIB occurred most frequently during the contingent attention condition and rarely occurred during the free play or contingent escape conditions. These data suggested that noncompliance served an escape function, and SIB served an attention function. As is often the case for low-frequency, high-intensity behaviors, it was diffi- cult to form hypotheses about the function of aggression because the behavior occurred rarely in any of the FBA conditions.

532

Functional Behavior Assessment

Developing an Intervention.

Different interventions were developed for the problem behaviors because results of the FBA suggested that the behaviors served different functions. To address non- compliance, Lorraine was taught to request breaks from difficult tasks. Tasks were broken down into very small steps. Lorraine was presented with only one step of a task at a time. Each time a task request was made, Lorraine was reminded that she could request a break (either by saying “Break, please” or by touching a break card). If she requested a break, the task materials were removed for a brief period of time. Then they were presented again. Also, if Lorraine engaged in noncompliance, she was not allowed to escape the task. Instead she was prompted through one step of the task and then another step of the task was presented. Initially, Lorraine was allowed to completely escape the task if she appropriately requested

a break each time the task was presented. Over time, how- ever, she was required to complete increasing amounts of work before a break was allowed.

Intervention for aggression consisted of teaching Lorraine appropriate ways to gain attention (e.g., tapping someone on the arm and saying, “Excuse me”) and teach- ing group home staff to regularly attend to Lorraine when she made such requests. In addition, because her articu- lation was so poor, a picture communication book was created to assist Lorraine in having conversations with others. This communication book could be used to clar- ify words that staff could not understand. Finally, staff were encouraged to ignore Lorraine’s SIB when it did occur. In the past, staff had approached Lorraine and stopped her from engaging in SIB when it occurred. The functional analysis demonstrated that this intervention may have increased the occurrence of SIB, so this prac- tice was discontinued.

Summary Functions of Behavior

1. Many problem behaviors are learned and maintained via positive, negative, and/or automatic reinforcement. In this respect, problem behavior can be said to have a “function” (e.g., to gain access to stimuli or to escape stimuli).

2. The topography, or form, of a behavior often reveals little useful information about the conditions that account for it. Identifying the conditions that account for a behavior (its function) suggests what conditions need to be altered to change the behavior. Assessment of the function of a be- havior can therefore yield useful information with respect to intervention strategies that are likely to be effective.

Role of Functional Behavior Assessment in Intervention and Prevention

3. FBA can lead to effective interventions in at least three ways: (a) it can identify antecedent variables that can be altered to prevent problem behavior, (b) it can identify reinforcement contingencies that can be altered so that problem behavior no longer receives reinforcement, and (c) it can help identify reinforcers for alternative re- placement behaviors.

4. FBA can decrease reliance on default technologies (in- creasingly intrusive, coercive, and punishment-based in- terventions) and contribute to more effective interventions. When FBAs are conducted, reinforcement-based inter- ventions are more likely to be implemented than are in- terventions that include a punishment component.

Overview of FBA Methods

5. FBA methods can be classified into three types: (a) func- tional (experimental) analysis, (b) descriptive assess- ment, and (c) indirect assessment. The methods can be

ordered on a continuum with respect to considerations such as ease of use and the type and precision of infor- mation they yield.

6. Functional analysis involves systematically manipulating environmental events thought to maintain problem be- havior within an experimental design. The primary ad- vantage of functional analysis is its ability to yield a clear demonstration of the variable(s) that relate to the occur- rence of a problem behavior. However, this assessment method requires a certain amount of expertise to imple- ment and interpret.

7. Descriptive assessment involves observation of the prob- lem behavior in relation to events that are not arranged in a systematic manner and includes ABC recording (both continuous and narrative) and scatterplots. The primary advantages to these assessment methodologies are that they are easier to do than functional analyses and they rep- resent contingencies that occur within the individual’s nat- ural routine. Caution must be exercised when interpreting information from descriptive assessments, however, be- cause they can be biased and unreliable, and they provide correlations (as opposed to functional relations).

8. Indirect functional assessment methods use structured in- terviews, checklists, rating scales, or questionnaires to ob- tain information from persons who are familiar with the individual exhibiting the problem behavior (e.g., teachers, parents, caregivers, and/or the individual him- or herself) to identify possible conditions or events in the natural en- vironment that correlate with the problem behavior. Again, these forms of FBA are easy to conduct, but they are lim- ited in their accuracy. As such, they are probably best re- served for hypothesis formulation. Further assessment of these hypotheses is often necessary.

533

Conducting a Functional Behavior Assessment

9. Given the strengths and limitations of the different FBA procedures, FBA can best be viewed as a four-step process:

a. Gather information via indirect and descriptive assess- ment.

b. Interpret information from indirect and descriptive as- sessment and formulate hypotheses about the purpose of problem behavior.

c. Test hypotheses using functional analysis.

d. Develop intervention options based on the function of problem behavior.

10. A brief functional analysis can be used to test hypotheses when time is very limited.

11. When teaching an alternative behavior as a replacement for problem behavior, the replacement behavior should be functionally equivalent to the problem behavior (i.e., it should be reinforced using the same reinforcers that pre- viously maintained the problem behavior).

Case Examples Illustrating the FBA Process

12. A person can display one problem behavior for more than one reason, as shown in Brian’s case example. In such cases, intervention may need to consist of multiple com- ponents to address each function of problem behavior.

13. Functional analyses can be tailored to test specific and/or idiosyncratic hypotheses, as shown in Kaitlyn’s example.

14. Undifferentiated problem behavior during a functional analysis may indicate an automatic reinforcement function and warrants further evaluation, as shown in DeShawn’s case example. In such cases, alternative reinforcers can sometimes be identified and used effectively in an inter- vention to decrease problem behavior and improve adaptive responding.

15. Sometimes a person displays multiple topographies of problem behavior (e.g., self-injury and aggression), with each topography serving a different function, as shown in Lorraine’s example. In such cases, a different intervention for each topography of problem behavior is warranted.

Functional Behavior Assessment

534

535

Verbal Behavior

Key Terms audience autoclitic automatic punishment automatic reinforcement convergent multiple control copying a text divergent multiple control echoic formal similarity

generic (tact) extension impure tact intraverbal listener mand metaphorical (tact) extension metonymical (tact) extension multiple control point-to-point correspondence

private events solistic (tact) extension speaker tact textual transcription verbal behavior verbal operant

Behavior Analyst Certification Board® BCBA® & BCABA® Behavior Analyst Task List,© Third Edition

Content Area 3: Principles, Processes and Concepts.

3-15 Define and provide examples of echoics and imitation.

3-16 Define and provide examples of mands.

3-17 Define and provide examples of tacts.

3-18 Define and provide examples of intraverbals.

Content Area 9: Behavior Change Procedures

9-25 Use language acquisition programs that employ Skinner’s analysis of verbal behavior (i.e., echoics, mands, tacts, intraverbals).

9-26 Use language acquisition and communication training procedures.

This chapter was written by Mark L. Sundberg.

536

Why should applied behavior analysts be con- cerned with verbal behavior? A review of the de- finition of applied behavior analysis can provide

an answer to this question.

Applied behavior analysis is the science in which tactics derived from the principles of behavior are applied to improve socially significant behavior and experimenta- tion is used to identify the variables responsible for be- havior change (p. 20).

Note the point to improve socially significant behavior. The most socially significant aspects of human behavior involve verbal behavior. Language acquisition, social in- teraction, academics, intelligence, understanding, think- ing, problem solving, knowledge, perception, history, science, politics, and religion are all directly relevant to verbal behavior. In addition, many human problems, such as autism, learning disabilities, illiteracy, antisocial be- havior, marital conflicts, aggression, and wars, involve verbal behavior. In short, verbal behavior plays a central role in most of the major aspects of a person’s life, and in the laws, conventions, archives, and activities of a so- ciety. These topics are the main subject topics of most introductory psychology textbooks. These are socially significant behaviors that applied behavior analysts need to address. However, the verbal analysis of these topics has just begun, and a substantial amount of work has yet to be accomplished.

Verbal Behavior and Properties of Language Form and Function of Language

It is important in the study of language to distinguish be- tween the formal and functional properties of language (Skinner, 1957). The formal properties involve the topog- raphy (i.e., form, structure) of the verbal response, whereas the functional properties involve the causes of the response. A complete account of language must con- sider both of these elements.

The field of structural linguistics specializes in the formal description of language. The topography of what is said can be measured by (a) phonemes: the individual speech sounds that comprise a word; (b) morphemes: the units with an individual piece of meaning; (c) lexicon: the total collection of words that make up a given lan- guage; (d) syntax: the organization of words, phrases, or clauses in sentences; (e) grammar: the adherence to es- tablished conventions of a given language; and (f) se- mantics: what words mean (Barry, 1998; Owens, 2001).

The formal description of a language can be accom- plished also by classifying words as nouns, verbs, prepo- sitions, adjectives, adverbs, pronouns, conjunctions, and 1For more detail see Mabry (1994, 1995) and Novak (1994).

articles. Other aspects of a formal description of language include prepositional phrases, clauses, modifiers, gerunds, tense markers, particles, and predicates. Sentences then are made up of the syntactical arrangement of the lexical categories of speech with adherence to the grammatical conventions of a given verbal community. The formal properties of language also include articulation, prosody, intonation, pitch, and emphasis (Barry, 1998).

Language can be formally classified without the pres- ence of a speaker or any knowledge about why the speaker said what he did. Sentences can be analyzed as grammatical or ungrammatical from a text or from a tape recorder. For example, incorrect use of word tense can be identified easily from a recording of a child saying, “Juice all goned.”

A common misconception about Skinner’s analysis of verbal behavior is that he rejected the formal classi- fications of language. However, he did not find fault with classifications or descriptions of the response, but rather with the failure to account for the “causes” or functions of the classifications. The analysis of how and why one says words is typically relegated to the field of psychology combined with linguistics; hence the field of psycholinguistics.

Theories of Language

A wide variety of theories of language attempt to identify the causes of language. These theories can be classified into three separate, but often overlapping, views: biolog- ical, cognitive, and environmental. The basic orientation of the biological theory is that language is a function of physiological processes and functions. Chomsky (1965), for example, maintained that language is innate to hu- mans.1 That is, a human’s language abilities are inher- ited and present at birth.

Perhaps the most widely accepted views of the causes of language are those derived from cognitive psy- chology (e.g., Bloom, 1970; Piaget, 1952). Proponents of the cognitive approach to language propose that lan- guage is controlled by internal processing systems that ac- cept, classify, code, encode, and store verbal information. Spoken and written language is considered to be the struc- ture of thought. Distinguishing between the biological and cognitive views is often difficult; many are mixed (e.g., Pinker, 1994) and invoke cognitive metaphors such as storage and processing as explanations of language behaviors, or interchange the words brain and mind (e.g., Chomsky, 1965).

Verbal Behavior

537

Development of Verbal Behavior Skinner began working on a behavioral analysis of lan- guage in 1934 as a result of a challenge from Alfred North Whitehead,2 which Whitehead made when he was seated next to Skinner at a dinner at the Harvard Society of Fel- lows. Skinner (1957) described the interaction as follows:

We dropped into a discussion of behaviorism which was then still very much an “ism” and of which I was a zeal- ous devotee. Here was an opportunity which I could not overlook to strike a blow for the cause . . . . White- head . . . agreed that science might be successful in accounting for human behavior provided one made an exception of verbal behavior. Here, he insisted some- thing else must be at work. He brought the discussion to a close with a friendly challenge: “Let me see you,” he said, “account for my behavior as I sit here saying ‘No black scorpion is falling upon this table.’” The next morning I drew up the outline of the present study. (p. 457)

It took Skinner 23 years to fill in the details of his outline, which he published in his book Verbal Behavior (1957). The end result was so significant to Skinner (1978) that he believed Verbal Behavior would prove to be his most important work. However, Skinner’s use of the phrase prove to be 20 years after the book was pub- lished indicated that his analysis of verbal behavior had not yet had the impact that he thought it would.

There are several reasons for the slow appreciation of Verbal Behavior. Soon after the book was published, it was met with immediate challenges from the field of lin- guistics and the emerging field of psycholinguistics. Most notably was a review by Noam Chomsky (1959), a young linguist from MIT who had published his own account of language (Chomsky, 1957) the same year Verbal Behav- ior was published. Chomsky maintained that Skinner’s analysis was void of any value. Chomsky criticized every aspect of the analysis, but more so, he criticized the phi- losophy of behaviorism in general. However, a reading of Chomsky’s review will reveal to those who comprehend Verbal Behavior that Chomsky, like many scholars, gravely misunderstood Skinner’s radical behaviorism, which provided the philosophical and epistemological foundations for Verbal Behavior (Catania, 1972; Mac- Corqoudale, 1970).

Skinner never responded to Chomsky’s review, and many felt this lack of response was responsible for the

2Whitehead was perhaps the most prominent philosopher of the time, known best for his landmark three-volume set coauthored with Bertrand Russell titled Principia Mathematica (1910, 1912, 1913).

widely held conclusion that Chomsky’s review was unan- swerable and that Chomsky made valid criticisms. MacCorquodale (1970) pointed out that the reason no one challenged Chomsky’s review was the condescend- ing tone of the review, in addition to the clear misunder- standings of Skinner’s behaviorism.

Skinner was not at all surprised by this reaction from linguists because of their emphasis on the structure of language rather than its function. More recently however, a favorable review of Skinner’s book from within the field of linguistics was published, recognizing that Skinner has changed the history of linguistics (Andresen, 1991).

Although Skinner anticipated criticism from outside the field of behavior analysis, he probably did not expect the general disinterest and often outspoken negative re- action to Verbal Behavior from within the field. A num- ber of behaviorists have examined this issue and have collectively provided a list of reasons behavior analysts did not immediately embrace Verbal Behavior (e.g., Esh- leman, 1991; Michael, 1984; E. Vargas, 1986). Perhaps most troublesome to the behavior analysts of the time was that Verbal Behavior was speculative and did not contain experimental data (Salzinger, 1978).

The lack of research on verbal behavior continued to concern behavior analysts well into the 1980s (e.g., McPherson, Bonem, Green, & Osborne, 1984). However, this situation now appears to be changing, and a number of advances in research and applications directly relate to Verbal Behavior (Eshleman, 2004; Sundberg, 1991, 1998). Many of these advances are published in the jour- nal, The Analysis of Verbal Behavior.

Defining Verbal Behavior Skinner (1957) proposed that language is learned behav- ior, and that it is acquired, extended, and maintained by the same types of environmental variables and principles that control nonlanguage behavior (e.g., stimulus con- trol, motivating operations, reinforcement, extinction). He defined verbal behavior as behavior that is reinforced through the mediation of another person’s behavior. For example, the verbal response “Open the door” can pro- duce the reinforcer of an open door mediated through the behavior of a listener. This reinforcer is indirectly ob- tained, but is the same reinforcer that could be obtained nonverbally by opening the door.

Skinner defined verbal behavior by the function of the response, rather than by its form. Thus, any response form can become verbal based on Skinner’s functional definition. For example, the early differential crying of a 2-month-old infant may be verbal, as would other re- sponses such as pointing, clapping for attention, gestures

Verbal Behavior

538

such as waving one’s arm for attention, writing, or typ- ing. In other words, verbal behavior involves a social in- teraction between a speaker and a listener.

Speaker and Listener

The definition of verbal behavior makes a clear distinc- tion between the behavior of the speaker and that of the listener. Verbal behavior involves social interactions be- tween speakers and listeners, whereby speakers gain ac- cess to reinforcement and control their environment through the behavior of listeners. In contrast with most approaches to language, Skinner’s verbal behavior is pri- marily concerned with the behavior of the speaker. He avoided terms such as expressive language and receptive language because of the implication that these are merely different manifestations of the same underlying cogni- tive processes.

The listener must learn how to reinforce the speaker’s verbal behavior, meaning that listeners are taught to re- spond to words and interact with speakers. It is important to teach a child to react appropriately to the verbal stim- uli provided by speakers, and to behave verbally as a speaker. These are different functions, however. In some cases learning one type of behavior (i.e., speaker or lis- tener) facilitates learning another, but this must also be understood in terms of motivating operations, antecedent stimuli, responses, and consequences rather than in terms of learning the meanings of words as a listener and then using the words in various ways as a speaker.

Verbal Behavior: A Technical Term

In searching for what to call the subject matter of his analysis of language, Skinner wanted a term that (a) em- phasized the individual speaker, (b) referred to behavior that was selected and maintained by consequences; and (c) was relatively unfamiliar in the professions of speech and language. He selected the term verbal behavior. How- ever, in recent years verbal behavior has acquired a new meaning, independent from Skinner’s usage. In the field of

speech pathology, verbal behavior has become synony- mous with vocal behavior. Also, in psychology, the term nonverbal communication, which became popular in the 1970s, was contrasted with the term verbal behavior, im- plying that verbal behavior was vocal communication and nonverbal behavior was nonvocal communication. The term verbal has also been contrasted with quantitative as in GRE and SAT tests for college admissions. This dis- tinction suggests that mathematical behavior is not verbal. However, according to Skinner’s definition, much of math- ematical behavior is verbal behavior. Noting that verbal behavior includes vocal-verbal behavior and nonvocal- verbal behavior is sometimes confusing for those learn- ing to use Skinner’s analysis.

Unit of Analysis

The unit of analysis of verbal behavior is the functional relation between a type of responding and the same in- dependent variables that control nonverbal behavior, namely (a) motivating variables, (b) discriminative stim- uli, and (c) consequences. Skinner (1957) referred to this unit as a verbal operant, with operant implying a type or class of behavior as distinct from a particular response instance; and he referred to a set of such units of a par- ticular person as a verbal repertoire. The verbal reper- toire can be contrasted with the units in linguistics that consists of words, phrases, sentences, and the mean length of utterances.

Elementary Verbal Operants Skinner (1957) identified six elementary verbal oper- ants: mand, tact, echoic, intraverbal, textual, and tran- scription. He also included audience relation and copying a text as separate relations, but in this discussion the au- dience (or the listener) will be treated independently and copying a text will be considered a type of echoic be- havior. Table 1 presents plain English descriptions of

Verbal Behavior

Table 1 Plain English Definitions of Skinner’s Six Elementary Verbal Operants

Mand Asking for reinforcers that you want. Saying shoe because you want a shoe.

Tact Naming or identifying objects, actions, events, etc. Saying shoe because you see a shoe.

Echoic Repeating what is heard. Saying shoe after someone else says shoe.

Intraverbal Answering questions or having conversations in which your words are controlled by other words. Saying shoe when someone else says, What do you wear on your feet?

Textual Reading written words. Saying shoe because you see the written word shoe.

Transcription Writing and spelling words spoken to you. Writing shoe because you hear shoe spoken.

539

these terms. Technical definitions and examples of each elementary verbal operant are provided in the following sections.

Mand

The mand is a type of verbal operant in which a speaker asks for (or states, demands, implies, etc.) what he needs or wants. For example, the behavior of asking for direc- tions when lost is a mand. Skinner (1957) selected the term mand for this type of verbal relation because the term is conveniently brief and is similar to the plain Eng- lish words command, demand, and countermand. The mand is a verbal operant for which the form of the re- sponse is under the functional control of motivating op- erations (MOs) and specific reinforcement (see Table 2). For example, food deprivation will (a) make food effec- tive as reinforcement and (b) evoke behavior such as the mand “cookie” if this behavior has produced cookies in the past.

The specific reinforcement that strengthens a mand is directly related to the relevant MO. For example, if there is an MO for physical contact with one’s mother, the specific reinforcement that is established is physical con- tact. The response form may occur in several topograph- ical variations such as crying, pushing a sibling, reaching up, and saying “hug.” All of these behaviors could be mands for physical contact if functional relations exist among the MO, the response, and the specific reinforce- ment history. However, the response form alone is insuf- ficient for the classification of a mand, or any other verbal operant. For example, crying could also be a respondent behavior if it were elicited by a conditioned or uncondi- tioned stimulus.

Mands are very important for the early develop- ment of language and for the day-to-day verbal inter- actions of children and adults. Mands are the first verbal operant acquired by a human child (Bijou & Baer, 1965; Novak, 1996). These early mands usually occur in the form of differential crying when a child is hungry, tired, in pain, cold, or afraid; or wants toys, at- tention, help, movement of objects and people, direc- tions, or the removal of aversive stimuli. Typically developing children soon learn to replace crying with words and signs or other standard forms of communi- cation. Manding not only lets children control the de- livery of reinforcers, but it begins to establish the speaker and listener roles that are essential for further verbal development.

Skinner (1957) pointed out that the mand is the only type of verbal behavior that directly benefits the speaker, meaning that the mand gets the speaker reinforcers such as edibles, toys, attention, or the removal of aversive stim-

uli. As a result, mands often become strong forms of ver- bal behavior because of specific reinforcement, and this reinforcement often satisfies an immediate deprivation condition or removes some aversive stimulus. For exam- ple, young children often engage in a very high rate of manding because of its effects on listeners. In addition, much of the problem behaviors of children who have weak, delayed, or defective verbal repertoires may be mands (e.g., Carr & Durand, 1985). Eventually, a child learns to mand for verbal information with who, what, and where questions, and the acquisition of new verbal behavior accelerates rapidly (Brown, Cazden, & Bellugi, 1969). Ultimately, mands become quite complex and play a critical role in social interaction, conversation, academic behavior, employment, and virtually every aspect of human behavior.

Tact

The tact is a type of verbal operant in which a speaker names things and actions that the speaker has direct contact with through any of the sense modes. For ex- ample, a child saying “car” because he sees a car is a tact. Skinner (1957) selected the term tact because it suggests making contact with the physical environment. The tact is a verbal operant under the functional con- trol of a nonverbal discriminative stimulus, and it pro- duces generalized conditioned reinforcement (see Table 2). A nonverbal stimulus becomes a discriminative stimulus (SD) with the process of discrimination train- ing. For example, a shoe may not function as an SD for the verbal response “shoe” until after saying “shoe” in the presence of a shoe produces differential reinforce- ment.

A wide variety of nonverbal stimuli evoke tact re- lations. For example, a cake produces nonverbal visual, tactile, olfactory, and gustatory stimuli, any or all of which can become SDs for the tact “cake.” Nonverbal stimuli can be, for example, static (nouns), transitory (verbs), relations between objects (prepositions), prop- erties of objects (adjectives), or properties of actions (adverbs); that is, nonverbal stimuli can be as simple as a shoe, or as complex as a cancerous cell. A stimulus configuration may have multiple nonverbal properties, and a response may be under the control of those mul- tiple properties, as in the tact “The red truck is on the little table.” Nonverbal stimuli may be observable or unobservable (e.g., pain), subtle or salient (e.g., neon lights), relational to other nonverbal stimuli (e.g., size), and so on. Given the variation and ubiquity of nonver- bal stimuli, it is no surprise that the tact is a primary topic in the study of language.

Verbal Behavior

540

Echoic

The echoic is a type of verbal operant that occurs when a speaker repeats the verbal behavior of another speaker. For example, a child saying “cookie” after hearing the word spoken by her mother is echoic. Repeating the words, phrases, and vocal behavior of others, which is common in day-to-day discourse, is echoic also. The echoic operant is controlled by a verbal discriminative stimulus that has point-to-point correspondence and for- mal similarity with the response (Michael, 1982) (see Table 2).

Point-to-point correspondence between the stim- ulus and the response or response product occurs when the beginning, middle, and end of the verbal stimulus matches the beginning, middle, and end of the response. Formal similarity occurs when the controlling an- tecedent stimulus and the response or response product (a) share the same sense mode (e.g., both stimulus and response are visual, auditory, or tactile) and (b) physi- cally resemble each other (Michael, 1982). In the echoic relation the stimulus is auditory and the response pro- duces an auditory product (echoing what one hears), and the stimulus and the response physically resemble each other.

Echoic behavior produces generalized conditioned reinforcement such as praise and attention. The ability to echo the phonemes and words of others is essential for learning to identify objects and actions. A parent might say, “That’s a bear, can you say bear?” If the child can re- spond “bear,” then the parent says “Right!” Eventually, the child learns to name a bear without the echoic prompt. This often occurs in a few trials. For example, if a child can say “bear” (or a reasonable approximation) after a parent says “bear,” then it becomes possible to teach the child to say “bear” in the presence of a picture of a bear or a bear at the zoo. The echoic repertoire is very impor- tant for teaching language to children with language de- lays, and it serves a critical role in the process of teaching more complex verbal skills (e.g., Lovaas, 1977; Sund- berg & Partington, 1998).

Motor imitations can have the same verbal prop- erties as echoic behavior as demonstrated by their role in the acquisition of sign language by children who are deaf. For example, a child may learn to imitate the sign for cookie first, and then mand for cookie without an imitative prompt. Imitation is also critical for teaching sign language to hearing children who are nonvocal. For the many children who do not have an adequate echoic repertoire for vocal language instruction, time is spent teaching echoic behavior rather than more useful types of verbal behavior. A strong imitative repertoire permits a teacher to use sign language immediately to

instruct more advanced forms of language (e.g., mands, tacts, and intraverbals). This allows a child to learn quickly to communicate with others without using in- appropriate behavior (e.g., a tantrum) to get what he wants.

Skinner also presented copying a text as a type of verbal behavior in which a written verbal stimulus has point-to-point correspondence and formal similarity with a written verbal response. Because this relation has the same defining features as echoic and imitation as it relates to sign language, the three will be treated as one cate- gory, echoic.

Intraverbal

The intraverbal is a type of verbal operant in which a speaker differentially responds to the verbal behavior of others. For example, saying “the Buckeyes” as a result of hearing someone else say “Who won the game Satur- day?” is intraverbal behavior. Typically developing chil- dren emit a high frequency of intraverbal responses in the form of singing songs, telling stories, describing ac- tivities, and explaining problems. Intraverbal responses are also important components of many normal intellec- tual repertoires, such as saying “Sacramento” as a result of hearing “What is the capital of California?”; saying “sixty-four” as a result of hearing “eight times eight”; or saying “antecedent, behavior, and consequence” when asked, “What is the three-term contingency?” The in- traverbal repertoires of typical adult speakers include hundreds of thousands of such relations.

The intraverbal operant occurs when a verbal dis- criminative stimulus evokes a verbal response that does not have point-to-point correspondence with the verbal stimulus (Skinner, 1957). That is, the verbal stimulus and the verbal response do not match each other, as they do in the echoic and textual relations. Like all verbal oper- ants except the mand, the intraverbal produces general- ized conditioned reinforcement. For example, in the educational context, the reinforcement for correct an- swers usually involves some form of generalized condi- tioned reinforcement such as “Right!” or points, or the opportunity to move to the next problem or item (see Table 2).

An intraverbal repertoire facilitates the acquisition of other verbal and nonverbal behavior. Intraverbal be- havior prepares a speaker to respond rapidly and accu- rately with respect to further stimulation, and plays an important role in continuing a conversation. For exam- ple, a child hears an adult speaker say “farm” in some context. If the stimulus farm evokes several relevant in- traverbal responses, such as “barn,” “cow,” “rooster,” or “horse,” then a child is better able to react to other parts

Verbal Behavior

541

of an adult’s verbal behavior that may be related to a re- cent trip to a farm. One might say that the child is now thinking about farms and now has relevant verbal re- sponses at strength for further responses to the adult’s verbal behavior. An intraverbal stimulus probes the lis- tener’s repertoire and gets it ready for further stimulation. Collectively, mands, tacts, and intraverbals contribute to a conversation in the following ways: (a) A mand reper- toire allows a speaker to ask questions, (b) a tact reper- toire permits verbal behavior about an object or event that is actually present, and (c) an intraverbal repertoire allows a speaker to answer questions and to talk about (and think about) objects and events that are not physi- cally present.

Textual

Textual behavior (Skinner, 1957) is reading, without any implications that the reader understands what is being read. Understanding what is read usually involves other verbal and nonverbal operants such as intraverbal behav- ior and receptive language (e.g., following instructions, compliance). For example, saying “shoe” upon seeing the written word shoe is textual behavior. Understanding that shoes go on a person’s feet is not textual. Understanding is typically identified as reading comprehension. Skinner chose the term textual because the term reading refers to many processes at the same time.

The textual operant has point-to-point correspon- dence, but not formal similarity, between the stimulus and the response product. For example, (a) the verbal stimuli are visual or tactual (i.e., in one modality) and the response is auditory (i.e., another modality) and (b) the auditory response matches the visual or tactual stimuli. Table 2 presents a diagram of the textual relation.

Textuals and echoics are similar in three respects: (a) They both produce generalized conditioned rein- forcement, (b) both are controlled by antecedent verbal stimuli, and (c) there is point-to-point correspondence between the antecedent stimulus and the response. The important difference between textuals and echoics is that the response product of textual behavior (e.g., the spoken word) is not similar to its controlling stimulus (e.g., the written word evokes a spoken response or au- ditory response product). The textual operant does not have formal similarity, meaning that the SDs are not in the same sense mode and do not physically resemble the textual response. Words are visual and comprised of individual letters, whereas the reading response pro- duces an auditory response product (which often is covert) comprising phonemes. The echoic response product, however, does have formal similarity with its controlling verbal stimulus.

Transcription

Transcription consists of writing and spelling words that are spoken (Skinner, 1957). Skinner also referred to this behavior as taking dictation, with the key repertoires in- volving not only the manual production of letters, but also accurate spelling of the spoken word. In technical terms, transcription is a type of verbal behavior in which a spoken verbal stimulus controls a written, typed, or finger-spelled response. Like the textual operant, there is point-to-point correspondence between the stimulus and the response product, but no formal similarity (see Table 2). For example, when asked to spell the spoken word “hat,” a response h-a-t is a transcription. The stimulus and the response product have point-to-point correspon- dence, but they are not in the same sense mode or do not physically resemble each other. Spelling English words

Table 2 Antecedent and Consequent Controlling Variables for Six Elementary Verbal Operants

Antecedent Variables Response Consequence

Motivating operations (4 hours without water) Mand (“water, please”) Specific reinforcement (glass of water)

Nonverbal stimulus (see toy truck) Tact (“truck”) Generalized conditioned reinforcement (GCSR) (praise and approval)

Verbal stimulus with point-to-point correspondence Echoic (say “book”) GCSR and formal similarity (hear “book”)

Verbal stimulus without point-to point correspon- Intraverbal (say “dogs”) GCSR dence or formal similarity (hear “cats and . . .”)

Verbal stimulus with point-to-point correspondence, Textual (say “apple”) GCSR without formal similarity (see apple written)

Verbal stimulus with point-to-point correspondence, Transcription (write apple) GCSR without formal similarity (hear “apple”)

Verbal Behavior

542

is a difficult repertoire to acquire. Because many words in the English language are not spelled the way they sound, shaping an appropriate discriminative repertoire is often difficult.

Role of the Listener Skinner’s analysis of verbal behavior focuses on the speaker, whereas most linguists and psycholinguistic ac- counts of language emphasize the listener. Skinner sug- gested that the listener’s role is less significant than typically assumed because much of what is often de- scribed as listener behavior (e.g., thinking, understanding) is more correctly classified as speaker behavior. It is just that often the speaker and listener reside within the same skin (as we will see in the following discussion).

What role does the listener then play in Skinner’s account of language? In his analysis of listener behav- ior, Skinner pointed out that a verbal episode requires a speaker and a listener. The listener not only plays a critical role as a mediator of reinforcement for the speaker’s behavior, but also becomes a discriminative stimulus for the speaker’s behavior. In functioning as a discriminative stimulus, the listener is an audience for verbal behavior. “An audience, then, is a discrimina- tive stimulus in the presence of which verbal behavior is characteristically reinforced and in the presence of which, therefore, it is characteristically strong” (Skin- ner, 1957, p. 172). When Skinner (1978) wrote “very little of the behavior of the listener is worth distin- guishing as verbal” (p. 122), he was referring to when the listener serves as a discriminative stimulus in the role of an audience.

A listener functions in additional roles, other than as a mediator of reinforcement and a discriminative stim- ulus. For example, verbal behavior functions as dis- criminative stimuli (i.e., stimulus control) when a speaker talks to a listener. The question is, What are the effects of verbal behavior on listener behavior? A verbal discriminative stimulus may evoke echoic, textual, tran- scription, or intraverbal operants of a listener. The lis- tener becomes a speaker when this occurs. This is Skinner’s point: The speaker and listener can and often do reside within the same skin, meaning that a listener behaves simultaneously as a speaker. The most signifi- cant and complex responses to verbal stimuli occur when they evoke covert intraverbal behavior from a listener who becomes a speaker and functions as her own audi- ence. For example, a speaker’s verbal discriminative stimuli related to Pavlov’s work on respondent condi- tioning, such as, “What was Pavlov’s technique?” may evoke a listener’s covert intraverbal behavior such as

3In three of his books, Skinner devoted a full chapter to the topic of think- ing: Science and Human Behavior (1953, Chapter 16), Verbal Behavior (1957, Chapter 19), and About Behaviorism (1974, Chapter 7) with several sections dedicated to the topic of understanding (e.g., Verbal Behavior, pp. 277–280; About Behaviorism, pp. 141–142). A behavior analysis of thinking and understanding involves, in large part, situations in which both listener and speaker reside within the same skin.

thinking, “He paired the sound of a metronome with meat powder.”3

Verbal stimulus control may also evoke a listener’s nonverbal behavior. For example, when someone says “Shut the door,” the behavior of shutting a door is non- verbal, but shutting the door is evoked by verbal stim- uli. Skinner (1957) identified this type of listener behavior as understanding. “The listener can be said to understand a speaker if he simply behaves in an appro- priate fashion” (p. 277).

Verbal stimuli can become quite complex because separating the verbal and nonverbal behaviors of the lis- tener is difficult (Parrott, 1984; Schoenberger, 1990, 1991). For example, in following a directive to buy a cer- tain type and style of pipe fitting at the hardware store, success will involve both nonverbal behavior such as dis- criminating among pipe fittings and verbal behavior such as self-echoic prompts (e.g., “I need a three-quarter-inch fitting, three-quarter-inch”), tacts of the fittings (e.g., “This looks like three-quarter-inch”), and mands for in- formation (e.g., “Can you tell me if this will fit a three- quarter-inch pipe?”).

Identifying Verbal Operants The same word (i.e., topography or form of the behavior) can appear in the definitions of all of the elementary ver- bal operants because controlling variables define verbal operants, not the form of the verbal stimuli. Verbal be- havior is not classified or defined by its topography or form (i.e., by the words themselves). The classification of verbal operants can be accomplished by asking a series of questions regarding the relevant controlling variables that evoke a specific response form (see Figure 1). A sam- ple of verbal behavior classification exercise is presented in Table 3

1. Does an MO control the response form? If yes, then the operant is at least part mand.

2. Does an SD control the response form? If yes, then:

3. Is the SD nonverbal? If yes, then the operant is at least part tact.

4. Is the SD verbal? If yes, then:

Verbal Behavior

543

5. Is there point-to-point correspondence between the verbal SD and the response? If not, then the operant is at least part intraverbal. If there is point-to-point correspondence, then:

6. Is there formal similarity between the verbal SD

and the response. If yes, then the operant must be echoic, imitative, or copying a text). If not, then the operant must be textual or transcription.

Analyzing Complex Verbal Behavior Analysis of more complex verbal behavior includes au- tomatic reinforcement, tact extensions (generalization), and private events. These topics are presented in the fol- lowing sections.

Automatic Reinforcement

A common misconception about reinforcement in the context of verbal behavior is that it occurs only when the listener mediates the reinforcer. When behavior occurs without the apparent delivery of reinforcement, it is often assumed that higher mental processes are at work (e.g., Brown, 1973; Neisser, 1976). Intermittent reinforcement can explain some behavior that occurs in the absence of an observed consequence, but not all such behavior. Some behavior is strengthened or weakened, not by external consequences, but by its response products, which have reinforcing or punishing effects. Skinner used the terms automatic reinforcement and automatic punishment in a number of his writings simply to indicate that an ef- fective consequence can occur without someone provid- ing it (cf., Vaughan & Michael, 1982).

Verbal Behavior

Figure 1 Verbal behavior classification chart.

Controlling variables Verbal relation

UMO/CMO Yes Mand

Nonverbal discrimination stimulus Yes Tact

Verbal discrimination stimulus Yes Point-to-point correspondence No Intraverbal

Yes

Formal similarity Yes Echoic Imitation

Copy a text

No Textual Transcription

544

Table 3 Verbal Behavior Classification Exercises

As a result of . . . One has a tendency to . . . This is a(n) . . .

1. seeing a dog say “dog” __________

2. hearing an airplane say “airplane” __________

3. wanting a drink say “water” __________

4. hearing “How are you?” say “I’m fine” __________

5. smelling cookies baking say “cookies” __________

6. tasting soup say “pass the salt” __________

7. hearing “book” write book __________

8. hearing “book” sign “book” __________

9. hearing “book” say “book” __________

10. hearing “book” say “read” __________

11. hearing “book” sign “read” __________

12. hearing “book” finger-spell “book” __________

13. seeing a book write book __________

14. wanting a book write book __________

15. signing “book” write book __________

16. hearing “what color is this?” say “red” __________

17. seeing a dog on the table say “get off” __________

18. seeing stop written hit the brakes __________

19. hearing “Skinner” write behavior __________

20. smelling smoke say “fire” __________

21. being hungry go to a store __________

22. seeing apple written sign “apple” __________

23. seeing 5 say “five” __________

24. wanting things say “thanks” __________

25. hearing “write your name” write your name __________

26. hearing “run” finger-spell “run” __________

27. seeing “home” signed sign “Battle Creek” __________

28. hearing a phone ring say “phone” __________

29. smelling a skunk say “skunk” __________

30. hearing “table” say “mesa” __________

31. being happy smile __________

32. hoping a pilot sees it writing SOS __________

33. wanting blue say “blue” __________

34. hearing “Red, white, and” say “blue” __________

35. tasting candy say “mmmm” __________

Provide examples of verbal behavior.

36. Give an example of a mand involving an adjective.

37. Give an example of a tact of a smell.

38. Give an example of a response that is part mand and part tact.

39. Give an example of a response that is part tact and part intraverbal.

40. Give an example of a tact involving multiple responses.

41. Give an example of an intraverbal using writing.

42. Give an example of receptive language using sign language.

Answers to verbal behavior classification exercises in Table 3: 1. T; 2. T; 3. M; 4. IV; 5. T/M; 6. M; 7. TR; 8. IV; 9. E; 10. IV; 11. IV; 12. TR; 13. T; 14. M; 15. IV; 16. IV/T; 17. M; 18. NV; 19. IV; 20. T/M; 21. NV; 22. IV; 23. IV; 24. M; 25. IV; 26. TR; 27. IV; 28. T/M; 29. T; 30. IV; 31. NV; 32. M; 33. M; 34. IV; 35. T; 36. “I want the red one.”; 37. “Someone is smoking.”; 38. “My throat is dry.”; 39. “Macy’s.” When asked “Where did you buy that?”; 40. “That’s a big burger!”; 41. A response to an e-mail; 42. Stopping when someone signs “stop.”

545

Verbal behavior can produce automatic reinforce- ment, which has a significant role in the acquisition and maintenance of verbal behavior. For example, automatic reinforcement may explain why a typically developing infant engages in extensive babbling without the appar- ent delivery of reinforcement. Skinner (1957) pointed out that the exploratory vocal behavior of young children could produce automatic reinforcement when those ex- ploratory sounds match the speech sounds of parents, caregivers, and others.

Skinner (1957) described a two-stage conditioning history in establishing vocal responses as automatic re- inforcers. First, a neutral verbal stimulus is paired with an existing form of conditioned or unconditioned rein- forcement. For example, a mother’s voice is paired with conditions such as presenting food and warmth and re- moving aversive stimuli (e.g., medication on a diaper rash). As a result, the mother’s voice, a previously neu- tral stimulus, becomes a conditioned reinforcer. The mother’s voice will now strengthen whatever behavior precedes it. Second, a child’s vocal response as either random muscle movement of the vocal cords or reflexive behavior produces an auditory response that on occasion may sound somewhat like the mother’s words, intona- tions, and vocal pitches. Thus, a vocal response can func- tion as reinforcement by automatically increasing the frequency of a child’s vocal behavior.4

Automatic reinforcement also plays an important role in the development of more complex aspects of verbal behavior, such as the acquisition of syntax and gram- matical conventions. For example, Donahoe and Palmer (1994) and Palmer (1996, 1998) suggested that a child’s use of grammar produces automatic reinforcement when it sounds like the grammar used by others in the envi- ronment, but is automatically punished when it sounds odd or unusual. Palmer (1996) referred to this as achieving parity.

The stimulus conditions that evoke automatically re- inforced behavior may be encountered everywhere be- cause each time a response is automatically reinforced it may alter the evocative effect of any stimulus condition that might be present. For example, a person may persist in singing or humming a theme song from a movie while driving home from the movie because of the two-stage conditioning process described previously. However, the song may be periodically evoked several hours, or even days, after the movie because each time the song is re-

4Miller and Dollard (1941) were perhaps the first to suggest that a process like automatic reinforcement might be partially responsible for an infant’s high rate of babbling. Since then, many others have discussed and re- searched the role of automatic reinforcement in language acquisition (e.g., Bijou & Baer, 1965; Braine, 1963; Miguel, Carr, & Michael, 2002; Mowrer, 1950; Novak, 1996; Osgood, 1953; Smith, Michael, & Sundberg, 1996; Spradlin, 1966; Staats & Staats, 1963; Sundberg, Michael, Partington, & Sundberg, 1996; Vaughan & Michael, 1982; Yoon & Bennett, 2000).

peated, a new stimulus such as a traffic light, a street cor- ner, or a neon sign might acquire some degree of stimu- lus control. The next time the person comes in contact with, say, a red light, there could be some tendency to sing or hum the song. This effect might explain what is often termed delayed echolalia observed with children who have autism (Sundberg & Partington, 1998). At pre- sent, however, there have been no empirical investiga- tions of the stimulus control involved in automatic consequences, although it certainly seems like an inter- esting and important area of study.

Tact Extensions

Contingencies that establish stimulus and response classes and generalization allow a variety of novel and different discriminative stimuli to evoke verbal behavior. Skinner (1957) said it this way:

[A] verbal repertoire is not like a passenger list on a ship or plane, in which one name corresponds to one person with no one omitted or named twice. Stimulus control is by no means so precise. If a response is reinforced upon a given occasion or class of occasions, any feature of that occasion or common to that class appears to gain some measure of control. A novel stimulus possessing one such feature may evoke a response. There are sev- eral ways in which a novel stimulus may resemble a stimulus previously present when a response was rein- forced, and hence there are several types of what we may call “extended tacts” (p. 91).

Skinner (1957) distinguished four types of extended tacts: generic, metaphorical, metonymical, and solistic. The distinction is based on the degree to which a novel stimulus shares the relevant or irrelevant features of the original stimulus.

Generic Extension

In generic extension, the novel stimulus shares all of the relevant or defining features of the original stimulus. For example, a speaker who learns to tact “car” in the pres- ence of a white Pontiac Grand Am emits the tact “car” in the presence of a novel blue Mazda RX-7. A generic tact extension is evoked by simple stimulus generalization.

Metaphorical Extension

In metaphorical extension, the novel stimulus shares some but not all of the relevant features associated with the original stimulus. For example, Romeo was experi- encing a beautiful, sunny, warm day, and the exceptional weather elicited respondent behaviors (e.g., good feel- ings). When Romeo saw Juliet, whose presence elicited

Verbal Behavior

546

similar respondent behaviors like the sunny day, he said, “Juliet is like the sun.” The sun and Juliet evoked a sim- ilar effect on Romeo, controlling the metaphorical tact extension “Juliet is like the sun.”

Metonymical Extension

Metonymical extensions are verbal responses to novel stimuli that share none of the relevant features of the orig- inal stimulus configuration, but some irrelevant but re- lated feature has acquired stimulus control. Simply, one word substitutes for another in metonymical tact exten- sions, meaning that a part is used for a whole. As exam- ples: Saying “car” when shown a picture of a garage, or saying “the White House requested” in place of “Presi- dent Lincoln requested.”

Solistic Extension

Solistic extensions occur when a stimulus property that is only indirectly related to the tact relation evokes sub- standard verbal behavior such as malaprops. For instance, using the solistic tact extension, a person may say “You read good” instead of “You read well.” Saying “car” when referring to the driver of the car is a solistic tact extension.

Private Events

In 1945 Skinner first described radical behaviorism, his philosophy. At the core of radical behaviorism is the analysis of private stimuli (see also Skinner, 1953, 1974). Verbal behavior under the control of private stimuli has been a main topic of theoretical and philosophical analy- ses of behavior ever since. In 1957 Skinner stated, “A small but important part of the universe is enclosed within the skin of each individual. . . . It does not follow that . . . it is any way unlike the world outside the skin or inside another’s skin” (p. 130).

A significant amount of day-to-day verbal behavior is controlled in part by private events. What is commonly referred to as thinking involves overt stimulus control and private events (e.g., covert stimulus control). The analysis of private stimulation and how it acquires stim- ulus control is complex because of two problems: (a) The participant can directly observe the private stimuli, but the applied behavior analyst cannot (a limiting factor in the prediction and control of behavior), and (b) private stimulus control of verbal episodes in the natural envi- ronment will likely remain private, no matter how sensi- tive instruments will become in detecting private stimuli and behaviors. Skinner (1957) identified four ways that caregivers teach young persons to tact their private stim-

ulation: Public accompaniment, collateral responses, common properties, and response reduction.

Public Accompaniment

Public accompaniment occurs when an observable stim- ulus accompanies a private stimulus. For example, a fa- ther may observe a child bump his head on a table while chasing a ball. The public stimuli are available to the fa- ther, but not the private and more salient painful stimuli experienced by the child. The father can assume that the child is experiencing pain because of his own history of bumping objects, and may say “Ouch,” or “You hurt your- self.” In this way, the father is using the bump (observable stimulus) as an opportunity to develop verbal behavior under the stimulus control of a private stimulus. This may occur with an echoic to the private event; later, the stim- ulus control transfers to the private stimuli. Specifically, the child may echo the father’s “ouch” while the painful stimuli are present, and quickly (depending on the child’s history of echoic to tact transfer) the painful stimuli alone evokes the tact “ouch.”

Collateral Responses

Caregivers also teach young persons to tact their private stimuli by using collateral responses (i.e., observable be- havior) that reliably occur with private stimuli. For ex- ample, the father may not observe the child bump his head, but may see the child holding his head and crying. These collateral behaviors inform the father that a painful stimulus is present. The same training procedures with public accompaniment can be used with collateral re- sponses. Because the painful private stimuli are salient, only one trial may be needed for the private stimuli to acquire stimulus control of the tact relation.

Parents should use public accompaniment and col- lateral responses during the beginning stages of tact train- ing. However, even after developing a repertoire of tacting private events, a parent or listener will have difficulty confirming the actual presence of the private event as in “My stomach hurts,” or “I have a headache now.”

Also, learning to tact private behaviors is acquired probably with public accompaniment and collateral re- sponses; for example, private stimuli that evoke private emotions (behaviors) that we tact with words such as happiness, sadness, fear, and being upset. Learning to tact such private events is difficult if the private stimula- tion is not present during training. For example, proce- dures that use pictures of people smiling and frowning (i.e., public stimuli) for teaching children to tact emo- tions will be less effective than procedures that use vari- ables to evoke pleasure or displeasure (i.e., private stimuli) during training.

Verbal Behavior

547

Common Properties

The two procedures described earlier use public stimuli to establish tacting of private events. Common proper- ties also involve public stimuli, but in a different way. A speaker may learn to tact temporal, geometrical, or de- scriptive properties of objects and then generalize those tact relations to private stimuli. As Skinner (1957) noted, “[M]ost of the vocabulary of emotion is metaphorical in nature. When we describe internal states as ‘agitated,’ ‘depressed,’ or ‘ebullient,’ certain geometrical, temporal, and intensive properties have produced a metaphorical extension of responses” (p. 132). Much of our verbal be- havior regarding emotional events is acquired through this type of stimulus generalization.

Response Reduction

Most speakers learn to tact features of their own bodies such as movements and positions. The kinesthetic stim- uli arising from the movement and positions can acquire control over the verbal responses. When movements shrink in size (become covert), the kinesthetic stimuli may remain sufficiently similar to those resulting from the overt movements that the learner’s tact occurs as an instance of stimulus generalization. For example, a child can report imagining swimming, or can report self-talk about a planned conversation with someone, or can report thinking of asking for a new toy (Michael & Sundberg, 2003). Responses produced by private covert verbal be- havior can evoke other verbal behavior and will be dis- cussed later in further detail.

Multiple Control All verbal behavior contains multiple functional relations among antecedents, behavior, and consequences. “[A]ny sample of verbal behavior will be a function of many variables operating at the same time” (Skinner, 1957, p. 228). The functional units of mands, tacts, echoics, in- traverbals, and textual relations form the foundation of a verbal behavior analysis. A working knowledge of these functional units is essential for understanding the analy- sis of multiple control and complex verbal behavior.

Convergent Multiple Control

Michael (2003) used the term convergent multiple con- trol to identify when the occurrence of a single verbal response is a function of more than one variable. The task of an applied behavior analyst is to identify the relevant sources that control an instance of verbal behavior. For example, saying “Why did the United States enter World War II?” may be evoked by (a) MOs (making it part

mand), (b) verbal discriminative stimuli (making it part echoic, intraverbal, or textual), (c) nonverbal stimuli (making it part tact), or the presence of a specific audi- ence. For example, it’s possible that an audience with contempt for war (i.e., the MO) evoked a mand, “Why did the United States enter World War II?” The question may be more a function of this variable than an intraverbal re- lated to the ongoing conversation or the nonverbal or tex- tual stimuli that may be present in the room. On the other hand, the speaker may have no strong MO for the an- swer, but asks the question because of its relation to MOs for social reinforcement related to political involvement.

Divergent Multiple Control

Multiple control also occurs when a single antecedent variable affects the strength of many responses. For ex- ample, a single word (e.g., football) will evoke a variety of intraverbal responses from different people, and from the same person at different times. Michael (2003) used the term divergent multiple control to identify this type of control. Divergent multiple control can occur also with mand and tact relations. A single MO may strengthen a variety of responses, such as food deprivation strength- ening the response “I’m hungry,” or “Let’s go to a restau- rant.” A single nonverbal stimulus can also strengthen several response forms, as when a picture of a car strengthens the responses “car,” “automobile,” or “Ford.”

Thematic and Formal Verbal Operants

Skinner (1957) identified thematic and formal verbal op- erants that function as sources of multiple control. The thematic verbal operants are mands, tacts, and intraver- bals and involve different response topographies con- trolled by a common variable. For an intraverbal example, the SD “blue” can evoke verbal responses such as “lake,” “ocean,” and “sky.” The formal verbal oper- ants are echoic (imitation, copying a text) and textual (and transcription) and are controlled by a common vari- able, with point-to-point correspondence. For example, the SD “ring” can evoke verbal responses such as “sing,” “wing,” and “spring.”

Multiple Audiences

The role of the audience raises the issue of multiple au- diences. Different audiences may evoke different re- sponse forms. For example, two applied behavior analysts talking (i.e., a technical speaker and a technical audience) will likely use different response forms than will a behavior analyst speaking with a parent (i.e., a

Verbal Behavior

548

technical speaker and a nontechnical audience). A positive audience has special effects, especially a large positive audience (e.g., as in a rally for a certain cause), as does a negative audience. When the two audiences are combined, the effects of the negative audience are most obvious: “[When a] seditious soapbox orator sees a po- liceman approaching from a distance, his behavior de- creases in strength as the negative audience becomes more important” (Skinner, 1957, p. 231).

Elaborating Multiple Control

Convergent multiple control occurs in most instances of verbal behavior. An audience is always a source of stim- ulus control related to verbal behavior, even when a speaker serves as his own audience. In addition, it is also the case that more than one of the controlling variables re- lated to the different verbal operants may be relevant to a specific instance of verbal behavior. Convergent control often occurs with MOs and nonverbal stimuli, resulting in a response that is part mand and part tact. For exam- ple, saying “You look great” may be partly controlled by the nonverbal stimuli in front of a speaker (a tact), but also by MOs related to wanting to leave soon or wanting to avoid potential aversive events (a mand). Skinner (1957) identified this particular blend of controlling vari- ables as evoking impure tacts (i.e., impure because an MO affects the tact relation).

Verbal and nonverbal stimuli can also share control over a particular response. For example, a tendency to say “green car” may be evoked by the verbal stimulus “What color is the car?” and the nonverbal stimulus of the color green.

Multiple sources of control can be any combination of thematic or formal sources, even multiple sources from within a single verbal operant, such as multiple tacts or multiple intraverbals. Skinner pointed out that because these separate sources may be additive, the “multiple cau- sation produces many interesting verbal effects, includ- ing those of verbal play, wit, style, the devices of poetry, formal distortions, slips, and many techniques of verbal thinking” (pp. 228–229). Additional sources of control often reveal themselves—for example, when a speaker in the presence of an obese friend who is wearing a new hat emits the “Freudian slip,” “I like that fat on you.” Multiple sources of control often provide the basis for verbal humor and listener enjoyment.

Autoclitic Relation This chapter has emphasized that a speaker can, and often does, function as her own listener. The analysis of how and why a speaker becomes a listener of her own verbal behavior and then manipulates her verbal behavior with

additional verbal behavior addresses the topic of the auto- clitic relation. Skinner (1957) introduced the term auto- clitic to identify when a speaker’s own verbal behavior functions as an SD or an MO for additional speaker ver- bal behavior. In other words, the autoclitic is verbal be- havior about a speaker’s own verbal behavior. The consequences for this behavior involve differential rein- forcement from the ultimate listener, meaning that the listener discriminates whether to serve or not serve as a mediator of reinforcement for those verbal stimuli. A speaker becomes a listener, and observer, of his own ver- bal behavior and its controlling variables, and then in turn becomes a speaker again. This effect can be very rapid and typically occurs in the emission of a single sentence composed of the two levels of responses.

Primary and Secondary Verbal Operants

Michael (1991, 1992) suggested that applied behavior analysts classify verbal behavior about a speaker’s own verbal behavior as primary (Level 1) and secondary (Level 2) verbal operants. In Level 1, MOs and/or SDs are present and affect the primary verbal operant. The speaker has something to say. In Level 2, the speaker ob- serves the primary controlling variables of her own ver- bal behavior and her disposition to emit the primary verbal behavior. The speaker discriminates these con- trolling variables and describes them to the listener. A secondary verbal operant enables the listener’s behavior as a mediator of reinforcement. For example, an MO or SD evokes the response, “She is in Columbus, Ohio.” It is important for listeners as reinforcement mediators to discriminate the primary variables controlling the speaker’s behavior. The verbal operant “She is in Colum- bus, Ohio,” does not inform the listener as to why the speaker said it. “I read in the Columbus Dispatch that she is in Columbus, Ohio,” informs the listener of the pri- mary controlling variable. The first level is “She is in Columbus, Ohio” (the primary verbal operant), and the second level is “I read in the Columbus Dispatch” (the autoclitic).

Autoclitic Tact Relations

Some autoclitics inform the listener of the type of pri- mary verbal operant the autoclitic accompanies (Peter- son, 1978). The autoclitic tact informs the listener of some nonverbal aspect of the primary verbal operant and is therefore controlled by nonverbal stimuli. For example, a child’s statement, “I see Mommy” may contain an au- toclitic tact. The primary verbal operant (i.e., the tact) is the nonverbal SD of (a) the child’s mother; (b) the re- sponse, “Mommy”; and (c) the associated reinforcement

Verbal Behavior

549

history. The secondary verbal operant (i.e., the autoclitic tact) is the speaker’s tact informing that a nonverbal SD

evoked the primary verbal operant. In this case, the non- verbal SD was the visual stimulus of the child’s mother, and the response “I see” informs the listener of the source of control that evoked the primary tact. If the child heard his mother, but did not see his mother, the autoclitic tact “I hear” would be appropriate.

The listener may challenge the existence and nature of the autoclitic tact—for example, “How do you know it’s mommy?” The challenge is one way that effective autoclitic behavior is shaped and brought under appro- priate stimulus control.

Autoclitic tacts also inform the listener of the strength of the primary operant. In the examples of the verbal stim- uli, “I think it is Mommy” and “I know it is Mommy,” “I think” informs the listener that the source of control for the primary tact “Mommy” is weak; “I know” is strong.

Autoclitic Mand Relations

Speakers use autoclitic mands frequently to help the lis- tener present effective reinforcers (Peterson, 1978). A specific MO controls the autoclitic mand, and its role is to mand the listener to react in some specific way to the primary verbal operant. “I know it is Mommy” may con- tain an autoclitic mand. For example, if “I know” is not a tact of response strength, it may be an MO in the same sense as “Hurry up.”

Autoclitic mands occur everywhere, but listeners have difficulty recognizing the MO controlling the auto- clitic mand because the sources of control are private. For example, hidden agendas as autoclitic mands often revel themselves only to the careful observer. For exam- ple, a primary intraverbal such as an answer to a question about the sale of a product may contain autoclitic mands, as in “I’m sure you will be pleased with the sale,” in which “I’m sure you will be pleased” is controlled by the same MO that might control the response “Don’t ask me for any details about the sale.”

Developing Autoclitic Relations

Speakers develop autoclitic relations in several ways. For example, a father is wrapping a gift for his child’s mother, and the child nearby says, “Mommy.” The father may ask the child to identify the primary variables controlling the response by asking, “Did you see her?” The father may differentially respond to “I see” indicating that clearly “Mommy” is a tact and hide the gift; rather than a mand for “Mommy,” in which case, he keeps wrapping the gift. The source of control for the response “Mommy” could be the gift, as in, “That is for Mommy.” “That is for” (i.e., the autoclitic) informs the father that the gift is a nonver-

bal stimulus controlling the primary tact, “Mommy,” and the father continues wrapping the gift. As Skinner (1957) pointed out, “[A]n autoclitic affects the listener by indi- cating either a property of the speaker’s behavior or the circumstances responsible for that property” (p. 329).

Early language learners seldom emit autoclitic re- sponses. Skinner was clear on this point: “In the absence of any other verbal behavior whatsoever autoclitics can- not occur. . . . It is only when [the elementary] verbal op- erants have been established in strength that the speaker finds himself subject to the additional contingencies which establish autoclitic behavior” (p. 330). Thus, early language intervention program should not include auto- clitic training.

Applications of Verbal Behavior Skinner’s analysis of verbal behavior provides a concep- tual framework of language that can be quite beneficial for applied behavior analysts. Viewing language as learned behavior involving a social interaction between speakers and listeners, with the verbal operants as the basic units, changes how clinicians and researchers ap- proach and ameliorate problems related to language. Skinner’s theory of language has been successfully ap- plied to an increasing number of human areas. For ex- ample, the analysis has been used for typical language and child development (e.g., Bijou & Baer, 1965), ele- mentary and high school education (e.g., Johnson & Layng, 1994), college education (e.g., Chase, Johnson, & Sulzar-Azaroff, 1985), literacy (e.g., Moxley, 1990), composition (e.g., J. Vargas, 1978), memory (e.g., Palmer, 1991), second language acquisition (e.g., Shimamune & Jitsumori, 1999), clinical interventions (e.g., Layng & Andronis, 1984), behavior problems (e.g., McGill, 1999), traumatic brain injury (e.g., Sundberg, San Juan, Dawdy, & Arguelles, 1990), artificial intelligence (e.g., Stephens & Hutchison, 1992), ape language acquisition (e.g., Sav- age-Rumbaugh, 1984), and behavioral pharmacology (e.g., Critchfield, 1993). The most prolific application of Skinner’s analysis of verbal behavior has been to lan- guage assessment and intervention programs for children with autism or other developmental disabilities. This area of application will be presented in more detail in the fol- lowing sections.

Language Assessment

Most standardized language assessments designed for children with language delays seek to obtain an age equivalent score by testing a child’s receptive and expressive language abilities (e.g., Peabody Picture

Verbal Behavior

550

Vocabulary Test III [Dunn & Dunn, 1997], Comprehen- sive Receptive and Expressive Vocabulary Test [Hammill & Newcomer, 1997]). Although this information is help- ful in many ways, the tests do not distinguish among the mand, tact, and intraverbal repertoires, and important lan- guage deficits cannot be identified. For example, these tests assess language skills under the control of discrim- inative stimuli (e.g., pictures, words, questions); how- ever, a substantial percentage of verbal behavior is under the functional control of MOs. Manding is a dominating type of verbal behavior, yet rarely is this repertoire as- sessed in standardized testing. It is quite common to find children with autism or other developmental disabilities who are unable to mand, but have extensive tact and re- ceptive repertoires. If a language assessment fails to iden- tify delayed or defective language skills that are related to MO control, an appropriate intervention program may be difficult to establish. A similar problem is the failure to adequately assess the intraverbal repertoire with most standardized assessments.

If a child with language delays is referred for a lan- guage assessment, the behavior analyst should examine the current effectiveness of each verbal operant in addi- tion to obtaining a standardized test from a speech and language pathologist. The behavior analyst would start by obtaining information about the child’s mand reper- toire. When known motivating operations are at strength, what behavior does the child engage in to obtain the re- inforcer? When the reinforcer is provided, does the mand behavior cease? What is the frequency and complexity of the various mand units? Information regarding the quality and strength of the echoic repertoire can reveal potential problems in producing response topographies that are essential for other verbal interactions. A thor- ough examination of the tact repertoire will show the na- ture and extent of nonverbal stimulus control over verbal responses, and a systematic examination of the receptive and intraverbal repertoires will show the control by ver- bal stimuli. Thus, a more complete understanding of a language deficit, and hence a more effective language in- tervention program, can be obtained by determining the strengths and weaknesses of each of the verbal operants, as well as a number of other related skills (e.g., Parting- ton & Sundberg, 1998; Sundberg, 1983; Sundberg & Partington, 1998).

Language Intervention

Skinner’s analysis suggests that a complete verbal reper- toire is composed of each of the different elementary op- erants, and separate speaker and listener repertoires. The individual verbal operants are then seen as the bases for building more advanced language behavior. Therefore, a language intervention program may need to firmly es-

tablish each of these repertoires before moving on to more complex verbal relations such as autoclitics or multiply controlled responses. Procedures for teaching mand, echoic, tact, and intraverbal repertoires will be presented briefly in the following sections, which will also include, some discussion of the relevant research.

Mand Training

As previously stated, mands are very important to early language learners. They allow a child to control the de- livery of reinforcers when those reinforcers are most valu- able. As a result, a parent or language trainer’s behavior (especially vocal behavior) can be paired with the rein- forcer at the right time (i.e., when the relevant MO for an item is strong). Mands also begin to establish a child’s role as a speaker, rather than just a listener, thus giving the child some control of the social environment. If mands fail to develop in a typical manner, negative behaviors such as tantrums, aggression, social withdrawal, or self- injury that serve the mand function (and thereby control the social environment) commonly emerge. Therefore, a language intervention program for a nonverbal child must include procedures to teach appropriate manding. The other types of verbal behavior should not be neglected, but the mand allows a child to get what he wants, when he wants it.

The most complicated aspect of mand training is the fact that the response needs to be under the functional control of the relevant MO. Therefore, mand training can only occur when the relevant MO is strong, and ulti- mately the response should be free from additional sources of control (e.g., nonverbal stimuli). Another com- plication of mand training is that different response forms need to be established and brought under the control of each MO. Vocal words are of course the most common re- sponse form, but sign language, pictures, or written words can also be used.

The basic procedure for establishing mands consists of using prompting, fading, and differential reinforce- ment to transfer control from stimulus variables to moti- vative variables (Sundberg & Partington, 1998). For example, if a child demonstrates an MO for watching bubbles by reaching for the bottle of bubbles, then smil- ing and laughing as he watches the bubbles in the air, the timing is probably right to conduct mand training. If the child can echo the word “bubbles” or an approximation such as “ba,” teaching a mand can be easy (see Table 4). The trainer should first present the bottle of bubbles (a nonverbal stimulus) along with an echoic prompt (a verbal stimulus) and differentially reinforce succes- sive approximations to “bubble” with blowing bubbles (specific reinforcement). The next step is to fade the echoic prompt to establish the response “bubble” under

Verbal Behavior

551

Table 4 Teaching a Mand by Transferring Stimulus Control to MO Control

Antecedent Behavior Consequences

Motivating operation

Nonverbal stimulus “Bubbles” Blow bubbles

Echoic prompt

Motivating operation

Nonverbal stimulus “Bubbles” Blow bubbles

Motivating operation “Bubbles” Blow bubbles

the multiple control of the MO and the nonverbal stimu- lus (the bottle of bubbles). The final step is to fade the nonverbal stimulus to bring the response form under the sole control of the MO.

The easiest mands to teach in an early language in- tervention program are usually mands for items for which the MO is frequently strong for the child and satiation is slow to occur (e.g., food, toys, videos). It is always im- portant to assess the current strength of a supposed MO by using choice procedures, observation of a child’s behavior in a free operant situation (i.e., no demands), latency to contacting the reinforcer, immediate con- sumption, and so forth. The goal of early mand training is to establish several different mands by bringing dif- ferent response forms (i.e., words) under the functional control of different MOs. It is important to note that MOs vary in strength across time, and the effects may be mo- mentary. In addition, the response requirement placed on a child may weaken the strength of an MO, making mand training more difficult. Many additional strategies exist for teaching early mands to more difficult learners, such as augmentative communication, physical prompts, verbal prompts, and more careful fading and differential reinforcement procedures (see Sundberg & Partington, 1998).

Manding continues to be an important part of a ver- bal repertoire as other verbal operants are acquired. Soon after mands for edible and tangible reinforcers are ac- quired, a typical child learns to mand for actions (verbs), attention, removal of aversive stimuli, movement to cer- tain locations (prepositions), certain properties of items (adjectives) and actions (adverbs), verbal information (WH-questions), and so on. These mands are often more difficult to teach to a child with language delays because the relevant MO often must be captured or contrived for training purposes (Sundberg, 1993, 2004). Fortunately, Michael’s (1993) classification of the different types of MOs provides a useful guide for capturing or contriv- ing MOs. For example, capturing a transitive conditioned motivative operation (CMO-T) in the natural environ-

ment involves capitalizing on a situation in which one stimulus increases the value of a second stimulus. A child who likes fire trucks may see a fire truck parked outside the window. This stimulus condition increases the value of a second stimulus condition, an opened door, and will evoke behavior that has resulted in doors opening in the past. A skilled trainer would be watchful for these events and would be quick to conduct a mand trial for the word “open” or “out.” The work of Hart and Risley (1975) and their incidental teaching model exemplify this teach- ing strategy.

Transitive CMOs can also be contrived to conduct mand training (e.g., Hall & Sundberg, 1987; Sigafoos, Doss, & Reichele, 1989; Sundberg, Loeb, Hale, & Eigen- heer, 2002). For example, Hall and Sundberg (1987) used a contrived CMO-T procedure with a deaf teenager with autism by presenting highly desired instant coffee with- out hot water. The coffee altered the value of hot water and thereby evoked behavior that had been followed by hot water in the past. During baseline this behavior con- sisted of tantrums. Appropriate mands (i.e., signing “hot water”) were easy to teach when this CMO-T was in ef- fect by using the transfer of control procedure described earlier. In fact, a number of mands were taught by using this procedure, and often the procedure led to the emis- sion of untrained mands and a substantial reduction in negative behavior.

Mand training should be a significant part of any in- tervention program designed for children with autism or other severe language delays. Without an appropriate mand repertoire, a child cannot obtain reinforcement when MOs are strong, or have much control of the so- cial environment. As a result, people who interact with the child may become conditioned aversive stimuli, and/or problem behaviors may be acquired that serve the mand function. These behaviors and social relationships can become hard to change until replacement mands are es- tablished. Teaching mands early in a language interven- tion program may help to prevent the acquisition of negative behaviors as mands. In addition, parents and

Verbal Behavior

552

Table 5 Teaching Echoics by Using a Mand Frame and Transferring Control from Multiple Control to Echoic Control

Antecedent Behavior Consequences

Motivating operation

Nonverbal prompt “Bubbles” Blow bubbles

Echoic stimulus

Nonverbal prompt

Echoic stimulus “Bubbles” Praise (GCR)

Verbal Behavior

teachers are paired with successful manding and can be- come conditioned reinforcers. If people become more re- inforcing to a child, social withdrawal, escape and avoidance, and noncompliance may be reduced.

Echoic Training

For an early language learner the ability to repeat words when asked to do so plays a major role in the develop- ment of other verbal operants (as in the bubbles example earlier). If a child can emit a word under echoic stimulus control, then transfer of stimulus control procedures can be used to bring that same response form under the con- trol of not only MOs, but also stimuli such as objects (tacts) and questions (intraverbal). Because many chil- dren with autism and other language delays are unable to emit echoic behavior, special training procedures are required to develop the echoic repertoire.

The first goal of echoic training is to teach the child to repeat the words and phrases emitted by parents and teachers when asked to do so. Once echoic control is ini- tially established, the goal becomes to establish a gener- alized repertoire in which the child can repeat novel words and combinations. But the ultimate goal with the echoic repertoire is to transfer the response form to other verbal operants. This transfer process can begin imme- diately and is not dependent on the acquisition of a gen- eralized repertoire. Several procedures will be described to achieve the first goal of establishing initial echoic stim- ulus control.

The most common form of echoic training is direct echoic training in which a vocal stimulus is presented and successive approximations to the target response are differentially reinforced. This procedure involves a com- bination of prompting, fading, shaping, extinction, and reinforcement techniques. Speech therapists commonly use prompts such as pointing to the mouth, exaggerated movements, physical lip prompting, and mirrors to watch lip movement. Successive approximations to a target vo- calization are reinforced, and others are ignored. The prompts are then faded, and pure echoic responses are

reinforced. For many children these procedures are ef- fective in establishing and strengthening echoic control and improving articulation. However, for some children the procedures are ineffective, and additional measures are necessary.

Placing an echoic trial within a mand frame can often be a more effective procedure for establishing echoic stimulus control. The MO is a powerful inde- pendent variable in language training and can be tem- porally used to establish other verbal operants (e.g., Carroll & Hesse, 1987; Drash, High, & Tudor, 1999; Sundberg, 2004; Sundberg & Partington, 1998). For echoic training, an MO and nonverbal stimulus can be added to the target echoic antecedent as a way to evoke the behavior (see Table 5). For example, if a child demonstrates a strong MO for bubbles, an echoic trial can be conducted while that MO is strong, and in the presence of the nonverbal stimulus of the bottle of bub- bles. These additional sources of control can help to evoke the vocal response along with the echoic prompt “Say bubbles.” The specific reinforcement of blowing bubbles is then contingent on any successive approxi- mation to “bubbles.” These additional antecedent vari- ables must be faded out, and the reinforcement changed from the specific reinforcement to generalized condi- tioned reinforcement. For some learners, the transfer from MO to echoic control may occur more quickly if a picture of the object is used rather than the actual object (this reduces the MO evocative effect).

Children with a low frequency of vocal behaviors may have difficulty establishing echoic control. For these children procedures to simply increase any vocal behav- ior may facilitate the ultimate establishment of echoic control. One method is to directly reinforce all vocal be- haviors. Taking this procedure one step further, if a child randomly emits a particular sound, the behavior analyst can reinforce this behavior and conduct an echoic trial with that sound immediately after the delivery of rein- forcement. Some children will repeat what they initially emitted, and this interaction sets up some of the basic variables that may facilitate echoic control.

553

Verbal Behavior

Table 6 Teaching Tacts by Using a Mand Frame and Trans- ferring Control from Multiple Control to Nonverbal Control

Antecedent Behavior Consequences

Motivating operation

Nonverbal prompt “Bubbles” Blow bubbles

Echoic stimulus

Nonverbal prompt

Echoic stimulus “Bubbles” Praise (GCR)

Nonverbal stimulus “Bubbles” Praise (GCR)

Automatic reinforcement procedures can also be used to increase the frequency of vocal behavior. By pair- ing a neutral stimulus with an established form of rein- forcement, the neutral stimulus can become a conditioned reinforcer. For example, if just prior to blowing bubbles the trainer emits the word bubbles, bubbles can become a reinforcer. Research has shown that this pairing pro- cedure can increase the rate of a child’s vocal play and result in the emission of targeted sounds and words that had never occurred echoically (Miguel, Carr, & Michael, 2002; Sundberg, Michael, Partington, & Sundberg, 1996; Smith, Michael, & Sundberg, 1996; Yoon & Bennett, 2000). For example, Yoon and Bennett (2000) demon- strated that this pairing procedure was more successful than direct echoic training in producing targeted sounds. Individual children who have difficulty acquiring an echoic repertoire may benefit from this procedure or a combination of all the procedures described in this section.

Tact Training

The tact repertoire is extensive and often the primary focus of many language intervention programs. A child must learn to tact objects, actions, properties of objects and ac- tions, prepositional relations, abstractions, private events, and so on. The goal of the teaching procedures is to bring a verbal response under nonverbal stimulus control. If a child has a strong echoic repertoire, then tact training can be quite simple. A language trainer can present a nonver- bal stimulus along with an echoic prompt, differentially re- inforce a correct response, and then fade the echoic prompt. However, for some children tact training is more difficult, and special procedures may be required.

A mand frame can also be used to establish tacting (Carroll & Hesse, 1987). The procedure is similar to that described for teaching an echoic response. Training be- gins with an MO for a desired object, the nonverbal ob- ject, and an echoic prompt (see Table 6). Using the

bubbles example, the first and second steps are the same, with the goal being to free the response from motiva- tional control by providing generalized conditioned re- inforcement rather than specific reinforcement. As with echoic training, at this point in the procedure the transfer may occur more quickly if a picture of the object is used rather than the actual object. Also, with some children it may be more effective to fade the echoic prompt before the MO is faded. The third step in the procedure involves fading out the echoic prompt and bringing the response under the sole control of the nonverbal stimulus, thus a tact. Additional nonechoic verbal prompts may also be helpful, such as “What is that?,” but these too are verbal prompts that are additional sources of control that need to be accounted for in the analysis of tact acquisition (Sundberg & Partington, 1998).

Methods for teaching more complex tacts can also make use of the transfer of stimulus control procedure. For example, teaching tacts of actions requires that the nonverbal stimulus of movement be present and a re- sponse such as “jump” be brought under the control of the action of jumping. Teaching tacts involving prepo- sitions, adjectives, pronouns, adverbs, and so on, also involves the establishment of nonverbal stimulus con- trol. However, these advanced tacts are often more complex than they appear, and frequently the type of stimulus control established in formal training may not be the same type of stimulus control that evokes simi- lar tacts for typically developing children (Sundberg & Michael, 2001). For example, some training programs with early learners attempt to bring verbal behavior under the control of private stimuli, such as those in- volved in emotional states (sad, happy, afraid), pains, itches, a full bladder, hunger pangs, nausea, and so forth. Such verbal behavior is an important part of any person’s repertoire, but because the controlling vari- ables that are affecting the learner cannot be directly contacted by the teacher or parent, accurate tact rela- tions are difficult to develop. An instructor cannot pre- sent the relevant private stimulus that is inside a

554

Verbal Behavior

person’s body, and therefore cannot differentially re- inforce correct tact responses in the same way that cor- rect tacts to objects and actions can be reinforced. Teaching a child to correctly say “itch” with respect to a stimulus coming from a portion of the child’s arm is trained indirectly as the teacher reacts to common pub- lic accompaniments of such stimuli (observing a skin rash) and collateral responses by the learner (observing the child’s scratching). However, this method is fraught with difficulties (the rash may not itch, the scratching may be imitated), and such repertoires even in typical adults are often quite imprecise.

Intraverbal Training

Many children with autism, developmental disabilities, or other language delays suffer from defective or nonexis- tent intraverbal repertoires, even though some can emit hundreds of mands, tacts, and receptive responses. For example, a child may (a) say “bed” when hearing “bed” spoken by another person (echoic), (b) say “bed” when he sees a bed (tact), and even (c) ask for bed when he is tired (mand), but (d) may not say “bed” when someone asks, “Where do you sleep?” or says, “You sleep in a . . . .” In cognitive terms, this type of language disorder may be described as a child’s failure to process the auditory stim- ulus, or explained in terms of other hypothesized internal processes. However, verbal stimulus control is not the same as nonverbal stimulus control, and a response ac- quired as a tact for an early language learner may not au- tomatically occur as an intraverbal without special training (e.g., Braam & Poling, 1982; Luciano, 1986; Partington & Bailey, 1993; Watkins, Pack-Teixteiria, & Howard, 1989).

In general, verbal stimulus control over verbal re- sponding is more difficult to establish than nonverbal control. This does not mean that all intraverbals are harder than all tacts. Some intraverbal behavior is simple and easy to acquire. However, formal training on intraverbal

behavior for a language-delayed child should not occur until the child has well-established mand, tact, echoic, imitation, receptive, and matching-to-sample repertoires (Sundberg & Partington, 1998). A common mistake in early intraverbal training is to attempt to teach intraver- bal relations too early, or intraverbals that are too complex and out of developmental sequence such as personal in- formation (e.g., “What is your name and phone num- ber?”). Some of the easiest intraverbal relations are fill-in-the-blanks with songs (e.g., “The wheels on the. . .)” and other fun activities (e.g., Peek-a- ). The goal of early intraverbal training is to begin to break verbal respond- ing free from mand, echoic, and tact sources of control. That is, no new response topographies are taught; rather, known words are brought under a new type of stimulus control.

MOs can be helpful independent variables in facili- tating the transfer of stimulus control, as in the Peek-a- Boo example, and the mand frame procedures described earlier for the echoic and tact. However, ultimately, a child needs to learn to emit intraverbal responses that are free from MO control. For example, if a child likes bub- bles and the target intraverbal is for the verbal stimulus “You blow. . .” to evoke the verbal response “bubbles,” then intraverbal training should involve the transfer of control from MOs and nonverbal stimuli to verbal stim- uli (echoic prompts can also be used). The language trainer can present the verbal stimulus (e.g., “You blow . . .”) when the MO is strong along with the nonverbal stimu- lus (e.g., a bottle of bubbles). Then, the trainer can begin providing generalized conditioned reinforcement rather than specific reinforcement, use a picture of the item rather than the actual item, and finally, fade the nonver- bal prompt (see Table 7)

The intraverbal repertoire becomes increasing more valuable to a child as the verbal stimuli and related re- sponses become more varied and complex. Common as- sociations (e.g., “Mommy and . . .”), fill-in-the-blanks (e.g., “You bounce a . . .”), animal sounds (e.g., “A kitty goes . . .”), and eventually what questions (e.g., “What

Table 7 Teaching Intraverbals by Using a Mand Frame and Transferring Control from Multiple Control to a Verbal Stimulus

Antecedent Behavior Consequences

Motivating operation

Nonverbal prompt “Bubbles” Blow bubbles

Verbal stimulus

Nonverbal prompt

Verbal stimulus “Bubbles” Praise (GCR)

555

do you eat?”) will help to strengthen intraverbal behav- ior by expanding the content and variation of the verbal stimuli and the verbal responses. In addition, these pro- cedures can help to develop verbal stimulus and response classes and more fluent intraverbal responding that is free from tact and echoic sources of stimulus control. More advanced intraverbal training can be accomplished in a variety of ways (Sundberg & Partington, 1998). For ex- ample, the verbal stimulus can have multiple components involving conditional discriminations in which one ver- bal stimulus alters the evocative effect of another, as in “What do you eat for breakfast?” versus “What do you eat for dinner?” Expansion prompts can also be used, such as “What else do you eat for lunch?,” as well as additional WH-questions, such as “Where do you eat lunch?,” and “When do you eat lunch?” As with the other verbal reper- toires, typical developmental sequences can be a helpful guide to the progression of increasingly complex in- traverbal behavior, as can the task analysis of the verbal

operants presented in the Assessment of Basic Learning and Language Skills: The ABLLS (Partington & Sund- berg, 1998).

Additional Aspects of Language Training

In addition to these four basic repertoires, there are sev- eral other components of a verbal behavior program and curriculum, such as receptive language training, matching-to-sample, mixing and varying trials, multiple response training, sentence construction, conversational skills, peer interaction, reading, and writing (Sundberg & Partington, 1998). Although a description of these pro- grams is beyond the scope of this chapter, many of the procedures for teaching these skills involve the same basic elements of the transfer of stimulus control meth- ods described here.

Summary Verbal Behavior and Properties of Language

1. Verbal behavior is defined as behavior that is reinforced through the mediation of another person’s behavior.

2. The formal properties of verbal behavior involve the topog- raphy (i.e., form, structure) of the verbal response.

3. The functional properties of verbal behavior involve the causes (i.e., antecedents and consequences) of the response.

4. Skinner’s analysis of verbal behavior was met with strong opposition from the field of linguistics, and with indiffer- ence within the field of behavior analysis. However, Skin- ner predicted in 1978 that Verbal Behavior would prove to be his most important work.

Defining Verbal Behavior

5. Verbal behavior involves a social interaction between speakers and listeners, whereby speakers gain access to reinforcement and control their environment through the behavior of listeners.

6. The verbal operant is the unit of analysis of verbal behav- ior and is the functional relation between a type of re- sponding and (a) motivating variables, (b) discriminative stimuli, and (c) consequences.

7. A verbal repertoire is a set of verbal operants emitted by a particular person.

Elementary Verbal Operants

8. The mand is a verbal operant in which the form of the re- sponse is under the functional control of motivating oper- ations (MOs) and specific reinforcement.

9. The tact is a verbal operant under the functional control of nonverbal discriminative stimulus, and it produces gener- alized conditioned reinforcement.

10. The echoic is a verbal operant that consists of a verbal dis- criminative stimulus that has point-to-point correspon- dence and formal similarity with a verbal response.

11. Point-to-point correspondence between the stimulus and the response or response product occurs when the begin- ning, middle, and end of the verbal stimulus matches the beginning, middle, and end of the verbal response.

12. Formal similarity occurs when the controlling antecedent stimulus and the response or response product (a) share the same sense mode (e.g., both stimulus and response are visual, auditory, or tactile) and (b) physically resemble each other.

13. The intraverbal is a verbal operant that consists of a ver- bal discriminative stimulus that evokes a verbal response that does not have point-to-point correspondence.

14. The textual relation is a verbal operant that consists of a verbal discriminative stimulus that has point-to-point cor- respondence between the stimulus and the response prod- uct, but does not have formal similarity.

15. The transcription relation is a verbal operant that con- sists of a verbal discriminative stimulus that controls a written, typed, or finger-spelled response. Like the tex- tual relation, there is point-to-point correspondence be- tween the stimulus and the response product, but no formal similarity.

Verbal Behavior

556

Role of the Listener

16. The listener not only mediates reinforcement, but func- tions as a discriminative stimulus for verbal behavior. Often, much of the behavior of a listener is covert verbal behavior.

17. An audience is a discriminative stimulus in the presence of which verbal behavior is characteristically reinforced.

18. Classifying verbal responses as mands, tacts, intraverbals, etc., can be accomplished by an analysis of the relevant controlling variables.

Analyzing Complex Verbal Behavior

19. Automatic reinforcement is a type of conditioned rein- forcement in which a response product has reinforcing properties as a result of a specific conditioning history.

20. Automatic punishment is a type of conditioned punish- ment in which a response product has punishing proper- ties as a result of a specific conditioning history.

21. In generic tact extension, the novel stimulus shares all of the relevant or defining features associated with the orig- inal stimulus.

22. In metaphorical tact extension, the novel stimulus shares some, but not all, of the relevant features of the original stimulus.

23. In metonymical tact extension, the novel stimulus shares none of the relevant features of the original stimulus con- figuration, but some irrelevant but related feature has ac- quired stimulus control.

24. In solistic tact extension, a stimulus property that is only indirectly related to the tact relation evokes substandard verbal behavior.

25. Private events are stimuli that arise from within some- one’s body.

26. Public accompaniment occurs when a publicly observable stimulus accompanies a private stimulus.

27. Collateral responses are publicly observable behaviors that reliably occur with private stimuli.

28. Common properties involve a type of generalization in which private stimuli share some of the features of public stimuli.

29. Response reduction is also a type of generalization in which kinesthetic stimuli arising from movement and po- sitions acquire control over the verbal responses. When movements shrink in size (become covert), the kinesthetic stimuli may remain sufficiently similar to those resulting from the overt movements.

Multiple Control

30. Convergent multiple control occurs when a single verbal response is a function of more than one controlling variable.

31. Divergent multiple control occurs when a single antecedent variable affects the strength of many responses.

32. The thematic verbal operants are mand, tact, and in- traverbal, and involve different response topographies con- trolled by a common variable.

33. The formal verbal operants are echoic (and imitation as it relates to sign language and copying a text), textual, and transcription, and involve control by a common variable with point-to-point correspondence.

34. Multiple audiences consist of two or more different audi- ences that may evoke different response forms.

35. Impure tacts occur when an MO shares control with a non- verbal stimulus.

Autoclitic Relation

36. The autoclitic relation involves two related but separate three-term contingencies in which some aspect of a speaker’s own verbal behavior functions as an SD or an MO for additional speaker verbal behavior.

37. Primary verbal behavior involves the elementary verbal operants emitted by a speaker.

38. Secondary verbal behavior involves verbal responses con- trolled by some aspect of the speaker’s own ongoing ver- bal behavior.

39. The autoclitic tact informs the listener of some nonverbal aspect of the primary verbal operant and is therefore con- trolled by nonverbal stimuli.

40. The autoclitic mand is controlled by an MO and enjoins the listener to react in some specific way to the primary verbal operant.

Applications of Verbal Behavior

41. The verbal operants can be used to assess a wide variety of language deficits.

42. Mand training involves bringing verbal responses under the functional control of MOs.

43. Echoic training involves bringing verbal responses under the functional control of verbal discriminative stimuli that have point-to-point correspondence and formal similarity with the response.

44. Tact training involves bringing verbal responses under the functional control of nonverbal discriminative stimuli.

45. Intraverbal training involves bringing verbal responses under the functional control of verbal discriminative stimuli that lack point-to-point correspondence with the response.

Verbal Behavior

557

Contingency Contracting, Token Economy, and Group Contingencies

Key Terms

backup reinforcer behavioral contract contingency contract dependent group contingency

group contingency hero procedure independent group contingency interdependent group contingency

level system self-contract token token economy

Content Area 9: Behavior Change Procedures

9-18 Use contingency contracting (e.g., behavioral contracts).

9-19 Use token economy procedures, including levels systems.

9-20 Use independent, interdependent, and dependent group contingencies.

558

Contingency Contracting, Token Economy, and Group Contingencies

This chapter addresses contingency contract- ing, token economy, and group contingencies as special applications of behavioral proce-

dures. Each application will be defined; its relationship to behavioral principles will be explained; requisite components of the procedure will be addressed; and guidelines for designing, implementing, and evaluating each will be presented. These topics are grouped be- cause they have several features in common. First, they have an effective and robust literature supporting their inclusion. Second, they can be combined with other ap- proaches in package programs to provide an additive effect. Also, each can be used in individual and group arrangements. The flexibility that these three special ap- plications provide makes them an attractive option for practitioners.

Contingency Contracting Definition of Contingency Contract

A contingency contract, also called a behavioral con- tract, is a document that specifies a contingent relation- ship between the completion of a specified behavior and access to, or delivery of, a specified reward such as free time, a letter grade, or access to a preferred activity.

Typically, contracts specify how two or more people will behave toward each other. Such quid pro quo agree- ments make one person’s behavior (e.g., preparing din- ner) dependent on the other person’s behavior (e.g., washing and putting away the dishes by a prescribed time the night before). Although verbal agreements may be considered contracts in the legal sense, they are not contingency contracts because the degree of specificity in designing, implementing, and evaluating a contin- gency contract far exceeds what is likely to occur in a verbal arrangement between parties. In addition, the physical act of signing the contract and its prominent visibility during execution are integral parts of contin- gency contracts.

Contingency and behavior contracts have been used to modify academic performance (Newstrom, McLaugh- lin, & Sweeney, 1999; Wilkinson, 2003), weight control (Solanto, Jacobson, Heller, Golden, & Hertz, 1994), ad- herence to medical regimens (Miller & Stark, 1994), and athletic skills (Simek, O’Brien, & Figlerski, 1994). In- deed, a compelling advantage of contingency contracts is their ability to be implemented alone or in packaged programs that incorporate two or more interventions con- currently (De Martini-Scully, Bray, & Kehle, 2000).

Components of Contingency Contracts

There are three major parts in most contracts—a de- scription of the task, a description of the reward, and the task record. Essentially, the contract specifies the per- son(s) to perform the task, the scope and sequence of the task, and the circumstances or criterion for task comple- tion. Figure 1 shows a contingency contract imple- mented by the parents of a 10-year-old boy to help him learn to get up and get ready for school each day.

Task

The task side of the contract consists of four parts. Who is the person who will perform the task and receive the reward—in this case, Mark. What is the task or behav- ior the person must perform—in this example, getting ready for school. When identifies the time that the task must be completed—every school day. How well is the most important part of the task side, and perhaps of the entire contract. It calls for the specifics of the task. Sometimes it is helpful to list a series of steps or sub- tasks so that the person can use the contract as a check- list of what must be done. Any exceptions should be written in this part.

Reward

The reward side of a contract must be as complete and accurate as the task side (Ruth, 1996). Some people are very good at specifying the task side of a contract; they know what they want the other person to do. When it comes to the reward side, however, specificity is lost and problems arise. Reward statements such as “Can watch some television” or “Will play catch when I get a chance” are not explicit, specific, or fair to the person completing the task.

On the reward side, Who is the person who will judge task completion and control delivery of the reward. With Mark’s getting-ready-for-school contract, those persons are his parents. What is the reward. When specifies the time that the reward can be received by the person earn- ing it. With any contract it is crucial that the reward come after successful task completion. However, many rewards cannot be delivered immediately following task comple- tion. In addition, some rewards have built-in, limited avail- ability and can be delivered only at certain times (e.g., seeing the home-town baseball team play). Mark’s con- tract specifies that his reward, if earned, will be received on Friday nights. How much is the amount of reward that

559

Contingency Contracting, Token Economy, and Group Contingencies

Feb. 12, 2007 Feb. 12, 2007

Figure 1 Example of a contingency contract. From Sign Here: A Contracting Book for Children and Their Parents (2nd ed., p. 31) by J. C. Dardig and W. I. Heward, 1981, Bridgewater, NJ: Fournies and Associates. Copyright 1981 by Fournies and Associates. Reprinted by permission.

can be earned by completing the task. Any bonus contin- gencies should be included; for example, “By adhering to her contract Monday through Friday, Eileen receives an extra reward on Saturday and Sunday.”

Task Record

Including on the contract a place to record task comple- tion serves two purposes. First, recording task completion and reward delivery on the contract sets the occasion for all parties to review the contract regularly. Second, if a certain number of task completions are required to earn the reward (e.g., if a child must dress herself each morn- ing before school for 5 days in a row), a check mark, smi- ley face, or star can be placed on the task record each time the task is completed successfully. Marking the con- tract in this manner helps the person remain focused until the assignment is completed and the reward is earned. Mark’s parents used the top row of boxes in the task record to record the days of the school week. In the mid- dle row of boxes they placed gummed stars each day Mark met the conditions of his contract. In the bottom row, Mark’s parents wrote comments about the progress of his contract.

Implementing Contingency Contracts

How Do Contracts Work?

At first glance the principle of behavior behind contin- gency contracts seems deceptively simple: A behavior is followed by a contingent reward—surely a case of positive reinforcement. Yet, in most contracts the re- ward, although contingent, is much too delayed to re- inforce the specified behavior directly; and many successful contracts specify a reward that would not, in fact, function as a reinforcer for the task even if it was presented immediately after the task completion. Fur- ther, behavioral contracting is not a single procedure with a single associated behavior and a single reinforcer. Contracting is more accurately conceptualized as an in- tervention package that combines several behavior prin- ciples and procedures.

So how do contracts work? Several principles, pro- cedures, and factors are likely to apply. Certainly rein- forcement is involved, but not in as simple or direct a fashion as it might seem at first. Rule-governed behavior is probably involved (Malott, 1989; Malott & Garcia, 1991; Skinner, 1969). A contract describes a rule: A

560

Contingency Contracting, Token Economy, and Group Contingencies

100

Pe rc

en t

C o

rr ec

t C

ap it

al iz

at io

5 1510

Spelling

Baseline Contingency Contracting

100

Pe rc

en t

C o

rr ec

t P

u n

ct u

at io

5 1510

Written Sentences

Sessions

specified behavior will be followed by a specified (and reasonably immediate) consequence. The contract serves as a response prompt to perform the target behavior and enables the effective use of a consequence (e.g., going to the movies Saturday night), too delayed in itself to rein- force certain behaviors (e.g., practicing the trumpet on Tuesday). Delayed consequences can help exert control over behaviors performed hours and even days before if they are associated with and linked by verbal behavior to the rule (e.g., “I’ve just finished my trumpet practice— that’s another check mark toward the movies on Satur- day”), or to interim token reinforcers (e.g., the check mark on the contract after practicing). The physical vis- ibility of the contract may also function as a response prompt for escaping “guilt” (Malott & Garcia, 1991). At this stage of knowledge development within the field, it cannot be said that contracting is simply positive rein- forcement loosely based on the Premack principle. It is more likely a complex package intervention of related positive and negative reinforcement contingencies and rule-governed behavior that operate alone and together.

Applications of Contingency Contracting

Contracting in the Classroom

The use of contracting in classrooms is well established. For example, teachers have employed contracting to ad- dress specific discipline, performance, and academic challenges (Kehle, Bray, Theodore, Jenson, & Clark, 2000; Ruth, 1996). Newstrom and colleagues (1999), for instance, used a contingency contract with a middle school student with behavior disorders to improve the written mechanics associated with spelling and written language. After collecting baseline data on the percent- age of capitalization and punctuation marks used cor- rectly across spelling and sentence writing, a contingency contract was negotiated and signed with the student that specified that improved performance would yield free time on the classroom computer. The student was re- minded of the terms of the contract before each language arts class that included spelling worksheets and journal writing (i.e., sentences).

Figure 2 shows the results of the contingency contracting intervention. When baseline was in effect for spelling and written sentences respectively, mean scores for both variables were in the 20% correct range. When contracting was initiated, the student’s perfor- mance for spelling and written sentences increased im- mediately to an average of approximately 84% correct. Because the percentage of correct performance increased

immediately when contingency contracting was imple- mented for spelling (Sessions 4 through 12), but did not increase for written sentences until Session 11, a func- tional relation was demonstrated between contracting and improved performance. The authors also reported positive anecdotal evidence related to spelling and writ- ten language from other teachers with whom this stu- dent interacted.

Wilkinson (2003) used a contingency contract to reduce the disruptive behavior of a first-grade student. Disruptive behaviors included being off-task, refusals to comply with work assignments and instructions, fighting with peers, and temper tantrums. A behavioral consultation effort was launched with the classroom teacher that included problem identification, analysis, intervention, and evaluation. Contingency contracting consisted of the student earning preferred rewards and social praise from the teacher for three behaviors: in- creased time on-task, appropriate interactions with other children, and compliance with teacher requests. Observations of her behavior over 13 sessions of base- line and contingency contracting showed a decrease in

Figure 2 Percentage of capitalization and punctua- tion marks used correctly on spelling worksheets and journal writing during baseline and contingency contracting. From “The Effects of Contingency Contracting to Improve the Mechanics of Written Language with a Middle School Student with Behavior Disorders” by J. Newstrom, T. F. McLaughlin, & W. J. Sweeney, 1999, Child & Family Behavior Therapy, 21 (1), p. 44. Copyright 1999 by The Haworth Press, Inc. Reprinted by permission.

561

Contingency Contracting, Token Economy, and Group Contingencies

the percentage of intervals with disruptive behavior when the contingency contract was in effect. Wilkinson reported that the student’s disruptive behavior de- creased substantially and remained low during a 4-week follow-up period.

Ruth (1996) conducted a 5-year longitudinal study with emotionally disturbed students that blended con- tingency contracting with goal setting. After students negotiated their contracts with their teachers, a goal- setting component was added that included statements about their daily and weekly goals and the criterion levels for success. Results for the 37 out of 43 students who finished the program after 5 years showed that 75% of daily goals, 72% of weekly goals, and 86% of total goals were reached. Ruth summarized the benefi- cial effects of combining strategies: “When [goal- setting] methods are incorporated into a contract, the motivational aspects of behavior contracting and goal setting may combine to produce maximum effort and success” (p. 156).

Contracting in the Home

Miller and Kelley (1994) combined contingency con- tracting and goal setting to improve the homework per- formance of four preadolescent students with histories of poor homework completion and who were at risk for other academic problems (e.g., procrastination, being off-task, errors with submitted work). During baseline, parents recorded their children’s homework time on- task, type and accuracy of problem completion, and number of problems completed correctly. Subsequently, parents and children entered into a goal-setting and con- tingency contracting phase that was preceded by parent training on how to set and negotiate goals and write con- tracts. Each night, parents and children established their respective goals and negotiated a compromise goal based on that interaction. Each week, renegotiations oc- curred for tasks, rewards, and sanctions should the con- tract not be met. A recording sheet was used to measure progress.

Figure 3 shows the results of the study. When goal setting was combined with contingency contracting, ac- curacy performance increased for all students. Miller and Kelley’s findings reaffirm the notion that contingency contracting can be combined successfully with other strategies to produce functional outcomes.

Clinical Applications of Contracting

Flood and Wilder (2002) combined contingency con- tracting with functional communication training to re-

duce the off-task behavior of an elementary-aged stu- dent diagnosed with attention-deficit/hyperactivity dis- order (ADHD) who had been referred to a clinic-based program because his off-task behavior had reached alarming levels. Antecedent assessment, functional communication training, and contingency contracting were conducted in a therapy room located in the clini- cal facility. Specifically, antecedent assessments were conducted to determine the level of off-task behavior when the difficulty of academic tasks varied from easy to difficult, and therapist attention varied from low to high. A preference assessment was also conducted. Using discrete trial training, the student was taught to raise his hand for assistance with tasks (e.g., “Can you help me with this problem?”). The therapist sat nearby and responded to appropriate requests for assistance and ignored other vocalizations. Once requesting as- sistance was mastered, the contingency contract was established whereby the student could earn preferred items, identified through the assessment, contingent on accurate task completion. The results showed that dur- ing baseline, off-task performance was high for math division and word problems. When the intervention was introduced, an immediate reduction in off-task behav- ior was noted for division and word problems. Also, the student’s accuracy in completing the division and word problems improved as well. Whereas during base- line conditions he solved correctly 5% and 33% of the division and word problems, respectively, during in- tervention he solved 24% and 92% of the problems correctly.

Using Contracting to Teach Self-Management to Children

Ideally, contingency contracting involves the active par- ticipation of the child throughout the development, im- plementation, and evaluation of the contract. For many children, contracting is a first experience in identify- ing specific ways they would like to act and then ar- ranging certain aspects of their environment to set the occasion for and reward those acts. If more of the de- cision making for all parts of the contracting process is turned over to children gradually and systematically, they can become skilled at self-contracting. A self- contract is a contingency contract that a person makes with herself, incorporating a self-selected task and re- ward as well as personal monitoring of task comple- tion and self-delivery of the reward. Self-contracting skills can be achieved by a multistep process of having an adult prescribe virtually all of the elements of the task and reward and gradually shifting the design of the elements to the child.

562

Contingency Contracting, Token Economy, and Group Contingencies

100

BL GS + CC BL GS + CC

Richard

100

0 Jenny

100

0 Adam

100

0 Ann

2 4 6 8 10 12 14 16 18 20 22 24 26 28

Sessions

P er

ce n

ta ge

o f

A cc

u ra

Figure 3 Percentage of homework problems completed accurately during baseline and a treatment condition consisting of goal setting and contingency contracting. Sessions correspond to sequential school days (i.e., Monday through Thursdays) on which subjects were assigned homework. Data were not collected on days on which homework was not assigned. From “The Use of Goal Setting and Contingency Contracting for Improving Children’s Homework Performance” by D. L. Miller & M. L. Kelley, 1994, Journal of Applied Behavior Analysis, 27, p. 80. Copyright 1994 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

Developing Contingency Contracts

Although teachers, therapists, or parents can unilaterally determine a contract for a child or client, contracting is usually more effective when all of the parties involved play an active role in developing the contract. Several methods and guidelines have been proposed for devel- oping contingency contracts (Dardig & Heward, 1981;

Downing, 1990; Homme Csanyi, Gonzales & Rechs, 1970). Contract development involves the specification of tasks and rewards in a fashion agreeable and beneficial to each party. Dardig and Heward (1981) described a five- step procedure for task and reward identification that can be used by teachers and families.

Step 1: Hold a meeting. To get the entire group (family or class) involved in the contracting process, a

563

Contingency Contracting, Token Economy, and Group Contingencies

meeting should be held. At this meeting members can discuss how contracts work, how they can help the group cooperate and get along better, and how con- tracts can help individuals meet personal goals. Par- ents or teachers should emphasize that they will participate in all of the steps leading up to and includ- ing implementation of contracts. It is important that children view contracting as a behavior-exchange process shared by all members of the group, not as something adults impose on them. The field-tested list-making procedures described in the following steps provide a simple and logical framework for the selection of tasks and rewards for family and class- room contracts. Most groups can complete the proce- dure within 1 to 2 hours.

Step 2: Fill Out List A. Each member completes three lists prior to the actual writing of the contract. List A (see Figure 4) is designed to help each member identify not only those tasks he can perform within the context of a contract, but also those tasks he already does to help the group. In this way positive attention can be focused on appropriate behaviors that individual members are currently completing satisfactorily.

Each member should be given a copy of List A. Every- one should be careful to describe all tasks as specifically as possible. Then the completed lists can be put aside, and

the group can proceed to the next step. If a member is un- able to write, that person’s list can be completed orally.

Step 3: Fill Out List B. List B (see Figure 5) is de- signed to help group members identify possible contract tasks for other group members and helpful behaviors currently being completed by those persons. List B can also identify areas where disagreement exists between group members as to whether certain tasks are actually being completed properly and regularly.

Each member should be given a copy of List B and asked to write his or her name in all three blanks at the top. These lists can then be passed around the table so that everyone has a chance to write at least one behavior on each side of everyone else’s list. Everyone writes on every List B except his or her own, and each person should be required to write at least one positive behavior on everyone else’s List B. After completion, these lists should be set aside before moving to the next step.

Step 4: Fill Out List C. List C (see Figure 6) is simply a sheet with numbered lines on which each group member identifies potential rewards he would like to earn by completing contracted tasks. Participants should list not only everyday favorite things and activi- ties, but also special items and activities they may have wanted for a long time. It is all right if two or more

Figure 4 A form for self-identification of possible tasks for contingency contracts.

List A Name: Jean

Things I Do to Help My Family Other Ways I Could Help My Family and Myself

1. Feed Queenie and 1. Be on time for Chippy supper

2. Clean up my 2. Turn off the lights bedroom when I leave a room

3. Practice my piano 3. Dust the living room ____________________ ____________________

4. Wash dishes 4. Clean up the back ____________________ yard

5. Help Dad with the 5. Hang up my coat when laundry I get home from school

6. ____________________ 6. ____________________ ____________________ ____________________

7. ____________________ 7. ____________________ ____________________ ____________________

From Sign Here: A Contracting Book for Children and Their Parents (2nd ed., p. 111) by J. C. Dardig and W. L. Heward, 1981. Bridgewater, NJ: Fournies and Associates. Copyright 1981 by Fournies and Associates. Reprinted by permission.

564

Contingency Contracting, Token Economy, and Group Contingencies

Figure 5 A form for identifying potential contracting tasks for others.

List B Name: Bobby

THINGS Bobby DOES OTHER WAYS Bobby TO HELP THE FAMILY COULD HELP THE FAMILY

1. Vacuums when asked 1. Put his dirty clothes in___ ____________________ hamper_________________

2. Makes his bed________ 2. Do homework at night____ ____________________ without being asked______

3. Reads stories to little__ 3. Make his own sandwiches sister________________ for his school lunch______

4. Empties trash_________ 4. Clean and sponge off _____ ____________________ table after supper _____

5. Rakes leaves_________ 5. _______________________ ____________________ _______________________

6. ____________________ 6. _______________________ ____________________ _______________________

7. ____________________ 7. _______________________ ____________________ ________________________

From Sign Here: A Contracting Book for Children and Their Parents (2nd ed., p. 113) by J. C. Dardig and W. L. Heward, 1981, Bridgewater, NJ: Fournies and Associates. Copyright 1981 by Fournies and Associates. Reprinted by permission.

Figure 6 A form for self-identification of possible rewards for contingency contracts.

List C Name: Sue Ann

MY FAVORITE THINGS, ACTIVITIES, AND SPECIAL TREATS

1. Listening to records 2. Movies 3. Playing pinball 4. Miniature golf 5. Swimming 6. Ice skating 7. Ice cream sundaes 8. Aquarium and fish 9. Picnics

10. Coin collection 11. Riding a horse 12. Fishing with Dad 13. _____ _ 14. _____ _ 15. _____ _

From Sign Here: A Contracting Book for Children and Their Parents (2nd ed., p. 115) by J. C. Dardig and W. L. Heward, 1981, Bridgewater, NJ: Fournies and Associates. Copyright 1981 by Fournies and Associates. Reprinted by permission.

565

Contingency Contracting, Token Economy, and Group Contingencies

Figure 7 A contingency contract for a nonreader.

people indicate the same reward. After List C is com- pleted, each person should collect his two other lists and read them carefully, talking over any misunder- stood items.

Step 5: Write Contracts. The final step begins with choosing a task for each person’s first contract. Discus- sion should move around the group with members trying to help each other decide which task is the most impor- tant to start doing first. Everyone should write who is going to perform the task, exactly what the task is, how well and when it has to be done, and any possible ex- ceptions. Everyone should also look at List C and choose a reward that is neither excessive nor insignifi- cant but is fair for the selected task. Each member should write who will control the reward, what the re- ward is, when it is to be given, and how much is to be given. Everyone in the group should write one contract during the first meeting.

Guidelines and Considerations in Implementing Contracts

In determining whether contingency contracting is an appropriate intervention for a given problem, the prac- titioner should consider the nature of the desired behav- ior change, the verbal and conceptual skills of the participant, the individual’s relationship with the per- son(s) with whom the contract will be made, and the available resources. The target behavior to be changed by a contingency contract must be in the person’s repertoire already and must typically be under proper stimulus con- trol in the environment in which the response is desired. If the behavior is not in the individual’s repertoire, other

behavior-building techniques should be attempted (e.g., shaping, chaining). Contracting is most effective with behaviors that produce permanent products (e.g., com- pleted homework assignment, cleaned bedroom) or that occur in the presence of the person who is to deliver the reward (e.g., the teacher or parent).

Reading ability by the participant is not a prerequi- site for successful contracting; however, the individual must be able to come under the control of the visual or oral statements (rules) of the contract. Contracting with nonreaders involves three types of clients: (a) preschool- ers with good verbal skills, (b) school-aged children with limited reading skills, and (c) adults with adequate language and conceptual skills but who lack reading and writing skills. Contracts using icons, symbols, pic- tures, photographs, audiotapes, or other nonword char- acterizations can be developed to suit the individual skills of children and adults in all three nonreader groups (see Figure 7).

Persons who refuse to enter into a contingency con- tract are another consideration. Whereas many children are eager, or at least willing, to try a contract, some want nothing to do with the whole idea. Using contingency contracting in a collaborative approach (Lassman, Jo- livette, & Wehby, 1999) may reduce the likelihood of noncompliance, and following a step-by-step method may help to ensure that consensus is reached at each de- cision point in the contract (Downing, 1990). However, the reality is that some nonsigners may not agree to par- ticipate in a contingency contract, even with the best of positive approaches built into the system. In such cases, another behavior change strategy would likely be a bet- ter alternative for dealing with the target behavior. Nu- merous lists of rules and guidelines for effective

566

Contingency Contracting, Token Economy, and Group Contingencies

Table 1 Guidelines and Rules for Contingency Contracts

Guidelines and rules for contracts Comment

Write a fair contract. There must be a fair relationship between the difficulty of the task and the amount of the reward. The goal is to achieve a win–win situation for both parties, not for one party to gain an advantage over the other.

Write a clear contract. In many instances a contract’s greatest advantage is that it specifies each person’s expectations. When a teacher’s or parent’s expectations are explicit, performance is more likely to improve. Contingency contracts must say what they mean and mean what they say.

Write an honest contract. An honest contract exists if the reward is delivered at the time and in the amount specified when the task is completed as agreed. In an honest contract the reward is not delivered if the task has not been completed as specified.

Build in several layers of rewards. Contracts can include bonus rewards for beating the best daily, weekly, or monthly performance. Adding these bonuses increases the motivational effect.

Add a response cost contingency, Occasionally, it may be necessary to incorporate a “fine”—the removal of rewards—if the agreed-upon task is not completed.

Post the contract in a visible place. Public posting allows all parties to see progress toward achieving the goals of the contract.

Renegotiate and change a con- Contracting is designed to be a positive experience for all parties, not a tedious endurance tract when either party is consis- contest to determine survivors. If the contract is not working, reconsider the task, the reward tently unhappy with it. components, or both.

Terminate a contingency contract. A contingency contract is a means to an end, not the end product. Once independent and proficient performance is achieved, the contract can be terminated. Further, a contract can and should be terminated when one party or both parties consistently fail to live up to the terms of the contract.

contingency contracting have been published (e.g., Dardig & Heward, 1976; Downing, 1990; Homme et al., 1970). Table 1 provides a list of the frequently cited guidelines and rules.

Evaluating Contracts

The evaluation of a contingency contract should focus on the objective measurement of the target behavior. The simplest way to evaluate a contract is to record the oc- curance of task completion. Including a task record on the contract helps make evaluation a natural by-product of the contracting process. By comparing the task record with a precontract baseline of task completion, an ob- jective determination can be made as to whether im- provement has occurred. A good outcome is one that results in the specified task being completed more often than it was before the contract.

Sometimes the outcome data indicate that the task is being completed more often and more consistently than it was before the contract, but the parties involved are still not happy. In such cases, either the original problem or goal that prompted the development of the contract is not being attained, or one or more of the participants do not like the way in which the contract is being carried out. The first possibility results from se- lecting the wrong behavior for the task part of the con-

tract. For example, let us suppose that John, a ninth- grader, wants to improve on the Ds and Fs he has been getting in algebra and writes a contract with his par- ents specifying as the task “studying his math” for 1 hour each school night. After several weeks John has failed to study for the required 1 hour on only two nights, but his in-school algebra performance remains unchanged. Has John’s contract worked? The correct answer is both yes and no. John’s contract was suc- cessful in that he was consistently completing the spec- ified task—1 hour of study each day. However, in terms of his original objective—better algebra grades—the contract was a failure. John’s contract helped him change the behavior he specified, but he specified the wrong task. Studying for 1 hour, for John at least, was not directly related to his goal. By changing his contract to require that he solve correctly 10 algebra equations each night (the behavior required to get good grades on algebra tests), his goal of obtaining better grades may be- come a reality.

It is also important to consider the participant’s re- actions to the contract. A contract that produces desired change in the specified target behavior but causes other maladaptive or emotional responses may be an unac- ceptable solution. Having the client in the negotiation development of the contract and jointly conducting reg- ular progress checks helps to avoid this situation.

567

Contingency Contracting, Token Economy, and Group Contingencies

Token Economy The token economy is a highly developed and researched behavior change system. It has been applied successfully in virtually every instructional and therapeutic setting pos- sible. The usefulness of a token economy in changing be- haviors that have been resistant to instruction or therapy is well established in the literature (Glynn, 1990; Musser, Bray, Kehle, & Jenson, 2001). In this section we describe and define token economy and outline effective proce- dures for using it in applied settings.

Definition of a Token Economy

A token economy is a behavior change system con- sisting of three major components: (a) a specified list of target behaviors; (b) tokens or points that partici- pants receive for emitting the target behaviors; and (c) a menu of backup reinforcer items—preferred items, activities, or privileges—that participants obtain by ex- changing tokens they have earned. Tokens function as generalized conditioned reinforcers for the target be- haviors. First, behaviors to be reinforced are identified and defined. Second, a medium of exchange is selected; that medium of exchange is a symbol, object, or item, called a token. Third, backup reinforcers are provided that can be purchased with the token. Store-based or manufacturer coupons are analogous to a token econ- omy. When a customer purchases an item from a par- ticipating store, the cashier provides the purchaser with the coupon—the medium of exchange—that serves as the token. The coupon is traded later for another item at a reduced price, or it is redeemed immediately for the backup reinforcer. Money is another example of a token that can be exchanged at a later time for backup objects and activities (e.g., food, clothing, transporta- tion, entertainment).

Carton and Schweitzer (1996), for example, imple- mented a token economy to increase the compliance be- havior of a 10-year-old boy who was hospitalized for

severe renal disease and who required regular hemodial- ysis. The patient developed a noncompliant repertoire that affected his interactions with nurses and caretakers. During baseline, the number of 30-minute intervals of noncompliance was measured by dividing a 4-hour time block into eight, 30-minute segments. When the token economy was introduced, the boy was told that he could earn one token for each 30-minute period that passed without a noncompliant episode. Tokens were exchanged weekly for baseball cards, comics, and toys.

Carton and Schweitzer reported a functional relation between the onset of the token economy and the reduction in noncompliant behavior. When tokens were in effect, noncompliant behavior was virtually eliminated. Follow- up data collected 3 months and 6 months after termination of token reinforcement yielded continued evidence of low noncompliance to nurse and caretaker requests.

Higgins, Williams, and McLaughlin (2001) used a token economy to decrease the disruptive behaviors of an elementary-age student with learning disabilities. The student exhibited high levels of out-of-seat behavior, talk- ing out, and poor sitting posture. After collecting baseline data on the number of out-of-seat, talking out, and poor posture responses, a token economy was implemented. The student earned a check mark that was exchangeable for free time after each minute if an alternative behavior was being emitted instead of the behaviors targeted for re- duction. Maintenance checks were taken on two subse- quent occasions to determine duration effects. Figure 8 displays the results of the study across the three depen- dent variables. A functional relation was established be- tween the onset of the token economy and the reduction of the behaviors. Further, maintenance checks indicated that the behavior remained at low levels beyond the ter- mination of the token economy.

Level Systems

A level system is a type of token economy in which par- ticipants move up (and sometimes down) a hierarchy of levels contingent on meeting specific performance crite- ria with respect to the target behaviors. As participants move “up” from one level to the next level, they have ac- cess to more privileges and are expected to demonstrate more independence. The schedule of token reinforcement is gradually thinned so that participants at the highest lev- els are functioning on schedules of reinforcement that are similar to those in natural settings.

According to Smith and Farrell (1993) level systems are an outgrowth of two major educational advances that began in the late 1960s and 1970s: (a) Hewett’s Engineered Classroom (1968), and (b) Phillips, Phillips, Fixen, and Wolf’s Achievement Place (1971). In both

A token is an example of a generalized conditioned reinforcer. It can be exchanged for a wide variety of backup reinforcers. Generalized conditioned rein- forcers are independent of specific states of motivation because they are associated with a wide variety of backup reinforcers. However, generalized conditioned reinforcement is a relative concept: Effectiveness de- pends to a large extent on the extensiveness of the backup reinforcers. Tokens exchangeable for a wide variety of backup reinforcers have considerable utility in schools, clinics, and hospitals where it is difficult for personnel to control the deprivation states of their clients.

568

Contingency Contracting, Token Economy, and Group Contingencies

T al

k- O

u ts

Baseline Token Economy Maintenance

O u

t- o

f-S ea

P o

o r

P o

st u

1 3 5 7 9 11 13 15 17 19

Observation Days

Figure 8 The number of talk-outs, out-of- seat behavior, and poor posture during baseline, token economy, and maintenance conditions. From “The Effects of a Token Economy Employing Instructional Consequences for a Third-Grade Student with Learning Disabilities: A Data-Based Case Study” by J. W. Higgins, R. L. Williams, and T. F. McLaughlin, 2001, Education and Treatment of Children, 24 (1), p. 103. Copyright 2001 by The H. W. Wilson Company. Reprinted by permission.

cases, systematic academic and social programming com- bined token reinforcement, tutoring systems, student self- regulation, and managerial arrangements. Smith and Farrell stated that level systems are designed to

foster a student’s improvement through self-manage- ment, to develop personal responsibility for social, emo- tional, and academic performance . . . and to provide student transition to a less restrictive mainstream set- ting. . . . Students advance through the various levels as they show evidence of their achievement. (p. 252)

In a level system, students must acquire and achieve increasingly more refined repertoires while tokens, social praise, or other reinforcers are simultaneously decreased. Level systems have built-in mechanisms for participants to advance through a series of privileges, and they are based on at least three assumptions: (a) combined tech- niques—so called “package programs”—are more ef-

fective than individual contingencies introduced alone, (b) student behaviors and expectations must be stated ex- plicitly, and (c) differential reinforcement is necessary to reinforce closer and closer approximations to the next level (Smith & Farrell, 1993).

Lyon and Lagarde (1997) proposed a three-level group of reinforcers that placed less desirable reinforcers at Level 1. At this level, students had to earn 148 points or 80% of the 185 maximum points that could be earned during the week to purchase certain items. At Level 3, the highly desirable reinforcers could be purchased only if the students had accumulated at least 167 points, or 90% of the total points possible. As the levels progress, the expectations for performance increase.

Cavalier, Ferretti, and Hodges (1997) incorporated a self-management approach with an existing level sys- tem to improve the academic and social behavior of two adolescent students with learning disabilities for whom

569

Contingency Contracting, Token Economy, and Group Contingencies

increased participation in a general education classroom was an individualized education program (IEP) goal. Basically, other students in the classroom were making adequate progress through the six-level point system that the teacher had devised, but the inappropriate verbaliza- tions of the two target students during class stalled their progress at Level 1. After collecting baseline data on the number of inappropriate verbalizations, the researchers trained the students to self-record occurrences of these behaviors during two 50-minute intervals in the day. In- appropriate verbalizations were defined explicitly, and during mock trials the students practiced self-recording and received feedback based on teacher observations during the same interval. An emphasis was placed on the accuracy of the students’ recording, and desirable reinforcers were awarded for accurate recording. During the intervention (level system plus self-recording), stu- dents were told that they would be observed during the two 50-minute intervals for accuracy. If they met the cri- terion for each level (e.g., a decrease of five inappropri- ate verbalizations from the previous session), a reinforcer was delivered. As the students progressed through the levels, the reinforcer changed to more highly desirable items. Results showed a high number of verbalizations during baseline; however, when the packaged intervention was initiated for Student 1, inappropriate verbalizations decreased. This outcome was replicated for Student 2, confirming a functional relation between the the inter- vention and the decrease in inappropriate verbalizations.

Designing a Token Economy

The basic steps in designing and preparing to implement a token economy are as follows:

1. Select tokens that will serve as a medium of ex- change (e.g., points, stickers, plastic chips).

2. Identify target behaviors and rules.

3. Select a menu of backup reinforcers.

4. Establish a ratio of exchange.

5. Write procedures to specify when and how tokens will be dispensed and exchanged and what will hap- pen if the requirements to earn a token are not met. Will the system include a response cost procedure?

6. Field-test the system before full-scale implementa- tion.

Selecting Tokens

A token is a tangible symbol that can be given immedi- ately after a behavior and exchanged later for known rein- forcers. Frequently used tokens include washers, checkers,

coupons, poker chips, points or tally marks, teacher ini- tials, holes punched in a card, and strips of plastic. Crite- ria to consider in selecting the token itself are important. First, the token should be safe; it should not be harmful to the learners. If a very young child or a person with severe learning or behavioral problems is to receive the token, it should not be an item that can be swallowed or used to cause injury. Second, the analyst should control token pre- sentation; learners should not be able to bootleg the deliv- ery of the tokens. If tally marks are used, they should be on a special card or made with a special marking pen that is available only to the analyst. Likewise, if holes are punched in a card, the paper punch should be available only to the analyst to avoid counterfeiting.

Tokens should be durable because they may have to be used for an extended period of time, and they should be easy to carry, handle, bank, store, or accumulate. Tokens should also be readily accessible to the practitioner at the moment they are to be dispensed. It is important that they be provided immediately after the target behavior. Tokens should be inexpensive; there is no need to spend a large sum of money to purchase tokens. Rubber stamps, stars, check marks, and buttons are all inexpensive items that can be used as tokens. Finally, the token itself should not be a desirable object. One teacher used baseball cards as tokens, but the students spent so much time interacting with the tokens (e.g., reading about players) that the to- kens distracted students from the purpose of the system.

For some students, the object of an obsession can serve as a token reinforcer (Charlop-Christy & Haymes, 1998). In their study, three children with autism who at- tended an after-school program served as participants. All three children were consistently off task during activities, preoccupied with some objects, and engaged in self-stim- ulatory behaviors. During baseline, the students earned stars for appropriate behavior. Inappropriate behaviors or incorrect responses were addressed by saying “try again” or “no.” When the students earned five stars, the tokens could be exchanged for backup reinforcers (e.g., food, pencils, tablets). During the token condition, an “obses- sion” object (one the children had been previously pre- occupied with) was used as the token. Once they had earned five obsession objects, children could exchange them for food items or other known reinforcers. Charlop- Christy and Haymes reported that the overall pattern of re- sponding demonstrated that when the token was an obsession object, student performance improved.

Identifying Target Behaviors and Rules

The criteria presented

570

Contingency Contracting, Token Economy, and Group Contingencies

also apply to the selection and definition of rules and tar- get behaviors for a token economy. Generally, the guide- lines for selecting behaviors for a token economy include (a) selecting only measurable and observable behaviors; (b) specifying criteria for successful task completion; (c) starting with a small number of behaviors, including some that are easy for the individual to accomplish; and (d) being sure the individual possesses the prerequisite skills for any targeted behaviors (Myles, Moran, Ormsbee, & Downing, 1992).

After rules and behaviors that apply to everyone are defined, then criteria and behaviors specific to individual learners should be established. Many token economy failures can be traced to requiring the same behaviors and setting the same criteria for all learners. Token economies usually need to be individualized. For example, in a class- room setting the teacher may want to select different be- haviors for each student. Or perhaps the token economy should not be applied to all students in the classroom. Perhaps only the lowest functioning students in the class- room should be included. However, students who do not need a token system should still continue to receive other forms of reinforcement.

Selecting a Menu of Backup Reinforcers

Most token economies can use naturally occurring ac- tivities and events as backup reinforcers. For example, in a classroom or school setting, tokens can be used to buy time with popular games or materials, or they can be ex- changed for favorite classroom jobs such as office mes- senger, paper passer, teacher assistant, or media operator. Tokens can also be used for schoolwide privileges such as a library or study hall pass; a special period (e.g., phys- ical education) with another class; or special responsi- bilities such as school patrol, cafeteria monitor, or tutor. Higgins and colleagues (2001) used naturally occurring activities and events as backup reinforcers for a token economy in their study (i.e., the students had access to computer games and leisure books). However, play ma- terials, hobby-type games, snacks, television time, al- lowance, permission to go home or downtown, sports events, and coupons for gifts or special clothing could also be used as backup reinforcers because these objects or items tend to occur in many settings.

If naturally occurring activities and events fail, then backup items not ordinarily present in a particular pro- gram can be considered (e.g., pictures of movie or sports stars, CDs or DVDs, magazines, or edibles). Such items should generally be considered only when more naturally occurring activities have proven ineffective. Using the least intrusive and most naturally occurring reinforcers is recommended.

Selection of backup reinforcers should follow con- sideration of ethical and legal issues, as well as state and local education agency policies. Token reinforcement contingencies that would deny the learner basic needs (e.g., food) or access to personal or privileged informa- tion or events (e.g., access to mail, telephone privileges, attending religious services, medical care, etc.) should not be used. Furthermore, general comforts that are as- sociated with basic rights afforded to all citizens should not be used in a token program (e.g., clean clothing, ad- equate heating, ventilation, hot water).

Establishing a Ratio of Exchange

Initially, the ratio between the number of tokens earned and the price of backup items should be small to provide immediate success for learners. Thereafter, the ratio of exchange should be adjusted to maintain the responsive- ness of the participants. Following are several general guidelines for establishing the ratio between earned to- kens and the price of backup items:

1. Keep initial ratios low.

2. As token-earning behaviors and income increase, increase the cost of backup items, devalue tokens, and increase the number of backup items.

3. With increased earnings, increase the number of luxury backup items.

4. Increase the prices of necessary backup items more than those of luxury items.

Myles and colleagues (1992) provided guidelines for es- tablishing and maintaining a token economy, including the distribution and redemption of tokens. The next section addresses frequently asked questions related to tokens.

What procedure will be used to dispense tokens? If objects such as tally marks or holes punched in a card are selected as the tokens, how the learner will receive them is obvious. If objects such as coupons or poker chips are used, there should be some container for stor- ing the accumulated tokens before they are exchanged for the backup items. Some practitioners have learners construct individual folders or containers for storing their tokens. Another suggestion is to deposit the tokens through cut slots in the plastic tops of coffee cans. With younger learners tokens can be chained to form a neck- lace or bracelet.

How will the tokens be exchanged? A menu of the backup items should be provided with a given price for

571

Contingency Contracting, Token Economy, and Group Contingencies

each item. Learners can then select from the menu. Many teachers have a table store with all the items dis- played (e.g., games, balloons, toys, certificates for priv- ileges). To avoid noise and confusion at shopping time, individual orders can be filled by writing in or checking off the items to be purchased. Those items are then placed in a bag with the order form stapled to the top and are returned to the purchaser. Initially, the store should be open frequently, perhaps twice per day. Lower functioning learners may need more frequent ex- change periods. Later, exchange periods might be avail- able only on Wednesdays and Fridays, or only on Fridays. As quickly as possible, token exchange should occur on an intermittent basis.

Writing Procedures to Specify What Happens if Token Requirements Are Not Met

Occasionally, token requirements will not be met for one reason or another. One approach is to nag the individual: “You didn’t do your homework. You know your home- work must be completed to earn tokens. Why didn’t you do it?” A better approach is a matter-of-fact restatement of the contingency: “I’m sorry. You haven’t enough to- kens to exchange at this time. Try again.” It is important to know whether the individual has the skills required to earn tokens. A learner should always be able to meet the response requirements.

What Should Be Done When a Learner Tests the Sys- tem? How should a practitioner respond when a learner says she doesn’t want any tokens or backup items? One approach is to argue, debate, or cajole the learner. A better approach is to say something neutral (e.g., “That is your decision”) and then walk away, pre- cluding any argument or debate. In this way a con- frontation is avoided, and the occasion remains set for token delivery for the learner. Most learners can and should have input in selecting the backup items, gener- ating the rules for the economy, establishing the price for the backup items, and performing general duties in managing the system. A learner can be a salesperson for the store or a bookkeeper to record who has how many tokens and what items are purchased. When learners are involved and their responsibilities for the economy are emphasized, they are less likely to test the system.

Will the Token Economy Include a Response Cost Procedure? Most token economies do include a token loss contingency for inappropriate behaviors and rule infractions (Musser et al., 2001). Any behaviors subject to response cost should be defined and stated clearly in the rules. Learners need to be aware of what actions

will result in token loss and how much the behavior will cost. The more serious the inappropriate behavior is, the greater the token loss should be. Clearly, fight- ing, acting out, or cheating should result in greater token loss than minor infractions (e.g., out-of-seat be- havior or talk-outs). Token loss should never be applied to a behavior if the learner does not have tokens. Stu- dents should not be allowed to go into debt, which would likely decrease the reinforcement value of the to- kens. A learner should always earn more tokens than she loses.

Field-Testing the System

The final step before actually implementing a token sys- tem is to field-test it. For 3 to 5 days token delivery is tallied exactly as if tokens were being earned, but no to- kens are actually awarded during the field test. Data from the field test are used for assessment. Are learners actu- ally deficient in the targeted skills? Are some learners demonstrating mastery of behaviors targeted for inter- vention? Are some learners not receiving tokens? Based on answers to questions such as these, final adjustments in the system can be made. For some learners, more dif- ficult behaviors may need to be defined; others may need less demanding target behaviors. Perhaps more or fewer tokens need to be delivered relative to the price of the backup reinforcers.

Implementing a Token Economy

Initial Token Training

The manner in which initial training is conducted to im- plement a token economy depends on the functioning level of the learners. For high-functioning learners and those with mild disabilities, initial training might require minimal time and effort and consist primarily of verbal instructions or modeling. Usually the initial token train- ing for these individuals can be accomplished in one 30- to 60-minute session. Three steps are normally sufficient. First, an example of the system should be given. The prac- titioner might describe the system as follows:

This is a token and you can earn it by [specify behav- ior]. I will watch your behavior; and when you accom- plish [specify behavior], you will earn a token. Also, as you continue [specify behavior], you will earn more to- kens. At [specify time period] you will be able to ex- change the tokens you have earned for whatever you want and can afford on this table. Each item is marked with the number of tokens needed for purchase. You can spend only the tokens you have earned. If you want an

572

Contingency Contracting, Token Economy, and Group Contingencies

item that requires more tokens than you have earned, you will have to save your tokens over several [specify time period].

The second step is to model the procedure for token delivery. For instance, each learner might be directed to emit the specified behavior. Immediately following the occurrence of the behavior, the learner should be praised (e.g., “Enrique, I’m pleased to see how well you are working by yourself!”) and the token delivered.

The third step is to model the procedure for token exchange. Learners should be taken to the store and shown the items for purchase. All learners should already have one token, which was acquired during the modeling of token delivery. At this time, several items should be able to be purchased for one token (the price may go up later)—a game, 5 minutes of free time, a pencil-sharp- ening certificate, or teacher helper privilege. Students should actually use their tokens in this exchange. Lower functioning learners may require several sessions of ini- tial token training before the system is functional for them. Further response prompts may be needed.

Ongoing Token Training

During token reinforcement training, the practitioner and the students should follow the guidelines for effective use of reinforcement. For example, tokens should be dis- pensed contingently and immediately after the occur- rence of the desired behavior. Procedures for delivery and exchange should be clear and should be followed consistently. If a booster session is needed to improve student understanding of how tokens are earned and ex- changed, practitioners should do so early in the program. Finally, the focus should be on building and increasing desirable behaviors through token delivery rather than decreasing undesirable behaviors through response cost.

As part of the overall training, the behavior analyst may choose to take part in the token economy as well. For example, the analyst could pinpoint a personal behavior to be increased and then model how to act if her perfor- mance of that behavior does not meet the criterion to earn a token, how to save tokens, and how to keep track of progress. After 2 to 3 weeks, a revision in the token econ- omy system may be needed. It is usually desirable to have learners discuss behaviors they want to change, the backup items they would like to have available, or the schedule of exchange. If some participants rarely earn tokens, a sim- pler response or prerequisite skill may be necessary. On the other hand, if some participants always earn all possi- ble tokens, requirements may need to be changed to a more complex skill.

Management Issues During Implementation

Students must be taught how to manage the tokens they earn. For instance, once received, tokens should be placed in a safe, but accessible container so that they are out of the way, but readily available when needed. If the tokens are in clear view and readily accessible, some students might play with them at the expense of performing aca- demic tasks assigned by the teacher. Also, placing the to- kens in a secure location reduces the risk of other students counterfeiting or stealing them. Preemptive measures should be taken to ensure that tokens are not easily coun- terfeited or reachable by anyone other than the recipient. If counterfeiting or stealing occurs, switching to different tokens will help reduce the likelihood of these tokens being exchanged under false pretenses.

Another management issue, however, relates to stu- dents’ token inventories. Some students may hoard their tokens and not exchange them for backup reinforcers. Other students may try to exchange their tokens for a backup reinforcer, but they lack the requisite number of tokens to do so. Both extremes should be discouraged. That is, students should be required to exchange at least some of their earned tokens periodically, and students without the requisite number of tokens should not be per- mitted to participate in an exchange. That is, they should not be permitted to buy backup reinforcers on credit.

A final management issue relates to chronic rule breakers or students who test the system at every turn. Practitioners can minimize this situation by (a) ensuring that the token does serve as a generalized conditioned re- inforcer, (b) conducting a reinforcer assessment to deter- mine that the backup reinforces are preferred by the students and function as reinforcers, and (c) applying re- sponse cost procedures for chronic rule breakers.

Withdrawing the Token Economy

Strategies for promoting generalization and maintenance of target behaviors to settings in which a token or level system is not used should be considered in the design and implementation of a token or level system. Before ap- plying the initial token program, analysts should plan how they will remove the program. One goal of the token program should be to have the descriptive verbal praise that is delivered simultaneously with the token acquire the reinforcing capability of the token. From the begin- ning, a systematic goal of the token economy should be to withdraw the program. Such an approach, aside from having functional utility for practitioners (i.e., they will not have to issue tokens forever), also has advantages for the learner. For example, if a special education teacher is using a token economy with a student scheduled for

573

Contingency Contracting, Token Economy, and Group Contingencies

full-time placement in a regular fourth grade classroom, the teacher wants to be certain that the student’s responses can be maintained in the absence of the token economy. It is unlikely that the student would encounter a similar token system in the general education classroom.

Various methods have been used to withdraw token reinforcers gradually after criterion levels of behavior have been reached. The following six guidelines allow the practitioner to develop, and later withdraw, token re- inforcers effectively. First, the token presentation should always be paired with social approval and verbal praise. This should increase the reinforcing effect of the social approval and serve to maintain behaviors after token withdrawal.

Second, the number of responses required to earn a token should be gradually increased. For instance, if a student receives a token initially after reading only one page, he should be required to read more pages later for token delivery.

Third, the duration the token economy is in effect should be gradually decreased. For example, during Sep- tember the system might be in effect all day; in October the time might be 8:30 A.M. to 12:00 P.M. and 2:00 P.M. to 3:00 P.M.; and in November, 8:30 A.M. to 10:00 A.M. and 2:00 P.M. to 3:00 P.M. In December the times might be the same as those in November, but on only 4 days a week, and so on.

Fourth, the number of activities and privileges that serve as backup items and are likely to be found in the un- trained setting should be increased gradually. For exam- ple, the analyst should start taking away tangible items in the store that might not be present in the regular class- room. Are edibles available at the store? They are usually not available for reinforcement in regular classrooms. Gradually, items should be introduced that would be com- mon in a regular class (e.g., special award sheets, gold stars, positive notes home).

Fifth, the price of more desirable items should be in- creased systematically while keeping a very low price on less desirable items for exchange. For example, in a token system with adolescent girls with moderate to severe mental retardation, the price of candy bars and trips to the canteen and grooming aids (e.g., comb, deodorant) was initially about the same. Slowly, the cost of items such as candy bars was increased to such a high level that the girls no longer saved tokens to purchase them. More girls used their tokens to purchase grooming aids, which cost substantially less than candy.

Sixth, the physical evidence of the token should be faded over time. The following sequence illustrates how the physical evidence of the token can be faded.

• Learners earn physical tokens such as poker chips or washers.

• The physical tokens are replaced with slips of paper.

• The slips of paper are replaced with tally marks on an index card that is kept by the learners.

• In a school setting the index card can now be taped to the learner’s desk.

• The index card is removed from the learner and is kept by the analyst, but participants can check their balance at any time.

• The analyst keeps tallies, but no checking is al- lowed during the day. Totals are announced at the end of the day, and then every other day.

• The token system is no longer operative. The be- havior analyst does not announce point totals even though they are still kept.

Evaluating the Token Economy

Token economies can be evaluated using any number of reliable, valid, field-tested, best practice designs. Given that most token economy programs are conducted with small groups, we recommend that single-subject evalua- tion designs be used such that the participant serves as his or her own control. Further, we suggest that social vali- dation data on the target participants and significant oth- ers who come into contact with the person be collected before, during, and after token intervention.

Reasons for the Effectiveness of the Token Economy

A token economy is often effective in applied settings for three reasons. First, tokens bridge the time gap be- tween the occurrence of a behavior and delivery of a backup reinforcer. For example, a token may be earned during the afternoon, but the backup reinforcer is not awarded until the next morning. Second, tokens bridge the setting gap between the behavior and the delivery of the backup reinforcer. For instance, tokens earned at school could be exchanged for reinforcers at home, or tokens earned in a general education classroom in the morning could be exchanged in a special education class- room in the afternoon. Finally, as generalized conditioned reinforcers, tokens make the management of motivation less critical for the behavior analyst.

Further Considerations

Intrusive. Token systems can be intrusive. It takes time, energy, and resources to establish, implement, and evaluate token programs. Also, because most natural

574

Contingency Contracting, Token Economy, and Group Contingencies

environments do not reinforce a person’s behavior with tokens, careful thought must be given to how to thin the token schedule while simultaneously maintaining per- formance. In any case, token economy programs can have lots of “moving parts,” and practitioners must be prepared to deal with them.

Self-Perpetuating. A token economy can be an effective procedure for managing behavior, and ana- lysts can be so encouraged by the results that they do not want to remove the system. Learners then continue working for reinforcement that is not normally avail- able in the natural environment.

Cumbersome. Token economies can be cumber- some to implement, especially if there are multiple par- ticipants with multiple schedules of reinforcement. The system may require additional time and effort from the learner and the behavior analyst.

Federal Mandates. When tokens are introduced within the context of a level system, practitioners should exercise caution that the explicit and uniform re- quirements for students to earn a specific number of to- kens before progressing to the next level does not violate the spirit or intent of federal mandates that call for individualized programs. Scheuermann and Webber (1996) suggested that tokens and other programs em- bedded within level systems be individualized and that self-management techniques be combined with the level system to increase the likelihood of a successful inclusion program.

Group Contingencies Thus far in the text we have focused primarily on how contingencies of reinforcement can be applied to change the future frequency of certain behaviors of individual persons. Applied research has also demonstrated how contingencies can be applied to groups, and behavior an- alysts have increasingly turned their attention toward group contingencies in areas such as leisure activities for adults (Davis & Chittum, 1994), schoolwide applications (Skinner, Skinner, Skinner, & Cashwell, 1999), class- rooms (Brantley & Webster, 1993; Kelshaw-Levering, Sterling-Turner, Henry, & Skinner, 2000; Skinner, Cash- well, & Skinner, 2000), and playgrounds (Lewis, Powers, Kelk, & Newcomer, 2002). Each of these applications has shown that group contingencies, properly managed, can be an effective and practical approach to changing

the behavior of many people simultaneously (Stage & Quiroz, 1997).

Definition of a Group Contingency

A group contingency is one in which a common conse- quence (usually, but not necessarily, a reward intended to function as reinforcement) is contingent on the behavior of one member of the group, the behavior of part of the group, or the behavior of everyone in the group. Group contingencies can be classified as dependent, indepen- dent, or interdependent (Litow & Pumroy, 1975).

Rationale for and Advantages of Group Contingency

There are a number of reasons for using a group contin- gency in applied settings. First, it can save time during ad- ministration (Skinner, Skinner, Skinner, & Cashwell, 1999). Instead of repeatedly administering a consequence to each member of a group, the practitioner can apply one consequence to all members of the group. From a lo- gistical perspective, a practitioner’s workload may be re- duced. Group contingencies have been demonstrated to be effective in producing behavior change (Brantley & Webster, 1993). A group contingency can be effective and economical, requiring fewer practitioners or less time to implement.

Another advantage is that a practitioner can use a group contingency in a situation in which an individual contingency is impractical. For example, a teacher at- tempting to reduce disruptive behaviors of several stu- dents might have difficulty administering an individual program for each pupil in the classroom. A substitute teacher, in particular, might find the use of a group con- tingency a practical alternative because her knowledge of the students’ previous histories of reinforcement would be limited, and the group contingency could be applied across a variety of behaviors, settings, or students.

A group contingency can also be used in cases in which the practitioner must resolve a problem quickly, as when serious disruptive behavior occurs. The prac- titioner might be interested not only in decreasing the disruptive behavior rapidly, but also in building im- proved levels of appropriate behavior (Skinner et al., 2000).

Furthermore, a practitioner can use a group contin- gency to capitalize on peer influence or peer monitoring because this type of contingency sets the occasion for peers to act as change agents (Gable, Arllen, & Hen- drickson, 1994; Skinner et al., 1999). Admittedly, peer

575

Contingency Contracting, Token Economy, and Group Contingencies

bonus

Figure 9 An independent group contingency.

A B C

Criterion stated for one

individual or small group*

Criterion met SR+ for whole group

*(E.g., “When all students at Table 2 finish their math assignments, the class will have 5 minutes of free time.”)

Figure 10 A dependent group contingency.

pressure can have a detrimental effect on some people; they may become scapegoats, and negative effects may surface (Romeo, 1998). However, potentially harmful or negative outcomes can be minimized by structuring the contingency elements randomly (Kelshaw-Levering et al., 2000; Poplin & Skinner, 2003).

Practitioners can establish a group contingency to facilitate positive social interactions and positive be- havioral supports within the group (Kohler, Strain, Maretsky, & DeCesare, 1990). For example, a teacher might establish a group contingency for a student or a group of students with disabilities. The students with disabilities might be integrated into the general education classroom, and a contingency could be arranged in such a way that the class would be awarded free time contin- gent on the performance of one or more of the students with disabilities.

Independent Group Contingency Applications

An independent group contingency is an arrangement in which a contingency is presented to all members of a group, but reinforcement is delivered only to those group members who meet the criterion outlined in the contin- gency (see Figure 9) Independent group contingencies are frequently combined with contingency contracting and token reinforcement programs because these pro- grams usually establish reinforcement schedules inde- pendent of the performance of other members of the group.

Brantley and Webster (1993) used an independent group contingency in a general education classroom to decrease the disruptive behavior of 25 fourth-grade stu- dents. After collecting data on off-task behavior, call-outs, and out-of-seat behavior, the teachers posted rules related to paying attention, seeking the teachers’ permission be- fore talking, and remaining in their seats. An independent group contingency was established whereby each student

could earn a check mark next to his or her name on a list that was posted publicly in the room during any of the in- tervals that marked the observation periods during the day. When a student emitted an appropriate or prosocial behavior, a check mark was registered. The criterion for earning a reward was increased from four to six check marks over 4 out of 5 days per week.

Results showed that after 8 weeks, the total number of combined disruptions (e.g., off-task behavior, call- outs, and out-of-seat behavior) decreased by over 70%, and some off-task behaviors (e.g., not keeping hands to self) were eliminated completely. The teacher’s satisfac- tion with the approach was positive, and parents reported that they were able to understand the procedures that were in place for their children at school. Brantley and Web- ster (1993) concluded:

The independent contingency added structure for stu- dents by using clear time intervals and clarified teacher expectations by limiting and operationally defining rules to be followed, monitoring behavior consistently, and setting attainable criteria for students. (p. 65)

Dependent Group Contingency Applications

Under a dependent group contingency the reward for the whole group is dependent on the performance of an individual student or small group. Figure 10 il- lustrates the dependent group contingency as a three- term contingency. The contingency operates like this: If an individual (or small group within the total group) performs a behavior to a specific criterion, the group shares the reinforcer. The group’s access to the reward depends on the individual’s (or small group’s) perfor- mance. If the individual performs below the criterion, the reward is not delivered. When an individual, or small group, earns a reward for a class, the contingency is sometimes referred to as the hero procedure.

576

Contingency Contracting, Token Economy, and Group Contingencies

A B C

Criterion stated for all

group members*

Criterion met SR+ for whole group if all

members of the group

achieve criterion

*(E.g., "Each student must complete at least four science projects by the 6th week in the term in order for the class to go on the field trip.")

Figure 11 An interdependent group contingency.

According to Kerr and Nelson (2002), the hero proce- dure can facilitate positive interactions among students because the class as a whole benefits from the improved behavior of the student targeted for the group con- tingency.

Gresham (1983) conducted a dependent group con- tingency study in which the contingency was applied at home, but the reward was delivered at school. In this study an 8-year-old boy who was highly destructive at home (e.g., set fires, destroyed furniture) earned good notes for nondestructive behavior at home. Billy received a good note—a daily report card—each day that no de- structive acts took place. Each note was exchangeable for juice, recess, and five tokens at school the next day. After Billy received five good notes, the whole class re- ceived a party, and Billy served as the host. Gresham re- ported that the dependent group contingency reduced the amount of destructive behavior and represented the first application of a dependent group contingency in a com- bined home–school setting.

Allen, Gottselig, and Boylan (1982) used an inter- esting variation of the dependent group contingency. In their study eight disruptive third-graders from a class of 29 students served as target students. On the first day of intervention the teacher posted and explained classroom rules for hand raising, leaving a seat, dis- turbing others, and getting help. Contingent on reduced amounts of disruptive behavior during 5-minute inter- vals in math and language arts, the class earned 1 extra minute of recess time. If a disruptive behavior occurred during the 5-minute interval, the teacher cited the dis- rupter for the infraction (e.g., “James, you disturbed Sue”) and reset the timer for another 5-minute inter- val. The teacher also posted the accumulated time on an easel in full view of the class. The results indicated that reduced disruptive behavior occurred under the depen- dent group contingency.

Interdependent Group Contingencies

An interdependent group contingency is one in which all members of a group must meet the criterion of the contingency (individually and as a group) before any member earns the reward (Elliot, Busse, & Shapiro, 1999; Kelshaw-Levering et al., 2000; Lewis et al., 2002; Skinner et al., 1999; Skinner et al., 2000). Theoretically, interdependent group contingencies have a value-added advantage over dependent and independent group con- tingencies insofar as they yoke students to achieve a common goal, thereby capitalizing on peer pressure and group cohesiveness.

The effectiveness of dependent and interdependent group contingencies may be enhanced by randomly ar- ranging some or all of the components of the contingency (Poplin & Skinner, 2003). That is, randomly selected stu- dents, behaviors, or reinforcers are targeted for the con- tingency (Kelshaw-Levering et al., 2000; Skinner et al., 1999). Kelshaw-Levering and colleagues (2000) demon- strated that randomizing either the reward alone or mul- tiple components of the contingency (e.g., students, behaviors, or reinforcers) was effective in reducing dis- ruptive behavior.

Procedurally, an interdependent group contingency can be delivered (a) when the group as a whole meets the criterion, (b) when the group achieves a mean group score, or (c) based on the results of the Good Behavior Game or the Good Student Game. In any case, interde- pendent group contingencies represent an “all or none” arrangement. That is, all students earn the reward or none of them do (Poplin & Skinner, 2003). Figure 11 il- lustrates the interdependent group contingency as a three- term contingency.

Total Group Meets Criterion

Lewis and colleagues (2002) used the total group meets criterion variation to reduce the problem playground behaviors of students enrolled in a suburban elementary school. After a faculty team conducted an assessment of problematic playground behaviors, social skill in- struction in the classroom and on the playground was coupled with a group contingency. During social skills instruction, students learned how to get along with friends, cooperate with each other, and be kind. During the group contingency, students earned elastic loops that they could affix to their wrists. After recess, stu- dents placed the loops in a can on the teacher’s desk. When the can was full, the group earned a reinforcer.

577

Contingency Contracting, Token Economy, and Group Contingencies

Baseline Intervention

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Recess 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Recess 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Recess 3

Daily Sessions

Fr eq

u en

cy o

f P

ro b

le m

B eh

av io

r Figure 12 Frequency of problem behaviors across recess periods. Recess 1 was composed of second- and fourth-grade students, Recess 2 was composed of first- and third-grade students, and Recess 3 was composed of fifth- and sixth-grade students. Kindergarten students were on the play- ground across Recess 1 and 2. From “Reducing Problem Behaviors on the Playground: An Investigation of the Application of School-Wide Positive Behavior Supports” by T. J. Lewis, L. J. Powers, M. J. Kelk, and L. L. Newcomer, 2002, Psychology in the Schools, 39 (2), p. 186. Copyright 2002 by Wiley Periodicals, Inc. Reprinted by permission of John Wiley & Sons, Inc.

Figure 12 shows the results of the social skills plus group contingency intervention across three recess pe- riods during the day.

Group Averaging

Baer and Richards (1980) used a group averaging inter- dependent group contingency to improve the math and English performance of 5 elementary-aged students. In their study all of the students in a class of 10, including the 5 target students, were told that they would earn 1 extra minute of recess for each point of class improve- ment beyond the previous weekly average. Also, all stu- dents were given a contract stating this same contingency. The extra recess was awarded every day of the following week. For example, if the students’ weekly averages ex- ceeded their previous weekly averages by 3 points, they would receive 3 minutes of extra recess every day of the following week. The results of this 22-week study showed that all students improved when the group contingency was in effect. Anecdotal data indicated that all students

participated and received extra recess time during the course of the study.

Good Behavior Game

Barrish, Saunders, and Wolf (1969) used the Good Be- havior Game to describe an interdependent group contin- gency in which a group is divided into two or more teams. Prior to the game being played, the teams are told that whichever team has the fewest marks against it at the end of the game will earn a privilege. Each team is also told that it can win a privilege if it has fewer than a specified num- ber of marks (a DRL schedule). Data reported by the au- thors show that this strategy can be an effective method of reducing disruptive behavior in the classroom. When game conditions were in effect during math or reading, talking- out and out-of-seat behaviors occurred at low levels. When game conditions were not in effect, disruptive behaviors oc- curred at much higher levels (see Figure 13).

In the Good Behavior Game teacher attention is di- rected toward observing and recording occurrences of

578

M at

h P

er io

d R

ea d

in g

P er

io d

Sessions Observed

P er

ce n

ta ge

o f

In te

rv al

s in

w h

ic h

B eh

av io

r O

cc u

rr ed

Figure 13 Percentage of 1-minute intervals containing talking-out and out-of-seat behaviors in a classroom of 24 fourth-grade children during math and reading periods. From “Good Behavior Game: Effects of Individual Contingencies for Group Consequences on Disruptive Behavior in a Classroom” by H. H. Barrish, M. Saunders, and M. M. Wolf, 1969, Journal of Applied Behavior Analysis, 2, p. 122. Copyright 1969 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

misbehavior, with the incentive that if one or more teams have fewer than the criterion number of infractions, a re- inforcer is delivered. The advantage of the Good Behav- ior Game is that competition occurs within teams and against the criterion, not across groups.

Good Student Game

The Good Student Game combines an interdependent group contingency (like the Good Behavior Game) with self-monitoring tactics (Babyak, Luze, & Kamps, 2000). Basically, the Good Student Game is intended for im- plementation during independent seatwork periods when problematic behaviors surface. In the Good Student

Game, the teacher (a) chooses target behaviors to mod- ify, (b) determines goals and rewards, and (c) determines whether group or individual monitoring (or both) will occur.

Students are trained in the Good Student Game using a model-lead-test instructional sequence: students are arranged in clusters of four to five, the target behaviors are defined, examples and nonexamples are provided, prac- tice occurs under teacher supervision, and one or more students record their own or the group’s performance. Table 2 shows how the Good Student Game compares to the Good Behavior Game. Note that two distinctions relate to target behaviors, the delivery of rewards, and feedback.

Contingency Contracting, Token Economy, and Group Contingencies

579

Table 2 Components of the Good Behavior Game and the Good Student Game

Component Good Behavior Game Good Student Game

Organization Students play in teams. Students play in teams or as individuals.

Management Teacher monitors and records behavior. Students self-monitor and record their behavior.

Target behaviors Behaviors are stated as rule breaking or Behaviors are stated as rule following. rule following.

Recording Teachers records incidents of rule-breaking Students record rule-following behaviors on a behaviors as they occur. variable-interval schedule.

System of reinforcement Positive. Positive.

Criterion for reinforcement Teams must not exceed a set number of Groups or individuals achieve or exceed a set rule-breaking behaviors. percentage of rule-following behaviors.

Delivery of reinforcement Dependent on group performance. Dependent on individual or group performance.

Feedback Teacher provides feedback when rule- Teacher provides feedback at intervals. Praise and breaking behaviors occur. encouragement presented during the game to

reinforce positive behaviors.

From “The Good Student Game: Behavior Management for Diverse Classrooms” by A. E. Babyak, G. J. Luze, and D. M. Kamps, 2000, Intervention in School and Clinic, 35 (2), p. 217. Copyright 2000 by PRO ED, Inc. Reprinted by permission.

Implementing a Group Contingency

Implementing a group contingency requires as much pre- planning as any other behavior change procedure. Pre- sented here are six guidelines to follow before and during the application of a group contingency.

Choose an Effective Reward

One of the most important aspects of a group contingency is the strength of the consequence; it must be strong enough to serve as an effective reward. Practitioners are advised to use generalized conditioned reinforcers or re- inforcer menus at every opportunity. Both of these strate- gies individualize the contingency, thereby increasing its power, flexibility, and applicability.

Determine the Behavior to Change and Any Collateral Behaviors That Might Be Affected

Let us suppose a dependent group contingency is estab- lished in which a class receives 10 minutes of extra free time, contingent on the improved academic performance of a student with developmental disabilities. Obviously, the teacher will need to collect data on the student’s aca- demic performance. However, data might also be col- lected on the number of positive interactions between the student and her classmates without disabilities within and outside the room. An additional benefit of using a group contingency might be the positive attention and encour-

agement that the student with developmental disabilities receives from her classmates.

Set Appropriate Performance Criteria

If a group contingency is used, the persons to whom the contingency is applied must have the prerequisite skills to perform the specified behaviors. Otherwise, they will not be able to meet the criterion and could be subject to ridicule or abuse (Stolz, 1978).

According to Hamblin, Hathaway, and Wodarski (1971), the criteria for a group contingency can be estab- lished by using the average-, high-, or low-performance levels of the groups as the standard. In an average-per- formance group contingency, the mean performance of the group is averaged, and reinforcement is contingent on the achievement of that mean score or a higher score. If the average score for a math exercise were 20 correct prob- lems, a score of 20 or more would earn the reward. In a high-performance group contingency, the high score de- termines the level of performance needed to receive the re- ward. If the high score on a spelling test was 95%, only students achieving a score of 95% would receive the re- ward. In the low-performance group contingency, the low performance score determines the reinforcer. If the low score on a social studies term paper were C, then students with a C or better would receive the reinforcer.

Hamblin and colleagues (1971) indicated that dif- ferential effects can be noted with these performance con- tingencies. Their data show that slow students performed

Contingency Contracting, Token Economy, and Group Contingencies

580

worse under a high-performance contingency, whereas gifted students performed best under this contingency. The data of Hamblin and colleagues suggest that a group contingency can be effective in improving behavior, but it should be applied with the understanding that effec- tiveness may vary for different members of the group.

Combine with Other Procedures When Appropriate

According to LaRowe, Tucker, and McGuire (1980), a group contingency can be used alone or in combination with other procedures to systematically change perfor- mance. The LaRowe and colleagues study was designed to reduce the excessive noise levels in an elementary school lunchroom; their data suggest that differential reinforce- ment of low rates of behavior (DRL) can be incorporated easily into a group contingency. In situations in which higher levels of group performance are desired, differen- tial reinforcement of high rates of behavior (DRH) can be used. With either DRL or DRH the use of a changing crite- rion design may facilitate the analysis of treatment effects.

Select the Most Appropriate Group Contingency

The selection of a specific group contingency should be based on the overall programmatic goals of the practi-

tioner, the parents (if applicable), and the participants whenever possible. For instance, if the group contin- gency is designed to improve the behavior of one per- son or a small group of individuals, then perhaps a dependent group contingency should be employed. If the practitioner wants to differentially reinforce appro- priate behavior, then an independent group contingency should be considered. But if the practitioner wants each individual within a group to perform at a certain level, then an interdependent group contingency should be chosen. Regardless of which group contingency is se- lected, the ethical issues must be addressed.

Monitor Individual and Group Performance

With a group contingency practitioners must observe both group and individual performance. Sometimes the group’s performance improves but some members within the group do not improve, or at least do not improve as fast. Some members of the group might even attempt to sabotage the group contingency, preventing the group from achieving the reinforcement. In these cases indi- vidual contingencies should be arranged for the sabo- teurs, in combination with the group contingency.

Summary Contingency Contracting

1. A contingency contract, also called a behavioral contract, is a document that specifies a contingent relation between the completion of a specified behavior and access to, or delivery of, a specified reward such as free time, a letter grade, or access to a favorite activity.

2. Every contract has three major parts: a description of the task, a description of the reward, and the task record. Under task, who, what, when, and how well should be specified. Under reward, who, what, when, and how much should be specified. A task record provides a place for recording the progress of the contract and providing in- terim rewards.

3. Implementing a contract involves a complex package in- tervention of related positive and negative reinforcement contingencies and rule-governed behaviors that operate alone and together.

4. Contracting has been used widely in classroom, home, and clinical settings.

5. Contracts have also been used to teach self-management to children.

6. A self-contract is a contingency contract that an individ- ual makes with herself, incorporating a self-selected task and reward as well as self-monitoring of task completion and self-delivery of the reward.

Token Economy

7. A token economy is a behavior change system consisting of three major components: (a) a specified list of target behaviors to be reinforced; (b) tokens or points that par- ticipants receive for emitting the target behaviors; and (c) a menu of items or activities, privileges, and backup re- inforcers from which participants choose and obtain by exchanging tokens they have earned.

8. Tokens function as generalized conditioned reinforcers be- cause they have been paired with a wide variety of backup reinforcers.

9. A level system is a type of token economy in which participants move up or down a hierarchy of levels

Contingency Contracting, Token Economy, and Group Contingencies

581

Contingency Contracting, Token Economy, and Group Contingencies

contingent on meeting specific performance criteria with respect to the target behaviors.

10. There are six basic steps in designing a token economy: (1) selecting the tokens that will serve as a medium of ex- change; (2) identifying the target behaviors and rules; (3) selecting a menu of backup reinforcers; (4) establish- ing a ratio of exchange; (5) writing procedures to specify when and how tokens will be dispensed and exchanged and what will happen if the requirements to earn a token are not met; and (6) field testing the system before full- scale implementation.

11. When establishing a token economy, decisions must be made regarding how to begin, conduct, maintain, evalu- ate, and remove the system.

Group Contingencies

12. A group contingency is one in which a common conse- quence is contingent on the behavior of one member of

the group, the behavior of part of the group, or the behav- ior of everyone in the group.

13. Group contingencies can be classified as independent, de- pendent, and interdependent.

14. Six guidelines can assist the practitioner in implement- ing a group contingency: (a) choose a powerful reward, (b) determine the behavior to change and any collateral behaviors that might be affected, (c) set appropriate per- formance criteria, (d) combine with other procedures when appropriate, (e) select the most appropriate group contingency, and (f) monitor individual and group performance.

582

habit reversal massed practice self-control

self-evaluation self-instruction self-management

self-monitoring systematic desensitization

Content Area 9: Behavior Change Procedures

9-27 Use self-management strategies.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and questions about this document must be submitted directly to the BACB.

Self-Management

Key Terms

583

Self-Management

Raylene used to be so forgetful, always failing to do the things she needed and wanted to do. Her days were just so busy! But Raylene is beginning to get a handle on her hectic life. For example, this morning a Post-It Note on her clothes closet door reminded her to wear her gray suit for her luncheon meeting. A note on the refrigerator prompted her to take her completed sales report to work. And when Raylene got into her car—sales report in hand and looking sharp in her gray suit—and sat on the library book she’d placed on the driver’s seat the night before, the chances were very good that she would return the book to the library that day and avoid another overdue fine.

Almost a year has passed since Daryl collected the last data point for his master’s thesis. It’s a solid study on a topic Daryl enjoys and believes to be important, but Daryl is struggling mightily to get his thesis written. Al- though Daryl knows that if he could sit down and write for an hour or two each day his thesis would get written, the size and difficulty of the task are daunting. He wishes his ability to sit and write were half as good as his ability to avoid those behaviors.

“Some people [self-manage their behavior] fre- quently and well, others do it rarely and poorly” (Epstein, 1997a, p. 547). Raylene has recently

discovered self-management—behaving to alter the oc- currence of other behaviors she wants to control—and feels on top of the world. Daryl, in sore need of some self- management, is feeling worse each passing week. This chapter defines self-management, identifies uses for self- management and the benefits of teaching and learning self- management skills, describes a variety of self-management tactics, and offers guidelines for conducting a successful self-management program. We begin with a discussion of the role of the self as a controller of behavior.

The “Self” as Controller of Behavior A fundamental precept of radical behaviorism is that the causes of behavior are found in the environment (Skinner, 1974). Throughout the evolution of the human species, causal variables selected by the contingencies of survival have been transmitted through genetic endowment. Other causes of behavior can be found in the contingencies of reinforcement that describe behavior–environment inter- actions during the lifetime of an individual person. What role, then, if any, is left for the self to play?

Locus of Control: Internal or External Causes of Behavior

Proximate causes of some behaviors are apparent to any- one observing the events as they unfold. A mother picks up and cuddles her crying infant, and the baby stops cry-

ing. A highway worker leaps off the road when he sees a fast-approaching car not slowed by the warning signs. An angler casts his lure toward the spot where his previ- ous cast produced a strike. A behavior analyst likely would suggest the involvement of escape, avoidance, and positive reinforcement contingencies, respectively, for these events. Although a nonbehaviorist might offer a mentalistic explanation for why the person in each sce- nario responded as he or she did (e.g., the infant’s crying triggered the mother’s nurturing instinct), most people, ir- respective of their education and orientation, professional and layperson alike, would identify the same antecedent events—a baby cried, a speeding car approached, a fish struck the lure—as functional variables in the three sce- narios. An analysis of the three episodes would reveal al- most certainly that the immediately preceding events did have functional roles—the crying baby and speeding car as motivating operations that evoked escape or avoidance responses; the biting fish as a powerful reinforcer.

But much human behavior does not immediately fol- low such obviously related antecedent events. Nonethe- less, we humans have a long history of assigning causal status to events that immediately precede behavior. As Skinner (1974) noted, “We tend to say, often rashly, that if one thing follows another, it was probably caused by it—following the ancient precept of post hoc, ergo propter hoc (after this, therefore because of this)” (p. 7). When causal variables are not readily apparent in the im- mediate, surrounding environment, the tendency to point to internal causes of behavior is particularly strong. As Skinner explained:

The person with whom we are most familiar is ourself; many of the things we observe just before we behave occur within our body, and it is easy to take them as the causes of our behavior. . . . Feelings occur at just the right time to serve as causes of behavior, and they have been cited as such for centuries. We assume that other people feel as we feel when they behave as we behave. (pp. 7, 8)

Why does one college student maintain a regular study schedule from the first week of the semester, while her roommates are out partying night after night? When a group of people joins a weight-loss or a smoking ces- sation program in which each group member is exposed to the same intervention, why do some people meet their self-determined goals but many others do not? Why does a high school basketball player with limited physical abil- ities consistently outperform his more athletically gifted teammates? The hardworking student is said to have more willpower than her less studious roommates; the group members who lost weight or stopped smoking are thought to possess more desire than their peers who failed to meet their goals or dropped out; the athlete’s superior play is considered a result of his exceptional drive. Although

584

Self-Management

some psychological theories grant causal status to hypo- thetical constructs such as willpower, desire, and drive, these explanatory fictions lead to circular reasoning and bring us no closer to understanding the behaviors they claim to explain.1

Skinner’s Two-Response Conceptualization of Self-Control

Skinner was the first to apply the philosophy and theory of radical behaviorism to actions typically considered to be controlled by the self. In his classic textbook, Science and Human Behavior, Skinner (1953) devoted a chapter to self-control.

When a man controls himself, chooses a course of ac- tion, thinks out the solution to a problem, or strives to- ward an increase in self-knowledge, he is behaving. He controls himself precisely as he would control the be- havior of anyone else—through the manipulation of variables of which behavior is a function. His behavior in so doing is a proper object of analysis, and eventually it must be accounted for with variables lying outside the individual himself. (pp. 228–229)

Skinner (1953) continued by conceptualizing self- control as a two-response phenomenon:

One response, the controlling response, affects variables in such a way as to change the probability of the other, the controlled response. The controlling response may manipulate any of the variables of which the controlled response is a function; hence there are a good many dif- ferent forms of self-control. (p. 231)

Skinner (1953) provided examples of a wide variety of self-control techniques, including using physical re- straint (e.g., clapping a hand over one’s mouth to prevent yawning at an embarrassing moment), changing the an- tecedent stimulus (e.g., putting a box of candy out of sight to reduce overeating), and “doing something else” (e.g., talking about another topic to avoid talking about a par- ticular topic), to name a few. Since Skinner’s initial list of techniques, a variety of taxonomies and catalogs of self-control tactics have been described (e.g., Agran, 1997; Kazdin, 2001; Watson & Tharp, 2007). All self- control—or self-management—tactics can be opera- tionalized in terms of two behaviors: (a) the target behavior a person wants to change (Skinner’s controlled response) and (b) the self-management behavior (Skin- ner’s controlling response) emitted to control the target behavior. For example, consider the following table:

1Explanatory fictions and circular reasoning are discussed in Chapters en- titled “Definition and Characteristics of Applied Behavior Analysis” and “Positive Reinforcement.”

2There is also nothing new about self-management. As Epstein (1997) pointed out, many of the self-control techniques outlined by Skinner were described by ancient Greek and Roman philosophers and have appeared in the teachings of many organized religions for millennia (cf. Bolin & Gold- berg, 1979; Shimmel, 1977, 1979).

Target Behavior Self-Management Behavior

• Save money instead of • Enroll in a payroll deduction spending it on frivolous plan. items.

• Take garbage and recycling • After backing the car out cans from garage to the of the garage for work curb on Thursday nights for Thursday morning, pull the pickup early Friday morning. trash and recycling cans

onto the space in the garage where you will park the car in the evening.

• Ride exercise bike for 30 • Make a chart of the minutes minutes each evening. you ride, and show it to a

coworker each morning.

• Write 20-page paper. • (1) Outline paper and divide it into five 4-page parts; (2) specify a due date for each part; (3) give room- mate five $10 checks made out to a despised organi- zation; (4) on each due date show a completed part of the paper to room- mate and get one $10 check back.

A Definition of Self-Management

There is nothing mystical about self-management.2 As the previous examples show, self-management is simply behavior that a person emits to influence another behav- ior. But many of the responses a person emits each day affect the frequency of other behaviors. Putting tooth- paste on a toothbrush makes it very likely that tooth brushing will occur soon. But we do not think of squeez- ing toothpaste on a brush as an act of “self-management” just because it alters the frequency of tooth brushing. What gives some responses the special status of self- management? How can we distinguish self-management from other behavior?

Numerous definitions of self-control or self-man- agement have been proposed, many of which are similar to the one offered by Thoresen and Mahoney (1974). They suggested that self-control occurs when, in the “rel- ative absence” of immediate external controls, a person emits a response designed to control another behavior. For example, a man shows self-control if, when home alone and “otherwise free” to do whatever he wants, he forgoes his usual peanuts and beer and instead rides his

585

Self-Management

exercise bike for 20 minutes. According to definitions such as Thoresen and Mahoney’s, self-control is not in- volved when immediate and obvious external events set the occasion for or reinforce the controlling response. The man would not be credited with self-control if his wife were there reminding him of his overeating, prais- ing him for riding the exercise bike, and marking his time and mileage on a chart.

But what if the man had asked his wife to prompt, praise, and chart his exercise? Would his having done so constitute a form of self-control? One problem with con- ceptualizing self-control as occurring only in the absence of “external control” is that it excludes those situations in which the person designs and puts into place the contin- gencies entailing external control for a behavior he desires to change. An additional problem with the “relative ab- sence of external controls” concept is that it creates a false distinction between internal and external control- ling variables when, in fact, all causal variables for be- havior ultimately reside in the environment.

Definitions of self-control such as Kazdin’s (2001), “those behaviors that a person deliberately undertakes to achieve self-selected outcomes” (p. 303), are more func- tional for applied behavior analysis. With this definition self-control occurs whenever a person purposely emits behavior that changes the environment in order to mod- ify another behavior. Self-control is considered pur- poseful in the sense that a person labels (or tacts) her responses as designed to attain a specified result (e.g., to reduce the number of cigarettes smoked each day).

We define self-management as the personal appli- cation of behavior change tactics that produces a desired change in behavior. This is an intentionally broad, func- tional definition of self-management. It encompasses one- time self-management events, such as Raylene’s taping a note to her closet door to remind herself to wear her gray suit the next day, as well as complex, long-running self- directed behavior change programs in which a person plans and implements one or more contingencies to change his behavior. It is a functional definition in that the desired change in the target behavior must occur for self- management to be demonstrated.

Self-management is a relative concept. A behavior change program may entail a small degree of self- management or be totally conceived, designed, and im- plemented by the person. Self-management occurs on a continuum along which the person controls one or all components of a behavior change program. When a be- havior change program is implemented by one person (e.g., teacher, therapist, parent) on behalf of another (e.g., student, client, child), the external change agent manip- ulates motivating operations, arranges discriminative stimuli, provides response prompts, delivers differential consequences, and observes and records the occurrence

3For a variety of conceptual analyses of self-control/self-management from a behavioral perspective, see Brigham (1983); Catania (1975, 1976); Goldiamond (1976); Hughes & Lloyd (1993); Kanfer and Karoly (1972); Malott (1984, 1989, 2005a, b); Nelson and Hayes (1981); Newman, Buff- ington, Hemmes, and Rosen (1996); Rachlin (1970, 1974, 1995); and Wat- son and Tharp (2007).

or nonoccurrence of the target behavior. Some degree of self-management is involved whenever a person performs (that is to say, controls) any element of a program that changes his behavior.

It is important to recognize that defining self- management as the personal application of behavior change tactics that results in a desired change in behav- ior does not explain the phenomenon. Our definition of self-management is descriptive only, and then so in a very broad sense. Although self-management tactics can be classified according to their emphasis on a given com- ponent of the three- or four-term contingency, or by their structural similarity with a particular principle of behav- ior (e.g., stimulus control, reinforcement), it is likely that all self-management tactics involve multiple principles of behavior. Thus, when researchers and practitioners de- scribe self-management tactics, they should provide a de- tailed statement of the exact procedures used. Behavior analysts should not ascribe the effects of self-management interventions to specific principles of behavior in the ab- sence of an experimental analysis demonstrating such re- lations. Only through such research will a more complete understanding of the mechanisms that account for the ef- fectiveness of self-management be attained.3

Terminology: Self-Control or Self-Management?

Although self-control and self-management often appear interchangeably in the behavioral literature, we recom- mend that self-management be used in reference to a per- son acting “in some way in order to change subsequent behavior” (Epstein, 1997, p. 547, emphasis in original). We have three reasons for making this recommendation. First, self-control is an “inherently misleading” term that implies that the ultimate control of behavior lies within the person (Brigham, 1980). Although Skinner (1953) ac- knowledged that a person could achieve practical control over a given behavior by acting in ways that manipulate variables that affect the frequency of that behavior, he ar- gued that the controlling behaviors themselves are learned from the person’s interactions with the environment.

A man may spend a great deal of time designing his own life—he may choose the circumstances in which he is to live with great care, and he may manipulate his daily en- vironment on an extensive scale. Such activity appears to exemplify a high order of self-determination. But it is also behavior, and we account for it in terms of other

586

Self-Management

variables in the environment and history of the individ- ual. It is these variables which provide the ultimate con- trol. (p. 240)

In other words, the causal factors for “self-control” (i.e., controlling behaviors) are to be found in a person’s experiences with his environment. Beginning with the example of a person setting an alarm clock (controlling behavior) so as to get out of bed at a certain time (con- trolled behavior) and concluding with an impressive in- stance of self-control by the main character in Homer’s Odyssey, Epstein (1997) described the origin of self- control as follows:

As is true of all operants, any number of phenomena might have produced the [controlling] behavior origi- nally: instructions, modeling, shaping, or generative process (Epstein, 1990, 1996), for example. Its occur- rence might have been verbally mediated by a self- generated rule (“I’ll bet I’d get up earlier if I set an alarm clock”), and that rule, in turn, might have had any number of origins. Odysseus had his men tie him to his mast (controlling behavior) to lower the probability that he would steer toward the Siren’s song (controlled be- havior). This was an elegant instance of self-control, but it was entirely instruction driven: Circe had told him to do it. (p. 547)

Second, attributing the cause of a given behavior to self-control can serve as an explanatory fiction. As Baum (2005) pointed out, self-control “seems to suggest con- trolling a [separate] self inside or [that there is] a self in- side controlling external behavior. Behavior analysts reject such views as mentalistic. Instead they ask, ‘What is the behavior that people call “self-control”?’” (p. 191, words in brackets added). Sometimes self is synonymous with mind and “not far from the ancient notion of ho- munculus—an inner person who behaves in precisely the ways necessary to explain the behavior of the outer per- son in whom he dwells” (Skinner, 1974, p. 121).4

Third, laypeople and behavioral researchers alike often use self-control to refer to a person’s ability to “delay gratification” (Mischel & Gilligan, 1964). In op- erant terms, this connotation of self-control entails re- sponding to achieve a delayed, but larger or higher quality reward instead of acting to obtain an immediate, less- valuable reward (Schweitzer & Sulzer-Azaroff, 1988).5

Using the same term to refer to a tactic for changing be- havior and to a certain type of behavior that may be an

4In early biological theory, homunculus was a fully formed human being thought to exist inside an egg or spermatozoon. 5Behavior that fails to exhibit such self-control is often labeled “impul- sive.” Neef and colleagues (e.g., Neef, Bicard, & Endo, 2001; Neef, Mace, & Shade, 1993) and Dixon and colleagues (e.g., Binder, Dixon, & Ghezzi, 2000; Dixon et al., 1998; Dixon, Rehfeldt, & Randich, 2003) have devel- oped procedures for assessing impulsiveness and for teaching self-control in the form of tolerance for delayed rewards.

outcome of the tactic is at best confusing and logically faulty. Restricting the use of self-control to describe a certain type of behavior may reduce the confusion caused by using self-control as the name for both an indepen- dent variable and dependent variable. In this sense, self- control can be a possible goal or outcome of a behavior change program irrespective of whether that behavior is the product of an intervention implemented by an exter- nal agent or the subject. Thus, a person can use self- management to achieve, among other things, self-control.

Applications, Advantages, and Benefits of Self-Management In this section we identify four basic applications of self- management and explain the numerous advantages and benefits that accrue to people who use self-management, to practitioners who teach it to others, and to society as a whole.

Applications of Self-Management

Self-management can help a person be more effective and efficient in his daily life, replace bad habits with good ones, accomplish difficult tasks, and achieve per- sonal goals.

Living a More Effective and Efficient Daily Life

Raylene writing notes to herself and placing her library book on her car seat are examples of self-management techniques that can be used to overcome forgetfulness or lack of organization. Most people use simple self- management techniques such as making a shopping list before going to the store or creating “to-do” lists as ways to organize their day; but few probably consider what they are doing to be “self-management.” Although many of the most widely used self-management techniques can be considered common sense, a person with an under- standing of basic principles of behavior can use that knowledge to apply those common sense techniques more systematically and consistently in her life.

Breaking Bad Habits and Acquiring Good Ones

Many behaviors that we want to do more (or less) often and know that we should do (or not do) are caught in re- inforcement traps. Baum (2005) suggested that impul- siveness, bad habits, and procrastination are products of

587

Self-Management

naturally existing reinforcement traps in which immedi- ate but smaller consequences have greater influence on our behavior than more significant outcomes that are de- layed. Baum gave this description of a reinforcement trap for smoking:

Acting impulsively leads to a small but relatively imme- diate reinforcer. The short-term reinforcement for smok- ing lies in the effects of nicotine and social reinforcers such as appearing grown-up or sophisticated. The trou- ble with impulsive behavior lies in long-term ill effects. It may be months or years before the bad habit takes its toll in consequences such as cancer, heart disease, and emphysema.

The alternative to impulsiveness, refraining from smoking, also leads to both short-term and long-term con- sequences. The short-term consequences are punishing but relatively minor and short-lived: withdrawal symp- toms (e.g., headaches) and possibly social discomfort. In the long run, however, . . . refraining from smoking re- duces the risk of cancer, heart disease, and emphysema; ultimately, it promotes health. (pp. 191–192)

Reinforcement traps are two-sided contingencies that work to promote bad habits while simultaneously work- ing against the selection of behavior that is beneficial in the long-term. Even though we may know the rules de- scribing such contingencies, the rules are difficult to fol- low. Malott (1984) described weak rules as those with delayed, incremental, or unpredictable outcomes. An ex- ample of a hard-to-follow, weak rule is, I’d better not smoke or I might get cancer someday and die. Even though the potential consequence—cancer and death—is a major one, it is far off in the future and not a sure thing even then (we all know someone who smoked two packs a day since the age of 15 and lived to be 85), which se- verely limits its effectiveness as a behavioral conse- quence. So the rule Don’t smoke or you might get cancer is hard to follow. The deleterious effect of any one ciga- rette is small, so small that it is not even noticed. Em- physema and lung cancer are probably years and thousands of puffs away. What will one more puff hurt?

Self-management provides one strategy for avoiding the deleterious effects of reinforcement traps. A person can use self-management tactics to arrange immediate consequences that will counteract the consequences cur- rently maintaining the self-destructive behavior.

Accomplishing D ifficult Tasks

Immediate outcomes that bring a person only “infinites- imally closer” to a significant outcome do not control be- havior (Malott, 1989, 2005a). Many self-management problems are the result of outcomes that are small, but of cumulative significance.

Malott (2005a) argued that our behavior is controlled by the outcome of each individual response, not by the cumulative effect of a large number of responses.

The natural contingency between a response and its out- come will be ineffective if the outcome of each individ- ual response is too small or too improbable, even though the cumulative impact of those outcomes is significant. So, we have trouble managing our own behavior, when each instance of that behavior has only an insignificant outcome, though the outcomes of many repetitions of that behavior will be highly significant. And we have lit- tle trouble managing our own behavior, when each in- stance of that behavior has a significant outcome, though that outcome might be greatly delayed.

Most obese people in the United States know the rule describing that contingency: If you repeatedly overeat, you will become overweight. The problem is, knowing the overweight rule does not suppress this one instance of eating that delicious fudge sundae topped with the whipped cream and the maraschino cherry, be- cause a single instance of eating this dessert will cause no significant harm, and will, indeed, taste great. Knowledge of this rule describing the natural contin- gency exerts little control over gluttony.

But the outcome of our behavior also needs to be probable. Many people have trouble following the buckle-up rule, even though they might get in a serious auto accident as soon as they pull onto the street. Even though the delay between the failure to buckle up and the accident could be only a few seconds, they fail to buckle up because the probability of the accident is so low. However, if they were driving in a dangerous auto race or a dangerous stunt-driving demonstration, they would always buckle up, because the probability of a se- rious accident would be fairly high. (pp. 516–517)

Just as a person who smokes another cigarette or eats just one more hot fudge sundae cannot detect that lung cancer or obesity is a noticeably closer outcome, gradu- ate student Daryl does not perceive that writing one more sentence brings him closer to his long-term goal of a com- pleted master’s thesis. Daryl is capable of writing one more sentence but, like many of us, he often has trouble getting started and working steadily on difficult tasks where each response produces little or no apparent change in the size of the remaining task. He procrastinates.

Applying self-management to this type of perfor- mance problem involves designing and implementing one or more contrived contingencies to compete with the in- effective natural contingencies. The self-management con- tingency provides immediate consequences or short-term outcomes for each response or small set of responses. These contrived consequences increase the frequency of the target responses, which over time produce the cumu- lative effects necessary to complete the task.

588

Self-Management

Achieving Personal Goals

People can use self-management to achieve personal goals, such as learning to play a musical instrument, learning a foreign language, running a marathon, sticking to a schedule of daily yoga sessions (Hammer-Kehoe, 2002), or simply taking some time each day to relax (Har- rell, 2002) or listen to enjoyable music (Dams, 2002). For example, a graduate student who wanted to become a better guitarist used self-management to increase the time he practiced scales and chords (Rohn, 2002). He paid a $1 fine to a friend for each day that he failed to practice playing scales and chords for a criterion number of minutes before going to bed. Rohn’s self- management program also included several contingen- cies based on the Premack principle. For example, if he practiced scales (a low-frequency behavior) for 10 min- utes, he could play a song (a high-frequency behavior).

Advantages and Benefits of Self-Management

A dozen potential advantages and benefits accrue to peo- ple who learn to use self-management skills and to prac- titioners who teach such skills to their clients and students.

Self-Management Can Influence Behaviors Not Accessible to External Change Agents

Self-management can be used to change behaviors with topographies that make them inaccessible to observation by others. Behaviors such as thoughts of self-doubt, ob- sessive thoughts, and feelings of depression are private events for which a self-managed treatment approach may be needed (e.g., Kostewicz, Kubina, & Cooper, 2000; Kubina, Haertel, & Cooper, 1994).

Even publicly observable target behaviors are not al- ways emitted in settings and situations accessible to an external change agent. Sometimes the behavior a person wishes to change must be prompted, monitored, evalu- ated, or reinforced or punished on a day-to-day or even minute-to-minute basis in all of the situations and envi- ronments that the person encounters. Even though a per- son enrolls in a behavior change program planned and directed by others, most successful smoking, weight-loss, exercise, and habit-reversal programs rely heavily on par- ticipants using various self-management techniques when they are away from the treatment setting. Many behavior change goals brought to clinicians pose the same chal- lenge: How can an effective contingency be arranged that follows the client at all times wherever she goes? For ex- ample, target behaviors designed to increase a secretary’s 6Producing generalized behavior change is a defining goal of applied be-

havior analysis and the focus of Chapter entitled “ Generalization and Maintenance of Behavior Change.”

self-esteem and assertiveness can be identified and prac- ticed in the therapist’s office, but implementation of an ac- tive contingency in the workplace is apt to require self-management techniques.

External Change Agents Often Miss Important Instances of Behavior

In most education and treatment settings, particularly in large group situations, many important responses go un- noticed by the person responsible for applying behavior change procedures. Classrooms, for example, are espe- cially busy places where desired behaviors by students often go unnoticed because the teacher is engaged with other tasks and students. As a result, students invariably miss opportunities to respond, or they respond and re- ceive no feedback because, in a behavioral sense, the teacher is not there. However, students who have been taught to evaluate their own performance and provide their own feedback in the form of self-delivered rewards and error correction, including seeking assistance and praise from their teacher when needed (e.g., Alber & Heward, 2000; Bennett & Cavanaugh, 1998; Olympia, Sheridan, Jenson, & Andrews, 1994), are not dependent on the teacher’s direction and feedback for every learn- ing task.

Self-Management Can Promote the Generalization and Maintenance of Behavior Change

A behavior change that (a) continues after treatment has ended, (b) occurs in relevant settings or situations other than the one(s) in which it was learned originally, or (c) spreads to other related behaviors has generality (Baer, Wolf, & Risley, 1968). Important behavior changes with- out such generalized outcomes must be supported indefi- nitely by continued treatment.6 As soon as the student or client is no longer in the setting in which a behavior change was acquired, she may no longer emit the desired response. Certain aspects of the original treatment environment, in- cluding the person (teacher, parent) who administered the teaching program, may have become discriminative stim- uli for the newly learned behavior, enabling the learner to discriminate the presence or absence of certain contin- gencies across settings. Generalization to nontreatment environments is also hampered when the naturally exist- ing contingencies in those settings do not provide rein- forcement for the target behavior.

589

Self-Management

Such challenges to achieving generalized outcomes may be overcome if the learner has self-management skills. Baer and Fowler (1984) posed and answered a prag- matic question related to the problem of promoting the generalization and maintenance of newly learned skills.

What behavior change agent can go with the student to every necessary lesson, at all times, to prompt and rein- force every desirable form of the behavior called for by the curriculum? The student’s own “self” can always meet these specifications. (p. 148)

A Small Repertoire of Self-Management Skills Can Control Many Behaviors

A person who learns how to apply a few self-management tactics can control a potentially unlimited range of behav- iors. For example, self-monitoring—observing and record- ing one’s own behavior—has been used to increase on-task behavior (e.g., Blick & Test, 1987), academic productivity and accuracy (e.g., Maag, Reid, & DiGangi, 1993), em- ployee productivity and job success (e.g., Christian & Pol- ing, 1997), and independence (e.g., Dunlap, Dunlap, Koegel, & Koegel, 1991), and to decrease undesired be- haviors such as habits or tics (e.g., Koegel & Koegel, 1991).

People with Diverse Abilities Can Learn Self-Management Skills

People of wide-ranging ages and cognitive abilities have successfully used self-management tactics. Preschoolers (e.g., Sainato, Strain, Lefebvre, & Rapp, 1990; DeHaas- Warner, 1992), typically developing students from pri- mary grades through high school (e.g., Sweeney, Salva, Cooper, & Talbert-Johnson, 1993), students with learning disabilities (e.g., Harris, 1986), students with emotional and behavior disorders (e.g., Gumpel & Shlomit, 2000), children with autism (e.g., Koegel, Koegel, Hurley, & Frea, 1992; Newman, Reinecke, & Meinberg, 2000), and children and adults with mental retardation and other de- velopmental disabilities (e.g., Grossi & Heward, 1998) have all used self-management successfully. Even col- lege professors are capable of using self-management to improve their performance (Malott, 2005a)!

Some People Perform Better under Self-Selected Tasks and Performance Criteria

Most studies that have compared self-selected and other- selected reinforcement contingencies have found that under certain conditions self-selected contingencies can be as effective in maintaining behavior as contingencies determined by others (e.g., Felixbrod & O’Leary, 1973, 1974; Glynn, Thomas, & Shee, 1973). However, some studies have found better performance with self-selected

work tasks and consequences (e.g., Baer, Tishelman, Degler, Osnes, & Stokes, 1992; Olympia et al., 1994; Parsons, Reid, Reynolds, & Bumgarner, 1990). For ex- ample, in three brief experiments with a single student, Lovitt and Curtiss (1969) found pupil-selected rewards and contingencies to be more effective than teacher- selected performance standards. In the first phase of Ex- periment 1, the subject, a 12-year-old boy in a special education classroom, earned a teacher-specified number of minutes of free time based on the number of math and reading tasks he completed correctly. In the next phase of the study, the student was allowed to specify the num- ber of correct math and reading items needed to earn each minute of free time. During the final phase of the study the original teacher-specified ratios of academic production to reinforcement were again in effect. The student’s median academic response rate during the self- selected contingency phase was 2.5 correct responses per minute (math and reading tasks reported together), com- pared to correct rates of 1.65 and 1.9 in the two teacher- selected phases.

An experiment by Dickerson and Creedon (1981) employed a yoked-control (Sidman, 1960/1988) and between-group comparison experimental design with 30 second- and third-graders and found that pupil-selected standards resulted in significantly higher academic pro- duction of both reading and math tasks. Both the Lovitt and Curtiss and the Dickerson and Creedon studies demon- strated that self-selected reinforcement contingencies can be more effective than teacher-selected contingencies. However, research in this area also shows that simply let- ting children determine their own performance standards does not guarantee high levels of performance; some stud- ies have found that children select too-lenient standards when given the opportunity (Felixbrod & O’Leary, 1973, 1974). But interestingly, in a study of student-managed homework teams by Olympia and colleagues (1994), al- though students assigned to teams who could select their own performance goals chose more lenient accuracy cri- teria than the 90% accuracy criterion required of students on the teams that worked toward teacher-selected goals, the overall performance of the student-selected-goal teams was slightly higher than that of the teams working toward teacher-selected goals. More research is needed to deter- mine the conditions under which students will self-select and maintain appropriate standards of performance.

People with Good Self-Management Skills Contribute to More Efficient and Effective Group Environments

The overall effectiveness and efficiency of any group of people who share a working environment is limited when a single person is responsible for monitoring, supervising,

590

Self-Management

and providing feedback for the performance of each group member. When individual students, teammates, band members, or employees have self-management skills that enable them to work without having to rely on the teacher, coach, band director, or shift manager for every task, the overall performance of the entire group or organization improves. In the classroom, for example, teachers traditionally have assumed full responsibility for setting performance criteria and goals for students, eval- uating students’ work, delivering consequences for those performances, and managing students’ social behavior. The time required for these functions is considerable. To the extent that students can self-score their work, provide their own feedback using answer keys or self-checking materials (Bennett & Cavanaugh, 1998; Goddard & Heron, 1998), and behave appropriately without teacher assistance, the teacher is freed to attend to other aspects of the curriculum and perform other instructional duties (Mitchem, Young, West, & Benyo, 2001).

When Hall, Delquadri, and Harris (1977) conducted observations of elementary classrooms and found low lev- els of active student response, they surmised that higher rates of academic productivity might actually be punishing to teachers. Even though there is considerable evidence linking high rates of student opportunity to respond with academic achievement (e.g., Ellis, Worthington, & Larkin, 2002; Greenwood, Delquadri, & Hall, 1984; Heward, 1994), in most classrooms generating more student aca- demic responses results in the teacher having to spend time grading papers. The time savings produced by students with even the most simple self-management skills can be significant. In one study a class of five elementary special education students completed as many arithmetic prob- lems as they could during a 20-minute session each day (Hundert & Batstone, 1978). When the teacher graded the papers, an average of 50.5 minutes was required to pre- pare and conduct the session and score and record each student’s performance. When the students scored their own papers, the total teacher time needed to conduct the arith- metic period was reduced to an average of 33.4 minutes.

Teaching Students Self-Management Skills Provides Meaningful Practice for Other Areas of the School Curriculum

When students learn to define and measure behavior and to graph, evaluate, and analyze their own responses, they are practicing various math and science skills in a relevant way. When students are taught how to conduct self- experiments such as using A-B designs to evaluate their self-management projects, they receive meaningful prac- tice in logical thinking and scientific method (Marshall & Heward, 1979; Moxley, 1998).

Self-Management Is an Ultimate Goal of Education

When asked to identify what education should accom- plish for its students, most people—educators and lay- people alike—include in their reply the development of independent, self-directed people who are capable of be- having appropriately and constructively without the su- pervision of others. John Dewey (1939), one of this country’s most influential educational philosophers, said that “the ideal aim of education is the creation of self- control” (p. 75). A student’s ability to be self-directed, with the teacher as a guide or facilitator, and his ability to evaluate his own performance have long been consid- ered cornerstones of humanistic education (Carter, 1993).

As Lovitt (1973) observed more than 30 years ago, the fact that systematic instruction of self-management skills is not a regular part of most schools’ curricula is a paradox because “one of the expressed objectives of the educational system is to create individuals who are self- reliant and independent” (p. 139).

Although self-control is a social skill expected and valued by society, it is seldom addressed directly by the school curriculum. Teaching self-management as an in- tegral part of the curriculum requires that students learn a fairly sophisticated sequence of skills (e.g., Marshall & Heward, 1979; McConnell, 1999). However, system- atic teaching of self-management skills, if successful, is worthy of the effort required; students will have an ef- fective means of dealing with situations in which there is little, if any, external control. Materials and lesson plans are available that help students become independent, self- directed learners (e.g., Agran, 1997; Daly & Ranalli, 2003; Young, West, Smith, & Morgan, 1991). Box 1 de- scribes free software programs children can use to create self-management tools for use in school and at home.

Self-Management Benefits Society

Self-management serves two important functions for so- ciety (Epstein, 1997). First, citizens with self-management skills are more likely to fulfill their potential and make greater contributions to society. Second, self-management helps people behave in ways (e.g., purchasing fuel-efficient automobiles, taking public transportation) that forgo im- mediate reinforcers that are correlated with devastating outcomes that are so long deferred that only future gener- ations will experience them (e.g., depletion of natural re- sources and global warming). People who have learned self-management techniques that help them to save re- sources, recycle, and burn less fossil fuel help create a bet- ter world for everyone. Self-management skills can provide people with the means of meeting the well-intentioned but difficult-to-follow rule, Think globally but act locally.

591

Self-Management

KidTools, KidSkills, and StrategyTools are software pro- grams that help children create and use a variety of self- management tactics, organizational skills, and learning strategies. The programs enable children to take responsi- bility for changing and managing their own social and aca- demic behaviors. eKidTools and eKidSkills are designed for children ages 7 to 10; iKidTools and iKidSkills are de- signed for children ages 11 to 14; StrategyTools includes en- hanced self-management and learning strategy tools, transition planning tools, and self-training modules for secondary students. All three programs feature easy-to- complete templates that let children be as independent as possible in identifying personal behaviors or academic learning needs, selecting a self-management tool to support the task, developing the steps of their personalized plan, and printing the tool for immediate use in the classroom or at home. The remainder of this essay provides additional in- formation on the KidTools and KidSkills programs.

KidTools

The KidTools program supports children’s self-manage- ment skills to identify behaviors, make plans, and mon- itor implementation of the plans. KidTools includes self-management interventions for three types, or levels, of control. In all three levels, there is an emphasis on what children say and think to themselves as they execute their behavior change plans.

External control procedures may be necessary to establish control over problematic behaviors before the child moves to shared control or self-control interven- tions. In these procedures, the adult provides direction and structure for appropriate behaviors. KidTools avail- able for this level of intervention are point cards.

Shared control techniques provide a transition step for encouraging the child to develop self-control. There is an emphasis on problem solving and making plans to change or learn a behavior. The child and adult jointly participate in these procedures. KidTools available for this level of intervention are contracts, make-a-plan cards, and problem-solving planning tools.

Internal control techniques assist children through cues and structure provided by a variety of self-monitoring procedures.

KidSkills

KidSkills helps children be successful in school by get- ting organized, completing tasks, and using learning strategies. KidSkills provides computer-generated tem- plates for getting organized, learning new stuff, organiz- ing new information, preparing for tests, doing homework, and doing projects. KidSkills programs are complemen- tary to the KidTools programs and operate the same in terms of selecting tools, entering content, recordkeeping, and adult supports.

Box 1 Self-Management Software for Children

592

Self-Management

Tool Resources and Skill Resources

In addition to the computer programs for children, in- formation databases about the tools and strategies are provided for parents and teachers who assist children with self-management and learning strategies. Each data- base includes guidelines and tips for each procedure, steps for implementation, age-appropriate examples for children, troubleshooting tips, and references to related resources. The user can navigate through the informa- tion, return to a menu to select new strategies, or search the entire database using the Find function.

Rationale for KidTools and KidSkills

KidTools and KidSkills encourage children to express their social and academic behaviors in positive terms. They focus on doing positive behaviors and being suc- cessful, rather than stopping negative behaviors and fail- ure. For example, they decide “I will use my inside voice” rather than be reminded “Don’t talk loudly in class.” Children are taught organizational and learning strate- gies that can be used in general and special settings. The computerized tools serve as bridges to support students in behaving and learning. This form of assistive tech- nology works because it provides direct instruction as well as scaffolds to support learning at the “right time, right form, and right place.”

For these materials to be successful, teachers or par- ents must first provide instruction in behavior change techniques and self-monitoring. Children need to under- stand the concept of a behavior as something they do or

think, be able to name behaviors, understand how to mon- itor whether behaviors occur or not, and assume some level of responsibility for their own behaviors. Children need to learn how and when to use organizational and learning strategies and to use self-monitoring techniques in applying the strategies across settings.

Helping Children Use KidTools and KidSkills

Children will need to be taught to use self-talk cues and inner speech to give themselves directions. Each self- management procedure in KidTools has variations that will need to be taught, demonstrated, and practiced with children before they can be used independently. The or- ganizational and learning strategies in KidSkills are generic but require teaching, practice, and teacher help in applying the tools prior to independent use. Once chil- dren know the prerequisite skills, the computer program can be used with a minimal amount of teacher help. On- going support by the teacher or parent will ensure that the process remains effective and positive for children.

A teacher can use the following steps to help chil- dren learn to use KidTools and KidSkills.

Discuss behavioral and task expectations. The teacher can use setting or task demands to initiate dis- cussion about expectations. Students can verbalize what they must do to be successful and identify challenges in meeting the expectations. Students should be able to demonstrate the behaviors and do the tasks prior to using the tools.

Introduce the software tools. The tool programs can be introduced to the whole class using a computer and

593

Self-Management

projection device. The teacher can demonstrate the menu of available tools and highlight a variety of them, in- cluding how to navigate the program, how to enter con- tent into the tools, and how to save and print out completed tools.

Model and demonstrate use of the tools. Using pre- selected tools for specific purposes, the teacher can so- licit student input while demonstrating how the software works. The finished tools can be displayed and printed out for students to have copies. A discussion of their ac- tual use follows.

Provide guided practice. Using available computer equipment, in a lab for group practice or in the classroom for independent or small group practice, the teacher can guide and oversee student development of tools. It will be helpful to have students verbalize their content prior to filling in the tool. The purpose of the guided practice step is to ensure correct software usage.

Provide independent practice. The teacher can help students identify authentic uses for the tools and provide opportunities for creating and using tools. During inde- pendent usage, teachers should provide encouragement and reinforcement for student tool usage and successful outcomes in supervised situations.

Facilitate generalization. The teacher can help stu- dents identify authentic needs and uses of tools across school and home environments. The teacher can assist by prompting tool usage, problem solving, and com- municating with other teachers and parents about the procedures.

Programs Available Free

The programs can be downloaded at no cost in Windows or Macintosh versions from the KTSS Web site at http://kidtools.missouri.edu. Training modules for teach- ers with video demonstrations and practice materials can also be downloaded at this site. All three programs are also available on CD-ROM at cost from Instructional Materials Laboratory at www.iml.coe.missouri.edu. The KTSS programs have been developed with partial fund- ing by the U.S. Department of Education, Office of Spe- cial Education Programs.

Adapted from “KidTools and KidSkills: Self-Management Tools for Children” by Gail Fitzgerald and Kevin Koury, 2006. In W. L. Heward, Exceptional Children: An Introduction to Special Education (8th ed., pp. 248–250). Copyright 2006 by Pearson Education. Used by permission.

Self-Management Helps a Person Feel Free

Baum (2005), noted that people caught in reinforcement traps who recognize the contingency between their ad- dictive behavior, impulsiveness, or procrastination and the eventual consequence that is likely to occur as a result of that behavior are unhappy and do not feel free. How- ever, “Someone who escapes from a reinforcement trap, like someone who escapes from coercion, feels free and happy. Ask anyone who has kicked an addiction” (p. 193).

Ironically, a person who skillfully uses self-man- agement techniques derived from a scientific analysis of behavior–environment relations predicated on an as- sumption of determinism is more likely to feel free than is a person who believes that her behavior is a product of free will. In discussing the apparent disconnect be- tween philosophical determinism and self-control, Ep- stein (1997) compared two people: one who has no self-management skills and one who is very skilled at managing his own affairs.

First, consider the individual who has no self-control. In Skinner’s view, such a person falls prey to all immediate stimuli, even those that are linked to delayed punish- ment. Seeing a chocolate cake, she eats it. Handed a cig- arette, she smokes it. . . . She may make plans, but she has no ability to carry them out, because she is entirely at the mercy of proximal events. She is like a sailboat blowing uncontrollably in a gale. . . .

At the other extreme, we have a skillful self- manager. He, too, sets goals, but he has ample ability to meet them. He has skills to set dangerous reinforcers aside. He identifies conditions that affect his behavior and alters them to suit him. He takes temporally remote possibilities into account in setting his priorities. The wind is blowing, but he sets the boat’s destination and directs it there.

These two individuals are profoundly different. The first is being controlled in almost a linear fashion by her immediate environment. The second is, in a nontrivial sense, controlling his own life. . . .

The woman who lacks self-control skills feels con- trolled. She may believe in free will (in fact, in our cul- ture, it’s a safe bet that she does) but her own life is out of control. A belief in free will only exacerbate her frus- tration. She should be able to will herself out of any jam, but “willpower” proves to be highly unreliable. In con- trast, the self-manager feels that he is in control. Ironi- cally, like Skinner, he may believe in determinism, but he not only feels that he is in control, he is in fact exer- cising considerably more control over his life than our impulsive subject. (p. 560, emphasis in original)

Self-Management Feels Good

A final, but by no means trivial, reason for learning self-management is that being in control of one’s life feels good. A person who arranges her environment in

594

Self-Management

purposeful ways that support and maintain desirable behaviors will not only be more productive than she would otherwise be, but will also feel good about her- self. Seymour (2002) noted her personal feelings about a self-management intervention she implemented with the objective of running for 30 minutes three times per week.

The guilt I had been feeling for the past years had been burning strong within me. [As a former scholar- ship softball player] I used to exercise 3 hours per day (6 days per week) because that exercise paid for my schooling. . . . But now the contingencies were missing, and my exercising had gone completely down the tubes. Since the start of my project, I’ve run 15 out of the 21 times; so I’ve gone from 0% during baseline to 71% during my intervention. I’m happy with that. Because the intervention’s successful, the guilt I have felt for the past years has been lifted. I have more energy and my body feels new and strong. It’s amazing what a few contingencies can do to improve the quality of life in a big way. (p. 7–12)

Antecedent-Based Self-Management Tactics In this and the three sections that follow we describe some of the many self-management tactics that behav- ioral researchers and clinicians have developed. Al- though no standard set of labels or classification scheme for self-management tactics has emerged, the techniques are often presented according to their relative emphasis on antecedents or consequences for the target behavior. An antecedent-based self-management tactic is one whose primary feature is the manipulation of events or stimuli antecedent to the target (controlled) behavior. Sometimes lumped under general terms such as envir- onmental planning (Bellack & Hersen, 1977; Thoresen & Mahoney, 1974) or situational inducement (Martin & Pear, 2003), antecedent-based approaches to self-man- agement encompass a wide range of tactics, including the following:

• Manipulating motivating operations to make a de- sired (or undesired) behavior more (or less) likely

• Providing response prompts

• Performing the initial steps of a behavior chain to ensure being confronted later with a discriminative stimulus that will evoke the desired behavior

• Removing the materials required for an unde- sired behavior

• Limiting an undesired behavior to restricted stimu- lus conditions

2 1 2

7Motivating operations are described in detail in Chapter entitled “ Moti- vating Operations.”

• Dedicating a specific environment for a desired behavior

Manipulating Motivating Operations

People with an understanding of the dual effects of mo- tivating operations can use that knowledge to their ad- vantage when attempting to self-manage their behavior. A motivating operation (MO) is an environmental con- dition or event that (a) alters the effectiveness of some stimulus, object, or event as a reinforcer and (b) alters the current frequency of all behaviors that have been re- inforced by that stimulus, object, or event in the past. An MO that increases the effectiveness of a reinforcer and has an evocative effect on behaviors that have produced that reinforcer is called an establishing operation (EO); an MO that decreases the effectiveness of a reinforcer and has an abative (i.e., weakening) effect on the frequency of behaviors that have produced that reinforcer is called an abolishing operation (AO).7

The general strategy for incorporating an MO into a self-management intervention is to behave in a way (con- trolling behavior) that creates a certain state of motivation that, in turn, increases (or decreases as desired) the sub- sequent frequency of the target behavior (controlled be- havior). Imagine that you have been invited to dinner at the house of someone known to be a first-rate cook. You would like to obtain maximum enjoyment from what promises to be a special meal but are worried you will not be able to eat everything. By purposively skipping lunch (controlling behavior), you create an EO that would increase the likelihood of your being able to enjoy the evening meal from appetizer through dessert (controlled behavior). Conversely, eating a meal just before going grocery shopping (controlling behavior) could serve as an AO that decreases the momentary value of ready-to- eat foods with high sugar and fat content as reinforcers and results in purchasing fewer such items at the super- market (controlled behavior).

Providing Response Prompts

Creating stimuli that later function as extra cues and re- minders for desired behaviors is one of the simplest, most effective, and most widely applied self-management tech- niques. Response prompts can take a wide variety of forms (e.g., visual, auditory, textual, symbolic), be per- manent for regularly occurring events (e.g., entering “Put

595

Self-Management

ZAP!

garbage out tonight!” each Thursday on one’s calendar or PDA), or be one-time affairs such as when Raylene wrote, “gray suit today” on a Post-It Note and attached it to her closet door where she was sure to see it when dressing for work in the morning. Dieters often put pictures of over- weight people, or perhaps even unattractive pictures of themselves, on the refrigerator door, near the ice cream, in the cupboard—anyplace where they might look for food. Seeing the pictures may evoke other controlling responses, such as moving away from the food, phoning a friend, going for a walk, or marking a point on an “I Didn’t Eat It!” chart.

The same object can be used as a generic response prompt for various behaviors, such as putting a rubber band around the wrist as a reminder to do a certain task at some later time. This form of response prompt, how- ever, will be ineffective if the person is unable later to re- member what specific task the physical cue was intended to prompt. Engaging in some self-instruction when “putting on” a generic response prompt may help the per- son recall the task to be completed when seeing the cue later. For example, one of the authors of this text uses a small carabineer he calls a “memor-ring” as a generic re-

sponse prompt. When the thought of an important task that can only be completed in another setting later in the day occurs to him, he clips the carabineer to a belt loop or the handle of his briefcase and “activates the memor- ring” by stating the target task three times to himself (e.g., “borrow Nancy’s journal at Arps Hall,” “borrow Nancy’s journal at Arps Hall,” “borrow Nancy’s journal at Arps Hall”). When he sees the “memor-ring” in the relevant setting later, it almost always functions as an effective response prompt for completing the targeted task.

Cues can also be used to prompt the occurrence of a behavior the person wants to repeat in a variety of set- tings and situations. In this case, the person might salt his environment with supplemental response prompts. For example, a father who wants to increase the number of times he interacts with and praises his children might put ZAP! cues like those shown in Figure 1 in various places around the house where he will see them rou- tinely—on the microwave, on the T V remote control, as a bookmark. Every time he sees a ZAP! cue, the father is reminded to make a comment or ask a question of his child, or to look for some positive behavior for which he can provide attention and praise.

Instructions: Cut out these ZAP! cues and place them in conspicuous locations throughout your home or apartment, such as the refrigerator, bathroom mirror, TV remote control, and the door to your child’s bedroom. Each time you see a ZAP cue, you’ll be reminded to do one of your self-selected target behaviors. Figure 1 Contrived cues used by parents to remind themselves to emit self- selected parenting behaviors. Originally developed by Charles Novak. From Working with Parents of Handicapped Children by W. L. Heward, J. C. Dardig, and A. Rossett, 1979, p. 26, Columbus, OH: Charles E. Merrill. Copyright 1979 by Charles E. Merrill. Reprinted by permission.

596

Self-Management

Other people can also provide supplemental response prompts. For example, Watson and Tharp (2007) de- scribed the self-management procedure used by a man who had been off cigarettes for about 2 months and wanted to stop smoking permanently.

He had stopped several times in the past, but each time he had gone back to smoking. He had been successful this time because he had identified those situations in which he had relapsed in the past and had taken steps to deal with them. One of the problem situations was being at a party. The drinks, the party atmosphere, and the feel- ing of relaxation represented an irresistible temptation to “smoke just a few,” although in the past this usually led to a return to regular smoking. One night, as the man and his wife were getting ready to go out to a party, he said, “You know, I am really going to be tempted to smoke there. Will you do me a favor? If you see me bumming a cigarette from someone, remind me of how much the kids want me to stay off cigarettes.” (pp. 153–154)

Performing the Initial Steps of a Behavior Chain

Another self-management tactic featuring manipulation of antecedent events entails behaving in a manner that ensures being confronted later with a discriminative stim- ulus (SD ) that reliably evokes the target behavior. Al- though operant behavior is selected and maintained by its consequences, the occurrence of many behaviors can be controlled on a moment-to-moment basis by the pres- entation or removal of discriminative stimuli.

A self-management tactic more direct than adding response prompts to the environment is to behave in such a way so that your future behavior makes contact with a powerful discriminative stimulus for the desired behav- ior. Many tasks consist of chains of responses. Each re- sponse in a behavior chain produces a change in the environment that serves both as a conditioned reinforcer for the response that preceded it and as an SD for the next response in the chain. Each response in a behavior chain must occur reliably in the presence of its related dis- criminative stimulus for the chain to be completed suc- cessfully. By performing part of a behavioral chain (the self-management response) at one point in time, a person has changed her environment so that she will be con- fronted later with an SD that will evoke the next response in the chain and will lead to the completion of the task (the self-managed response). Skinner (1983b) provided an excellent example of this tactic.

Ten minutes before you leave your house for the day you hear a weather report: It will probably rain before you return. It occurs to you to take an umbrella (the sen- tence means quite literally what it says: The behavior of taking an umbrella occurs to you), but you are not yet

able to execute it. Ten minutes later you leave without the umbrella. You can solve that kind of problem by exe- cuting as much of the behavior as possible when it oc- curs to you. Hang the umbrella on the doorknob, or put it through the handle of your briefcase, or in some other way start the process of taking it with you. (p. 240)

Removing Items Necessary for an Undesired Behavior

Another antecedent-based self-management tactic is to alter the environment so that an undesired behavior is less likely or, better yet, impossible to emit. The smoker who discards her cigarettes and the dieter who removes all cookies and chips from his house, car, and office, have, for the moment at least, effectively controlled their smoking and junk food eating. Although other self-management efforts will probably be needed to refrain from seeking and reobtaining the harmful material (e.g., the ex-smoker who asked his wife to provide response prompts if she saw him asking for cigarettes at a party), ridding oneself of the items needed to engage in an undesired behavior is a good start.

Limiting Undesired Behavior to Restricted Stimulus Conditions

A person may be able to decrease the frequency of an undesired behavior by limiting the setting or stimulus conditions under which he engages in the behavior. To the extent that the restricted situation acquires stimulus control over the target behavior and access to the situa- tion is infrequent or otherwise not reinforcing, the target behavior will occur less frequently. Imagine a man who habitually touches and scratches his face (he is aware of this bad habit because his wife often nags him about it and asks him to stop). In an effort to decrease the fre- quency of his face touching, the man decides that he will do two things. First, whenever he becomes aware that he is engaging in face touching, he will stop; second, at any time he can go to a bathroom and touch and rub his face for as long as he wants.

Nolan (1968) reported the case of a woman who used restricted stimulus conditions in an effort to quit smok- ing. The woman noted that she most often smoked when other people were around or when she watched televi- sion, read, or lay down to relax. It was decided that she would smoke only in a prescribed place, and that she would eliminate from that place the potentially reinforc- ing effects of other activities. She designated a specific chair as her “smoking chair” and positioned it so that she could not watch TV or easily engage in conversation while sitting in it. She asked other family members not to approach her or talk to her while she was in her smoking

597

Self-Management

chair. She followed the procedure faithfully, and her smoking decreased from a baseline average of 30 ciga- rettes per day to 12 cigarettes per day. Nine days after beginning the program, the woman decided to try to re- duce her smoking even further by making her smoking chair less accessible. She put the chair in the basement, and her smoking decreased to five cigarettes per day. A month after beginning the smoking chair program, the woman had quit smoking completely.

Goldiamond (1965) used a similar tactic to help a man who continually sulked when interacting with his wife. The husband was instructed to limit his sulking be- havior to his “sulking stool,” which was placed in the garage. The man went to the garage whenever he felt like sulking, sat on his stool and sulked for as long as he wanted, and left when he was finished sulking. The man found that his sulking decreased considerably when he had to sit on a stool in the garage to do it.

Dedicating a Specific Environment for a Desired Behavior

A person may achieve some degree of stimulus control over a behavior that requires diligence and concentration by reserving or creating an environment where he will only engage in that behavior. For example, students have improved their study habits and professors their schol- arly productivity when they have selected a specific place in which to study, free of other distractions, and have not engaged in any other behaviors such as daydreaming or letter writing in that place (Goldiamond, 1965). Skinner (1981b) suggested this kind of stimulus control strategy when he offered the following advice to aspiring writers:

Equally important are the conditions in which the be- havior occurs. A convenient place is important. It should have all the facilities needed for the execution of writ- ing. Pens, typewriters, recorders, files, books, a comfort- able desk and chair. . . . Since the place is to take control of a particular kind of behavior, you should do nothing else there at any time. (p. 2)

People who do not have the luxury of devoting a cer- tain environment to a single activity can create a special stimulus arrangement that can be turned on and off in the multipurpose setting, as in this example provided by Wat- son and Tharp (2007):

A man had in his room only one table, which he had to use for a variety of activities, such as writing letters, paying bills, watching TV, and eating. But when he wanted to do concentrated studying or writing, he al- ways pulled his table away from the wall and sat on the other side of it. In that way, sitting on the other side of his table became the cue associated only with concen- trated intellectual work. (p. 150)

Most students have a single personal computer that they use for their academic work and scholarly writing but also for completing administrative tasks, writing and reading personal e-mails, conducting family business, playing games, shopping online, browsing the Internet, and so on. In a manner similar to the man who sat on the other side of his table when studying and writing, a per- son could display a particular background on the com- puter’s desktop that signals that only academic or scholarly work is to be done on the computer. For ex- ample, when he sits down to study or write, a student could replace the standard display on his computer desk- top, a photo of his dog, let’s say, with a solid green back- ground. The green desktop signals that only scholarly work will be done. Over time the “scholarly work” desk- top may acquire a degree of stimulus control over the desired behavior. If the student wants to quit working and do anything else on the computer, he must first change the background on the desktop. Doing so before he has fulfilled his time or productivity goal for a work session may generate some guilt that he can escape by re- turning to work.

This tactic can also be used to increase a desired be- havior that is not being emitted at an acceptable rate be- cause of competing undesired behaviors. In one case an adult insomniac reportedly went to bed about midnight but did not fall asleep until 3:00 or 4:00 A.M. Instead of falling asleep, the man would worry about several mun- dane problems and turn on the television. Treatment con- sisted of instructing the man to go to bed when he felt tired, but he was not to stay in bed if he was unable to sleep. If he wanted to think about his problems or watch television, he could do so, but he was to get out of bed and go into another room. When he again felt sleepy, he was to return to bed and try again to go to sleep. If he still could not sleep, he was to leave the bedroom again. The man reported getting up four or five times a night for the first few days of the program, but within 2 weeks he went to bed, stayed there, and fell asleep.

Self-Monitoring Self-monitoring has been the subject of more research and clinical applications than any other self-management strategy. Self-monitoring (also called self-recording or self-observation) is a procedure whereby a person ob- serves his behavior systematically and records the oc- currence or nonoccurrence of a target behavior. Originally conceived as a method of clinical assessment for col- lecting data on behaviors that only the client could ob- serve and record (e.g., eating, smoking, fingernail biting), self-monitoring soon became a major therapeutic inter- vention in its own right because of the reactive effects it

598

Self-Management

At the top of the page are several rows of squares. At different times during the period (whenever you think of it but don’t fill them all in at the same time) put down a “+” if you were studying and a “−” if you weren’t. If, for example, you were ready to mark a square, you would ask yourself if, for the last few minutes you had been studying and then you would put down a “+” if you had been studying or a “−” if you hadn’t been studying.

Date

+ + − − + + − + − +

− + −

Figure 2 Example of a self-recording form used by an eighth-grade girl. From “The Effect of Self-Recording on the Classroom Behavior of Two Eighth-Grade Students” by M. Broden, R. V. Hall, and B. Mitts, 1971, Journal of Applied Behavior Analysis, 4, p. 193. Copyright 1971 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

often produced. Reactivity refers to the effects on a per- son’s behavior produced by an assessment or measure- ment procedure. In general, the more obtrusive an observation and measurement method, the greater will be the likelihood of reactivity (Haynes & Horn, 1982; Kazdin, 2001). When the person observing and recording the target behavior is the subject of the behavior change program, maximum obtrusiveness exists and reactivity is very likely.

Although reactivity caused by a researcher’s mea- surement system represents uncontrolled variability in a study and must be minimized as much as possible, the reactive effects of self-monitoring are usually welcome from a clinical perspective. Not only does self-monitoring often change behavior, but also the change is typically in the educationally or therapeutically desired direction (Hayes & Cavior, 1977, 1980; Kirby, Fowler, & Baer, 1991; Malesky, 1974).

Behavior therapists have had adult clients use self- monitoring to reduce overeating, decrease smoking (Lip- inski, Black, Nelson, & Ciminero, 1975; McFall, 1977), and stop biting their nails (Maletzky, 1974). Self- monitoring has helped students with and without dis- abilities be on task more often in the classroom (Blick & Test, 1987; Kneedler & Hallahan, 1981; Wood, Murdock, Cronin, Dawson, & Kirby, 1998), decrease talk-outs and aggression (Gumpel & Shlomit, 2000; Martella, Leonard, Marchand-Martella, & Agran, 1993), improve their per- formance in a variety of academic subject areas (Harris, 1986; Hundert & Bucher, 1978; Lee & Tindal, 1994; Maag, Reid, & DiGangi, 1993; Moxley, Lutz, Ahlborn, Boley, & Armstrong, 1995; Wolfe, Heron, & Goddard, 2000), and complete homework assignments (Trammel,

Schloss, & Alper, 1994). Classroom teachers have used self-monitoring to increase their use of positive state- ments during classroom instruction (Silvestri, 2004).

In one of the first published accounts of self-moni- toring in the classroom, Broden, Hall, and Mitts (1971) analyzed the effects of self-recording on the behavior of two eighth-grade students. Liza was earning a D– in his- tory and exhibited poor study behavior during the lecture format class. Using 10-second momentary time sampling for 30 minutes each day, an observer seated in the back of the classroom (Liza was not told she was being ob- served) produced a 7-day baseline showing that Liza ex- hibited study behaviors (e.g., faced the teacher, took notes when appropriate) an average of 30% of the observation intervals, despite two conferences with the school coun- selor in which she promised to “really try.” Prior to the eighth session, the counselor gave Liza a piece of paper with three rows of 10 squares (see Figure 2) and directed her to record her study behavior “when she thought of it” during her history class sessions. Some aspects of study behavior were discussed at this time, including a definition of what constituted studying.

Liza was instructed to take the slip to class each day and to record a “+” in the square if she was studying or had been doing so for the last few minutes, and a “�” if she was not studying at the time she thought to record. Sometime before the end of the school day she was to turn it in to the counselor. (p. 193)

Figure 3 shows the results of Liza’s self-monitoring. Her level of study behavior increased to 78% (as mea- sured by the independent observer) and stayed at ap- proximately that level during Self-Recording 1, quickly

599

Self-Management

100

P er

ce n

t St

u d

y B

eh av

io r

Baseline 1 Self- Recording 1

Baseline 2 Self Recording 2 Praise Only

Self-Recording Plus Praise

Baseline 3

No slip

6 5 4 3 2 1 0

T ea

ch er

A tt

en ti

5 10 15 20 25 30 35 40

Sessions

Figure 3 Percentage of observed intervals in which an eighth-grade girl paid attention in a history class. From “The Effect of Self-Recording on the Classroom Behavior of Two Eighth-Grade Students” by M. Broden, R. V. Hall, and B. Mitts, 1971, Journal of Applied Behavior Analysis, 4, p. 194. Copyright 1971 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

decreased to baseline levels when self-recording was ter- minated during Baseline 2, and averaged 80% during Self-Recording 2. During Self-Recording Plus Praise, Liza’s teacher attended to her whenever he could during history class and praised her for study behavior when- ever possible. Liza’s level of study behavior during this condition increased to 88%.

The bottom part of Figure 3 shows the number of times per session the observer recorded teacher attention toward Liza. The fact that the frequency of teacher at- tention did not correlate in any apparent way with Liza’s study behavior during the first four phases of the exper- iment suggests that teacher attention—which often ex- erts a powerful influence on student behavior—was not a confounding variable and that the improvements in Liza’s study behavior most likely can be attributed to the self-recording procedure. However, the effects of self- monitoring may have been confounded by the fact that Liza turned her self-recording slips in to the counselor before the end of each school day, and at weekly student– counselor conferences she received praise from the coun-

selor for recording slips with a high percentage of plus marks.

In the second experiment reported by Broden and colleagues (1971), self-monitoring was used with Stu, an eighth-grade student who talked out continually in class. Using a 10-second partial interval recording procedure to calculate the number of talk-outs per minute, an inde- pendent observer recorded Stu’s behavior during both parts of a math class that met before and after lunch. Self- recording was begun first during the before-lunch por- tion of the math class. The teacher handed Stu a slip of paper on which was marked a 2-inch by 5-inch rectan- gular box and the instruction, “Put a mark every time you talk out” (p. 196). Stu was told to use the recording slip and turn it in after lunch. At the top of the recording slip was a place for the student’s name and the date. No other instructions were given to Stu, nor were any conse- quences or other contingencies applied to his behavior. Self-recording was later implemented during the second half of the math class. During the prelunch portion of the math class, Stu’s baseline rate of 1.1 talk-outs per minute

600

Self-Management

was followed by a rate of 0.3 talk-outs per minute when he self-recorded. After lunch Stu talked out at an average rate of 1.6 times per minute before self-recording and 0.5 times per minute during self-recording.

The combination reversal and multiple baseline de- sign showed a clear functional relation between Stu’s re- duced talk-outs and the self-monitoring procedure. During the final phase of the experiment, however, Stu talked out at rates equal to initial baseline levels, even though self-recording was in effect. Broden and col- leagues (1971) suggested that the lack of effect produced by self-recording during the final phase may have re- sulted from the fact that “no contingencies were ever ap- plied to differential rates of talking out and the slips thus lost their effectiveness” (p. 198). It is also possible that the initial decreases in talk-outs attributed to self-recording were confounded by Stu’s expectations of some form of teacher reinforcement.

As can be seen from these two experiments, it is ex- tremely difficult to isolate self-monitoring as a straight- forward, “clean” procedure; it almost always entails other contingencies. Nevertheless, the various and combined procedures that comprise self-monitoring have often proved to be effective in changing behavior. However, the effects of self-monitoring on the target behavior are sometimes temporary and modest and may require the implementation of reinforcement contingencies to main- tain the desired behavior changes (e.g., Ballard & Glynn, 1975; Critchfield & Vargas, 1991).

Self-Evaluation

Self-monitoring is often combined with goal setting and self-evaluation. A person using self-evaluation (also called self-assessment) compares her performance with a predetermined goal or standard (e.g., Keller, Brady, & Taylor, 2005; Sweeney et al., 1993). For example, Grossi and Heward (1998) taught four restaurant trainees with developmental disabilities to select productivity goals, self-monitor their work, and self-evaluate their perfor- mance against a competitive productivity standard (i.e., the typical rate at which nondisabled restaurant employ- ees perform the task in competitive settings). Job tasks in this study included scrubbing pots and pans, loading a dishwasher, busing and setting tables, and mopping and sweeping floors.

Self-evaluation training with each trainee took ap- proximately 35 minutes across three sessions and con- sisted of five parts: (1) rationale for working faster (e.g., finding and keeping a job in a competitive setting), (2) goal setting (trainee was shown a simple line graph of his baseline performance compared to the competitive standard and prompted to set a productivity goal), (3) use

of timer or stopwatch, (4) self-monitoring and how to chart self-recorded data on a graph that showed compet- itive standard with a shaded area, and (5) self-evaluation (comparing their work against the competitive standard and making self-evaluative statements, e.g., “I’m not in the shaded area. I need to work faster.” or “Good, I’m in the area.”). When a trainee met his goal for three consec- utive days, a new goal was selected. At that time all four trainees had selected goals within the competitive range.

The work productivity of all four trainees increased as a function of the self-evaluation intervention. Figure 4 shows the results for one of the participants, Chad, a 20- year-old male with mild mental retardation, cerebral palsy, and a seizure disorder. For the 3 months prior to the study, Chad had been training at two work tasks at the dishwashing job station: scrubbing cooking pots and pans and racking dishes (a six-step behavior chain for loading an empty dishwasher rack with dirty dishes). Throughout the study, pot scrubbing was measured during 10-minute observation sessions both before and after the daily lunch operation. Racking dishes was measured using a stop- watch to time to the nearest second how long a trainee took to load a rack of dishes; 4 to 8 racks per session were timed during the busiest hour of the lunch shift. After each observation session during baseline, Chad re- ceived feedback on the accuracy and quality of his work. No feedback was given during baseline on the produc- tivity of his performance.

During baseline Chad scrubbed an average of 4.5 pots and pans every 10 minutes, and none of his 15 base- line trials were within the competitive range of 10 to 15 pots. Chad’s pot scrubbing rate increased to a mean of 11.7 pots during self-evaluation, with 76% of 89 self- evaluation trials at or above the competitive range. Dur- ing baseline it took Chad a mean of 3 minutes 2 seconds to load one rack of dishes, and 19% of his 97 baseline trials were within the competitive range of 1 to 2 min- utes. During self-evaluation, Chad’s time to load one rack improved to a mean of 1 minute 55 seconds, and 70% of the 114 self-evaluation trials were within the competitive range. At the end of the study, three of the four trainees stated that they liked to time and record their work per- formance. Although one trainee stated it was “too stress- ful” to time and record his work, he said that self-monitoring helped him show other people that he could do the work.

Self-Monitoring with Reinforcement

Self-monitoring is often part of an intervention package that includes reinforcement for meeting either self- or teacher-selected goals (e.g., Christian & Poling, 1997;

601

Self-Management

N u

m b

er o

f P

ot s

Sc ru

bb ed

5 10 15 20 25 30 35 40 45 50 55 60

Baseline Self-Evaluation

Attempted

Accepted

M ea

n N

u m

b er

a n

d R

an ge

o f

M in

u te

s to

R ac

k D

is h

5 10 15 20 25 30 35 40 45 50 55 60

SessionsChad

Figure 4 Number of pots scrubbed in 10 minutes and mean and range of minutes to load a rack of dishes in dishwasher during baseline and self-evaluation conditions. Shaded bands represent competitive performance range of restaurant workers without disabilities. Dashed hori- zontal line represents Chad’s self-selected goal; no dashed line means that Chad’s goal was within the competitive range. Vertical lines through data points on bottom graph show range of Chad’s performance across 4 to 8 trials. From “Using Self-Evaluation to Improve the Work Productivity of Trainees in a Community-Based Restaurant Training Program” by T. A. Grossi and W. L. Heward, 1998, Education and Training in Mental Retardation and Developmental Disabilities, 33, p. 256. Copyright 1998 by the Division on Developmental Disabilities. Reprinted by permission.

Dunlap & Dunlap, 1989; Olympia et al., 1994; Rhode, Morgan, & Young, 1983). The reinforcer may be self- administered or teacher delivered. For example, Koegel and colleagues (1992) taught four children with autism between the ages of 6 and 11 to obtain their own rein- forcers after they had self-recorded a criterion number of appropriate responses to questions from others (e.g., “Who drove you to school today?”).

Teacher-delivered reinforcement was a component of a self-monitoring intervention used by Martella and colleagues (1993) to help Brad, a 12-year-old student with mild mental retardation, reduce the number of neg- ative statements he made during classroom activities (e.g., “I hate this #*%@! calculator,” “Math is crappy”). Brad self-recorded negative statements during two class peri- ods, charted the number on a graph, and then compared his count with the count obtained by a student teacher. If Brad’s self-recorded data agreed with the teacher at a level of 80% or higher, he received his choice from a

menu of “small” reinforcers (items costing 25 cents or less). If Brad’s self-recorded data were in agreement with the trainer’s data and were at or below a gradually de- clining criterion level for four consecutive sessions, he was allowed to choose a “large” (more than 25 cents) re- inforcer (see Figure 13).

Why Does Self-Monitoring Work?

The behavioral mechanisms that account for the effec- tiveness of self-monitoring are not fully understood. Some behavior theorists suggest that self-monitoring is effective in changing behavior because it evokes self- evaluative statements that serve either to reinforce de- sired behaviors or to punish undesired behaviors. Cautela (1971) hypothesized that a child who records on a chart that he has completed his chores may emit covert verbal responses (e.g., “I am a good boy”) that serve to rein- force chore completion. Malott (1981) suggested that

602

Self-Management

self-monitoring improves performance because of what he called guilt control. Self-monitoring less-than- desirable behavior produces covert guilt statements that can be avoided by improving one’s performance. That is, the target behavior is strengthened through negative re- inforcement by escape and avoidance of the guilty feel- ings that occur when one’s behavior is “bad.”

Descriptions of the self-monitoring techniques used by two famous authors apparently agreed with Malott’s guilt control hypothesis. Novelist Anthony Trollope, writ- ing in his 1883 autobiography, stated:

When I have commenced a new book, I have always prepared a diary, divided into weeks, and carried on for the period which I have allowed myself for the comple- tion of the work. In this I have entered, day by day, the number of pages I have written, so that if at any time I have slipped into idleness for a day or two, the record of that idleness has been there staring me in the face, and demanding of me increased labour, so that the defi- ciency might be supplied. . . . I have allotted myself so many pages a week. The average number has been about 40. It has been placed as low as 20, and has risen to 112. And as a page is an ambiguous term, my page has been made to contain 250 words; and as words, if not watched, will have a tendency to straggle, I have had every word counted as I went. . . . There has ever been the record before me, and a week passed with an insuffi- cient number of pages has been a blister to my eye and a month so disgraced would have been a sorrow to my heart. (from Wallace, 1977, p. 518)

Guilt control also helped to motivate the legendary Ernest Hemingway. Novelist Irving Wallace (1977) re- ported the following extract from an article by George Plimpton about the self-monitoring technique used by the Hemingway.

He keeps track of his daily progress—“so as not to kid myself ”—on a large chart made out of the side of a cardboard packing case and set up against the wall under the nose of a mounted gazelle head. The numbers on the chart showing the daily output of words differ from 450, 575, 462, 1250, back to 512, the higher fig- ures on days Hemingway put in extra work so he won’t feel guilty spending the following day fishing on the gulf stream. (p. 518)

Exactly what principles of behavior are operating when self-monitoring results in a change in the target be- havior is not known because much of the self-monitoring procedure consists of private, covert behaviors. In addi- tion to the problem of access to these private events, self- monitoring is usually confounded by other variables. Self-monitoring is often part of a self-management pack- age in which contingencies of reinforcement, punish- ment, or both, are included, either explicitly (e.g., “If I run 10 miles this week, I can go to the movies”) or implicitly

(e.g., “I’ve got to show my record of calories consumed to my wife”). Regardless of the principles of behavior involved, however, self-monitoring is often an effective procedure for changing one’s behavior.

Guidelines and Procedures for Self-Monitoring

Practitioners should consider the following suggestions when implementing self-monitoring with their students and clients.

Provide Materials That Make Self-Monitoring Easy

If self-monitoring is difficult, cumbersome, or time- consuming, at best it will be ineffective and disliked by the participant, and at worst it may have negative effects on the behavior. Participants should be provided with ma- terials and devices that make self-monitoring as easy and efficient as possible. All of the devices and procedures for measuring behavior (e.g., paper and pencil, wrist counters, hand-tally counters, timers, stopwatches) can be used for self-monitoring. For example, a teacher who wants to self-monitor the number of praise statements she makes to students during a class period could put ten pennies in her pocket prior to start of class. Each time she praises a student’s behavior, she moves one coin to another pocket.

The recording forms used for most self-monitoring applications can and should be very simple. Self-recording forms consisting of little more than a series of boxes or squares such as the one shown in Figure 2 are often ef- fective. At various intervals the participant might write a plus or minus, circle yes or no, or put an X through a smil- ing face or sad face, in a kind of momentary time sam- pling procedure; or he could record a count of the number of responses made during a just-completed interval.

Recording forms can be created for self-monitoring specialized tasks or a chain of behaviors. Dunlap and Dunlap (1989) taught students with learning disabilities to self-monitor the steps they used to solve subtraction problems with regrouping. Each student self-monitored his work by recording a plus or a minus next to each step on an individualized checklist of steps (e.g., “I under- lined all the top numbers that were smaller than the bot- tom.”; “I crossed out only the number next to the underlined number and made it one less.” [p. 311]) de- signed to prompt the student from committing specific types of errors.

Lo (2003) taught elementary students who were at- risk for behavioral disorders to use the form shown in Figure 5 to self-monitor whether they worked quietly,

603

Self-Management

Figure 5 Form used by elementary school students to self-monitor whether they worked quietly and followed a prescribed sequence for recruiting teacher assistance. From “Functional Assessment and Individualized Intervention Plans: Increasing the Behavioral Adjustment of Urban Learners in General and Special Education Settings” by Y. Lo, 2003. Unpublished doctoral dissertation. Columbus, OH: The Ohio State University. Reprinted by permission.

evaluated their work, and followed a prescribed sequence for obtaining teacher assistance during independent seat- work activities. The form, which was taped to each stu- dent’s desk, functioned as a reminder to students of the expected behaviors and as a device on which they self- recorded those behaviors.

“Countoons” are self-monitoring forms that illustrate the contingency with a series of cartoon-like frames. A

countoon reminds young children of not only what be- havior to record, but also what consequences will follow if they meet their performance criteria. Daly and Ranalli (2003) created six-frame countoons that enable students to self-record an inappropriate behavior and an incom- patible appropriate behavior. In the countoon shown in Figure 6, Frames F1 and F4 show the student doing her math work, appropriate behavior that is counted in Frame F5. The criterion number of math problems to meet the contingency—in this case 10—is also indicated in Frame F5. Frame F2 shows the student talking with a friend, the inappropriate behavior to be counted in F3. The student must not chat more than six times to meet the contin- gency. The “What Happens” frame (F6) depicts the re- ward the student will earn by meeting both parts of the contingency. Daly and Ranalli have provided detailed steps for creating and using countoons to teach self- management skills to children.

Self-monitoring forms can also be designed to en- able self-recording of multiple tasks across days. For ex- ample, teachers working with high school-level students might employ the Classroom Performance Record (CPR) developed by Young and colleagues (1991) as a way to as- sist students with monitoring their assignments, comple- tion of homework, points earned, and citizenship scores. The form also provides students with information about their current status in the class, their likely semester grade, and tips for improving their performance.

Provide Supplementary Cues or Prompts

Although some self-recording devices—a recording form taped to a student’s desk, a notepad for counting calories carried in a dieter’s shirt pocket, a golf counter on a teacher’s wrist—serve as continual reminders to self- monitor, additional prompts or cues to self-monitor are often helpful. Researchers and practitioners have used a variety of auditory, visual, and tactile stimuli as cues and prompts for self-monitoring.

Auditory prompts in the form of prerecorded signals or tones have been widely used to cue self-monitoring in classrooms (e.g., Blick & Test, 1987; Todd, Horner, & Sugai, 1999). For example, second-graders in a study by Glynn, Thomas, and Shee (1973) placed a checkmark in a series of squares if they thought they were on task at the moment they heard a tape-recorded beep. A total of 10 beeps occurred at random intervals during a 30-minute class period. Using a similar procedure, Hallahan, Lloyd, Kosiewicz, Kauffman, and Graves (1979) had an 8-year- old boy place a checkmark next to Yes or No under the heading “Was I paying attention?” when he heard a tone played at random intervals by a tape recorder.

Ludwig (2004) used visual cues written on the class- room board to prompt kindergarten children to self-record

604

Self-Management

Do my math work

What I Do

F1 F2 F4 F6

Chat with my friend

Do my math work 5 minutes of my favorite

game on Friday

My Count

1 2 3 4

5 6 7 8

9 10 11 12

What I Do What Happens

1 2 3 4

7 8 9 10

12 13 14 15

My Count

Figure 6 Example of a countoon that can be taped to a student’s desk as a reminder of target behaviors, the need to self-record, and the conse- quence for meeting the contingency. From “Using Countoons to Teach Self-Monitoring Skills” by P. M. Daly and P. Ranalli, 2003, Teaching Exceptional Children, 35 (5), p. 32. Copyright 2003 by the Council for Exceptional Children. Reprinted by permission.

their productivity during a morning seatwork activity in which they wrote answers on individual worksheets to items, problems, or questions that the teacher had printed on a large whiteboard on the classroom wall. The work was divided into 14 sections by topic and covered a va- riety of curriculum areas (e.g., spelling, reading com- prehension, addition and subtraction problems, telling time, money). At the end of each section the experimenter drew a smiley face with a number from 1 to 14. The num- bered smiley faces drawn on the board corresponded to 14 numbered smiley faces on each student’s self- monitoring card.

Tactile prompts can also be used to signal self- recording opportunities. For example, the MotivAider (www.habitchange.com) is a small electronic device that vibrates at fixed or variable time intervals programmed by the user. This device is excellent for signaling people to self-monitor or perform other self-management tasks, or to prompt practitioners to attend to the behavior of stu- dents or clients.8

Whatever form they take, prompts to self-monitor should be as unobtrusive as possible so they do not dis- rupt the participant or others in the setting. As a general rule, practitioners should provide frequent prompts to self-monitor at the beginning of a self-management in- tervention and gradually reduce their number as the par- ticipant becomes accustomed to self-monitoring.

8Flaute, Peterson, Van Norman, Riffle, and Eakins (2005) described 20 ex- amples for using a MotivAider to improve behavior and productivity in the classroom.

Self-Monitor the Most Important Dimension of the Target Behavior

Self-monitoring is an act of measurement. But as we saw, there are a variety of dimensions by which a given target behavior could be measured. What dimension of the target behavior should be self-monitored? A person should self-monitor the target behavior dimension that, should desired changes in its value be achieved, would yield the most direct and significant progress toward the person’s goal for the self-management program. For example, a man wishing to lose weight by eating less food could measure the number of bites he takes throughout the day (count), the number of bites per minute during each meal (rate), how much time elapses from the time he sits down at the table to when he takes the first bite (latency), how long he pauses between bites (interresponse time), or the amount of time his meals last (duration). Although daily measures of each of these dimensions would provide the man with some quanti- tative information about his eating, none of the dimen- sions would be as directly related to his goal as would a count of the total number of calories he consumes each day.

Several studies have examined the question of whether students should self-monitor their on-task be- havior or their academic productivity. Harris (1986) con- ducted one of the most frequently cited studies in this line of research. She taught four elementary students with learning disabilities to self-monitor either their on-task behavior or the number of academic responses they

605

Self-Management

completed while practicing spelling. During the self- monitoring of attention condition, students asked them- selves “Was I paying attention?” and checked their answer under Yes and No columns on a card whenever they heard a tape-recorded tone. During the self- monitoring of productivity condition, students counted the number of spelling words written at the end of the practice period. Both self-monitoring procedures resulted in increased on-task behavior for all four students. How- ever, for three of the four students, self-monitoring of productivity resulted in higher academic response rates than did self-monitoring of on-task behavior.

Similar results have been reported in other studies comparing the differential effects of self-monitoring on- task behavior or productivity (e.g., Maag et al., 1993; Lloyd, Bateman, Landrum, & Hallahan, 1989; Reid & Harris, 1993). Although both procedures increase on-task behavior, students tend to complete more assignments when self-monitoring academic responses than when self- recording on-task behavior. Additionally, most students prefer self-monitoring academic productivity to on-task behavior.

In general we recommend teaching students to self- monitor a measure of academic productivity (e.g., num- ber of problems or items attempted, number correct) rather than whether they are on task. This is because in- creasing on-task behavior, whether by self-monitoring or by contingent reinforcement, does not necessarily result in a collateral increase in productivity (e.g., Marholin & Steinman, 1977; McLaughlin & Malaby, 1972). By con- trast, when productivity is increased, improvements in on-task behavior almost always occur. However, a stu- dent whose persistent off-task and disruptive behaviors are creating problems for him or others in the classroom may benefit more from self-monitoring on-task behav- ior, at least initially.

Self-Monitor Early and Often

In general, each occurrence of the target behavior should be self-recorded as soon as possible. However, the act of self-monitoring a behavior the person wants to increase should not disrupt “the flow of the behavior” (Critchfield, 1999). Self-monitoring target behaviors that produce nat- ural or contrived response products (e.g., answers on aca- demic worksheets, words written) can be conducted after the session by permanent product.

Relevant aspects of some behaviors can be self- monitored even before the target behavior itself has oc- curred. Self-recording a response early in a behavior chain that leads to an undesired behavior the person wants to decrease may be more effective in changing the target behavior in the desired direction than will recording on the terminal behavior in the chain. For example, Rozen-

sky (1974) reported that, when a woman who had been a heavy smoker for 25 years recorded the time and place she smoked each cigarette, there was little change in her rate of smoking. She then began to record the same in- formation each time she noticed herself beginning the chain of behaviors that would lead to smoking: reaching for her cigarettes, removing a cigarette from the pack, and so on. She stopped smoking within a few weeks of self-monitoring in this fashion.

Generally, a person should self-monitor more often at the beginning of a behavior change program. If and as performance improves, the frequency of self-monitoring can be decreased. For example, Rhode and colleagues (1983) had students with behavioral disorders begin self- evaluating (on a 0- to 5-point scale) the extent to which they followed classroom rules and completed academic work correctly at the end of each 15-minute interval throughout the school day. Over the course of the study, the self-evaluation intervals were increased gradually, first to every 20 minutes, then every 30 minutes, then once per hour. Eventually, the self-evaluation cards were withdrawn and students reported their self-evaluations verbally. In the final self-evaluation condition, the stu- dents verbally self-evaluated their academic work and compliance with classroom rules on an average of every 2 days (i.e., a VR 2-day schedule).

Reinforce Accurate Self-Monitoring

Some studies have found little correlation between the accuracy of self-monitoring and its effectiveness in changing the behavior being recorded (e.g., Kneedler & Hallahan, 1981; Marshall, Lloyd, & Hallahan, 1993). It appears that accurate self-monitoring is neither a suffi- cient nor a necessary condition for behavior change. For example, Hundert and Bucher (1978) found that, even though the students became highly accurate in self- scoring their arithmetic, their performance on the arith- metic itself did not improve. On the other hand, in the Broden and colleagues (1971) study, both Liza and Stu’s behavior improved even though their self-recorded data seldom matched the data from independent observers.

Nevertheless, accurate self-monitoring is desirable, especially when participants are using their self-recorded data as the basis for self-evaluation or self-administered consequences.

Although some studies have shown that young chil- dren can accurately self-record their behavior without spe- cific external contingencies for accuracy (e.g., Ballard & Glynn, 1975; Glynn, Thomas, & Shee, 1973), other re- searchers have reported low agreement between the self- recordings of children and the data collected by independent observers (Kaufman & O’Leary, 1972; Turkewitz, O’Leary & Ironsmith, 1975). One factor that

606

Self-Management

seems to affect the accuracy of self-scoring is the use of self-reported scores as a basis for reinforcement. San- togrossi, O’Leary, Romanczyk, and Kaufman (1973) found that, when children were allowed to evaluate their own work and those self-produced evaluations were used to determine levels of token reinforcement, their self- monitoring accuracy deteriorated over time. Similarly, Hundert and Bucher (1978) found that students who pre- viously had accurately self-scored their arithmetic assign- ments greatly exaggerated their scores when higher scores resulted in points that could be exchanged for prizes.

Rewarding children for producing self-recorded data that match the data of an independent observer and spot- checking students’ self-scoring reports are two proce- dures that have been used successfully to increase the accuracy of self-monitoring by young children. Drab- man, Spitalnik, and O’Leary (1973) used these proce- dures in teaching children with behavior disorders to self-evaluate their own classroom behavior.

Now something different is going to happen. If you get a rating within plus or minus one point of my rating, you can keep all your points. You lose all your points if you are off my rating by more than plus or minus one point. In addition, if you match my rating exactly you get a bonus point. (O’Leary, 1977, p. 204)

After the children demonstrated that they could eval- uate their own behavior reliably, the teacher began checking only 50% of the children’s self-ratings by pulling names from a hat at the end of the period, then 33%, then 25%, then 12%. During the last 12 days of the study she did not check any child’s self-rating. Throughout this period of reduced checks and finally no checks at all, the children continued to self-evaluate ac- curately. Rhode and colleagues (1983) used a similar faded matching technique.

Self-Administered Consequences Arranging to have specified consequences follow occur- rences (or nonoccurrences) of one’s behavior is one of the fundamental approaches to self-management. In this section we review some of the tactics people have used to self-reinforce and self-punish. First, we briefly exam- ine some of the conceptual issues raised by the concept of “self-reinforcement.”

Is Self-Reinforcement Possible?

Skinner (1953) pointed out that self-reinforcement should not be considered synonymous with the principle of op- erant reinforcement.

The place of operant reinforcement in self-control is not clear. In one sense, all reinforcements are self- administered since a response may be regarded as “pro- ducing” its reinforcement, but “reinforcing one’s own behavior” is more than this. . . . Self-reinforcement of operant behavior presupposes that the individual has it in his power to obtain reinforcement but does not do so until a particular response has been emitted. This might be the case if a man denied himself all social contacts until he had finished a particular job. Something of this sort unquestionably happens, but is it operant reinforce- ment? It is roughly parallel to the procedure in condi- tioning the behavior of another person. But it must be remembered that the individual may at any moment drop the work in hand and obtain the reinforcement. We have to account for his not doing so. It may be that such in- dulgent behavior has been punished—say, with disap- proval—except when a piece of work has just been completed. (pp. 237–38)

In his discussion of self-reinforcement, Goldiamond (1976) continued with Skinner’s example and stated that the fact that a person “does not cheat, but engages in the task, cannot be explained simply by resort to his self- reinforcement by social contact, which he makes avail- able contingent on his finishing the job” (p. 510). In other words, the variables influencing the controlling re- sponse—in this case not allowing oneself social contact until the job is complete—must still be accounted for; simply citing self-reinforcement as the cause is an ex- planatory fiction.

The issue is not whether procedures labeled as “self- reinforcement” often work to change behavior in ways that resemble the defining effect of reinforcement; they do. However, a careful examination of instances of self- reinforcement reveals that something more than, or dif- ferent from, a straightforward application of positive reinforcement is involved (e.g., Brigham, 1980; Catania, 1975, 1976; Goldiamond, 1976a, 1976b; Rachlin, 1977). As a technical term, self-reinforcement (as also, self- punishment) is a misnomer, and the issue is not merely a semantic one as some writers have suggested (e.g., Mahoney, 1976). Assigning the effectiveness of a be- havior change tactic to a well-understood principle of be- havior when something other than, or in addition to, that principle is operating overlooks relevant variables that may hold the key to a fuller understanding of the tactic. Considering the analysis of a behavioral episode com- plete once it has been identified as a case of self- reinforcement precludes further search for other relevant variables.

We agree with Malott (2005a; Malott & Harrison, 2002; Malott & Suarez, 2004) who argued that performance-management contingencies, whether designed and implemented by oneself or by others, are best viewed as rule-governed analogs of reinforcement and punishment

607

Self-Management

contingencies because the response-to-consequence delay is too great.9 Malott (2005a) provided the following exam- ples of self-management contingencies as analogs to neg- ative reinforcement and punishment contingencies:

[Consider the] contingency, Every day I eat more than 1,250 calories, I will give someone else $5 to spend friv- olously, like a despised roommate or despised charity, though for many of us, the loss of $5 to a beloved char- ity is sufficiently aversive that the thought of it will pun- ish overeating. [This] contingency is an analog to a penalty [punishment] contingency, an analog because the actual loss of the $5 will normally occur more than 1 minute after exceeding the 1,250 calorie limit. Such analog penalty contingencies are effective in decreasing undesirable behavior. And to increase desirable behav- ior, an analog to an avoidance [negative reinforcement] contingency works well: Every day I exercise for one hour, I avoid paying a $5 fine. But, if you have not finished your hour by midnight, you must pay the $5. (p. 519, emphasis in original, words in brackets added)

Self-Administered Consequences to Increase Desired Behavior

A person can increase the future frequency of a target be- havior in a self-management program by applying con- tingencies that are analogs to positive reinforcement and negative reinforcement.

Self-Management Analogs of Positive Reinforcement

Several self-reinforcement studies with schoolchildren have involved positive reinforcement in which partici- pants obtained a self-determined number of tokens, points, or minutes of free time based on a self-assessment of their performance (Ballard & Glynn, 1975; Bolstad & Johnson, 1972; Glynn, 1970; Koegel et al., 1992; Olympia et al., 1994).

The effects of treatments involving self-administered rewards are difficult to evaluate because they are typi- cally confounded by self-monitoring and self-evaluation. However, in a study by Ballard and Glynn (1975), third- graders were taught after a baseline condition to self- score and self-record several aspects of their writing— number of sentences, number of describing words, and number of action words. Self-monitoring had no effect on any of the variables measured even though the stu- dents handed in their counting sheets with their writing each day. The children were then given a notebook in which to record their points, which could be exchanged

9The critical importance of immediacy in reinforcement is discussed in Chapter entitled “ Positive Reinforcement.”

at the rate of one point per minute for each student’s choice of activities during an earned-time period each day. The self-reinforcement procedure resulted in large increases in each of the three dependent variables.

Self-administered reinforcement does not have to be self-delivered: the learner could make a response that re- sults in another person providing the reinforcer. For ex- ample, in studies on self-recruited reinforcement, students are taught to periodically self-evaluate their work and then show it to their teacher and request feedback or as- sistance (e.g., Alber, Heward, & Hippler, 1999; Craft, Alber, & Heward, 1998; Mank & Horner, 1987; Smith, & Sugai, 2000). In a sense, students administer their own reinforcer by recruiting the teacher’s attention, which often results in praise and other forms of reinforcement (see Alber & Heward, 2000 for a review).

Todd and colleagues (1999) taught an elementary student to use a self-management system that included self- monitoring, self-evaluation, and self-recruited reinforce- ment. Kyle was a 9-year-old boy diagnosed with a learning disability and receiving special education services in read- ing, math, and language arts. Kyle’s IEP also included sev- eral objectives for problem behavior (e.g., disrupting independent and group activities, teasing and taunting class- mates, and saying sexually inappropriate comments). The classroom teacher’s goal setting and daily evaluation had proven ineffective. An “action team” conducted a functional assessment and designed a support plan that included the self-management system.

Kyle was taught how to use the self-management system during two 15-minute one-on-one training ses- sions in which he practiced self-recording numerous role- played examples and nonexamples of on-task and off-task behavior and learned appropriate ways to recruit teacher attention and praise. For self-monitoring, Kyle listened with a single earplug to a 50-minute cassette tape on which 13 checkpoints (e.g., “check one,” “check two”) had been prerecorded on a VI 4-minute schedule (rang- ing from 3- to 5-minute intervals between checkpoints). Each time Kyle heard a checkpoint, he marked a plus (if he had been working quietly and keeping his hands, feet, and objects to himself) or a zero (teasing peers and/or not working quietly) on a self-recording card.

Todd and colleagues (1999) described the recruitment of teacher praise and how Kyle’s special program was in- tegrated into the reinforcement system for the entire class:

Each time Kyle marked three pluses on his card he raised his hand (during instruction) or walked up to the teacher (during group project time) and requested feedback on his performance. The teacher acknowledged Kyle’s good work and placed a mark on his self-monitoring card to show where he should begin a new count of three pluses. In addition to these within-session contingencies, Kyle could earn a self-manager sticker at the end of each class

608

Self-Management

period if he had no more than two zeroes for the period. Stickers were earned by all students in the class for ap- propriate behavior and were pooled weekly for class re- wards. Because these stickers were combined, Kyle’s stickers were valued by all the students and provided opportunities for positive peer attention. (p. 70)

Procedures for the second self-management phase (SM2) were the same as the first phase; during the third phase (SM3) Kyle used a 95-minute tape on which 16 checkpoints were prerecorded on a VI 5-minute schedule (ranging from 4 to 6 minutes). When Kyle used the self- management system, the percentage of intervals in which he engaged in problem behavior was much lower than baseline levels (see Figure 7). Large increases in on-task behavior and completion of academic work were also noted. An important outcome of this study was that Kyle’s teacher praised him more often. Significantly, the self- management intervention was initiated in Class Period B (bottom graph in Figure 7) at the teacher’s request be- cause she had noted an immediate and “dramatic change” in Kyle’s performance when he began self-monitoring and recruiting teacher praise in Class Period A. This out- come is strong evidence for the social validity of the self- management intervention.

Self-Management Analogs of Negative Reinforcement

Many successful self-management interventions involve self-determined escape and avoidance contingencies that are analogous to negative reinforcement. Most of the case studies featured in Malott and Harrison’s (2002) excellent book on self-management, I’ll Stop Procrastinating When I Get Around To It, feature escape and avoidance contin- gencies in which emitting the target behavior enabled the person to avoid an aversive event. For example:

Target Behavior/Objective Self-Management Contingency

Make a daily journal entry to Each day I failed to make a be able to remember the journal entry, I had to interesting things that had do my friend’s chores, happened to include in a including doing the dishes weekly letter to my parents. and laundry. (Garner, 2002)

Run 30 minutes, 3 times per I had to pay the piper $3 at week. 10:00 P.M. Sunday for every

day I’d run less than three times that week. I could count only one run per day. (Seymour, 2002)

Practice guitar for half an Do 50 sit-ups at 11:00 P.M. hour each day of the week on Sundays for every day before 11:00 P.M. that I did not practice during

the week. (Knittel, 2002)

In response to the uneasiness some people may feel about using negative reinforcement to control their behav- ior, Malott (2002) made the following case for building “pleasant aversive control” into self-management programs.

Aversive control doesn’t have to be aversive! . . . Here’s what I think you need in order to have a pleasant aver- sive control procedure. You need to make sure that the aversive consequence, the penalty, is small. And you need to make sure the penalty is usually avoided—the avoidance response needs to be one that the person will readily make, most of the time, as long as the avoidance procedure is in effect.

Our everyday life is full of such avoidance proce- dures, and yet they don’t make us miserable. You don’t have an anxiety attack every time you walk through a doorway, even though you might hurt yourself if you don’t avoid bumping into the doorjamb. And you don’t break out in a cold sweat every time you put the leftovers in the refrigerator, and thereby avoid leaving them out to spoil. So that leads us to the Self-Management Rule: Don’t hesitate to use an avoidance procedure to get yourself to do something you want to do anyway. Just be sure the aversive outcome is as small as possible (but not so small it’s ineffective) and the response is easy to make. (p. 8-2)

Self-Administered Consequences to Decrease Undesired Behavior

The frequency of an undesirable behavior can be de- creased by self-administered consequences analogous to positive punishment or negative punishment.

Self-Management Analogs of Positive Punishment

A person can decrease the frequency of an undesired be- havior by following each occurrence with the onset of painful stimulus or aversive activity. Mahoney (1971) re- ported a case study in which a man beleaguered by obses- sive thoughts wore a heavy rubber band around his wrist. Each time he experienced an obsessive thought, he snapped the rubber band, delivering a brief, painful sensation to his wrist. A 15-year-old girl who had compulsively pulled her hair for years, to the point of creating bald spots, also used contingent self-delivered snaps of a rubber band on her wrist to stop her habit (Mastellone, 1974). Another woman stopped her hair pulling by performing 15 sit-ups each time she pulled her hair or had the urge to do so (Mac- Neil & Thomas, 1976). Powell and Azrin (1968) designed a special cigarette case that delivered a one-second electric shock when opened. The controlling responses by a per- son using such a device as part of a self-management program is carrying the case and smoking only cigarettes that he has personally removed from it.

2 1 2

609

Self-Management

100

BL1 SM1 BL2 Self-Management 2 SM3

100

C la

ss P

er io

d A

C la

ss P

er io

d B

P er

ce n

ta ge

o f

In te

rv al

s of

P ro

b le

m B

eh av

io r

5 10 15 20 25 30 35 40 45 50 56 60 63 68 73

Sessions

Figure 7 Problem behaviors by a 9-year-old boy during 10-minute probes across two class periods during baseline and self-management conditions. From “Self-Monitoring and Self-Recruited Praise: Effects on Problem Behavior, Academic Engagement, and Work Completion in a Typical Classroom” by A. W. Todd, R. H. Horner, and G. Sugai, 1999, Journal of Positive Behavior Interventions, 1, p. 71. Copyright 1999 by Pro-Ed, Inc. Reprinted by permission.

Self-administering a positive practice overcorrection procedure also qualifies as an example of self-administered positive punishment. For example, a teenage girl who frequently said don’t instead of doesn’t with the third person singular (e.g., “She don’t like that one”) used this form of self-administered positive punishment to decrease the frequency of her speech error (Heward, Dardig, & Rossett, 1979). Each time the girl observed herself say- ing don’t when she should have said doesn’t, she repeated the complete sentence she had just spoken 10 times in a row using correct grammar. She wore a wrist counter that reminded her to listen to her speech and to keep track of the number of times she employed the positive practice procedure.

Self-Management Analogs of Negative Punishment

Self-administered analogs to negative punishment con- sist of arranging the loss of reinforcers (response cost) or denying oneself access to reinforcement for a specified period of time (time-out) contingent on the occurrence of the target behavior. Response cost and time-out con- tingencies are widely used self-management strategies. The most commonly applied self-administered response cost procedure is paying a small fine each time the target behavior occurs. In one study, smokers reduced their rate of smoking by tearing up a dollar bill each time they lit up (Axelrod, Hall, Weis, & Rohrer, 1971). Response cost

610

Self-Management

procedures have also been used effectively with elementary school children who have self-determined the number of tokens they should lose for inappropriate social behavior (Kaufman & O’Leary, 1972) or for poor academic work (Humphrey, Karoly, & Kirschenbaum, 1978).

James (1981) taught an 18-year-old man who had stuttered severely since the age of 6 to use a time-out from speaking procedure. Whenever he observed himself stut- tering, the young man immediately stopped talking for at least 2 seconds, after which he could begin speaking again. His frequency of dysfluencies decreased markedly. If talking is reinforcing, then this procedure might func- tion as time-out (e.g., not allowing oneself to engage in a preferred activity for a period of time).

Recommendations for Self-Administered Consequences

Persons designing and implementing self-administered consequences should consider the following recommendations.

Select Small, Easy-to-Deliver Consequences

Both rewards and penalties used in self-management pro- grams should be small and easily delivered. A common mistake in designing self-management programs is se- lecting consequences that are big and grand. Although a person may believe that the promise of a large reward (or threat of a severe aversive event) will motivate him to meet his self-determined performance criteria, large con- sequences often work against the program’s success. Self- selected rewards and punishing consequences should not be costly, elaborate, time-consuming, or too severe. If they are, the person may not be able (in the case of grand rewards) or willing (in the case of severe aversive events) to deliver them immediately and consistently.

In general, it is better to use small consequences that can be obtained immediately and frequently. This is par- ticularly important with consequences intended to func- tion as punishers, which—in order to be most effective— must be delivered immediately every time the behavior targeted for reduction is emitted.

Set a Meaningful But Easy-to-Meet Criterion for Reinforcement

When designing contingencies involving self-adminis- tered consequences, a person should guard against mak- ing the same two mistakes practitioners often make when implementing reinforcement contingencies with students and clients: (1) setting expectations so low that im- provement in the current level of performance is not nec- essary to obtain the self-administered reward, or

(2) making the initial performance criterion too high (the more common mistake), thereby effectively program- ming an extinction contingency that may cause the per- son to give up on self-management altogether. Keys to the effectiveness of any reinforcement-based interven- tion are setting an initial criterion that ensures that the person’s behavior makes early contact with reinforce- ment and that continued reinforcement requires im- provements over baseline levels. The criterion-setting formulae (Heward, 1980) provide guidelines for making those determinations.

Eliminate “Bootleg Reinforcement”

Bootleg reinforcement—access to the specified reward or to other equally reinforcing items or events without meeting the response requirements of the contingency— is a common downfall of self-management projects. A person who obtains a comfortable supply of smuggled rewards is less likely to work hard to earn response- contingent rewards.

Bootleg reinforcement is common when people use everyday preferred activities and treats as rewards in self-management programs. Although everyday expec- tations and enjoyments are easy to deliver, a person may find it difficult to withhold things he is used to enjoy- ing on a regular basis. For example, a man who is used to unwinding at the end of the day by watching Baseball Tonight with a beer and some peanuts may not be con- sistent in making those daily treats contingent on meet- ing the requirements of his self-management program.

One method for combating this form of bootleg re- inforcement is to no longer make access to the activities or items the person routinely enjoyed prior to the self- management program contingent on meeting any per- formance criteria and provide alternative items or activities that are a cut above the usual. For example, each time the man in the previous scenario meets his perfor- mance criteria, he could replace his everyday brand of beer with his choice of a specialty beer from a collection reserved in the back of the refrigerator.

If Necessary, Put Someone Else in Control of Delivering Consequences

Most ineffective self-management programs fail, not be- cause the controlling behavior is ineffective in controlling the controlled behavior, but because the contingencies that control the controlling behavior are not sufficiently strong. In other words, the person does not emit the controlling behavior consistently enough for its effects to be realized. What keeps a person from rationalizing that she did most of the behaviors she was supposed to and obtaining a

611

Self-Management

self-determined reward anyway? What keeps a person from failing to deliver a self-determined aversive consequence? Too often, the answer to both questions is, nothing.

A person who really wants to change her behavior but has difficulty following through with delivering self- determined consequences should enlist another person to serve as performance manager. A self-manager can en- sure that her self-designed consequences will be admin- istered faithfully by creating a contingency in which the consequence for failing to meet the performance criterion is aversive to the her but a reinforcer for the person she has asked to implement the consequence. And, if the first per- son put in charge of the contingency fails to implement it as planned, the self-manager should find another person who will do the job. Malott and Harrison (2002) wrote,

Christie wanted to walk on her dust-gathering treadmill 20 minutes a day, 6 days a week. She tried her husband as her performance contractor, but he wasn’t tough enough; he always felt sorry for her. So she fired him and hired her son; she made his bed every time she failed to do her 20 minutes, and he showed no mercy. (p. 18-7)

Kanfer (1976) called this kind of self-management decisional self-control: The person makes the initial de- cision to alter her behavior and plans how that will be ac- complished, but then turns over the procedure to a second party in order to avoid the possibility of not emit- ting the controlling response. Kanfer made a distinction between decisional self-control and protracted self- control, in which a person consistently engages in self- deprivation in order to effect the desired behavior change. Bellack and Hersen (1977) stated that decisional self-control “is generally considered to be less desirable than protracted self-control, as it does not provide the individual with an enduring skill or resource” (p. 111).

We disagree that a self-management program that in- volves the help of other people is less desirable than one in which the self-manager does everything. First, a self- management program in which the contingencies are turned over to someone else may be more effective than trying to “go it alone” because that other person is more consistent in applying the consequences. In addition, as a result of ex- periencing a successful self-management program in which he has chosen the target behavior, set performance criteria, determined a self-monitoring/self-evaluation system, and arranged to have someone else administer self-designed consequences, a person has acquired a considerable reper- toire of self-management skills for future use.

Keep It Simple

A person should not create elaborate self-management contingencies if they are not needed. The same general rule that applies to behavior change programs designed on behalf of others—the least complicated and intrusive

yet effective intervention should be employed—also ap- plies to self-management programs. With respect to the use of self-administered consequences, Bellack and Schwartz (1976) warned that,

Adding complicated procedures where they are not re- quired is more likely to have negative than positive effects. Second, our experience indicates that many individuals find explicit self-reinforcement procedures to be tedious, childish, and “gimmicky.” (p. 137)

There is no need for self-reinforcement procedures to be complicated. And in our experience, more people find creating and implementing their own performance management contingencies fun than find them tedious or childish.

Other Self-Management Tactics Other self-management strategies have been the sub- ject of behavior analysis research but are not so easily classified according to the four-term contingency. They include self-instruction, habit reversal, self-directed systematic desensitization, and massed practice.

Self-Instruction

People talk to themselves all the time, offering encour- agement (e.g., “You can do this; you’ve done it before”), congratulations (e.g., “Great shot, Daryl! You just crushed that 5-iron!”), and admonishment (e.g., “Don’t say that anymore; you hurt her feelings”) for their behavior, as well as specific instructions (e.g., “Pull the bottom rope through the middle”). Such self-statements can function as controlling responses—verbal mediators—that affect the occurrence of other behaviors. Self-instruction con- sists of self-generated verbal responses, covert or overt, that function as response prompts for a desired behavior. As a self-management tactic, self-instructions are often used to guide a person through a behavior chain or se- quence of tasks.

Bornstein and Quevillon (1976) conducted a study frequently cited as evidence of the positive and lasting effects of self-instruction. They taught three hyperactive preschool boys a series of four types of self-instructions designed to keep them on task with classroom activities:

1. Questions about the assigned task (e.g., “What does the teacher want me to do?”)

2. Answers to the self-directed questions (e.g., “I’m supposed to copy that picture”)

3. Verbalizations to guide the child through the task at hand (e.g., “OK, first I draw a line through here . . .”)

4. Self-reinforcement (e.g., “I really did that one well”)

612

Self-Management

During a 2-hour session the children were taught to use self-instructions using a sequence of training steps originally developed by Meichenbaum and Good- man (1971):

1. The experimenter modeled the task while talking aloud to himself.

2. The child performed the task while the experi- menter provided the verbal instructions.

3. The child performed the task while talking aloud to himself with the experimenter whispering the in- structions softly.

4. The child performed the task while whispering softly to himself with the experimenter moving his lips but making no sound.

5. The child performed the task while making lip movements but no sound.

6. The child performed the task while guiding his performance with covert instructions. (adapted from p. 117)

During the training session a variety of classroom tasks were used, ranging from simple motor tasks such as copying lines and figures to more complex tasks such as block design and grouping tasks. The children showed a marked increase in on-task behavior immediately after receiving self-instruction training, and their improved be- havior was maintained over a considerable period of time. The authors suggested that the generalization obtained from the training setting to the classroom could have been the result of telling the children during training to imag- ine that they were working with their teacher, not the ex- perimenter. The authors hypothesized that a behavioral trapping phenomenon (Baer & Wolf, 1970) may have been responsible for the maintenance of on-task behav- ior; that is, the self-instructions may have initially pro- duced better behavior, which in turn produced teacher attention that maintained the on-task behavior.

Although some studies evaluating self-instruction failed to reproduce the impressive results obtained by Bornstein and Quevillon (e.g. Billings & Wasik, 1985; Friedling & O’Leary, 1979), other studies have produced generally encouraging results (Barkley, Copeland, & Sivage, 1980; Burgio, Whitman, & Johnson, 1980; Hughes 1992; Kosiewicz, Hallahan, Lloyd, & Graves, 1982; Peters & Davies, 1981; Robin, Armel, & O’Leary, 1975). Self-instruction training has increased the fre- quency of high school students’ initiating conversations with familiar and unfamiliar peers (Hughes, Harmer, Kil- lian, & Niarhos, 1995; see Figure 5).

Employees with disabilities have learned to self- manage their work performance by providing their own verbal prompts and self-instructions (Hughes, 1997). For

example, Salend, Ellis, and Reynolds (1989) used a self- instruction strategy to teach four adults with severe men- tal retardation to “talk while you work.” Productivity increased dramatically and error rates decreased when the women verbalized to themselves, “Comb up, comb down, comb in bag, bag in box” while packaging combs in plastic bags. Hughes and Rusch (1989) taught two sup- ported employees working at a janitorial supply company how to solve problems by using a self-instruction proce- dure consisting of four statements:

1. Statement of the problem (e.g., “Tape empty”)

2. Statement of the response needed to solve the problem (e.g., “Need more tape”)

3. Self-report (e.g., “Fixed it”)

4. Self-reinforcement (e.g., “Good”)

O’Leary and Dubey (1979) summarized their review of self-instruction training by suggesting four factors that appear to influence its effectiveness with children.

Self-instructions appear to be effective self-controlling procedures if the children actually implement the in- structional procedure, if the children use them [the self- instructions] to influence behavior at which they are skilled, if children have been reinforced for adhering to their self-instructions in the past, and if the focus of the instructions is the behavior most subject to conse- quences. (p. 451)

Habit Reversal

In his initial discussion of self-control, Skinner (1953) included “doing something else” as a self-management tactic. In an interesting application of “doing something else,” Robin, Schneider, and Dolnick (1976) taught 11 primary-aged children with emotional and behavioral disorders to control their aggressive behaviors by using the turtle technique: The children pulled their arms and legs close to their bodies, put their heads down on their desks, relaxed their muscles, and imagined they were turtles. The children were taught to use the turtle re- sponse whenever they believed an aggressive exchange with someone else was about to take place, when they were angry with themselves and felt they were about to have a tantrum, or when the teacher or a classmate called out, “Turtle!”

Azrin and Nunn (1973) developed an intervention they called habit reversal, in which clients are taught to self-monitor their nervous habits and interrupt the be- havior chain as early as possible by engaging in behavior incompatible with the problem behavior (i.e., doing some- thing else). For example, when a nail biter observes her- self beginning to bite her fingernails, she might squeeze her hand into a tight fist for 2 or 3 minutes (Azrin, Nunn,

613

Self-Management

Figure 8 Sequence of imaginary scenes concerning fear of cats that could be used for systematic self-desensitization.

Directions 1.You’re sitting in a comfortable chair in the safety of your home watching TV. 2.You’re watching a commercial for cat food—no cat is visible. 3. The commercial continues and a cat is now eating the food. 4. A man is now petting the cat. 5. A man is holding the cat and fondling it. 6. A woman is holding the cat, and the cat is licking her hands and face. 7.You’re looking out the window of your home and you see a cat on the lawn across the

street. 8.You’re sitting in front of your house, and you see a cat walk by on the sidewalk across

the street. 9.You’re sitting in your yard, and you see a cat walk by on your sidewalk.

10. A cat walks within 15 feet of you. 11. A friend of yours picks the cat up and plays with it. 12.Your friend is 10 feet away from you, and the cat is licking his face. 13.Your friend comes within 5 feet of you while he’s holding the cat. 14.Your friend stands 2 feet away and plays with the cat. 15.Your friend asks if you’d like to pet the cat. 16.Your friend reaches out and offers you the cat. 17. He puts the cat on the ground, and it walks over to you. 18. The cat rubs up against your leg. 19. The cat walks between your legs purring. 20.You reach down and touch the cat. 21.You pet the cat. 22.You pick up the cat and pet it. (p. 71)

From “Self-Directed Systematic Desensitization: A Guide for the Student, Client, and Therapist” by W.W.Wen- rich, H. H. Dawley, and D. A. General, 1976, p. 71. Copyright 1976 by Behaviordelia, Kalamazoo, MI. Used by permission.

& Frantz, 1980) or sit on her hands (Long, Miltenberger, Ellingson, & Ott, 1999). As a clinical intervention, habit reversal is typically implemented as a multiple-component treatment package that includes self-awareness training in- volving response detection and procedures for identifying events that precede and trigger the response, competing response training, motivation techniques including self- administered consequences, social support systems, and procedures for promoting the generalization and mainte- nance of treatment gains (Long et al., 1999). Habit rever- sal has proven to be a highly effective self-management tactic for a wide variety of problem behaviors. For a re- view of habit reversal procedures and research, see Mil- tenberger, Fuqua, and Woods (1998).

Self-Directed Systematic Desensitization

Systematic desensitization is a widely used behavior ther- apy treatment for anxieties, fears, and phobias that fea- tures the self-management strategy of engaging in an alternative behavior (i.e., doing something else). Origi- nally developed by Wolpe (1958, 1973), systematic desensitization involves substituting one behavior, gen-

erally muscle relaxation, for the unwanted behavior—the fear and anxiety. The client develops a hierarchy of situ- ations from the least to the most fearful and then learns to relax while imagining these anxiety-producing situa- tions, first the least fearful situation, then the next fear- ful one, and so on. Figure 8 shows a anxiety-stimulus hierarchy that a person might develop in attempting to control a fear of cats. When a person is able to go com- pletely through his hierarchy, imagining each scene in detail while maintaining deep relaxation and feeling no anxiety, he begins to expose himself gradually to real- life (in vivo) situations.

Detailed procedures for achieving deep muscle re- laxation, constructing and validating a hierarchy of anx- iety- or fear-producing situations, and implementing a self-directed systematic desensitization program can be found in Martin and Pear (2003) and Wenrich, Dawley, and General (1976).

Massed Practice

Forcing oneself to perform an undesired behavior again and again, a technique called massed practice, will sometimes decrease the future frequency of the behavior.

614

Self-Management

Wolff (1977) reported an interesting case of this form of treatment by a 20-year-old woman who engaged in a compulsive, ritualized routine of 13 specific security checks every time she entered her apartment (e.g., look- ing under beds, checking the closets, looking in the kitchen). She began her program by deliberately going through the 13 steps in an exact order and then repeating the complete ritual four more times. After doing this for 1 week, she permitted herself to check the apartment if she wanted to but made herself go through the entire rou- tine five times whenever she did any checking at all. She soon quit performing her compulsive checking behavior.

Suggestions for Conducting an Effective Self-Management Program Incorporating the suggestions that follow into the design and implementation of self-management programs should increase the likelihood of success. Although none of these guidelines has been rigorously examined through exper- imental analyses—research on self-management has a long way to go—each suggestion is consistent with pro- cedures proven effective in other areas of applied behav- ior analysis and with “best practices” commonly reported in the self-management literature (e.g., Agran, 1997; Mal- ott & Harrison, 2002; Martin & Pear, 2003; Watson & Tharp, 2007).

1. Specify a goal and define the behavior to be changed.

2. Begin self-monitoring the behavior.

3. Contrive contingencies that will compete with nat- ural contingencies.

4. Go public with your commitment to change your behavior.

5. Get a self-management partner.

6. Continually evaluate your self-management pro- gram and redesign it as necessary.

Specify a Goal and Define the Target Behavior

A self-management program begins with identifying a personal goal or objective and the specific behavior changes necessary to accomplish that goal or objective. A person can use most of the questions and issues that practitioners should consider when selecting target be- haviors for students or clients to assess the social signif- icance and prioritize the importance of a list of self- determined target behavior changes.

Begin Self-Monitoring the Behavior

A person should begin self-monitoring as soon as he has defined the target behavior. Self-monitoring prior to im- plementing any other form of intervention yields the same benefits as does taking baseline data:

1. Self-monitoring makes a person observant of events occurring before and after the target behav- ior; information about antecedent-behavior- consequent correlations that may be helpful in designing an effective intervention.

2. Self-monitored baseline data provide valuable guidance in determining initial performance crite- ria for self-administered consequences.

3. Self-monitored baseline data provide an objective basis for evaluating the effects of any subsequent interventions.

Another reason to begin self-monitoring as soon as pos- sible without employing additional self-management tac- tics is that the desired improvement in behavior may be achieved by self-monitoring alone.

Create Contrived Contingencies That Will Compete with Ineffective Natural Contingencies

When self-monitoring alone does not result in the desired behavior changes, the next step is designing a contrived contingency to compete with the ineffective natural con- tingencies. A person who implements a contingency that provides immediate, definite consequences for each oc- currence (or perhaps, nonoccurrence) of the target be- havior greatly increases the probability of obtaining a previously elusive self-management goal. For example, a smoker who self-records and reports each cigarette smoked to his self-management partner, who provides contingent praise, rewards, and penalties, has arranged immediate, frequent, and more effective consequences for reducing his smoking behavior than those provided by the natural contingency: the threat of lung cancer and em- physema in the future, neither of which becomes appre- ciably more immediate or likely from one cigarette to the next.

Go Public

The effectiveness of a self-management effort may be enhanced by publicly sharing the intentions of the pro- gram. When a person shares a goal or makes a predic- tion to others about her future behavior, she has arranged potential consequences—praise or condemnation—for

615

Self-Management

her success or failure in meeting that goal. A person should state in specific terms what she intends to do and her deadline for completing it. Taking the idea of public commitment a step further, consider the powerful poten- tial of public posting (e.g., an untenured junior faculty member could post a chart of his writing where his de- partment chair or dean could see and comment on it).

Malott (1981) called this the “public spotlight prin- ciple of self-management.”

A public statement of goals improves performance. But just how do the social contingencies engaged by public commitment produce change? They increase the reward- ing value of success and the aversive value of failure, I presume. But those outcomes are probably too delayed to directly reinforce problem solving. Instead they must be part of the rules the student self-states at crucial times: “If I don’t make my goal, I’ll look like a fool; but if I do make it, I’ll look pretty cool.” Such rules then function as cues for immediate self-reinforcement of on- task behavior and self-punishment of off-task behavior. (Volume II, No. 18, p. 5)

Skinner (1953) also theorized about the principles of behavior that operate when a person shares his self- management goal with important others.

By making it in the presence of people who supply aver- sive stimulation when a prediction is not fulfilled, we arrange consequences which are likely to strengthen the behavior resolved upon. Only by behaving as predicted can we escape the aversive consequences of breaking our resolution. (p. 237)

Get a Self-Management Partner

Setting up a self-management exchange is a good way to involve another person whose differential feedback about how a self-management project is going can be effective as a behavioral consequence. Two people, each of whom has a long-range goal or a regular series of tasks to do, can agree to talk to each other on a daily or weekly basis, as determined by the target behaviors and by each person’s progress. They can share the data from their self- monitoring and exchange verbal praise or admonishments and perhaps even more tangible consequences contingent on performance. Malott (1981) reported a successful self- management exchange in which he and a colleague paid each other $1 every time either person failed to complete any one of a series of self-determined daily exercise, housekeeping, and writing tasks. Each morning they spoke on the telephone, reporting their performance dur- ing the previous 24 hours.

To help themselves and each other study for exams and complete research and writing tasks, one group of doc- toral students created a self-management group they called

the Dissertation Club (Ferreri et al., 2006). At weekly meet- ings each group member shared data on the “scholarly be- haviors” she had targeted (e.g., graphs of words written each day or number of hours studied). Group members re- ceived social support from one another in the form of en- couragement to continue working hard and praise for accomplishments, they served as behavioral consultants to one another on the deign of self-management interven- tions, and sometimes they administered rewards and fines to one another. All six members of the group wrote and successfully defended their dissertations within the time- lines they had set.

Continually Evaluate and Redesign Program as Needed

Your self-management project may not work the first time you try it. And it will certainly fall apart from time to time, so be prepared with some scotch tape and bub- ble gum to put it back together again.

— Malott and Harrison (2002, p. 18-7)

The development and evaluation of most self- management programs reflects a pragmatic, data-based approach to personal problem solving more than it does rigorous research with an emphasis on experimental analysis and control. Like the researcher, however, the self-manager should be guided by the data. If the data show the program is not working satisfactorily, the in- tervention should be redesigned.

An A-B design is sufficient to evaluate the effects of most self-management projects. People who have learned to define, observe, record, and graph their own behavior can easily be taught how to evaluate their self- management efforts with self-experiments. A simple A-B design provides a straightforward accounting of the results in a before-and-after fashion that is usually sufficient for self-evaluation. Experimentally deter- mining functional relations between self-managed in- terventions and their effects usually takes a back seat to the pragmatic goal of changing one’s behavior. How- ever, the changing criterion design lends itself nicely, not only to the stepwise increments in performance that are often a part of improving personal performance, but also to a clearer demonstration and understanding of the relation between the intervention and changes in the tar- get behavior.

In addition to data-based evaluation, people should evaluate their self-management projects in terms of so- cial validity dimensions (Wolf, 1978). A practitioner teaching self-management to others can aid his students’ or clients’ social validity assessments of their self- management efforts by providing checklists of questions

616

Self-Management

covering topics such as how practical their intervention was, whether they felt their self-management program affected their behavior in any unmeasured ways, and whether they enjoyed the project.

One of the most important aspects of social valid- ity for any behavior change program is the extent to which the results—measured changes in the target be- havior—actually made a difference in the lives of the participants. One approach to assessing the social va- lidity of the results of a self-management program is to collect data on what Malott and Harrison (2002) called benefit measures. For example, a person might measure:

• Number of pounds lost as a benefit measure of eat- ing less or exercising more

• Improved lung capacity (peak airflow volume measured as cubic centimeters per second on an inexpensive device available at many pharmacies) as an outcome of reducing the number of ciga- rettes smoked

• Decrease in time it takes to run a mile as a benefit of number of miles run each day

• Lower resting heart rate and faster recovery time as a benefit of aerobic exercise

• Higher scores on foreign language practice exams as an outcome of studying

In addition to assessing the social validity of self- management interventions, the positive results of bene- fit measures can serve as consequences that reward and strengthen continued adherence to self-management.

An account of a self-management weight loss pro- gram that incorporated many of the tactics and sugges-

tions described in this chapter is presented in Box 2, “Take a Load Off Me.”

Behavior Changes Behavior Referring to a book on self-management he wrote for a popular audience, Epstein (1997) wrote,

A young man whose life is in disarray (he smokes, drinks, overeats, loses things, procrastinates, and so on) seeks advice from his parents, teachers, and friends, but no one can help. Then the young man re- members his Uncle Fred (modeled shamelessly after Fred Skinner), whose life always seemed to be in per- fect harmony. In a series of visits, Uncle Fred reveals to him the three “secrets” of self-management, all Ms: Modify your environment, monitor your behaviors, and make commitments. Fred also reveals and explains the “self-management principle”: Behavior changes behavior. After each visit, the young man tries out a new technique, and his life is changed radically for the better. In one scene, he sees a classroom of remarkably creative and insightful children who have been trained in self-management techniques in a public school. It is fiction, of course, but the technology is well estab- lished and the possibilities are well within reach. (p. 563, emphasis in original)

Building upon Skinner’s (1953) conceptual analysis of self-control more than 50 years ago, behavior analysts have developed numerous self-management tactics and methods for teaching learners with diverse abilities how to apply them. Underlying all of these efforts and findings is the simple, yet profound principle that behavior changes behavior.

617

Joe was a 63-year-old man whose doctor had recently in- formed him that the 195 pounds on his 5-foot 11-inch frame must be reduced to 175 pounds or he was very likely to experience serious health problems. Although Joe obtained regular physical exercise by keeping up the lawn and garden, cutting and hauling firewood, feeding the rabbits, and doing myriad other chores that come with a farmhouse, his prodigious appetite—long con- sidered one of the “best in the county”—had caught up with him. His doctor’s warning and the recent death of a high school classmate had scared Joe enough to sit down with his son to plan a self-management weight- loss program.

Joe’s program included antecedent-based tactics, self-monitoring, self-evaluation, contingency con- tracting, self-selected and self-administered conse- quences, and contingency management by significant others.

Goal To reduce current weight of 195 pounds to 175 pounds at a rate of 1 pound per week.

Behavior Change Needed to Achieve Goal Reduce eating to a maximum of 2,100 calories per day.

Rules and Procedures

1. Mount bathroom scale each morning before dress- ing or eating and chart weight on “Joe’s Weight” graph taped to bathroom mirror.

2. Carry notepad, pencil, and calorie counter in pocket throughout the day and record the type and amount of all food and liquid (other than water) immediately after consumption.

3. Before going to bed each night, add total calories consumed during the day and chart that number on “Joe’s Eating” graph taped to bathroom mirror.

4. Do not waiver from Steps 1–3, regardless of weight loss or lack of weight loss.

Immediate Contingencies/Consequences

• Calorie counter, notepad, and pencil in pocket pro- vide continuously available response prompts.

• Recording all food and drink consumed provides immediate consequence.

• When you have not eaten any food you thought about eating, put a star on your notepad and give self-praise (“Way to go, Joe!”).

Daily Contingencies/ Consequences

• If the calories consumed during the day do not ex- ceed the 2,100 criterion, put 50 cents in “Joe’s Garden Jar.”

• If total calories exceed the criterion, remove $1.00 from “Joe’s Garden Jar.”

• For each 3 days in a row that you meet the calorie criterion, add a bonus of 50 cents to “Joe’s Garden Jar.”

• Have Helen initial your contract each day that you meet the calorie criterion.

Weekly Contingencies/ Consequences

• Each Sunday night write the calories consumed each day for the previous week on one of the pre- dated, addressed, and stamped postcards. Helen will verify your self-report by initialing the post- card and mailing it to Bill and Jill on Monday.

• Each Monday, if total calories consumed met the criterion for at least 6 of the past 7 days, obtain an item or activity from “Joe’s Reward Menu.”

Intermediate Contingencies/Consequences

• Eat within the daily calorie limit often enough and there will be sufficient money in “Joe’s Garden Jar” to buy seeds and plants for the spring veg- etable garden.

• If weight during May visit to Ohio represents a loss of at least 1 pound per week, be a guest of honor at a restaurant of your choice.

Long-Term Consequences

• Feel better.

• Look better.

• Be healthier.

• Live longer.

Results Joe often struggled with the rules and procedures, but he stuck with his self-management program and lost 22 pounds (achieving a weight of 173) in 16 weeks. As this report is being written, 26 years have passed since Joe’s adventure in self-management. Joe can still hold his own at the dinner table, but he has retained his weight loss and enjoys gardening, singing in a barbershop choir, and listening to the Cubs on the radio.

Box 2 Take a Load Off Me:

A Self-Managed Weight-Loss Program

618

Self-Management

Summary The “Self” As Controller of Behavior

1. We tend to assign causal status to events that immediately precede behavior, and when causal variables are not read- ily apparent in the immediate, surrounding environment, the tendency to turn to internal causes is strong.

2. Hypothetical constructs such as willpower and drive are explanatory fictions that bring us no closer to understand- ing the behaviors they claim to explain and lead to circu- lar reasoning.

3. Skinner (1953) conceptualized self-control as a two- response phenomenon: The controlling response affects variables in such a way as to change the probability of the other, the controlled response.

4. We define self-management as the personal application of behavior change tactics that produces a desired change in behavior.

5. Self-management is a relative concept. A behavior change program may entail a small degree of self-management or be totally conceived, designed, and implemented by the person.

6. Although self-control and self-management appear inter- changeably in the behavioral literature, we recommend that self-management be used in reference to a person acting in some way in order to change his subsequent behavior.

• Self-control implies that the ultimate control of behav- ior lies within the person, but the causal factors for “self- control” are to be found in a person’s experiences with his environment.

• Self-control “seems to suggest controlling a [separate] self inside or [that there is] a self inside controlling ex- ternal behavior” (Baum, 1994, p. 157).

• Self-control is also used to refer to a person’s ability to “delay gratification” by responding to achieve a delayed, but larger or higher quality reward instead of acting to obtain an immediate, less valuable reward.

Applications, Advantages, and Benefits of Self-Management

7. Four uses of self-management are to

• live a more effective and efficient daily life,

• break bad habits and acquiring good ones,

• accomplish difficult tasks, and

• achieve personal lifestyle goals.

8. Advantages and benefits of learning and teaching self- management skills include the following:

• Self-management can influence behaviors not accessi- ble to external change agents.

• External change agents often miss important instances of behavior.

• Self-management can promote the generalization and maintenance of behavior change.

• A small repertoire of self-management skills can control many behaviors.

• People with diverse abilities can learn self-manage- ment skills.

• Some people perform better under self-selected tasks and performance criteria.

• People with good self-management skills contribute to more efficient and effective group environments.

• Teaching students self-management skills provides meaningful practice for other areas of the curriculum.

• Self-management is an ultimate goal of education.

• Self-management benefits society.

• Self-management helps a person feel free.

• Self-management feels good.

Antecedent-Based Self-Management Tactics

9. Antecedent-based self-management tactics feature the ma- nipulation of events or stimuli antecedent to the target (controlled) behavior, such as the following:

• Manipulating motivating operations to make a desired (or undesired) behavior more (or less) likely

• Providing response prompts

• Performing the initial steps of a behavior chain to ensure being confronted later with a discriminative stimulus that will evoke the desired behavior

• Removing the materials required for an undesired behavior

• Limiting an undesired behavior to restricted stimulus conditions

• Dedicating a specific environment for a desired behavior

Self-Monitoring

10. Self-monitoring is a procedure whereby a person observes and responds to, usually by recording, the behavior he is trying to change.

11. Originally developed as a method of clinical assessment for collecting data on behaviors that only the client could observe, self-monitoring evolved into the most widely used and studied self-management strategy because it often re- sults in desired behavior change.

12. Self-monitoring is often combined with goal setting and self-evaluation. A person using self-evaluation compares her performance with a predetermined goal or standard.

13. Self-monitoring is often part of an intervention that in- cludes reinforcement for meeting either self- or teacher- selected goals.

619

14. It is difficult to determine exactly how self-monitoring works because the procedure necessarily includes, and is therefore confounded by, private events (covert verbal be- havior); it often includes either explicit or implicit con- tingencies of reinforcement.

15. Children can be taught to self-monitor and self-record their behavior accurately by means of a faded matching tech- nique, in which the child is rewarded initially for produc- ing data that match the teacher’s or parent’s data. Over time the child is required to match the adult’s record less often, eventually monitoring the behavior independently.

16. Accuracy of self-monitoring is neither necessary nor suf- ficient to achieve improvement in the behavior being monitored.

17. Suggested guidelines for self-monitoring are as follows:

• Provide materials that make self-monitoring easy.

• Provide supplementary cues or prompts.

• Self-monitor the most important dimension of the target behavior.

• Self-monitor early and often, but do not interrupt the flow of a desired behavior targeted for increase.

• Reinforce accurate self-monitoring.

Self-Administered Consequences

18. As a technical term, self-reinforcement (as also, self- punishment) is a misnomer. Although behavior can be changed by self-administered consequences, the variables influencing the controlling response make such self- management tactics more than a straightforward applica- tion of operant reinforcement.

19. Self-administered contingencies analogous to positive and negative reinforcement and positive and negative punish- ment can be incorporated into self-management programs.

20. When designing self-management programs involving self-administered consequences, a person should:

• Select small, easy-to-deliver consequences.

• Set a meaningful but easy-to-meet criterion for rein- forcement.

• Eliminate “bootleg reinforcement.”

• If necessary, put someone else in control of delivery consequences.

• Use the least complicated and intrusive contingencies that will be effective.

Other Self-Management Tactics

21. Self-instruction (talking to oneself ) can function as con- trolling responses (verbal mediators) that affect the oc- currence of other behaviors.

22. Habit reversal is a multiple-component treatment package in which clients are taught to self-monitor their unwanted habits and interrupt the behavior chain as early as possi- ble by engaging in a behavior incompatible with the prob- lem behavior.

23. Systematic desensitization is a behavior therapy treatment for anxieties, fears, and phobias that involves substituting one behavior, generally muscle relaxation, for the un- wanted behavior—the fear and anxiety. Self-directed sys- tematic desensitization involves developing a hierarchy of situations from the least to the most fearful and then learn- ing to relax while imagining these anxiety-producing sit- uations, first the least fearful situation, then the next fearful one, and so on.

24. Massed practice, forcing oneself to perform an undesired behavior again and again, can decrease the future fre- quency of the behavior.

Suggestions for Conducting an Effective Self-Management Program

25. Following are six steps in designing and implementing a self-management program:

Step 1: Specify a goal and define the behavior to be changed.

Step 2: Begin self-monitoring the behavior.

Step 3: Create contingencies that will compete with nat- ural contingencies.

Step 4: Go public with the commitment to change behavior.

Step 5: Get a self-management partner.

Step 6: Continually evaluate and redesign the program as needed.

Behavior Changes Behavior

26. The most fundamental principle of self-management is that behavior changes behavior.

Self-Management

620

621

Generalization and Maintenance of Behavior Change

Key Terms

behavior trap contrived contingency contrived mediating stimulus general case analysis generalization generalization across subjects generalization probe

generalization setting indiscriminable contingency instructional setting lag reinforcement schedule multiple exemplar training naturally existing contingency

programming common stimuli response generalization response maintenance setting/situation generalization teaching sufficient examples teaching loosely

Content Area 3: Principles, Processes, and Concepts

3-12 Define and provide examples of generalization and discrimination.

9-28 Use behavior change procedures to promote stimulus and response generalization.

9-29 Use behavior change procedures to promote maintenance.

622

Sherry’s teacher implemented an intervention that helped Sherry to complete each part of multiple-part, in- school assignments before submitting them and begin- ning another activity. Now, three weeks after the program ended, most of the work Sherry submits as “finished” is incomplete and her stick-with-a-task-until- it’s-finished behavior is as poor as it was before the in- tervention began.

Ricardo has just begun his first competitive job working as a copy machine operator in a downtown business of- fice. In spite of his long history of distractibility and poor endurance, Ricardo had learned to work indepen- dently for several hours at a time in the copy room at the vocational training center. His employer, however, is complaining that Ricardo frequently stops working after a few minutes to seek attention from others. Ricardo may soon lose his job.

Brian is a 10-year-old boy diagnosed with autism. In an effort to meet an objective on his individualized educa- tion program that targets functional language and com- munication skills, Brian’s teacher taught him to say, “Hello, how are you?” as a greeting. Now, whenever Brian meets anyone, he invariably responds with, “Hello, how are you?” Brian’s parents are concerned that their son’s language seems stilted and parrot-like.

Each of these three situations illustrates a com- mon type of teaching failure insofar as the most socially significant behavior changes are those

that last over time, are used by the learner in all relevant settings and situations, and are accompanied by changes in other relevant responses. The student who learns to count money and make change in the classroom today must be able to count and make change at the conve- nience store tomorrow and at the supermarket next month. The beginning writer who has been taught to write a few good sentences in school must be able to write many more meaningful sentences when writing notes or letters to family or friends. To perform below this stan- dard is more than just regrettable; it is a clear indication that the initial instruction was not entirely successful.

In the first scenario, the mere passage of time re- sulted in Sherry losing her ability to complete assign- ments. A change of scenery threw Ricardo off his game; the excellent work habits he had acquired at the voca- tional training center disappeared completely when he arrived at the community job site. Although Brian used his new greeting skill, its restricted form was not serving him well in the real world. In a very real sense, the in- struction they received failed all three of these people.

Applied behavior analysts face no more challeng- ing or important task than that of designing, imple- menting, and evaluating interventions that produce generalized outcomes. This chapter defines the major

types of generalized behavior change and describes the strategies and tactics researchers and practitioners use most often to promote them.

Generalized Behavior Change: Definitions and Key Concepts When Baer, Wolf, and Risley (1968) described the emerg- ing field of applied behavior analysis, they included gen- erality of behavior change as one of the discipline’s seven defining characteristics.

A behavior change may be said to have generality if it proves durable over time, if it appears in a wide variety of possible environments, or if it spreads to a wide vari- ety of related behaviors. (p. 96)

In their seminal review paper, “An Implicit Technol- ogy of Generalization,” Stokes and Baer (1977) also stressed those three facets of generalized behavior change—across time, settings, and behaviors—when they defined generalization as

the occurrence of relevant behavior under different, non- training conditions (i.e., across subjects, settings, peo- ple, behaviors, and/or time) without the scheduling of the same events in those conditions. Thus, generaliza- tion may be claimed when no extratraining manipula- tions are needed for extratraining changes; or may be claimed when some extra manipulations are necessary, but their cost is clearly less than that of the direct inter- vention. Generalization will not be claimed when simi- lar events are necessary for similar effects across conditions. (p. 350)

Stokes and Baer’s pragmatic orientation toward gen- eralized behavior change has proven useful for applied behavior analysis. They stated simply that if a trained behavior occurs at other times or in other places without it having to be retrained completely at those times or in those places, or if functionally related behaviors occur that were not taught directly, then generalized behavior change has occurred. The following sections provide def- initions and examples of the three basic forms of gener- alized behavior change: response maintenance, setting/ situation generalization, and response generalization. Box 1, “Perspectives on the Sometimes Confusing and Misleading Terminology of Generalization,” discusses the many and varied terms applied behavior analysts use to describe these outcomes.

Response Maintenance

Response maintenance refers to the extent to which a learner continues to perform the target behavior after a portion or all of the intervention responsible for the

Generalization and Maintenance of Behavior Change

623

*Response maintenance can be measured under extinction conditions, in which case the relative frequency of continued responding is de- scribed correctly in terms of resistance to extinction. However, using resistance to extinction to describe response maintenance in most applied situations is incorrect because reinforcement typically follows some oc- currences of the target behavior in the post-treatment environment.

Applied behavior analysts have used many terms to de- scribe behavior changes that appear as adjuncts or by- products of direct intervention. Unfortunately, the overlapping and multiple meanings of some terms can lead to confusion and misunderstanding. For example, maintenance, the most frequently used term for behavior changes that persist after an intervention has been with- drawn or terminated, is also the most common name for a condition in which treatment has been discontinued or partially withdrawn. Applied behavior analysts should distinguish between response maintenance as a measure of behavior (i.e., a dependent variable) and maintenance as the name for an environmental condition (i.e., an in- dependent variable). Other terms found in the behavior analysis literature for continued responding after pro- grammed contingencies are no longer in effect include durability, behavioral persistence, and (incorrectly) re- sistance to extinction.*

Terms used in the applied behavior analysis literature for behavior changes that occur in nontraining settings or stimulus conditions include stimulus generalization, set- ting generalization, transfer of training, or simply, gen- eralization. It is technically incorrect to use stimulus generalization to refer to the generalized behavior change achieved by many applied interventions. Stimulus gen- eralization refers to the phenomenon in which a response that has been reinforced in the presence of a given stim- ulus occurs with an increased frequency in the presence of different but similar stimuli under extinction condi- tions (Guttman & Kalish, 1956; see Chapter 17). Stimulus generalization is a technical term referring to a specific behavioral process, and its use should be restricted to those instances (Cuvo, 2003; Johnston, 1979).

Terms such as collateral or side effects, response variability, induction, and concomitant behavior change are often used to indicate the occurrence of behaviors that have not been trained directly. To further complicate matters, generalization is often used as a catchall term to refer to all three types of generalized behavior change.

Johnston (1979) discussed some problems caused by using generalization (the term for a specific behavioral process) to describe any desirable behavior change in a generalization setting.

This kind of usage is misleading in that it suggests that a single phenomenon is at work when actually a number of different phenomena need to be described, explained, and controlled. . . . Carefully designing procedures to opti- mize the contributions of stimulus and response general- ization would hardly exhaust our repertoire of tactics for getting the subject to behave in a desirable way in non- instructional settings. Our successes will be more frequent when we realize that maximizing behavioral influence in such settings requires careful consideration of all behav- ioral principles and processes. (pp. 1–2)

Inconsistent use of the “terminology of generaliza- tion” can lead researchers and practitioners to incorrect assumptions and conclusions regarding the principles and processes responsible for the presence or absence of generalized outcomes. Nevertheless, applied behavior analysts will probably continue to use generalization as a dual-purpose term, referring sometimes to types of be- havior change and sometimes to behavioral processes that can bring such changes about. Stokes and Baer (1977) clearly indicated their awareness of the differ- ences in definitions.

The notion of generalization developed here is an essen- tially pragmatic one; it does not closely follow the tradi- tional conceptualizations (Keller & Schoenfeld, 1950; Skinner, 1953). In many ways, this discussion will sidestep much of the controversy concerning terminology. (p. 350)

While discussing the use of naturally existing con- tingencies of reinforcement to maintain and extend pro- grammed behavior changes, Baer (1999) explained his preference for using the term generalization:

It is the best of the techniques described here and, inter- estingly, it does not deserve the textbook definition of “generalization.” It is a reinforcement technique, and the textbook definition of generalization refers to unreinforced behavior changes resulting from other directly reinforced behavior changes. . . . [But] we are dealing with the prag- matic use of the word generalization, not the textbook meaning. We reinforce each other for using the word prag- matically, and it serves us well enough so far, so we shall probably maintain this imprecise usage. (p. 30, emphasis in original)

In an effort to promote the precise use of the tech- nical terminology of behavior analysis and as a reminder that the phenomena of interest are usually products of multiple behavior principles and procedures, we use terms for generalized behavior change that focus on the type of behavior change rather than the principles or processes that bring it about.

Box 1 Perspectives on the Sometimes Confusing

and Misleading Terminology of Generalization

Generalization and Maintenance of Behavior Change

624

behavior’s initial appearance in the learner’s repertoire has been terminated. For example:

• Sayaka was having difficulty identifying the low- est common denominator (LCD) when adding and subtracting fractions. Her teacher had Sayaka write the steps for finding the LCD on an index card and told her to refer to the card when needed. Sayaka began using the LCD cue card, and the ac- curacy of her math assignments improved. After using the cue card for a week, Sayaka said she no longer needed it and returned it to her teacher. The next day Sayaka correctly computed the LCD for every problem on a quiz on adding and subtracting fractions.

• On Loraine’s first day on the job with a residential landscaping company, a coworker taught her how to use a long-handled tool to extract dandelions, root and all. Without further instruction, Loraine continues to use the tool correctly a month later.

• When he was in the seventh grade, one of Derek’s teachers taught him how to write down his assign- ments and keep materials for each class in sepa- rate folders. As a college sophomore, Derek continues to apply those organizational skills to his academic work.

These examples illustrate the relative nature of gen- eralized behavior change. Response maintenance was ev- ident in Sayaka’s performance on a math quiz one day after the cue card intervention ended and also in Derek’s continued use of the organizational skills he had learned years earlier. How long a newly learned behavior needs to be maintained depends on the importance of that be- havior in the person’s life. If covertly reciting a telephone number three times after hearing it enables a person to re- member the number long enough to dial it correctly when he locates a telephone a few minutes later, sufficient re- sponse maintenance has been achieved. Other behaviors, such as self-care and social skills, must be maintained in a person’s repertoire for a lifetime.

Setting/Situation Generalization

Setting/situation generalization occurs when a target be- havior is emitted in the presence of stimulus conditions other than those in which it was trained directly. We de- fine setting/situation generalization as the extent to which a learner emits the target behavior in a setting or stimulus situation that is different from the instructional setting. For example:

• While waiting for his new motorized wheelchair to arrive from the factory, Chaz used a computer sim- ulation program and a joystick to learn how to op-

erate his soon-to-arrive chair. When the new chair arrived, Chaz grabbed the joystick and immedi- ately began zipping up and down the hall and spin- ning perfectly executed donuts.

• Loraine had been taught to pull weeds from flower- beds and mulched areas. Although she had never been instructed to do so, Loraine has begun remov- ing dandelions and other large weeds from lawns as she crosses on her way to the flowerbeds.

• After Brandy’s teacher taught her to read 10 differ- ent C-V-C-E words (e.g., bike, cute, made), Brandy could read C-V-C-E words for which she had not received any instruction (e.g., cake, bite, mute).

A study by van den Pol and colleagues (1981) pro- vides an excellent example of setting/situation general- ization. They taught three young adults with multiple disabilities to eat independently in fast-food restaurants. All three students had previously eaten in restaurants but could not order or pay for a meal without assistance. The researchers began by constructing a task analysis of the steps required to order, pay for, and eat a meal appropri- ately in a fast-food restaurant. Instruction took place in the students’ classroom and consisted of role-playing each of the steps during simulated customer–cashier interac- tions and responding to questions about photographic slides showing customers at a fast-food restaurant per- forming the various steps in the sequence. The 22 steps in the task analysis were divided into four major compo- nents: locating, ordering, paying, and eating and exiting. After a student had mastered the steps in each compo- nent in the classroom, he was given “a randomly deter- mined number of bills equaling two to five dollars and instructed to go eat lunch” at a local restaurant (p. 64). Observers stationed inside the restaurant recorded each student’s performance of each step in the task analysis. The results of these generalization probes, which were also conducted before training (baseline) and after train- ing (follow-up) are shown in Figure 1. In addition to as- sessing the degree of generalization from the classroom, which was based on the specific McDonald’s restaurant used for most of the probes, the researchers conducted follow-up probes in a Burger King restaurant (also a mea- sure of maintenance).

This study is indicative of the pragmatic approach to assessing and promoting generalized behavior change used by most applied behavior analysts. The setting in which generalized responding is desired can contain one or more components of the intervention that was imple- mented in the instructional environment, but not all of the components. If the complete intervention program is required to produce behavior change in a novel environ- ment, then no setting/situation generalization can be claimed. However, if some component(s) of the training

Generalization and Maintenance of Behavior Change

625

100

1.0 2.0 3.0 4.0

Baseline Training Follow-Up

Covert Novel Restaurant

Overt Novel Restaurant

Covert 1-Year Follow-Up

100

1.0 2.0 3.0 4.0

100

2.0 3.0 4.0

5 10 15 20 25 30 35 Probe Sessions

P er

ce n

t C

or re

ct R

es p

on se

1.0

Student 1

Student 2

Student 3

Figure 1 Percentage of steps necessary to order a meal at a fast-food restaurant correctly performed by three students with disabilities before, during, and after instruction in the classroom. During follow-up, the closed triangles represent probes conducted at a Burger King restaurant using typical observation proce- dures, open triangles represent Burger King probes during which students did not know they were being observed, and open circles represent covert probes conducted in a different McDonald’s 1 year after training. From “Teaching the Handicapped to Eat in Public Places: Acquisition, Generalization and Maintenance of Restaurant Skills” by R. A. van den Pol, B. A. Iwata, M. T. Ivanic, T. J. Page, N. A. Neef, and F. P. Whitley, 1981, Journal of Applied Behavior Analysis, 14, p. 66. Copyright 1981 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

program results in meaningful behavior change in a gen- eralization setting, then setting/situation generalization can be claimed, provided it can be shown that the com- ponent(s) used in the generalization setting was insuffi- cient to produce the behavior change alone in the training environment.

For example, van den Pol and colleagues taught Stu- dent 3, who was deaf, how to use a prosthetic ordering device in the classroom. The device, a plastic laminated sheet of cardboard with a wax pencil, had preprinted ques- tions (e.g., “How much is . . . ?”), generic item names (e.g., large hamburger), and spaces where the cashier could write responses. Simply giving the student some money and the prosthetic ordering card would not have enabled him to order, purchase, and eat a meal indepen- dently. However, after classroom instruction that included guided practice, role playing, social reinforcement (“Good job! You remembered to ask for your change” [p. 64]),

1Because the majority of the examples in this chapter are school based, we have used the language of education. For our purposes here, instruction can be a synonym for treatment, intervention, or therapy, and instructional setting can be a synonym for clinical setting or therapy setting.

corrective feedback, and review sessions with the pros- thetic ordering card produced the desired behaviors in the instructional setting, Student 3 was able to order, pay for, and eat meals in a restaurant aided only by the card.

Distinguishing Between Instructional and Generalization Settings

We use instructional setting to denote the total envi- ronment where instruction occurs, including any aspects of the environment, planned or unplanned, that may in- fluence the learner’s acquisition and generalization of the target behavior.1 Planned elements are the stimuli and

Generalization and Maintenance of Behavior Change

626

Figure 2 Examples of an instructional setting and a generalization setting for six target behaviors.

Instructional Setting 1. Raising hand when special education

teacher asks a question in the resource room.

2. Practicing conversational skills with speech therapist at school.

3. Passing basketball during a team scrimmage on home court.

4. Answering addition problems in vertical format at desk at school.

5. Solving word problems with no distracter numbers on homework assignment.

6. Operating package sealer at community job site in presence of supervisor.

Generalization Setting 1. Raising hand when general education

teacher asks a question in the regular classroom.

2. Talking with peers in town. 3. Passing basketball during a game on

the opponent’s court. 4. Answering addition problems in

horizontal format at desk at school. 5. Solving word problems with distracter

numbers on homework assignment. 6. Operating package sealer at community

job site in absence of supervisor.

events the teacher has programmed in an effort to achieve initial behavior change and promote generalization. Planned elements of an instructional setting for a math lesson, for example, would include the specific math problems to be presented during the lesson and the for- mat and sequencing of those problems. Unplanned as- pects of the instructional setting are elements the teacher is not aware of or has not considered that might affect the acquisition and generalization of the target behavior. For example, the phrase, how much in a word problem may acquire stimulus control over a student’s use of ad- dition, even when the correct solution to the problem re- quires a different arithmetic operation. Or, perhaps a student always uses subtraction for the first problem on each page of word problems because a subtraction prob- lem has always been presented first during instruction.

A generalization setting is any place or stimulus situation that differs in some meaningful way from the in- structional setting and in which performance of the tar- get behavior is desired. There are multiple generalization settings for many important target behaviors. The student who learns to solve addition and subtraction word prob- lems in the classroom should be able to solve similar problems at home, at the store, and on the ball diamond with his friends.

Examples of instructional and generalization settings for six target behaviors are shown in Figure 2. When a person uses a skill in an environment physically re- moved from the setting where he learned it—as with Behaviors 1 through 3 in Figure 2—it is easy to under- stand that event as an example of generalization across settings. However, many important generalized outcomes occur across more subtle differences between the in- structional setting and generalization setting. It is a mis- take to think that a generalization setting must be somewhere different from the place where instruction is

Generalization and Maintenance of Behavior Change

provided. Students often receive instruction in the same place where they will need to maintain and generalize what they have learned. In other words, the instructional setting and generalization setting can, and often do, share the same physical location (as with Behaviors 4 through 6 in Figure 2).

Distinguishing between Setting/Situation Generalization and Response Maintenance

Because any measure of setting/situation generalization is conducted after some instruction has taken place, it might be argued that setting/situation generalization and response maintenance are the same, or are inseparable phenomena at least. Most measures of setting/situation generalization do provide information on response main- tenance, and vice versa. For example, the post-training generalization probes conducted by van den Pol and col- leagues (1981) at the Burger King restaurant and at the second McDonald’s provided data on setting/situation generalization (i.e., to novel restaurants) and on response maintenance of up to 1 year. However, a functional dis- tinction exists between setting/situation generalization and response maintenance, with each outcome presenting a somewhat different set of challenges for programming and ensuring enduring behavior change. When a behavior change produced in the classroom or clinic is not observed in the generalization environment, a lack of setting/ situation generalization is evident. When a behavior change produced in the classroom or clinic has occurred at least once in the generalization setting and then ceases to occur, a lack of response maintenance is evident.

An experiment by Koegel and Rincover (1977) illus- trated the functional difference between setting/situation generalization and response maintenance. Participants

627

100

Child 1

Instructional Setting

Generalization Setting

100

0 100

Child 2

Child 3

50 100 150 200 250 300 350 400 450 Trials

P er

ce n

t C

or re

ct T

ri al

Figure 3 Correct responding by three children on alternating blocks of 10 trials in the instructional setting and 10 trials in the generalization setting. From “Research on the Differences Between Generalization and Maintenance in Extra-Therapy Responding” by R. L. Koegel and A. Rincover, 1977, Journal of Applied Behavior Analysis, 10, p. 4. Copyright 1977 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

were three young boys with autism; each was mute, echolalic, or displayed no appropriate contextual speech. One-to-one instructional sessions were conducted in a small room with the trainer and child seated across from each other at a table. Each child was taught a series of imitative responses (e.g., the trainer said, “Touch your [nose, ear]” or “Do this” and [raised his arm, clapped his hands]). Each 40-minute session consisted of blocks of 10 training trials in the instructional setting alternated with blocks of 10 trials conducted by an unfamiliar adult stand- ing outside, surrounded by trees. All correct responses in the instructional setting were followed by candy and so- cial praise. During the generalization trials the children received the same instructions and model prompts as in the classroom, but no reinforcement or other conse- quences were provided for correct responses in the gen- eralization setting.

Figure 3 shows the percentage of trials in which each child responded correctly in the instructional setting and in the generalization setting. All three children learned to respond to the imitative models in the instructional setting. All three children showed 0% correct respond- ing in the generalization setting at the end of the experi- ment, but for different reasons. Child 1 and Child 3 began emitting correct responses in the generalization

setting as their performances improved in the instruc- tional setting, but their generalized responding was not maintained (most likely the result of the extinction con- ditions in effect in the generalization setting). The imita- tive responding acquired by Child 2 in the instructional setting never generalized to the outside setting. There- fore, the 0% correct responding at the experiment’s con- clusion represents a lack of response maintenance for Child 1 and Child 3, but for Child 2 it represents a fail- ure of setting generalization.

Response Generalization

We define response generalization as the extent to which a learner emits untrained responses that are functionally equivalent to the trained target behavior. In other words, in response generalization forms of behavior for which no programmed contingencies have been applied appear as a function of the contingencies that have been applied to other responses. For example:

• Traci wanted to earn some extra money by helping her older brother with his lawn mowing business. Her brother taught Traci to walk the mower up and down parallel rows that moved progressively from

Generalization and Maintenance of Behavior Change

628

one side of a lawn to the other. Traci discovered that she could mow some lawns just as quickly by first cutting around the perimeter of the lawn and then walking the mower in concentric patterns in- ward toward the center of the lawn.

• Loraine was taught to remove weeds with a long weed-removal tool. Although she has never been taught or asked to do so, sometimes Loraine removes weeds with a hand trowel or with her bare hands.

• Michael’s mother taught him how to take phone messages by using the pencil and notepaper next to the phone to write the caller’s name, phone num- ber, and message. One day, Michael’s mother came home and saw her son’s tape recorder next to the phone. She pushed the play button and heard Michael’s voice say, “Grandma called. She wants to know what you’d like her to cook for dinner Wednesday. Mr. Stone called. His number is 555- 1234, and he said the insurance payment is due.”

The study by Goetz and Baer (1973) of the block building by three preschool girls provides a good exam- ple of response generalization. During baseline the teacher sat by each girl as she played with the blocks, watching closely but quietly, and displaying neither enthusiasm nor criticism for any particular use of the blocks. During the next phase of the experiment, each time the child placed or rearranged the blocks to create a new form that had not appeared previously in that session’s constructions, the teacher commented with enthusiasm and interest (e.g., “Oh, that’s very nice—that’s different!”). Another phase followed in which each repeated construction of a given form within the session was praised (e.g., “How nice—an- other arch!”). The study ended with a phase in which de- scriptive praise was again contingent on the construction of different block forms. All three children constructed more new forms with the blocks when form diversity was reinforced than they did under baseline or under the rein- forcement-for-the-same-forms condition (see Figure 7).

Even though specific responses produced reinforce- ment (i.e., the actual block forms that preceded each in- stance of teacher praise), other responses sharing that functional characteristic (i.e., being different from block forms constructed previously by the child) increased in frequency as a function of the teacher’s praise. As a re- sult, during reinforcement for different forms, the children constructed new forms with the blocks even though each new form itself had never before appeared and therefore could not have been reinforced previously. Reinforcing a few members of the response class of new forms in- creased the frequency of other members of the same re- sponse class.

Generalization and Maintenance of Behavior Change

Generalized Behavior Change: A Relative and Intermixed Concept

As the examples presented previously show, generalized behavior change is a relative concept. We might think of it as existing along a continuum. At one end of the con- tinuum are interventions that might produce a great deal of generalized behavior change; that is, after all compo- nents of an intervention have been terminated, the learner may emit the newly acquired target behavior, as well as several functionally related behaviors not observed pre- viously in his repertoire, at every appropriate opportu- nity in all relevant settings, and he may do so indefinitely. At the other end of the continuum of generalized out- comes are interventions that yield only a small amount of generalized behavior change—the learner uses the new skill only in a limited range of nontraining settings and situations, and only after some contrived response prompts or consequences are applied.

We have presented each of the three primary forms of generalized behavior change individually to isolate its defining features, but they often overlap and occur in combination. Although it is possible to obtain response maintenance without generalization across settings/ situations or behaviors (i.e., the target behavior continues to occur in the same setting in which it was trained after the training contingencies have been terminated), any meaningful measure of setting generalization will entail some degree of response maintenance. And it is common for all three forms of generalized behavior change to be represented in the same instance. For example, during a relatively quiet shift at the widget factory on Monday, Joyce’s supervisor taught her to obtain assistance by call- ing out, “Ms. Johnson, I need some help.” Later that week (response maintenance) when it was very noisy on the factory floor (setting/situation generalization), Joyce sig- naled her supervisor by waving her hand back and forth (response generalization).

Generalized Behavior Change Is Not Always Desirable

It is hard to imagine any behavior that is important enough to target for systematic instruction for which re- sponse maintenance would be undesirable. However, un- wanted setting/situation generalization and response generalization occur often, and practitioners should de- sign intervention plans to prevent or minimize such unwanted outcomes. Undesirable setting/situation gen- eralization takes two common forms: overgeneralization and faulty stimulus control.

Overgeneralization, a nontechnical but effectively descriptive term, refers to an outcome in which the

629

behavior has come under the control of a stimulus class that is too broad. That is, the learner emits the target be- havior in the presence of stimuli that, although similar in some way to the instructional examples or situation, are inappropriate occasions for the behavior. For example, a student learns to spell division, mission, and fusion with the grapheme, –sion. When asked to spell fraction, the student writes f-r-a-c-s-i-o-n.

With faulty stimulus control, the target behavior comes under the restricted control of an irrelevant an- tecedent stimulus. For example, after learning to solve word problems such as, “Natalie has 3 books. Amy has 5 books. How many books do they have in total?” by adding the numerals in the problem, the student adds the numerals in any problem that includes the words “in total” (e.g., “Corinne has 3 candies. Amanda and Corinne have 8 candies in total. How many candies does Amanda have?”).2

Undesired response generalization occurs when any of a learner’s untrained but functionally equivalent re- sponses results in poor performance or undesirable out- comes. For example, although Jack’s supervisor at the widget factory taught him to operate the drill press with two hands because that is the safest method, sometimes Jack operates the press with one hand. One-handed re- sponses are functionally equivalent to two-handed re- sponses because both topographies cause the drill press to stamp out a widget, but one-handed responses com- promise Jack’s health and the factory’s safety record. Or, perhaps some of her brother’s customers do not like how their lawns look after Traci has mowed them in concen- tric rectangles.

Other Types of Generalized Outcomes

Other types of generalized outcomes that do not fit eas- ily into categories of response maintenance, setting/situ- ation generalization, and response generalization have been reported in the behavior analysis literature. For ex- ample, complex members of a person’s repertoire some- times appear quickly with little or no apparent direct conditioning, such as the stimulus equivalence relations (Sidman, 1994). Another type of such rapid learning that appears to be a generalized outcome of other events has been called contingency adduction, a process whereby a behavior that was initially selected and shaped under one set of conditions is recruited by a dif- ferent set of contingencies and takes on a new function

2Examples of faulty stimulus control caused by flaws in the design of in- structional materials, and suggestions for detecting and correcting those flaws, can be found in J. S. Vargas (1984).

in a person’s repertoire (Adronis, 1983; Johnson & Layng, 1992).

Sometimes an intervention applied to one or more people results in behavior changes in other people who were not directly treated by the contingencies. Genera- lization across subjects refers to changes in the behav- ior of people not directly treated by an intervention as a function of treatment contingencies applied to other peo- ple. This phenomenon, which has been described with a variety of related or synonymous terms—vicarious rein- forcement (Bandura, 1971; Kazdin, 1973), ripple effect (Kounin, 1970), and spillover effect (Strain, Shores, & Kerr, 1976)—provides another dimension for assessing the generalization of treatment effects. For example, Fan- tuzzo and Clement (1981) examined the degree to which behavior changes would generalize from one child who received teacher-administered or self-administered token reinforcement during a math activity to a peer seated next to the child.

Drabman, Hammer, and Rosenbaum (1979) combined four basic types of generalized treatment effects— (a) across time (i.e., response maintenance), (b) across set- tings (i.e., setting/situation generalization), (c) across be- haviors (i.e., response generalization), and (d) across subjects—into a conceptual framework they called the generalization map. By viewing each type of generalized outcome as dichotomous (i.e., either present or absent) and by combining all possible permutations of the four cate- gories, Drabman and colleagues arrived at 16 categories of generalized behavior change ranging from maintenance (Class 1) to subject-behavior-setting-time generalization (Class 16). Class 1 generalization is evident if the target be- havior of the target subject(s) continues in the treatment setting after any “experiment-controlled contingencies” have been discontinued. Class 16 generalization, which Drabman and colleagues (1979) called the “ultimate form” of generalization, is evidenced by “a change in a nontar- get subject’s nontarget behavior which endures in a dif- ferent setting after the contingencies have been withdrawn in the treatment setting” (p. 213).

Although Drabman and colleagues recognized that “with any heuristic technique the classifications may prove arbitrary” (p. 204), they provided objectively stated rules for determining whether a given behavioral event fits the requirements of each of their 16 classifications. Regardless of whether generalized behavior change con- sists of such distinctly separate and wide-ranging phe- nomena as detailed by Drabman and colleagues, their generalization map provided an objective framework by which the extended effects of behavioral interventions can be described and communicated. For example, Stevenson and Fantuzzo (1984) measured 15 of the 16 generalization map categories in a study of the effects of teaching a fifth-grade boy to use self-management

Generalization and Maintenance of Behavior Change

630

techniques. They not only measured the effects of the in- tervention on the target behavior (math performance) in the instructional setting (school), but they also assessed effects on the student’s math behavior at home, disruptive behavior at home and at school, both behaviors for a non- treated peer in both settings, and maintenance of all of the above.

Planning for Generalized Behavior Change

In general, generalization should be programmed, rather than expected or lamented.

—Baer, Wolf, and Risley (1968, p. 97)

In their review of 270 published studies relevant to generalized behavior change, Stokes and Baer (1977) concluded that practitioners should always “assume that generalization does not occur except through some form of programming . . . and act as if there were no such an- imal as ‘free’ generalization—as if generalization never occurs ‘naturally,’ but always requires programming” (p. 365). Of course, generalization of some type and de- gree does usually occur, whether or not it is planned. Such unplanned and unprogrammed generalization may be suf- ficient, but often it is not, particularly for many learners served by applied behavior analysts (e.g., children and adults with learning problems and developmental dis- abilities). And if left unchecked, unplanned generalized outcomes may be undesirable outcomes.

Achieving optimal generalized outcomes requires thoughtful, systematic planning. This planning begins with two major steps: (1) selecting target behaviors that will meet natural contingencies of reinforcement, and (2) specifying all desired variations of the target behav- ior and the settings/situations in which it should (and should not) occur after instruction has ended.

Selecting Target Behaviors That Will Meet Naturally Existing Contingencies of Reinforcement

The everyday environment is full of steady, dependable, hardworking sources of reinforcement for almost all of the behaviors that seem natural to us. That is why they seem natural to us.

—Donald M. Baer (1999, p. 15)

Numerous criteria have been suggested for determining whether a proposed teaching objective is relevant or functional for the learner. For example, the age- appropriateness of a skill and the degree to which it rep- resents normalization are often cited as important

Generalization and Maintenance of Behavior Change

criteria for choosing target behaviors for students with disabilities (e.g., Snell & Brown, 2006). In the end, how- ever, there is just one ultimate criterion of functionality: A behavior is functional only to the extent that it pro- duces reinforcement for the learner. This criterion holds no matter how important the behavior may be to the per- son’s health or welfare, or no matter how much teach- ers, family, friends, or the learner himself considers the behavior to be desirable. To repeat: A behavior is not functional if it does not produce reinforcement for the learner. Said another way: Behaviors that are not fol- lowed by reinforcers on at least some occasions will not be maintained.

Ayllon and Azrin (1968) recognized this fundamen- tal truth when they recommended that practitioners fol- low the relevance-of-behavior rule when selecting target behaviors. The rule: Choose only those behaviors to change that will produce reinforcers in the postinterven- tion environment. Baer (1999) believed so strongly in the importance of this criterion that he recommended that practitioners heed a similar rule:

A good rule is to not make any deliberate behavior changes that will not meet natural communities of rein- forcement. Breaking this rule commits you to maintain and extend the behavior changes that you want, by your- self, indefinitely. If you break this rule, do so knowingly. Be sure that you are willing and able to do what will be necessary. (p. 16, emphasis in original)

Programming for the generalization and maintenance of any behavior for which a natural contingency of rein- forcement exists, no matter the specific tactics employed, consists of getting the learner to emit the behavior in the generalization setting just often enough to contact the oc- curring contingencies of reinforcement. Generalization and maintenance of the behavior from that point forward, while not assured, is a very good bet. For example, after receiving some basic instruction on how to operate the steering wheel, gas pedal, and brakes on a car, the natu- rally existing reinforcement and punishment contingen- cies involving moving automobiles and the road will select and maintain effective steering, acceleration, and braking. Very few drivers need booster training sessions on the basic operation of the steering wheel, gas pedal, and brakes.

We define a naturally existing contingency as any contingency of reinforcement (or punishment) that oper- ates independent of the behavior analyst’s or practi- tioner’s efforts. This is a pragmatic, functional conception of a naturally existing contingency defined by the ab- sence of the behavior analyst’s efforts. Naturally exist- ing contingencies include contingencies that operate

631

without social mediation (e.g., walking fast on an icy sidewalk is often punished by a slip and fall) and socially mediated contingencies contrived and implemented by other people in the generalization setting. From the per- spective of a special educator who is teaching a set of targeted social and academic skills to students for whom the general education classroom represents the general- ization setting, a token economy operated by the general education classroom teacher is an example of the latter type of naturally existing contingency.3 Even though the token economy was contrived by the teacher in the gen- eral education classroom, it is a naturally existing con- tingency because it already operates in the generalization setting.

We define a contrived contingency as any contin- gency of reinforcement (or punishment) designed and implemented by a behavior analyst or practitioner to achieve the acquisition, maintenance, and/or generaliza- tion of a targeted behavior change. From the perspective of the teacher in the general education classroom who designed and implemented it, the token economy in the previous example is a contrived contingency.

In reality, practitioners are often charged with the difficult task of teaching important skills for which there are no dependable naturally existing contingencies of re- inforcement. In such cases, practitioners should realize and plan for the fact that the generalization and mainte- nance of target behaviors will have to be supported, per- haps indefinitely, with contrived contingencies.

Specifying All Desired Variations of the Behavior and the Settings/ Situations Where It Should (and Should Not) Occur

This stage of planning for generalized outcomes includes identifying all the desired behavior changes that need to be made and all the environments and stimulus condi- tions in which the learner should emit the target behav- ior(s) after direct training has ceased (Baer, 1999). For some target behaviors, the most important stimulus con- trol for each response variation is clearly defined (e.g., reading C-V-C-E words) and restricted in number (e.g., solving multiplication facts). For many important target behaviors, however, the learner is likely to encounter a multitude of settings and stimulus conditions where the behavior, in a wide variety of response forms, is desired. Only by considering these possibilities prior to instruction can the behavior analyst design an intervention with the best chance of preparing the learner for them.

In one sense, this component of planning for gener- alized outcomes is similar to preparing a student for a fu-

ture test without knowing the content or the format of all of the questions that will be on the test. The stimulus con- ditions and contingencies of reinforcement that exist in the generalization setting(s) will provide that test to the learner. Planning involves trying to determine what the final exam will cover (type and form of questions), whether there will be any trick questions (e.g., confus- ing stimuli that might evoke the target response when it should not occur), and whether the learner will need to use his new knowledge or skill in different ways (re- sponse generalization).

List All the Behaviors That Need to Be Changed

A list should be made of all the forms of the target be- havior that need to be changed. This is not an easy task, but a necessary one to obtain a complete picture of the teaching task ahead. For example, if the target behavior is teaching Brian, the young boy with autism, to greet people, he should learn a variety of greetings in addition to “Hello, how are you?” Brian may also need many other behaviors to initiate and participate in conversa- tions, such as responding to questions, taking turns, stay- ing on topic, and so forth. He may also need to be taught when and with whom to introduce himself. Only by hav- ing a complete list of all the desired forms of the behav- ior can the practitioner make meaningful decisions about which behaviors to teach directly and which to leave to generalization.

The practitioner should determine whether and to what extent response generalization is desirable for all of the behavior changes listed, and then, make a priori- tized list of the variations of the target behavior he would like to see as generalized outcomes.

List All the Settings and Situations in Which the Target Behavior Should Occur

A list should be made of all the desired settings and sit- uations in which the learner will emit the target behavior if optimal generalization is achieved. Will Brian need to introduce himself and talk with children his own age, to adults, to males and females? Will he need to talk with others at home, at school, in the lunchroom, on the play- ground? Will he be confronted with situations that may appear to be appropriate opportunities to converse but are not (e.g., an unknown adult approaches and offers candy) and for which an alternative response is needed (e.g., walking away, seeking out a known adult). (This kind of analysis often adds additional behaviors to the list of skills to be taught.)

When all of the possible situations and settings have been identified, they should be prioritized according to

Generalization and Maintenance of Behavior Change

3Token economies are described in Chapter entitled “Contingency Con- tracting, Token Economy, and Group Contingencies.

632

their importance and the client’s likelihood of encoun- tering them. Further analysis of the prioritized environ- ments should then be conducted. What discriminative stimuli usually set the occasion for the target behavior in these various settings and situations? What schedules of reinforcement for the target behavior are typical in these nontraining environments? What kinds of reinforcers are likely to be contingent on the emission of the target be- havior in each of the settings? Only when she has an- swered all of these questions, if not by objective observation then at least by considered estimation, can the behavior analyst begin to have a full picture of the teaching task ahead.

Is the Pre-Intervention Planning Worth It?

Obtaining all of the information just described requires considerable time and effort. Given limited resources, why not design an intervention and immediately begin trying to change the target behavior? It is true that many behavior changes do show generalization, even though the extension of the trained behavior across time, set- tings, and other behaviors was unplanned and unpro- grammed. When target behaviors have been chosen that are truly functional for the subject and when those be- haviors have been brought to a high level of proficiency under discriminative stimuli relevant to generalization settings, the chances of generalization are good. But what constitutes a high level of proficiency for certain behav- iors in various settings? What are all of the relevant dis- criminative stimuli in all of the relevant settings? What are all the relevant settings?

Without a systematic plan, a practitioner will usu- ally be ignorant of the answers to these vital questions. Few behaviors that are important enough to target have such limited needs for generalized outcomes that the an- swers to such questions are obvious. Just a cursory con- sideration of the behaviors, settings, and people related to a child introducing himself revealed numerous factors that may need to be incorporated into an instructional plan. A more thorough analysis would produce many more. In fact, a complete analysis will invariably reveal more behaviors to be taught, to one person or another, than time or resources would ever allow. And Brian—the 10-year-old who is learning to greet people and introduce himself—in all likelihood needs to learn many other skills also, such as self-help, academic, and recreation and leisure skills, to name just a few. Why then create all the lists in the first place when everything cannot be taught anyway? Why not just train and hope?4

Generalization and Maintenance of Behavior Change

4Teaching a new behavior without developing and implementing a plan to facilitate its maintenance and generalization is done so often that Stokes and Baer 1977) called it the “train and hope” approach to generalization.

Baer (1999) described six possible benefits of list- ing all the forms of behavior change and all the situations in which these behavior changes should occur.

1. You now see the full scope of the problem ahead of you, and thus see the corresponding scope that your teaching program needs to have.

2. If you teach less than the full scope of the problem, you do so by choice rather than by forgetting that some forms of the behavior could be important, or that there were some other situations in which the behavior change should or should not occur.

3. If less than a complete teaching program results in less than a complete set of behavior changes, you will not be surprised.

4. You can decide to teach less than there is to learn, perhaps because that is all that is practical or pos- sible for you to do.

5. You can decide what is most important to teach. You can also decide to teach the behavior in a way that encourages the indirect development of some of the other forms of the desired behavior, as well as the indirect occurrence of the behavior in some other desired situations, that you will not or cannot teach directly.

6. But if you choose the option discussed in number 5 above, rather than the complete program implicit in number 1, you will do so knowing that the de- sired outcome would have been more certain had you taught every desirable behavior change di- rectly. The best that you can do is to encourage the behavior changes that you do not cause directly. So, you will have chosen the option in number 5 either of necessity or else as a well-considered gamble after a thoughtful consideration of possibil- ities, costs, and benefits. (pp. 10–11)

After determining which behaviors to teach directly and in which situations and settings to teach those be- haviors, the behavior analyst is ready to consider strate- gies and tactics for achieving generalization to untrained behaviors and settings.

Strategies and Tactics for Promoting Generalized Behavior Change Various authors have described conceptual schemes and taxonomies of methods for promoting generalized be- havior change (e.g., Egel, 1982; Horner, Dunlap, & Koegel, 1988; Osnes & Lieblein, 2003; Stokes & Baer,

633

Figure 4 Strategies and tactics for promoting generalized behavior change.

Teach the Full-Range of Relevant Stimulus Conditions and Response Requirements 1. Teach sufficient stimulus examples 2. Teach sufficient response examples

Make the Instructional Setting Similar to the Generalization Setting 3. Program common stimuli 4. Teach loosely

Maximize Contact with Reinforcement in the Generalization Setting 5. Teach the target behavior to levels of performance required by naturally existing

contingencies of reinforcement 6. Program indiscriminable contingencies 7. Set behavior traps 8. Ask people in the generalization setting to reinforce the target behavior 9. Teach the learner to recruit reinforcement

Mediate Generalization 10. Contrive a mediating stimulus 11. Teach self-management skills

Train to Generalize 12. Reinforce response variability 13. Instruct the learner to generalize

1977; Stokes & Osnes, 1989). The conceptual scheme presented here is informed by the work of those authors and others, and by our own experiences in designing, implementing, and evaluating procedures for promot- ing generalized outcomes and in teaching practitioners to use them. Although numerous methods and tech- niques have been demonstrated and given a variety of names, most tactics that effectively promote general- ized behavior change can be classified under five strate- gic approaches:

• Teach the full range of relevant stimulus conditions and response requirements.

• Make the instructional setting similar to the gener- alization setting.

• Maximize the target behavior’s contact with rein- forcement in the generalization setting.

• Mediate generalization.

• Train to generalize.

In the following sections we describe and provide examples of 13 tactics applied behavior analysts have used to accomplish these five strategies (see Figure 4). Although each tactic is described individually, most ef- forts to promote generalized behavior change entail a combination of these tactics (e.g., Ducharme & Hol- born, 1997; Grossi, Kimball, & Heward, 1994; Hughes, Harmer, Killina, & Niarhos, 1995; Ninness, Fuerst, & Rutherford, 1991; Trask-Tyler, Grossi, & Heward, 1994).

Teach the Full Range of Relevant Stimulus Conditions and Response Requirements

The most common mistake that teachers make, when they want to establish a generalized behavior change, is to teach one good example of it and expect the student to generalize from that example.

— Donald M. Baer (1999, p. 15)

To be most useful, most important behaviors must be per- formed in various ways across a wide range of stimulus conditions. Consider a person skilled in reading, math, conversing with others, and cooking. That person can read thousands of different words; add, subtract, multiply, and divide any combination of numbers; make a multi- tude of relevant and appropriate comments when talking with others; and measure, combine, and prepare numer- ous ingredients in hundreds of recipes. Helping learners achieve such wide-ranging performances presents an enormous challenge to the practitioner.

One approach to this challenge would be to teach every desired form of a target behavior in every setting/ situation in which the learner may need that behavior in the future. Although this approach would eliminate the need to program for response generalization and setting/ situation generalization (response maintenance would re- main the only problem), it is seldom possible and never practical. A teacher cannot provide direct instruction on every printed word a student may encounter in the fu- ture, or teach a student every measuring, pouring, stir- ring, and sautéing movement needed to make every dish

Generalization and Maintenance of Behavior Change

634

he may want to make in the future. Even for most skill areas for which it would be possible to teach every pos- sible example (e.g., instruction could be provided on all 900 different single-digit-times-two-digit multiplication problems), to do so would be impractical for many rea- sons, not the least of which is that the student needs to learn not only many other types of math problems but also skills in other curriculum areas.

A general strategy called teaching sufficient exam- ples consists of teaching the student to respond to a sub- set of all of the possible stimulus and response examples and then assessing the student’s performance on untrained examples.5 For example, the generalization of a student’s ability to solve two-digit-minus-two-digit arithmetic prob- lems with regrouping can be assessed by asking the stu- dent to solve several problems of the same type for which no instruction or guided practice has been provided. If the results of this generalization probe show that the student responds correctly to untaught examples, then instruction can be halted on this class of problems. If the student per- forms poorly on the generalization probe, the practitioner teaches additional examples before again assessing the student’s performance on a new set of untaught examples. This cycle of teaching new examples and probing with untaught examples continues until the learner consistently responds correctly to untrained examples representing the full range of stimulus conditions and response require- ments found in the generalization setting.

Teach Sufficient Stimulus Examples

The tactic for promoting setting/situation generalization called teaching sufficient stimulus examples involves teaching the learner to respond correctly to more than one example of antecedent stimulus conditions and prob- ing for generalization to untaught stimulus examples. A different stimulus example is incorporated into the teach- ing program each time a change is made in any dimen- sion of the instructional item itself or the environmental context in which the item is taught. Examples of four di- mensions by which different instructional examples can be identified and programmed are the following:

• The specific item taught (e.g., multiplication facts: 7 � 2, 4 � 5; letter sounds: a, t)

• The stimulus context in which the item is taught (e.g., multiplication facts presented in vertical for- mat, in horizontal format, in word problems; say- ing the sound of t when it appears at the beginning and end of words: tab, bat)

Generalization and Maintenance of Behavior Change

5Other terms commonly used for this strategy for promoting generalized be- havior change are training sufficient exemplars (Stokes & Baer, 1977) and training diversely (Stokes & Osnes, 1989).

• The setting where instruction occurs (e.g., large- group instruction at school, collaborative learning group, home)

• The person doing the teaching (e.g., classroom teacher, peer, parent)

As a general rule, the more examples the practitioner uses during instruction, the more likely the learner will re- spond correctly to untrained examples or situations. The actual number of examples that must be taught before sufficient generalization occurs will vary as a function of factors such as the complexity of the target behavior being taught, the teaching procedures employed, the student’s opportunities to emit the target behavior under the vari- ous conditions, the naturally existing contingencies of reinforcement, and the learner’s history of reinforcement for generalized responding.

Sometimes teaching as few as two examples will pro- duce significant generalization to untaught examples. Stokes, Baer, and Jackson (1974) taught a greeting re- sponse to four children with severe mental retardation who seldom acknowledged or greeted other people. The senior author, working as a dormitory assistant, used un- conditioned reinforcers (potato chips and M&Ms) and praise to shape the greeting response (at least two back- and-forth waves of a raised hand). Then this initial trainer maintained the newly learned hand wave by contriving three to six contacts per day with each of the children in various settings (e.g., playroom, corridor, dormitory, courtyard). Throughout the study as many as 23 different staff members systematically approached the children during different times of the day in different settings and recorded whether the children greeted them with a hand waving response. If a child greeted a prober with the wav- ing response, the prober responded with “Hello, (name).” Approximately 20 such generalization probes were con- ducted each day with each child.

Immediately after learning the greeting response with just one trainer, one of the children (Kerry) showed good setting/situation generalization by using it appropriately in most of her contacts with other staff members. How- ever, the other three children usually failed to greet staff members most of the time, even though they continued to greet the original trainer on virtually every occasion. A second staff member then began to reinforce and main- tain the greeting responses of these three children. As a result of adding the second trainer, the children’s greet- ing behavior showed widespread generalization to the other staff members. Stokes and colleagues’ (1974) study is important for at least two reasons. First, it demonstrated an effective method for continual assessment of setting/ situation generalization across numerous examples (in this case, people). Second, the study showed that it is

635

sometimes possible to produce widespread generaliza- tion by programming only two examples.

Teach Sufficient Response Examples

Instruction that provides practice with a variety of re- sponse topographies helps to ensure the acquisition of desired response forms and also promotes response gen- eralization in the form of untrained topographies. Often called multiple exemplar training, this tactic typically incorporates both stimulus and response variations. Mul- tiple exemplar training was used to achieve the acquisi- tion and generalization of affective behavior by children with autism (Gena, Krantz, McClannahan, & Poulson, 1996); cooking skills by young adults with disabilities (Trask-Tyler, Grossi, & Heward, 1994); domestic skills (Neef, Lensbower, Hockersmith, DePalma, & Gray, 1990); vocational skills (Horner, Eberhard, & Sheehan, 1986); daily living skills (Horner, Williams, & Steveley, 1987); and requests for assistance (Chadsey-Rusch, Dras- gow, Reinoehl, Halle, & Collet-Klingenberg, 1993).

Four female high school students with moderate mental retardation participated in a study by Hughes and colleagues (1995) that assessed the effects of an inter- vention they called multiple-exemplar self-instructional training on the acquisition and generalization of the stu- dents’ conversational interactions with peers. The young women were recommended for the study because they initiated conversations and responded to peers efforts to talk with them at “low or nonexistent rates” and main- tained little eye contact. One of the students, Tanya, had recently been refused a job at a restaurant because of her “reticence and lack of eye contact during her job inter- view” (p. 202).

A key element of Hughes and colleagues’ interven- tion was practicing a wide variety of conversation starters and statements with different peer teachers. Ten volunteer peer teachers recruited from general education class- rooms helped teach conversation skills to the participants. The peer teachers were male and female, ranged in grade level from freshmen to seniors, and represented African American, Asian American, and Euro-American ethnic groups. Instead of learning a few scripted conversation openers, the participants practiced using multiple exam- ples of conversation starters selected from a pooled list of conversation openers used by general education students. Additionally, participants were encouraged to develop individual adaptations of statements, which further pro- moted response generalization by increasing the number and range of conversational statements that were likely to be used in subsequent conversations.

Before, during, and after multiple exemplar training, generalization probes were conducted of each partici- pant’s use of self-instructions, eye contact, and initiating

6Siegfried Engelmann and Douglas Carnine’s (1982) Theory of Instruc- tion: Principles and Applications is one of the most thorough and sophis- ticated treatments of the selection and sequencing of teaching examples for effective and efficient curriculum design.

and responding to conversation partners. The 23 to 32 different students who served as conversation partners for each participant represented the full range of student characteristics in the school population (e.g., gender, age, ethnicity, students with and without disabilities) and in- cluded students who were known and unknown to the participants prior to the study. The rate of conversation initiations by all four participants increased during mul- tiple exemplar training to levels approximating those of general education students and was maintained at those rates after the intervention was terminated completely (see Figure 5).

General Case Analysis

Teaching a learner to respond correctly to multiple ex- amples will not automatically produce generalized re- sponding to untaught examples. To achieve an optimal degree of generalization and discrimination, the behavior analyst must pay close attention to the specific examples used during instruction; not just any examples will do. Optimally effective instructional design requires select- ing teaching examples that represent the full range of stimulus situations and response requirements in the nat- ural environment.6 General case analysis (also called general case strategy) is a systematic method for select- ing teaching examples that represent the full range of stimulus variations and response requirements in the gen- eralization setting (Albin & Horner, 1988; Becker & En- gelmann, 1978; Engelmann & Carnine, 1982).

A series of studies by Horner and colleagues demon- strated the importance of teaching examples that sys- tematically sample the range of stimulus variations and response requirements the learner will encounter in the generalization setting (e.g., Horner, Eberhard, & Shee- han, 1986; Horner & McDonald, 1982; Horner, Williams, & Steveley, 1987). In a classic example of this line of re- search, Sprague and Horner (1984) evaluated the effects of general case instruction on the generalized use of vend- ing machines by six high school students with moderate to severe mental retardation. The dependent variable was the number of vending machines each student operated correctly during generalization probes of 10 different ma- chines located within the community. For a probe trial to be scored as correct, a student had to correctly perform a chain of five responses (i.e., insert the proper number of coins, activate the machine for the desired item, and so on). The researchers selected the 10 vending machines used to assess generalization because each student’s per- formance on those machines would serve as an index of

Generalization and Maintenance of Behavior Change

636

6 5 4 3 2 1 0

Follow-Up

Patti

MaintenanceMultiple Exemplar Self-Instruction Training

Baseline

6 5 4 3 2 1 0

Carrie Ann

6 5 4 3 2 1 0

Tanya

6 5 4 3 2 1 0 Melissa

Range of Expected Behavior

0 5 10 15 20 25 30 35 40 45 50 55 9 11

MonthsGeneralization Sessions Gym Classroom Lunchroom Workroom Participant Absence

Open symbols = untrained setting Filled symbols = trained setting

In it

ia ti

on s

p er

M in

u te

Figure 5 Conversation initiations per minute by four high school students with disabili- ties to conversation partners with and without disabilities during generalization sessions. The shaded bands represent typical performance by general education students. From “The Effects of Multiple-Exemplar Training on High-School Students’ Generalized Conversational Interactions” by C. Hughes, M. L. Harmer, D. J. Killina, and F. Niarhos, 1995. Journal of Applied Behavior Analysis, 28, p. 210. Copyright 1995 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

his performance “across all vending machines dispensing food and beverage items costing between $.20 and $.75 in Eugene, Oregon” (p. 274). None of the vending ma- chines used in the generalization probes was identical to any of the vending machines used during instruction.

After a single-baseline probe verified each student’s inability to use the 10 vending machines in the commu- nity, a condition the researchers called “single-instance instruction” began. Under this condition each student re- ceived individual training on a single vending machine located in the school until he used the machine in- dependently for three consecutive correct trials on each

Generalization and Maintenance of Behavior Change

of two consecutive days. Even though each student had learned to operate the training machine without errors, the generalization probe following single-instance in- struction revealed little or no success with the vending machines in the community (see Probe Session 2 in Figure 6). The continued poor performance of Students 2, 3, 5, and 6 on successive generalization probes that followed additional instruction with the single-instance training machine shows that overlearning on a single example does not necessarily aid generalization. Further evidence of the limited generalization obtained from single-instance instruction is the fact that seven of

637

Baseline

0 Su

b je

ct 1

Su b

je ct

Su b

je ct

Single- Instance

General Case

Baseline 10

Su b

je ct

Su b

je ct

Su b

je ct

Single- Instance

Multiple- Instance

General Case

1 2 3 4 5 6 7 Probe Sessions

N u

m b

er o

f V

en d

in g

M ac

h in

es O

p er

at ed

C or

re ct

Figure 6 The number of nontrained probe machines operated correctly by students across phases and probe sessions. From “The Effects of Single Instance, Multiple Instance, and General Case Training on Generalized Vending Machine Use by Moderately and Severely Handicapped Students” by J. R. Sprague and R. H. Horner, 1984, Journal of Applied Behavior Analysis, 17, p. 276. Copyright 1984 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

the eight total probe trials performed correctly by all stu- dents after single-instance instruction were on Probe Ma- chine 1, the vending machine that most closely resembled the machine on which the students had been trained.

Next, multiple-instance training was implemented with Students 4, 5, and 6. The teaching procedures and performance criteria for multiple-instance training were the same as those used in the single-instance condition ex- cept that each student received instruction until he reached criterion on all three new machines. Sprague and Horner (1984) purposely selected vending machines to use in the multiple-instance instruction that were similar

to one another and that did not sample the range of stim- ulus variations and response requirements that defined the vending machines in the community. After reaching training criterion on three additional machines, Students 4, 5, and 6 were still unable to operate the machines in the community. During the six probe sessions that followed multiple-instance instruction, these students correctly completed only 9 of the 60 total trials.

The researchers then introduced general case in- struction in multiple-baseline fashion across students. This condition was the same as multiple-instance in- struction except that the three different vending machines

Generalization and Maintenance of Behavior Change

638

used in general case training, when combined with the single-instance machine, provided the students with prac- tice across the total range of stimulus conditions and re- sponse variations found in vending machines in the community. None of the training machines, however, were exactly the same as any machine used in the gen- eralization probes. After reaching training criterion on the general case machines, all six students showed sub- stantial improvements in their performance on the 10 un- trained machines. Sprague and Horner (1984) speculated that Student 3’s poor performance on the first general- ization probe following general case instruction was caused by a ritualistic pattern of inserting coins that he had developed during previous probe sessions. After re- ceiving repeated practice on the coin insertion step dur- ing a training session between Probe Sessions 5 and 6, Student 3’s performance on the generalization machines improved greatly.

Negative, or “Don’t Do It,” Teaching Examples

Generalization across every possible condition or situation is rarely desirable. Teaching a student where and when to use a new skill or bit of knowledge does not mean that he also knows where and when not to use this newly learned behavior. Brian, for example, needs to learn that he should not say, “Hello, how are you?” to people that he greeted within the past hour or so. Learners must be taught to discriminate the stimulus conditions that signal when responding is appropriate from stimulus conditions that signal when responding is inappropriate.

Instruction that includes “don’t do it” teaching ex- amples intermixed with positive examples provides learn- ers with practice discriminating stimulus situations in which the target behavior should not be emitted (i.e., S�s) from those conditions when the behavior is appropriate. This sharpens the stimulus control necessary to master many concepts and skills (Engelmann & Carnine, 1982).7

Horner, Eberhard, and Sheehan (1986) incorporated “don’t do it” examples into training programs to teach four high school students with moderate to severe mental re- tardation how to bus tables in cafeteria-style restaurants. To correctly bus a table, the student had to remove all dishes, silverware, and garbage from tabletop, chairs, and the floor

Generalization and Maintenance of Behavior Change

7Teaching examples used to help students discriminate when not to re- spond (i.e., S�s) are sometimes called negative examples and contrasted with positive examples (i.e., SDs). However, in our work in teaching this concept, practitioners have told us that the term negative teaching exam- ple suggests that the teacher is modeling or showing the learner how not to perform the target behavior. Instruction on the desired topography for some behaviors may be aided by providing students with models on how not to perform certain behavior (i.e., negative examples), but the function of “don’t do it” examples is to help the learner discriminate antecedent conditions that signal an inappropriate occasion for responding.

under and around the table; wipe the tabletop; straighten chairs; and place dirty dishes and garbage in appropriate re- ceptacles. In addition, the students were taught to inquire, through the use of cards, if a customer was finished with empty dishes. The three settings, one for training and two for generalization probes, differed in terms of size and fur- niture characteristics and configurations.

Each training trial required the student to attend to the following stimulus features of a table: (a) the pres- ence or absence of people at a table, (b) whether people were eating at the table, (c) the amount and/or status of food on dishes, (d) the presence or absence of garbage at a table, and (e) the location of garbage and/or dirty dishes at a table. Training consisted of 30-minute sessions in- volving six table types that represented the range of con- ditions likely to be encountered in a cafeteria-style restaurant. A trainer modeled correct table bussing, ver- bally prompted correct responding and stopped the stu- dent when errors occurred, recreated the situation, and provided additional modeling and assistance. The six training examples consisted of four to-be-bussed tables and two not-to-be-bussed tables (see Table 1).

During generalization probe sessions, which were conducted in two restaurants not used for training, each student was presented with 15 tables selected to repre- sent the range of table types the students could be ex- pected to encounter if employed in a cafeteria-style restaurant. The 15 probe tables consisted of 10 to-be- bussed tables and 5 not-to-be-bussed tables. The results showed a functional relation between general case in- struction that included not-to-be-bussed tables and “im- mediate and pronounced improvement in the percentage of probe tables responded to correctly” (p. 467).

Negative teaching examples are necessary when dis- criminations must be made between appropriate and inappropriate conditions for a particular response. Prac- titioners should ask this question: Is responding always appropriate in the generalization setting(s)? If the an- swer is no, then “don’t do it now” teaching examples should be part of instruction.

Does the teaching setting or situation naturally or au- tomatically include a sufficient number and range of neg- ative examples? The teaching situation must be analyzed to answer this important question. Practitioners may need to contrive some negative teaching examples. Practition- ers should not assume the natural environment will read- ily reveal sufficient negative examples. Conducting training in the natural environment is no guarantee that learners will be exposed to stimulus situations that they are likely to encounter in the generalization environment after training. For example, in the study on teaching table bussing described earlier, Horner and colleagues (1986) noted that, “on some days the trainer needed actively to

639

Table 1 Six Training Examples Used to Teach Students with Disabilities How to Bus Tables in Cafeteria-Style Restaurants

Training Presence Garbage: Location Examples of People and People Eating Dishes: Empty/ Present or of Garbage Correct

Possessions or Not Eating Part/New Food Not Present and Dishes Response

1 0 People + N/A Partial Present Table Don’t Bus possessions Chairs

2 0 People N/A Partial Present Table, Floor, Chair Bus

3 2 People Eating New food Present Table, Chair, Floor Don’t Bus

4 0 People N/A Empty Present Table, Floor Bus

5 1 Person Not eating Empty Present Chair, Floor Bus

6 2 People Not eating Empty Present Table Bus

From “Teaching Generalized Table Bussing: The Importance of Negative Teaching Examples” by R. H. Horner, J. M. Eberhard, and M. R. Sheehan, 1986, Behavior Modification, 10, p. 465. Copyright 1986 by the Sage Publications, Inc. Used by permission.

set up one or more table types to ensure that a student had access to a table type that was not ‘naturally’ avail- able” (p. 464).

“Don’t do it” teaching examples should be selected and sequenced according to the degree to which they dif- fer from positive examples (i.e., SDs). The most effective negative teaching examples will share many of the rele- vant characteristics of the positive teaching examples (Horner, Dunlap, & Koegel, 1988). Such minimum dif- ference negative teaching examples help the learner to perform the target behavior with the precision required by the natural environment. Minimum difference teaching examples help eliminate “generalization errors” due to overgeneralization and faulty stimulus control. For ex- ample, the “don’t bus” tables used by Horner and col- leagues (1986) shared many features with the “bus” tables (see Table 1).

Make the Instructional Setting Similar to the Generalization Setting

Fresno State coach Pat Hill expects Ohio Stadium to be a new experience for the Bulldogs, who will visit for the first time. For a practice last week in FSU’s stadium, Hill hired a production company to blast the Ohio State fight song—at about 90 decibels—throughout the two- hour session. “We created some noise and atmosphere to give us a feel of a live game,” Hill said.

—Columbus Dispatch (August 27, 2000)

A basic strategy for promoting generalization is to incor- porate into the instructional setting stimuli that the learner is likely to encounter in the generalization setting. The greater the similarity between the instructional environ- ment and the generalization environment, the more likely the target behavior will be emitted in the generalization setting. The principle of stimulus generalization states that

a behavior is likely to be emitted in the presence of a stim- ulus very similar to the stimulus conditions in which the behavior was reinforced previously, but the behavior will likely not be emitted under stimulus conditions that differ significantly from the training stimulus.

Stimulus generalization is a relative phenomenon: The more a given stimulus configuration resembles the stimulus conditions present during instruction, the greater the probability that the trained response will be emitted, and vice versa. A generalization setting that differs sig- nificantly from the instructional setting may not provide sufficient stimulus control over the target behavior. Such a setting may also contain stimuli that impede the target behavior because their novelty confuses or startles the learner. Exposing the learner to stimuli during instruc- tion that are commonly found in the generalization set- ting increases the likelihood that those stimuli will acquire some stimulus control over the target behavior and also prepares the learner for the presence of stimuli in the generalization setting that have the potential of impeding performance. Two tactics used by applied be- havior analysts to implement this basic strategy are pro- gramming common stimuli and teaching loosely.

Program Common Stimuli

Programming common stimuli means including typical features of the generalization setting into the instructional setting. Although behavior analysts have attached a spe- cial term to this tactic, successful practitioners in many fields have long used this technique for promoting gen- eralized behavior change. For example, coaches, music teachers, and theater directors hold scrimmages, mock auditions, and dress rehearsals to prepare their athletes, musicians, and actors to perform important skills in set- tings that include the sights, sounds, materials, people,

Generalization and Maintenance of Behavior Change

640

and procedures that simulate as closely as possible those in the “real world.”

Van den Pol and colleagues (1981) programmed common stimuli when they taught three young adults with disabilities how to order and eat in fast-food restau- rants. The researchers used numerous items and photos from actual restaurants to make the classroom simulate the conditions found in actual restaurants. Plastic signs with pictures and names of various McDonald’s sand- wiches were posted on the classroom wall, a table was transformed into a mock “counter” for role-playing trans- actions, and the students practiced responding to 60 pho- tographic slides taken in actual restaurants showing both positive and negative (“don’t do it”) examples of situa- tions customers are likely to encounter.

Why go to all the trouble of simulating the gener- alization setting? Why not just conduct instruction in the generalization setting itself to ensure that the learner experiences all of the relevant aspects of the setting? First, conducting instruction in natural settings is not al- ways possible or practical. Lots of resources and time may be necessary to transport students to community- based settings.

Second, community-based training may not expose students to the full range of examples they are likely to encounter later in the same setting. For example, students who receive in situ instruction for grocery shopping or street crossing during school hours may not experience the long lines at the checkout counters or heavy traffic patterns typical of evening hours.

Third, instruction in natural settings may be less ef- fective and efficient than classroom instruction because the trainer cannot halt the natural flow of events to con- trive an optimal number and sequence of training trials needed (e.g., Neef, Lensbower, Hockersmith, DePalma, & Gray, 1990).

Fourth, instruction in simulated settings can be safer, particularly with target behaviors that must be performed in potentially dangerous environments or that have se- vere consequences if performed incorrectly (e.g., Mil- tenberger et al., 2005), or when children or people with learning problems must perform complex procedures. If the procedures involve invading the body or errors dur- ing practice are potentially hazardous, simulation train- ing should be used. For example, Neef, Parrish, Hannigan, Page, and Iwata (1990) had children with neu- rogenic bladder complications practice performing self- catheterization skills on dolls.

Programming common stimuli is a straightforward two-step process of (a) identifying salient stimuli that characterize the generalization setting(s) and (b) incor- porating those stimuli into the instructional setting. A practitioner can identify possible stimuli in the general- ization setting to make common by direct observation or

Generalization and Maintenance of Behavior Change

by asking people familiar with the setting. Practitioners should conduct observations in the generalization set- ting(s) and write down prominent features of the envi- ronment that might be important to include during training. When direct observation is not feasible, practi- tioners can obtain secondhand knowledge of the setting by interviewing or giving checklists to people who have firsthand knowledge of the generalization setting—those who live, work in, or are otherwise familiar with the gen- eralization setting(s) in question.

If a generalization setting includes important stimuli that cannot be recreated or simulated in the instructional setting, then at least some training trials must be con- ducted in the generalization setting. However, as pointed out previously, practitioners should not assume that community-based instruction will guarantee students’ ex- posure to all of the important stimuli common to the gen- eralization setting.

Teach Loosely

Applied behavior analysts control and standardize inter- vention procedures to maximize their direct effects, and so the effects of their interventions can be interpreted and replicated by others. Yet restricting teaching procedures to a “precisely repetitive handful of stimuli or formats may, in fact, correspondingly restrict generalization of the lessons being learned” (Stokes & Baer, 1977, p. 358). To the extent that generalized behavior change can be viewed as the opposite of strict stimulus control and dis- crimination, one technique for facilitating generalization is to vary as many of the noncritical dimensions of the an- tecedent stimuli as possible during instruction.

Teaching loosely, randomly varying noncritical as- pects of the instructional setting within and across teach- ing sessions, has two advantages or rationales for promoting generalization. First, teaching loosely reduces the likelihood that a single or small group of noncritical stimuli will acquire exclusive control over the target be- havior. A target behavior that inadvertently comes under the control of a stimulus present in the instructional set- ting but not always present in the generalization setting may not be emitted in the generalization setting. Here are two examples of this type of faulty stimulus control:

• Following teachers’ directions: A student with a history of receiving reinforcement for complying with teachers’ directions when they are given in a loud voice and accompanied by a stern facial ex- pression may not follow directions that do not con- tain one or both of those noncritical variables. The discriminative stimulus for the student’s compli- ance with teacher directions should be the content of the teacher’s statements.

641

• Assembling bicycle sprocket sets: A new employee at the bicycle factory inadvertently learns to assem- ble rear sprocket sets by putting a red sprocket on top of a green sprocket and a green sprocket on top of a blue sprocket because the sprocket sets on a particular bicycle model in production on the day she was trained were colored in that fashion. How- ever, proper assembly of a sprocket set has nothing to do with the colors of the individual sprockets; the relevant variable is the relative size of the sprockets (i.e., the biggest sprocket goes on the bottom, the next biggest on top of that one, and so on).

Systematically varying the presence and absence of noncritical stimuli during instruction greatly decreases the chance that a functionally irrelevant factor, such as a teacher’s tone of voice or sprocket color in these two ex- amples, will acquire control of the target behavior (Kirby & Bickel, 1988).

A second rationale for loose teaching is that including a wide variety of noncritical stimuli during instruction in- creases the probability that the generalization setting will in- clude at least some of the stimuli that were present during instruction. In this sense, loose teaching acts as a kind of catchall effort at programming common stimuli and makes it less likely that a student’s performance will be impeded or “thrown off” by the presence of a “strange” stimulus.

Loose teaching applied to the previous two exam- ples might entail the following:

• Following teachers’ directions: During instruction the teacher varies all of the factors mentioned ear- lier (e.g., tone of voice, facial expression), plus gives directions while standing, while sitting, from different places within the classroom, at different times of the day, while the student is alone and in groups, while looking away from the student, and so on. In each instance, reinforcement is contin- gent on the student’s compliance with the content of the teacher’s direction irrespective of the pres- ence or absence of any of the noncritical features.

• Assembling bicycle sprocket sets: During training the new employee assembles sprocket sets contain- ing sprockets of widely varying colors, after receiv- ing the component sprockets in varied sequences, when the factory floor is busy, at different times during a work shift, with and without music play- ing, and so forth. Irrespective of the presence, ab- sence, or values of any of these noncritical factors, reinforcement would be contingent on correct as- sembly of sprockets by relative size.

Seldom used as a stand-alone tactic, loose teaching is often a recognizable component of interventions when

generalization to highly variable and diverse settings or situations is desired. For example, Horner and colleagues (1986) incorporated loose teaching into their training pro- gram for table busing by systematically but randomly varying the location of the tables, the number of people at the tables, whether food was completely or partially eaten, the amount and location of garbage, and so forth. Hughes and colleagues (1995) incorporated loose teach- ing by varying the peer teachers and varying the loca- tions of training sessions. Loose teaching is often a recognizable feature of language training programs that use milieu, incidental, and naturalistic teaching methods (e.g., Charlop-Christy & Carpenter, 2000; McGee, Mor- rier, & Daly, 1999; Warner, 1992).

Few studies evaluating the effects of using loose teaching in isolation have been reported. One exception is an experiment by Campbell and Stremel-Campbell (1982), who evaluated the effectiveness of loose teaching as a tactic for facilitating the generalization of newly ac- quired language by two students with moderate mental re- tardation. The students were taught the correct use of the words is and are in “wh” questions (e.g., “What are you doing?”), yes/no reversal questions (e.g., “Is this mine?”), and statements (e.g., “These are mine?”). Each student re- ceived two 15-minute language training sessions con- ducted within the context of other instructional activities that were part of each child’s individualized education program, one during an academic task and the second during a self-help task. The student could initiate a lan- guage interaction based on the wide variety of naturally occurring stimuli, and the teacher could try to evoke a statement or question from the student by intentionally misplacing instructional materials or offering indirect prompts. Generalization probes of the students’ language use during two daily 15-minute free-play periods revealed substantial generalization of the language structures ac- quired during the loose teaching sessions.

The learner’s performance of the target behavior should be established under fairly restricted, simplified, and consistent conditions, before much “looseness” is in- troduced. This is particularly important when teaching complex and difficult skills. Only noncritical (i.e., func- tionally irrelevant) stimuli should be “loosened.” Practi- tioners should not inadvertently loosen stimuli that reliably function in the generalization setting as discrim- inative stimuli (SDs) or as “don’t do it” examples (S�s). Stimuli known to play important roles in signaling when and when not to respond should be systematically incor- porated into instructional programs as teaching examples. A stimulus condition that may be functionally irrelevant for one skill may be a critical SD for another skill.

Taking the notion of varying noncritical aspects of the instructional setting and procedures to its logical

Generalization and Maintenance of Behavior Change

642

limit, Baer (1999) offered the following advice for loose teaching:

• Use two or more teachers.

• Teach in two or more places.

• Teach from a variety of positions.

• Vary your tone of voice.

• Vary your choice of words.

• Show the stimuli from a variety of angles, using sometimes one hand and sometimes the other.

• Have other persons present sometimes and not other times.

• Dress quite differently on different days.

• Vary the reinforcers.

• Teach sometimes in bright light, sometimes in dim light.

• Teach sometimes in noisy settings, sometimes in quiet ones.

• In any setting, vary the decorations, vary the furni- ture, and vary their locations.

• Vary the times of day when you and everyone else teach.

• Vary the temperature in the teaching settings.

• Vary the smells in the teaching settings.

• Within the limits possible vary the content of what’s being taught.

• Do all of this as often and as unpredictably as pos- sible. (p. 24)

Of course, Baer (1999) was not suggesting that a teacher needs to vary all of these factors for every be- havior taught. But building a reasonable degree of ”loose- ness” into teaching is an important element of a teacher’s overall effort to program for generalization rather than train and hope.

Maximize Contact with Reinforcement in the Generalization Setting

Even though a practitioner is successful in getting the learner to emit a newly acquired target behavior in a gen- eralization setting with a naturally existing contingency of reinforcement, generalization and maintenance may be short-lived if the behavior makes insufficient contact with reinforcement. In such cases the practitioner’s ef- forts to promote generalization revolves around ensuring that the target behavior contacts reinforcement in the gen- eralization setting. Five of the 13 tactics for promoting

Generalization and Maintenance of Behavior Change

generalized behavior change described in this chapter in- volve some form of arranging or contriving for the target behavior to be reinforced in the generalization setting.

Teach Behavior to Levels Required by Natural Contingencies

Baer (1999) suggested that a common mistake practi- tioners make when attempting to employ natural contin- gencies of reinforcement is failing to teach the behavior change well enough so that it contacts the contingency.

Sometimes behavior changes that seem to need general- ization may only need better teaching. Try making the students fluent, and see if they still need further support for generalization. Fluency may consist of any or all of the following: high rate of performance, high accuracy of performance, fast latency, given the opportunity to re- spond, and strong response. (p. 17)

A new behavior may occur in the generalization set- ting but fail to make contact with the naturally existing contingencies of reinforcement. Common variables that diminish contact with reinforcement in the generaliza- tion setting include the accuracy of the behavior, the di- mensional quality of the behavior (i.e., frequency, duration, latency, magnitude), and the form (topography) of the behavior. The practitioner may need to enhance the learner’s performance in one or more of these vari- ables to ensure that the new behavior will meet the nat- urally existing contingencies of reinforcement. For example, when given a worksheet to complete at his desk, a student’s behavior that is consistent with the following dimensions is unlikely to contact reinforcement for com- pleting the task, even if the student has the ability to com- plete each worksheet item accurately.

• Latency too long. A student who spends 5 minutes “daydreaming” before he begins reading the direc- tions may not finish in time to obtain reinforcement.

• Rate too low. A student who needs 5 minutes to read the directions for an independent seatwork as- signment that his peers read in less than 1 minute may not finish in time to obtain reinforcement.

• Duration too brief. A student who can work with- out direct supervision for only 5 minutes at a time will not be able to complete any task requiring more than 5 minutes of independent work.

The solution for this kind of generalization problem, if not always simple, is straightforward. The behavior change must be made more fluent: The learner must be taught to emit the target behavior at a rate commensurate with the naturally occurring contingency, with more

643

accuracy, within a shorter latency, and/or at a greater mag- nitude. Generalization planning should include identifi- cation of the levels of performance necessary to access existing criteria for reinforcement.

Program Indiscriminable Contingencies

Applied behavior analysts purposely design and imple- ment interventions so the learner receives consistent and immediate consequences for emitting the target behav- ior. Although consistent and immediate consequences are often necessary to help the learner acquire new behav- ior, those very contingencies can impede generalization and maintenance. The clear, predictable, and immediate consequences that are typically part of systematic in- struction can actually work against generalized respond- ing. This is most likely to occur when a newly acquired skill has not yet contacted naturally existing contingen- cies of reinforcement, and the learner can discriminate when the instructional contingencies are absent in the generalization settings. If the presence or absence of the controlling contingencies in the generalization setting is obvious or predictable to the learner (“Hey, the game’s off. There’s no need to respond here/now”), the learner may stop responding in the generalization setting, and the behavior change the practitioner worked so hard to develop may cease to occur before it can contact the nat- urally existing contingency of reinforcement.

An indiscriminable contingency is one in which the learner cannot discriminate whether the next response will produce reinforcement. As a tactic for promoting generalization and maintenance, programming indis- criminable contingencies involves contriving a contin- gency in which (a) reinforcement is contingent on some, but not all, occurrences of the target behavior in the gen- eralization setting, and (b) the learner is unable to pre- dict which responses will produce reinforcement.

The basic rationale for programming indiscriminable contingencies is to keep the learner responding often enough and long enough in the generalization setting for the target behavior to make sufficient contact with the naturally existing contingencies of reinforcement. From that point on, the need to program contrived contingen- cies to promote generalization will be moot. Applied be- havior analysts use two related techniques to program indiscriminable contingencies: intermittent schedules of reinforcement and delayed rewards.

Intermittent Schedules of Reinforcement. A newly learned behavior often must occur repeatedly over a period of time in the generalization setting be- fore it contacts a naturally existing contingency of rein- forcement. During that time, an extinction condition

exists for responses emitted in the generalization set- ting. The current or most recent schedule of reinforce- ment for a behavior in the instructional setting plays a significant role in how many responses will be emitted in the generalization setting prior to reinforcement. Be- haviors that have been under continuous schedules of reinforcement (CRF) show very limited response main- tenance under extinction. When reinforcement is no longer available, responding is likely to decrease rapidly to prereinforcement levels. On the other hand, behaviors with a history of intermittent schedules of reinforcement often continue to be emitted for relatively long periods of time after reinforcement is no longer available (e.g., Dunlap & Johnson, 1985; Hoch, McComas, Thompson, & Paone, 2002).

An experiment by Koegel and Rincover (1977, Ex- periment II) showed the effects of intermittent schedules of reinforcement on response maintenance in a general- ization setting. The participants were six boys diagnosed with autism and severe to profound mental retardation, ages 7 to 12 years, who had participated in a previous study on generalization and had showed generalized re- sponding in the extra-therapy setting used in that exper- iment (Rincover & Koegel, 1975). As in Experiment I by Koegel and Rincover (1977) described earlier in this chapter, one-on-one training trials were conducted with each child and the trainer seated at a table in a small room, and generalization trials were conducted by an un- familiar adult standing outside on the lawn, surrounded by trees. Two types of imitative response class consisted of (a) nonverbal imitation (e.g., raising arm) in response to an imitative model and the verbal instruction, “Do this” and (b) touching a body part in response to verbal in- structions such as, “Touch your nose.” After acquiring an imitation response, each child was given additional trials on one of three randomly chosen schedules of reinforce- ment: CRF, FR 2, or FR 5. Only after these additional training trials were the children taken outside to assess re- sponse maintenance. Once outside, trials were conducted until the child’s correct responding had decreased to 0%, or was maintained at 80% correct or above for 100 con- secutive trials.

Behaviors that were most recently on a CRF sched- ule in the instructional setting underwent extinction quickly in the generalization setting (see Figure 7). Gen- eralized responding occurred longer for the FR 2 trained behavior, and longer still for behavior that had been shifted to an FR 5 schedule in the instructional setting. The results showed clearly that the schedule of rein- forcement in the instructional setting had a predictable effect on responding in the absence of reinforcement in the generalization setting: The thinner the schedule in the instructional setting, the longer the response maintenance in the generalization setting.

Generalization and Maintenance of Behavior Change

644

100

P er

ce n

t C

or re

ct R

es p

on d

in g

CRF Child 4

FR 2 Child 5

FR 2 Child 6

(Same Child)

FR 5FR 5

Child 6

Child 7

50 100 200 300 400 500 Trials

Figure 7 Percent of correct responses by three children in a generalization setting as a function of the schedule of reinforcement used during the final sessions in an instruc- tional setting. From “Research on the Differences between Generalization and Maintenance in Extra-Therapy Responding” by R. L. Koegel and A. Rincover, 1977, Journal of Applied Behavior Analysis, 10, p. 8. Copyright 1977 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

The defining feature of all intermittent schedules of reinforcement is that only some responses are reinforced, which means, of course, that some responses go unrein- forced. Thus, one possible explanation for the mainte- nance of responding during periods of extinction for behaviors developed under intermittent schedules is the relative difficulty of discriminating that reinforcement is no longer available. Thus, the unpredictability of an in- termittent schedule may account for the maintenance of behavior after the schedule is terminated.

Practitioners should recognize that although all in- discriminable contingencies of reinforcement involve in- termittent schedules, not all schedules of intermittent reinforcement are indiscriminable. For example, although the FR 2 and FR 5 schedules of reinforcement used by Koegel and Rincover (1977) were intermittent, many learners would soon be able to discriminate whether re- inforcement would follow their next response. In con- trast, a student whose behavior is being supported by a VR 5 schedule of reinforcement cannot determine whether his next response will be reinforced.

Delayed Rewards. Stokes and Baer (1977) sug- gested that not being able to discriminate in what set- tings a behavior will be reinforced is similar to not being able to discriminate whether the next response will be reinforced. They cited an experiment by Schwarz and Hawkins (1970) in which each day after school a sixth- grade girl was shown videotapes of her behavior in that

Generalization and Maintenance of Behavior Change

day’s math class and received praise and token rein- forcement for improvements in her posture, reducing the number of times she touched her face, and speaking with sufficient volume to be heard by others. Reinforce- ment after school was contingent on behaviors emitted during math class only, but comparable improvements were noted in spelling class as well. The generalization data were taken from videotapes that were made of the girl’s behavior in spelling class but were never shown to her. Stokes and Baer hypothesized that because rein- forcement was delayed (the behaviors that produced praise and tokens were emitted during math class but were not rewarded until after school), it may have been difficult for the student to discriminate when improved performance was required for reinforcement. They sug- gested that the generalization across settings of the target behaviors may have been a result of the indiscriminable nature of the response-to-reinforcement delay.

Delayed rewards and intermittent schedules of rein- forcement are alike in two ways: (a) Reinforcement is not delivered each time the target behavior is emitted (only some responses are followed by reinforcement), and (b) there are no clear stimuli to signal the learner which current responses will produce reinforcement. A delayed reward contingency differs from intermittent re- inforcement in that instead of delivering the consequence immediately following an occurrence of the target be- havior, the reward is provided after a period of time has elapsed (i.e., a response-to-reward delay). Receiving the

645

delayed reward is contingent on the learner having per- formed the target behavior in the generalization setting during an earlier time period. With an effective delayed reward contingency, the learner cannot discriminate when (or where, depending on the details of the contingency) the target behavior must be emitted in order to receive reinforcement. As a result, to have the best chance to re- ceive the reward later, the learner must “be good all day” (Fowler & Baer, 1981).

Two similar studies by Freeland and Noell (1999, 2002) investigated the effects of delayed rewards on the maintenance of students’ mathematics performance. Par- ticipants in the second study were two third-grade girls who had been referred by their teacher for help with mathematics. That target behavior for both students was writing answers to single-digit addition problems with sums to 18. The researchers used a multiple-treatment reversal design to compare the effects of five conditions on the number of correct digits written as answers to sin- gle-digit addition problems during daily 5-minute work periods (e.g., writing “11” as the answer to “5 + 6 = ?” counted as two digits correct).

• Baseline: Green worksheets; no programmed con- sequences; students were told they could attempt as many or as few problems as they wanted.

• Reinforcement: Blue worksheets with a goal num- ber at the top indicating the number of correct dig- its needed to choose a reward in the “goody box”; each student’s goal number was the median num- ber of correct digits on the last three worksheets; all worksheets were graded after each session.

• Delay 2: White worksheets with goal number; after every two sessions, one of the two worksheets completed by each student was randomly selected for grading; reinforcement was contingent on meeting highest median of three consecutive ses- sions up to that point in the study.

• Delay 4: White worksheets with goal number and same procedures as Delay 2 except that worksheets were not graded until four sessions had been completed, at which time one of each student’s previous four worksheets was randomly selected and graded.

• Maintenance: White worksheets with goal number as before; no worksheets were graded and no feed- back or rewards for performance were given.

The fact that different colored worksheets were used for each condition in this study made it easy for the stu- dents to predict the likelihood of reinforcement. A green worksheet meant no “goody bag”—and no feedback at all—no matter how many correct digits were written. However, meeting one’s performance criterion on a white

worksheet sometimes produced reinforcement. This study provides powerful evidence of the importance of having contingencies in the instructional setting “look like” the contingencies in effect in the generalization setting(s) in two ways: (a) Both students showed large decreases in performance when baseline conditions were reinstated, and immediate drops during a second return to baseline; and (b) the students continued completing math prob- lems at a high rate during the maintenance condition, even though no reinforcement was provided. (See Fig- ure 8.)

When the delayed (indiscriminable) contingencies were implemented, all students demonstrated levels of correct responding at or above levels during the rein- forcement phase. When the students were exposed to maintenance conditions, Amy maintained high levels of responding for 18 sessions with variable performance over the final six sessions, and Kristen showed a gradu- ally increasing rate of performance over 24 sessions. The results demonstrated that behavior with an indiscrim- inable contingency can be maintained at the same rate as with a predictable schedule, and with greater resistance to extinction.

Delayed consequences have been used to promote the setting/situation generalization and response mainte- nance of a wide range of target behaviors, including aca- demic and vocational tasks by individuals with autism (Dunlap, Koegel, Johnson, & O’Neill, 1987), young chil- dren’s toy play, social initiations, and selection of healthy snacks (R. A. Baer, Blount, Dietrich, & Stokes, 1987; R. A. Baer, Williams, Osnes, & Stokes, 1984; Osnes, Guevremont, & Stokes, 1986), restaurant trainees’ re- sponding appropriately to coworkers’ initiations (Grossi et al., 1994); and performance on reading and writing tasks (Brame, 2001; Heward, Heron, Gardner, & Prayzer, 1991).

The effective use of delayed consequences can re- duce (or even eliminate in some instances) the learner’s ability to discriminate when a contingency is or is not in effect. As a result, the learner needs to “be good” (i.e., emit the target behavior) all the time. If an effective con- tingency can be made indiscriminable across settings and target behaviors, the learner will also have to “be good” everywhere, with all of his or her relevant skills.

Following are four examples of classroom applica- tions of indiscriminable contingencies involving delayed rewards. Each of these examples also features an inter- dependent group contingency by making rewards for the whole class contingent on the performance of randomly selected students.

• Spinners and dice. A procedure such as the follow- ing can make academic seatwork periods more ef- fective. Every few minutes (e.g., on a VI 5-minute

Generalization and Maintenance of Behavior Change

646

D ig

it s

C or

re ct

p er

M in

u te

BL RF BL RF D2 BL D2 D4 Maintenance

Amy

10 20 30 40 50 60 70 Sessions

D ig

it s

C or

re ct

p er

M in

u te

BL RF BL RF D2 BL D2 D4 Maintenance

Kristen

10 20 30 40 50 60 Sessions

Digits Correct

Goal

Digits Correct

Goal

Figure 8 Number of correct digits per minute by two third-grade students while answering math problems during base- line (BL), reinforcement contingent on performance on one randomly selected worksheet after each session (RF), rein- forcement contingent on performance on one randomly selected worksheet after every two (D2) or four (D4) sessions, and maintenance conditions. From “Programming for Maintenance: An Investigation of Delayed Intermittent Reinforcement and Common Stimuli to Create Indiscriminable Contingencies” by J. T. Freeland and G. H. Noell, 2002, Journal of Behavioral Education, 11, p. 13. Copyright 2002 by Human Sciences Press. Reprinted by permission.

schedule), the teacher (a) randomly selects a stu- dent’s name, (b) walks to that student’s desk and has the student spin a spinner or roll a pair of dice, (c) counts backward from the worksheet problem or item the student is currently working on by the number shown on the spinner or dice, and (d) gives a token to the student if that problem or item is correct. Students who immediately begin to work on the assignment and work quickly but carefully throughout the seatwork period are most likely to obtain reinforcers under this indiscriminable contingency.

• Story fact recall game. Many teachers devote 20 to 30 minutes per day to sustained silent reading (SSR), a period when students can read silently from books of their choice. A story fact recall game can encourage students to read with a pur- pose during SSR. At the end of the SSR period the teacher asks several randomly selected students a question about the book they are reading. For ex-

Generalization and Maintenance of Behavior Change

ample, a student who is reading Chapter 3 of Eliza- beth Winthrop’s The Castle in the Attic might be asked, What did William give the Silver Knight to eat? (Answer: bacon and toast). A correct answer is praised by the teacher, applauded by the class, and earns a marble in a jar toward a reward for the whole class. Students do not know when they will be called on or what they might be asked (Brame, Bicard, Heward, & Greulich, 2007).

• Numbered heads together. Collaborative learning groups (small groups of students working together on a joint learning activity) can be effective, but teachers should use procedures that motivate all students to participate. A technique called num- bered heads together can ensure that all students actively participate (Maheady, Mallete, Harper, & Saca, 1991). Students are seated in heterogeneous groups of three or four, and each student is given the number 1, 2, 3, or 4. The teacher asks the class a question, and each group discusses the problem

647

and comes up with an answer. Next, the teacher randomly selects a number from 1 to 4 and then calls on one or more students with that number to answer. It is important that every person in the group knows the answer to the question. This strat- egy promotes cooperation within the group rather than competition. Because all students must know the answer, group members help each other under- stand not only the answer, but also the how and why behind it. Finally, this strategy encourages in- dividual responsibility.

• Intermittent grading. Most students do not receive sufficient practice writing, and when students do write, the feedback they receive is often ineffective. One reason for this may be that providing detailed feedback on a daily composition by each student in the class requires more time and effort than even the most dedicated teacher can give. A procedure called intermittent grading offers one solution to this problem (Heward, Heron, Gardner, & Prayzer, 1991). Students write for 10 to 15 minutes each day, but instead of reading and evaluating every student’s paper, the teacher provides detailed feed- back on a randomly selected 20 to 25% of students’ daily compositions. The students whose papers were graded earn points based on individualized performance criteria, and bonus points are given to the class contingent on the quality of the selected and graded papers (e.g., if the authors of four of the five papers that were graded met their individ- ual criteria). Students’ papers that were graded can be used as a source of instructional examples for the next lesson.

The success of a delayed reward tactic in promoting generalization and maintenance rests on (a) the indis- criminability of the contingency (i.e., the learner cannot tell exactly when emitting the target behavior in the gen- eralization setting will produce a reward at a later time), and (b) the learner understanding the relation between his emitting the target behavior at an earlier time and re- ceiving a reward later. A delayed rewards intervention may not be effective with some learners with severe cog- nitive disabilities.

Guidelines for Programming Indiscriminable Contingencies. Practioners should consider these guidelines when implementing indiscriminable con- tingencies:

• Use continuous reinforcement during the initial stages of acquiring new behaviors or when strengthening little-used behaviors.

• Systematically thin the schedule of reinforcement based on the learner’s performance. Remember

that the thinner the schedule of reinforcement is, the more indiscriminable it is (e.g., an FR 5 sched- ule is more indiscriminable than an FR 2 sched- ule); and variable schedules of reinforcement (e.g., VR and VI schedules) are more indiscriminable than fixed schedules are (e.g., FR and FI schedules).

• When using delayed rewards, begin by delivering the reinforcer immediately following the target be- havior and gradually increase the response-to- reinforcement delay.

• Each time a delayed reward is delivered, explain to the learner that he is receiving the reward for spe- cific behaviors he performed earlier. This helps to build and strengthen the learner’s understanding of the rule describing the contingency.

When selecting reinforcers to use during instruction, practitioners should try to use or eventually shift to the same reinforcers that the learner will acquire in the gen- eralization environment. The reinforcer itself may serve as a discriminative stimulus for the target behavior (e.g., Koegel & Rincover, 1977)

Set Behavior Traps

Some contingencies of reinforcement are especially pow- erful, producing substantial and long-lasting behavior changes. Baer and Wolf (1970) called such contingen- cies behavior traps. Using a mouse trap as an analogy, they described how a householder has only to exert a rel- atively small amount of behavioral control over the mouse—getting the mouse to smell the cheese—to pro- duce a behavior change with considerable (in this case, complete) generalization and maintenance.

A householder without a trap can, of course, still kill a mouse. He can wait patiently outside the mouse’s hole, grab the mouse faster than the mouse can avoid him, and then apply various forms of force to the unfortunate animal to accomplish the behavioral change desired. But this performance requires a great deal of competence: vast patience, super-coordination, extreme manual dexterity, and a well-suppressed squeamishness. By contrast, a householder with a trap need very few ac- complishments: If he can merely apply the cheese and then leave the trap where the mouse is likely to smell that cheese, in effect he has guaranteed general(ized) change in the mouse’s future behavior.

The essence of a trap, in behavioral terms, is that only a relatively simple response is necessary to enter the trap, yet once entered, the trap cannot be resisted in creating general behavior change. For the mouse, the entry response is merely to smell the cheese. Everything

Generalization and Maintenance of Behavior Change

648

proceeds from there almost automatically. (p. 321, em- phasis added)

Behavioral trapping is a fairly common phenomenon that everyone experiences from time to time. Behavior traps are particularly evident in the activities we “just can- not get (or do) enough of.” The most effective behavior traps share four essential features: (a) They are “baited” with virtually irresistible reinforcers that “lure” the student to the trap; (b) only a low-effort response already in the student’s repertoire is necessary to enter the trap; (c) in- terrelated contingencies of reinforcement inside the trap motivate the student to acquire, extend, and maintain tar- geted academic and/or social skills (Kohler & Green- wood, 1986); and (d) they can remain effective for a long time because students show few, if any, satiation effects.

Consider the case of the “reluctant bowler.” A young man is persuaded to fill in as a substitute for a friend’s bowling team. He has always regarded bowling as uncool. And bowling looks so easy on television that he does not see how it can be considered a real sport. Nevertheless, he agrees to go, just to help out this one time. During the evening he learns that bowling is not as easy as he had al- ways assumed (he has a history of reinforcement for ath- letic challenges) and that several people he would like to get to know are avid bowlers (i.e., it is a mixed doubles league). Within a week he has purchased a custom-fitted bowling ball, a bag, and shoes; practiced twice on his own; and signed up for the next league season.

The reluctant bowler example illustrates the funda- mental nature of behavior traps: easy to enter and diffi- cult to exit. Some naturally existing behavior traps can lead to maladaptive behaviors, such as alcoholism, drug addiction, and juvenile delinquency. The everyday term vicious circle refers to the natural contingencies of rein- forcement that operate in destructive behavior traps. Prac- titioners, however, can learn to create behavior traps that help students develop positive, constructive knowledge and skills. Alber and Heward (1996), who provided guidelines for creating successful traps, gave the follow- ing example of an elementary teacher creating a behav- ior trap that took advantage of a student’s penchant for playing with baseball cards.

Like many fifth graders struggling with reading and math, Carlos experiences school as tedious and unre- warding. With few friends of his own, Carlos finds that even recess offers little reprieve. But he does find solace in his baseball cards, often studying, sorting and playing with them in class. His teacher, Ms. Greene, long ago lost count of the number of times she had to stop an in- structional activity to separate Carlos and his beloved baseball cards. Then one day, when she approached Car- los’ desk to confiscate his cards in the middle of a lesson on alphabetization, Ms. Greene discovered that Carlos had already alphabetized all the left-handed pitchers in

Generalization and Maintenance of Behavior Change

the National League! Ms. Greene realized she’d found the secret to sparking Carlos’ academic development.

Carlos was both astonished and thrilled to learn that Ms. Greene not only let him keep his baseball cards at his desk, but also encouraged him to “play with them” during class. Before long, Ms. Greene had incorporated baseball cards into learning activities across the curricu- lum. In math, Carlos calculated batting averages; in geography, he located the hometown of every major leaguer born in his state; and in language arts, he wrote letters to his favorite players requesting an autographed photo. Carlos began to make significant gains academi- cally and an improvement in his attitude about school was also apparent.

But school became really fun for Carlos when some of his classmates began to take an interest in his knowl- edge of baseball cards and all the wonderful things you could do with them. Ms. Greene helped Carlos form a classroom Baseball Card Club, giving him and his new friends opportunities to develop and practice new social skills as they responded to their teacher’s challenge to think of new ways to integrate the cards into the curricu- lum. (p. 285)

Ask People in the Generalization Setting to Reinforce the Behavior

The problem may be simply that the natural community [of reinforcement] is asleep and needs to be awakened and turned on.

—Donald M. Baer (1999, p. 16).

Sometimes a potentially effective contingency of rein- forcement in the generalization setting is not operating in a form available to the learner no matter how often or well she performs the target behavior. The contingency is there, but dormant. One solution for this kind of prob- lem is to inform key people in the generalization setting of the value and importance of their attention to the learner’s efforts to acquire and use new skills and ask them to help.

For example, a special education teacher who has been helping a student learn how to participate in class discussions by providing repeated opportunities for prac- tice and feedback in the resource room could inform the teachers in the general education classes that the student attends about the behavior change program and ask them to look for and reinforce any reasonable effort by the stu- dent to participate in their classrooms. A small amount of contingent attention and praise from these teachers may be all that is needed to achieve the desired generalization of the new skill.

This simple but often effective technique for pro- moting generalized behavior was evident in the study by Stokes and colleagues (1974) in which staff members re- sponded to the children’s waving response by saying,

649

“Hello, (name).” Approximately 20 such generalization probes were conducted each day with each child.

Williams, Donley, and Keller (2000) gave mothers of two preschool children with autism explicit instruc- tions on providing models, response prompts, and rein- forcement to their children who were learning to ask questions about hidden objects (e.g., “What’s that?”, “Can I see it?”).

Contingent praise and attention from significant oth- ers can add to the effectiveness of other strategies already in place in the generalization environment. Broden, Hall, and Mitts (1971) showed this in the self-monitoring study. After self-recording had improved eighth-grader Liza’s study behavior during history class, the re- searchers asked Liza’s teacher to praise Liza for study behavior whenever he could during history class. Liza’s level of study behavior during this Self-Recording Plus Praise condition increased to a mean of 88% and was maintained at levels nearly as high during a subsequent Praise Only condition (see Figure 3).

Teach the Learner to Recruit Reinforcement

Another way to “wake up” a potentially powerful but dor- mant natural contingency of reinforcement is to teach the learner to recruit reinforcement from significant others. For example, Seymour and Stokes (1976) taught delin- quent girls to work more productively in the vocational training area of a residential institution. However, obser- vation showed that staff at the institution gave the girls no praise or positive interaction regardless of the quality of their work. The much-needed natural community of re- inforcement to ensure the generalization of the girls’ im- proved work behaviors was not functioning. To get around this difficulty, the experimenters trained the girls to use a simple response that called the attention of staff members to their work. With this strategy, staff praise for good work increased. Thus, teaching the girls an addi- tional response that could be used to recruit reinforce- ment enabled the target behavior to come into contact with natural reinforcers that would serve to extend and maintain the desired behavior change.

Students of various ages and abilities have learned to recruit teacher and peer attention for performing a wide range of tasks in classroom and community settings; preschoolers with developmental delays for completing pre-academic tasks and staying on task during transitions (Connell, Carta, & Baer, 1993; Stokes, Fowler, & Baer, 1978) as well as students with learning disabilities (Alber, Heward, & Hippler, 1999; Wolford, Alber, & Heward, 2001), students with behavioral disorders (Morgan, Young, & Goldstein, 1983), students with mental retar-

dation performing academic tasks in regular classrooms (Craft, Alber, & Heward, 1998), and secondary students with mental retardation for improved work performance in vocational training settings (Mank & Horner, 1987).

Craft and colleagues (1998) assessed the effects of recruitment training on academic assignments for which students recruited teacher attention. Four elementary stu- dents were trained by their special education teacher (the first author) when, how, and how often to recruit teacher attention in the general education classroom. Training consisted of modeling, role-playing, error correction, and praise in the special education classroom. The students were taught to show their work to the teacher or ask for help two to three times per work page, and to use appro- priate statements such as “How am I doing?” or “Does this look right?”

Data on the frequency of student recruiting and teacher praise statements were collected during a daily 20-minute homeroom period in a general education class- room. During this period, the general education students completed a variety of independent seatwork tasks (read- ing, language arts, math) assigned by the general educa- tion teacher, while the four special education students completed spelling worksheets assigned by the special education teacher, an arrangement that had been estab- lished prior to the experiment. If students needed help with their assignments during homeroom, they took their work to the teacher’s desk and asked for help.

The effects of recruitment training on the children’s frequency of recruiting and on the number of praise state- ments they received from the classroom teacher are shown in Figure 9. Recruiting across students increased from a mean rate of 0.01 to 0.8 recruiting responses per 20-minute session during baseline to a mean rate of 1.8 to 2.7 after training. Teacher praise statements received by the students increased from a mean rate of 0.1 to 0.8 praise statements per session during baseline to a mean rate of 1.0 to 1.7 after training. The ultimate meaning and outcome of the intervention was in the increased amount and improved accuracy of all four of the students’ acad- emic work (see Figure 9).

For a review of research on recruiting and sugges- tions for teaching children to recruit reinforcement from significant others, see Alber and Heward (2000). Box 2, “Look, Teacher, I’m All Finished!” provides suggestions for teaching students to recruit teacher attention.

Mediate Generalization

Another strategy for promoting generalized behavior change is to arrange for some thing or person to act as a medium that ensures the transfer of the target behavior from the instructional setting to the generalization setting.

Generalization and Maintenance of Behavior Change

650

Generalization and Maintenance of Behavior Change

Classrooms are extremely busy places, and even the most conscientious teachers can easily overlook students’ im- portant academic and social behaviors. Research shows that teachers are more likely to pay attention to a dis- ruptive student than to one who is working quietly and productively (Walker, 1997). It is hard for teachers to be aware of students who need help, especially low- achieving students who are less likely to ask for help (Newman & Golding, 1990).

Although teachers in general education classrooms are expected to adapt instruction to serve students with disabilities, this is not always the case. Most secondary teachers interviewed by Schumm and colleagues (1995) believed that students with disabilities should take re- sponsibility for obtaining the help they need. Thus, know- ing how to politely recruit teacher attention and assistance can help students with disabilities function more inde- pendently and actively influence the quality of instruction they receive.

Who Should Be Taught to Recruit?

Although most students would probably benefit from learning to recruit teacher praise and feedback, here are some ideal candidates for recruitment training:

Withdrawn Willamena. Willamena seldom asks a teacher anything. Because she is so quiet and well behaved, her teachers sometimes forget she’s in the room.

In-a-Hurry Harry. Harry is usually half-done with a task before his teacher finishes explaining it. Racing through his work allows him to be the first to turn it in. But his work is often incomplete and filled with errors, so he doesn’t hear much praise from his teacher. Harry would benefit from re- cruitment training that includes self-checking and self-correction.

Shouting Shelly. Shelly has just finished her work, and she wants her teacher to look at it—right now! But Shelly doesn’t raise her hand. She gets her teacher’s attention—and disrupts most of her classmates—by shouting across the room. Shelly should be taught appropriate ways to solicit teacher attention.

Pestering Pete. Pete always raises his hand, waits quietly for his teacher to come to his desk, and then politely asks, “Have I done this right?” But he repeats this routine a dozen or more times in a

20-minute period, and his teachers find it annoy- ing. Positive teacher attention often turns into rep- rimands. Recruitment training for Pete will teach him to limit the number of times he cues his teach- ers for attention.

How to Get Started

1. Identify target behaviors. Students should recruit teacher attention for target behaviors that are val- ued and therefore likely to be reinforced, such as writing neatly and legibly, working accurately, completing assigned work, cleaning up at transi- tions, and making contributions when working in a cooperative group.

2. Teach self-assessment. Students should self-assess their work before recruiting teacher attention (e.g., Sue asks herself, “Is my work complete?”). After the student can reliably distinguish between com- plete and incomplete work samples, she can learn how to check the accuracy of her work with an- swer keys or checklists of the steps or components of the academic skill, or how to spot-check two or three items before asking the teacher to look at it.

3. Teach appropriate recruiting. Teach students when, how, and how often to recruit and how to respond to the teacher after receiving attention.

• When? Students should signal for teacher atten- tion after they have completed and self-checked a substantial part of their work. Students should also be taught when not to try to get their teacher’s attention (e.g., when the teacher is working with another student, talking to another adult, or taking the lunch count).

• How? The traditional hand raise should be part of every student’s recruiting repertoire. Other methods of gaining attention should be taught depending on teacher preferences and routines in the general education classroom (e.g., have students signal they need help by standing up a small flag on their desks; expect students to bring their work to the teacher’s desk for help and feedback).

• How often? While helping Withdrawn Willa- mena learn to seek teacher attention, don’t turn her into a Pestering Pete. How often a student should recruit varies across teachers and activi- ties (e.g., independent seatwork, cooperative

Box 2 “Look, Teacher, I’m All Finished!”

Teaching Students to Recruit Teacher Attention

651

Generalization and Maintenance of Behavior Change

learning groups, whole-class instruction). Direct observation in the classroom is the best way to establish an optimal rate of recruiting; it is also a good idea to ask the regular classroom teacher when, how, and with what frequency she prefers students to ask for help.

• What to say? Students should be taught several statements that are likely to evoke positive feed- back from the teacher (e.g., “Please look at my work.” “Did I do a good job?” “How am I doing?”). Keep it simple, but teach the student to vary her verbal cues so she will not sound like a parrot.

• How to respond? Students should respond to their teacher’s feedback by establishing eye con- tact, smiling, and saying, “Thank you.” Polite appreciation is very reinforcing to teachers and will increase the likelihood of more positive at- tention the next time.

4. Model and role-play the complete sequence. Begin by providing students with a rationale for recruiting (e.g., the teacher will be happy you did a good job, you will get more work done, your grades might improve). Thinking aloud while modeling is a good way to show the recruiting se- quence. While performing each step, say, “Okay, I’ve finished my work. Now I’m going to check it. Did I put my name on my paper? Yes. Did I do all

the problems? Yes. Did I follow all the steps? Yes. Okay, my teacher doesn’t look busy right now. I’ll raise my hand and wait quietly until she comes to my desk.” Have another student pretend to be the regular classroom teacher and come over to you when you have your hand up. Say, “Mr. Patterson, please look at my work.” The helper says, “Oh, you did a very nice job.” Then smile and say, “Thank you, Mr. Patterson.” Role-play with praise and offer corrective feedback until the student correctly performs the entire sequence on several consecutive trials.

5. Prepare students for alternate responses. Of course, not every recruiting student attempt will result in teacher praise; some recruiting responses may even be followed by criticism (e.g., “This is all wrong. Pay better attention the next time.”). Use role-playing to prepare students for these pos- sibilities and have them practice polite responses (e.g., “Thank you for helping me with this”).

6. Promote generalization to the regular classroom. The success of any recruitment training effort de- pends on the student actually using his or her new skill in the regular classroom.

Adapted from “Recruit it or lose it! Training students to recruit con- tingent teacher attention” by S. R. Alber and W. L. Heward, 1997, Inter- vention in School and Clinic, 5, pp. 275–282. Used with permission.

Two tactics for implementing this strategy are contriving a mediating stimulus and teaching the learner to mediate her own generalization through self-management.

Contrive a Mediating Stimulus

One tactic for mediating generalization is to bring the target behavior under the control of a stimulus in the in- structional setting that will function in the generalization setting to reliably prompt or aid the learner’s performance of the target behavior. The stimulus selected for this im- portant role may exist already in the generalization envi- ronment, or it may be a new stimulus added to the instruction program that subsequently goes with the learner to the generalization setting. Whether it is a nat- urally existing component of the generalization setting or an added element to the instructional setting, to effec- tively mediate generalization, a contrived mediating stimulus must be (a) made functional for the target be- havior during instruction and (b) transported easily to the generalization setting (Baer, 1999). The mediating stim- ulus is functional for the learner if it reliably prompts or

aids the learner in performing the target behavior; the mediating stimulus is transportable if it easily goes with the learner to all important generalization settings.

Naturally existing features of generalization settings that are used as contrived mediating stimuli may be phys- ical objects or people. Van den Pol and colleagues (1981) used paper napkins, a common feature of any fast-food restaurant, as a contrived mediating stimulus. They taught students that a paper napkin was the only place to put food. In this way, the researchers eliminated the added challenge and difficulty of teaching the students to dis- criminate clean tables from dirty tables, to sit only at clean tables, or to wipe dirty tables and then program- ming the generalization and maintenance of those be- haviors. By contriving the special use of the napkin, only one response had to be trained, and napkins then served as mediating stimulus for that behavior.

When choosing a stimulus to be made common to both teaching and social generalization setting(s), prac- titioners should consider using people. In addition to being a requisite feature of social settings, people are transportable and important sources of reinforcement for

652

Generalization and Maintenance of Behavior Change

5 10 15 20 25 30 35 40

Latasha Praise

Recruiting

Baseline Training

Continuous Reinforcement

Generalization Programming

Intermittent Reinforcement

Maintenance

5 10 15 20 25 30 35 40

Olivia

5 10 15 20 25 30 35 40

Octavian

5 10 15 20 25 30 35 40

Kenny

Sessions

N u

m b

er p

er 2

0 m

in u

te s

Figure 9 Number of recruiting responses (data points) and teacher praise statement (bars) per 20-minute seatwork sessions. Target recruiting rate was two to three responses per session. Asterisks indicate when each student was trained in the resource room. From “Teaching Elementary Students with Developmental Disabilities to Recruit Teacher Attention in a General Education Classroom: Effects on Teacher Praise and Academic Productivity” by M. A. Craft, S. R. Alber, and W. L. Heward, 1998, Journal of Applied Behavior Analysis, 31, p. 407. Copyright 1998 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

many behaviors. A study by Stokes and Baer (1976) is a good example of the potential effects of the evocative ef- fects of the presence of a person in the generalization set- ting who had a functional role in the learner’s acquisition of the target behavior in the instructional setting. Two preschool children with learning disabilities learned word-recognition skills while working as reciprocal peer tutors for one another. However, neither child showed re- liable generalization of those new skills in nontraining settings until the peer with whom he had learned the skills was present in the generalization setting.

Some contrived mediating stimuli serve as much more than response prompts; they are prosthetic devices that assist the learner in performing the target behavior. Such devices can be especially useful in promoting gen-

eralization and maintenance of complex behaviors and extend response chains by simplifying a complex situa- tion. Three common forms are cue cards, photographic activity schedules, and self-operated prompting devices.

Sprague and Horner (1984) gave the students in their study cue cards to aid them in operating a vending ma- chine without another person’s assistance. The cue cards, which had food and drink logos on one side and pictures of quarters paired with prices on the other, not only were used during instruction and generalization probes, but also were kept by the students at the end of the program. A follow-up 18 months after the study was completed re- vealed that five of the six students still carried a cue card and were using vending machines independently.

653

Generalization and Maintenance of Behavior Change

MacDuff, Krantz, and McClannahan (1993) taught four boys with autism ages 9 to 14 to use photographic activity schedules when performing domestic-living skills such as vacuuming and table setting and for leisure ac- tivities such as using manipulative toys. Prior to training with photographic activity schedules, the boys

were dependent on ongoing supervision and verbal prompts to complete self-help, housekeeping, and leisure activities. . . . In the absence of verbal prompts from supervising adults, it appeared that stimulus con- trol transferred to photographs and materials that were available in the group home. When the study ended, all 4 boys were able to display complex home-living and recreational repertoires for an hour, during which time they frequently changed tasks and moved to different areas of their group home without adults’ prompts. Pho- tographic activity schedules, . . . became functional dis- criminative stimuli that promoted sustained engagement after training ceased and fostered generalized respond- ing to new activity sequences and novel leisure materi- als. (pp. 90, 97)

Numerous studies have shown that learners across a wide range of age and cognitive levels, including stu- dents with severe intellectual disabilities, can learn to use personal audio playback devices to independently per- form a variety of academic, vocational, and domestic tasks (e.g., Briggs et al., 1990; Davis, Brady, Williams, & Burta, 1992; Grossi, 1998; Mechling & Gast, 1997; Post, Storey, & Karabin, 2002; Trask-Tyler, Grossi, & Heward, 1994). The popularity of personal music devices such as “Walkman-style” tape players and iPods enables a person to listen to a series of self-delivered response prompts in a private, normalized manner that does not impose on or bother others.

Teach Self-Management Skills

The most potentially effective approach to mediating generalized behavior changes rests with the one element that is always present in every instructional and gener- alization setting—the learner herself. There are a vari- ety of self-management tactics that people can use to modify their own behavior. The logic of using self- management to mediate generalized behavior changes goes like this: If the learner can be taught a behavior (not the original target behavior, but another behavior—a controlling response from a self-management perspec- tive) that serves to prompt or reinforce the target behav- ior in all the relevant settings, at all appropriate times, and in all of its relevant forms, then the generalization of the target behavior is ensured. But as Baer and Fowler (1984) warned:

Giving a student self-control responses designed to mediate the generalization of some critical behavior

changes does not ensure that those mediating responses will indeed be used. They are, after all, just responses: they, too, need generalization and maintenance just as do the behavior changes that they are meant to general- ize and maintain. Setting up one behavior to mediate the generalization of another behavior may succeed— but it may also represent a problem in guaranteeing the generalization of two responses, where before we had only the problem of guaranteeing the generalization of one! (p. 149)

Train to Generalize If generalization is considered as a response itself, then a reinforcement contingency may be placed on it, the same as with any other operant.

—Stokes and Baer (1977, p. 362)

Training “to generalize” was one of the eight proactive strategies for programming generalized behavior change in Stokes and Baer’s (1977) conceptual scheme (which also included the nonstrategy, train and hope). The quotation marks around “to generalize” signified that the authors were hypothesizing about the possibilities of treating “to generalize” as an operant response and that they recog- nized “the preference of behaviorists to consider general- ization to be an outcome of behavioral change, rather than as a behavior itself” (p. 363). Since the publication of Stokes and Baer’s review, basic and applied research has demonstrated the pragmatic value of their hypothesis (e.g., Neuringer, 1993, 2004; Ross & Neuringer, 2002; Shahan & Chase, 2002). Two tactics that applied behavior analysts have used are reinforcing response variability and in- structing the learner to generalize.

Reinforce Response Variability

Response variability can help a person solve problems. A person who can improvise by emitting a variety of re- sponses is more likely to solve problems encountered when a standard response form fails to obtain reinforce- ment (e.g., Arnesen, 2000; Marckel, Neef, & Ferreri, 2006; Miller & Neuringer, 2000; Shahan & Chase, 2002). Response variability also may result in behavior that is valued because it is novel or creative (e.g., Goetz & Baer, 1973; Holman, Goetz, & Baer, 1977; Pryor, Haag, and O’Reilly, 1969). Response variability may expose a per- son to sources of reinforcement and contingencies not accessible by more constricted forms of responding. The additional learning that results from contacting those con- tingencies further expands the person’s repertoire.

One direct way to program desired response gener- alization is to reinforce response variability when it oc- curs. The contingency between response variability and reinforcement can be formalized with a lag reinforce-

654

Generalization and Maintenance of Behavior Change

ment schedule (Lee, McComas, & Jawar, 2002). On a lag reinforcement schedule, reinforcement is contingent on a response being different in some defined way from the previous response (a Lag 1 schedule) or a specified number of previous responses (Lag 2 or more). Cam- milleri and Hanley (2005) used a lag reinforcement contingency to increase varied selections of classroom activities by two typically developing girls, who were se- lected to participate in the study because they spent the vast majority of their time engaged in activities not sys- tematically designed to result in a particular set of skills within the curriculum.

At the beginning of each 60-minute session, the chil- dren were told they could select any activity and could

switch activities at any time. A timer sounded every 5 minutes to prompt activity choices. During baseline there were no programmed consequences for selecting any of the activities. Intervention consisted of a lag rein- forcement schedule in which the first activity selection and each subsequent novel selection were followed by the teacher handing the student a green card that she could exchange later for 2 minutes of teacher attention (result- ing in a Lag 12 contingency, which was reset if all 12 ac- tivities were selected within a session).

Activity selections during baseline showed little vari- ability, with both girls showing strong preferences for stack- able blocks (Figure 10 shows results for one of the girls). When the lag contingency was introduced, both girls

N u

m b

er o

f N

ov el

Se le

ct io

n s

Baseline Intervention

Tina

Baseline Intervention

RAS NPR

Prog. Reading CCC Math Prog. Math Math Facts Grammar

Free Reading Nobby Nuss

Monkey Maze Blocks Other

58 8 2 8 8 13 8 17 15 17 17 8 12 12 17 17 20 22 13 22 17 13

3 3 5 4 2

5 3 5 3 33 8 8 8 8 8 5 5 3 8 8 8 8 4 8 8 8 8 8 7

8 8 3 5 5 3 8 3 3

8 17 8 8 17 17 8 17 8 50 67 75 92 83 83

8 8 8 17 17 17 17 17 8 21 25 13 47 3 15 8 8 3 5 4

8 25 8 17 8

8 33 8 8 25 8 8 8 5 8 8

8 8 8 8

100 100 75 97 100 100

3 98 94 3 6

33 8

N u

m b

er o

f A

ca d

em ic

U n

it s

C om

p le

te d

2 4 6 8 10 12 14 16 18 20 22 24 Sessions

Figure 10 Number of novel activity selections (top panel), percentage of intervals of engagement in programmed (italicized) and nonprogrammed activities (middle panel; shaded cells indicate activities for which there was some engagement), and number of academic units completed (bottom panel). From “Use of a Lag Differential Reinforcement Contingency to Increase Varied Selections of Classroom Activities” by A. P. Cammilleri and G. P. Hanley, 2005, Journal of Applied Behavior Analysis, 38, p. 114. Copyright 2005 by the Society for the Experimental Analysis of Behavior, Inc. Reprinted by permission.

655

Generalization and Maintenance of Behavior Change

immediately selected and engaged in more diverse activi- ties. The researchers noted that “an indirect but important outcome of this shift in time allocation was a marked in- crease in the number of academic units completed” (p. 115).

Instruct the Learner to Generalize

The simplest and least expensive of all tactics for pro- moting generalized behavior change is to “tell the subject about the possibility of generalization and then ask for it” (Stokes & Baer, 1977, p. 363). For example, Ninness and colleagues (1991) explicitly told three middle school stu- dents with emotional disturbance to use self-management procedures they had learned in the classroom to self-assess and self-record their behavior while walking from the lunchroom to the classroom. Hughes and colleagues (1995) used a similar procedure to promote generaliza- tion: “At the close of each training session, peer teachers reminded participants to self-instruct when they wished to talk to someone” (p. 207). Likewise, at the conclusion of each training session conducted in the special education classroom on how to recruit assistance from peers during cooperative learning groups, Wolford and colleagues (2001) prompted middle school students with learning disabilities to recruit peer assistance at least two times but not more than four times during each cooperative learn- ing group in the language arts classroom.

To the extent that generalizations occur and are them- selves generalized, a person might then become skilled at generalizing newly acquired skills, or in the words of Stokes and Baer (1977), become a “generalized generalizer.”

Modifying and Terminating Successful Interventions With most successful behavior change programs it is im- possible, impractical, or undesirable to continue the in- tervention indefinitely. Withdrawal of a successful intervention should be carried out in a systematic fashion, guided by the learner’s performance of the target behav- ior in the most important generalization settings. Gradu- ally shifting from the contrived conditions of the intervention to the typical, everyday environment will in- crease the likelihood that the learner will maintain the new behavior patterns. When deciding how soon and how swiftly to withdraw intervention components, practi- tioners should consider factors such as the complexity of the intervention, the ease or speed with which behavior changed, and the availability of naturally existing con- tingencies of reinforcement for the new behavior.

This shift from intervention conditions to the postin- tervention environment can be made by modifying one or

more of the following components, each representing one part of the three-term contingency:

• Antecedents, prompts, or cue-related stimuli

• Task requirements and criteria

• Consequences or reinforcement variables

Although the order in which intervention compo- nents are withdrawn may make little or no difference, in most programs it is probably best to make all task-related requirements as similar as possible to those of the post- intervention environment before withdrawing significant antecedent or consequence components of the interven- tion. In this way the learner will be emitting the target behavior at the same level of fluency that will be required after the complete intervention has been withdrawn.

A behavior change program carried out many years ago by a graduate student in one of our classes illus- trates how the components of a program can be gradu- ally and systematically withdrawn. An adult male with developmental disabilities took an inordinate amount of time to get dressed each morning (40 to 70 minutes during baseline), even though he possessed the skills needed to dress himself. Intervention began with a con- struction paper clock hung by his bed with the hands set to indicate the time by which he had to be fully dressed to receive reinforcement. Although the man could not tell time, he could discriminate whether the position of the hands on the real clock nearby matched those on his paper clock. Two task-related intervention elements were introduced to increase the likelihood of initial success. First, he was given fewer and easier clothes to put on each morning (e.g., no belt, slip-on loafers instead of shoes with laces). Second, based on his baseline performance, he was initially given 30 min- utes to dress himself, even though the objective of the program was for him to be completely dressed within 10 minutes. An edible reinforcer paired with verbal praise was used first on a continuous schedule of reinforce- ment. Figure 11 shows how each aspect of the inter- vention (antecedent, behavior, and consequence) was modified and eventually withdrawn completely, so that by the program’s end the man was dressing himself completely within 10 minutes without the aid of extra clocks or charts or contrived reinforcement other than a naturally existing schedule of intermittent praise from staff members.

Rusch and Kazdin (1981) described a method for systematically withdrawing intervention components while simultaneously assessing response maintenance that they called “partial-sequential withdrawal.” Martella, Leonard, Marchand-Martella, and Agran (1993) used a partial-sequential withdrawal of various components of a self-monitoring intervention they had implemented to

656

Generalization and Maintenance of Behavior Change

Antecedents (cues, prompts)

Behavior (task criteria/ modifications)

Consequences (reinforcers, punishers)

A: Mock clock showing when must be dressed

B: No clock

A: Fewer clothes B: Full set of clothes C: Time gradually

reduced D: Criterion time

limit E: Variety of clothes

A: Edible reinforcer B: Token reinforcer C: Chart (self-record) D: Praise on

intermittent schedule

A B B C C C C D

A A B

B/C B/D B/D D/E D/E

A A A A A B B B

1 2 3 4 5 6 7 8

Phase

Figure 11 An example of modifying and with- drawing components of an independent morning dressing program for an adult with developmental disabilities to facilitate mainte- nance and generalization.

help Brad, a 12-year-old student with mild mental retar- dation, reduce the number of negative statements (e.g., “I hate this @#!%ing calculator,” “Math is crappy”) he made during classroom activities. The self-management intervention consisted of Brad (a) self-recording nega- tive statements on a form during two class periods, (b) charting the number on a graph, (c) receiving his choice from a menu of “small” reinforcers (items costing 25 cents or less), and (d) when his self-recorded data were in agreement with the trainer and at or below a gradually reducing criterion level for four consecutive sessions, he was allowed to choose a “large” reinforcer (more than 25 cents). After the frequency of Brad’s negative state- ments was reduced, the researchers began a four-phase partial-sequential withdrawal of the intervention. In the first phase, charting and earning of “large” reinforcers were withdrawn; in the second phase, Brad was required to have zero negative statements in both periods to re- ceive a daily “small” reinforcer; in the third phase, Brad used the same self-monitoring form for both class peri- ods instead of one form for each period as before, and the small reinforcer was no longer provided; and in the fourth phase (the follow-up condition), all components of the intervention were withdrawn except for the self- monitoring form without a criterion number highlighted. Brad’s negative statements remained low throughout the gradual and partial withdrawal of the intervention (see Figure 12).

A word of caution is in order regarding the termi- nation of successful behavior change programs. Achiev- ing socially significant improvements in behavior is a defining purpose of applied behavior analysis. Addi- tionally, those improved behaviors should be maintained and should show generalization to other relevant settings and behaviors. In most instances, achieving optimal gen-

eralization of the behavior change will require most, if not all, of the intervention components to be withdrawn. However, practitioners, parents, and others responsible for helping children learn important behaviors are some- times more concerned with whether and with how a po- tentially effective intervention will eventually be withdrawn than they are with whether it will produce the needed behavior change. Considering how a pro- posed intervention will lend itself to eventual withdrawal or blending into the natural environment is important and consistent with everything recommended throughout this book. And clearly, when the choice is between two or more interventions of potentially equal effectiveness, the intervention that is most similar to the natural envi- ronment and the easiest to withdraw and terminate should be given first priority. However, an important be- havior change should not go unmade because complete withdrawal of the intervention required to achieve it may never be possible. Some level of intervention may al- ways be required to maintain certain behaviors, in which case attempts to continue the necessary programming must be made.

The Sprague and Horner (1984) study on teaching generalized vending machine use provides another ex- ample of this point. The six students with moderate to severe mental retardation who participated in the pro- gram were given cue cards to aid them in operating a vending machine without another person’s assistance. The cue cards, which had food and drink logos on one side and pictures of quarters paired with prices on the other, not only were used during instruction and gener- alization probes, but also were kept by the students at the end of the program. Five of the six students still carried a cue card and were using vending machines indepen- dently 18 months after the study had ended.

657

Generalization and Maintenance of Behavior Change

Negative

Baseline

A B C D E I II III

Self-Monitoring Partial-Sequential Withdrawal

Follow-Up Probes (Weeks)

2 4 8

Period 2

Period 1

Positive

10 20 30 40 50

Sessions

N u

m b

er o

f St

at em

en ts

p er

M in

u te

Figure 12 Number of negative (solid data points) and positive (open data points) statements in two class periods by an adolescent boy with disabilities during baseline, self-monitoring, and partial-sequential withdrawal conditions. Horizontal lines during self-monitoring condition show changing criteria for teacher-delivered reinforcement. From “Self-Monitoring Negative Statements” by R. Martella, I. J. Leonard, N. E. Marchand-Martella, and M. Agran, 1993, Journal of Behavioral Education, 3, p. 84. Copyright 1993 by Human Sciences Press. Reprinted by permission.

Guiding Principles for Promoting Generalized Outcomes Regardless of the specific tactics selected and applied, we believe that practitioners’ efforts to promote general- ized behavior change will be enhanced by adherence to five guiding principles:

• Minimize the need for generalization as much as possible.

• Conduct generalization probes before, during, and after instruction.

• Involve significant others whenever possible.

• Promote generalization with the least intrusive, least costly tactics possible.

• Contrive intervention tactics as needed to achieve important generalized outcomes.

Minimize the Need for Generalization

Practitioners should reduce the need for generalization to untaught skills, settings, and situations as much as possi- ble. Doing so requires thoughtful, systematic assessment of what behavior changes are most important. Practition- ers should prioritize the knowledge and skills that will most often be required of the learner and the settings and

658

Generalization and Maintenance of Behavior Change

situations in which the learner will most often benefit from using those skills. In addition to the environment(s) in which the learner is currently functioning, practition- ers should consider the environments in which the learner will function in the immediate future and later in life.

The most critical behavior changes should not be relegated to the not-for-certain technology of general- ization. The most important skill-setting-stimulus com- binations should always be taught directly and, when possible, taught first. For example, a training program to teach a young adult with disabilities to ride the pub- lic bus system should use the bus routes the person will take most often (e.g., to and from her home, school, job site, and community recreation center) as teaching ex- amples. Instead of providing direct instruction on routes to destinations the learner may visit only occasionally after training, use those routes as generalization probes. Achieving a high level of response maintenance on the trained routes will still be a challenge.

Probe for Generalization Before, During, and After Instruction

Generalization probes should be conducted prior to, dur- ing, and after instruction.

Probe Before Instruction

Generalization probes conducted before instruction be- gins may reveal that the learner already performs some or all of the needed behaviors in the generalization setting, thereby reducing the scope of the teaching task. It is a mistake to assume that because a learner does not per- form a particular behavior in the instructional setting, she does not or cannot do it in the generalization setting(s). Preinstruction generalization probe data are the only ob- jective basis for knowing that the learner’s performance of the target behavior after instruction is in fact a gener- alized outcome.

Probes before instruction enable observation of the contingencies that operate in the generalization setting. Knowledge of such information may contribute to a more effective treatment or instruction.

Probe During Instruction

Generalization probes during instruction reveal if and when generalization has occurred and if and when in- struction can be terminated or shifted in focus from ac- quisition to maintenance. For example, a teacher who finds that a student can solve untaught examples of a par- ticular type of algebraic equation after receiving instruc- tion on just a few examples and shifts instruction to the next type of equation in the curriculum, will cover more

of the algebra program effectively than will a teacher who continues to present additional examples.

Results of probes during instruction also show whether generalization is not occurring, thereby indicat- ing that a change in instructional strategies is needed. For example, when Sprague and Horner (1984) observed that one student’s poor performance on the first generalization probe was caused by a ritualistic pattern of inserting coins, they incorporated repeated practice on the coin in- sertion step during a training session and the student’s performance on the generalization task improved greatly.

Probing can often be made more efficient by con- triving opportunities for the learner to use her new knowl- edge or skill. For example, instead of waiting for (and perhaps missing) naturally occurring opportunities for the learner to use her new conversational skills in the gen- eralization environment, a practitioner could enlist the assistance of a “confederate” peer to approach the learner. Ninness, Fuerst, and Rutherford, (1991) used contrived generalization probes in which students were provoked or distracted to test the extent to which they used a set of self-management skills.

Contrived generalization probes can also be used as primary measures of acquisition and generalization. For example, Miltenberger and colleagues (2005) contrived opportunities to measure and teach children’s use of gun safety skills by placing a gun in home and school loca- tions where the children would find it. If a child did not execute the target safety skills when finding a gun (i.e., do not touch it, get away from the gun, and tell an adult), the trainer entered the room and conducted an in situ training session, asking the child what he should have done and rehearsing the entire sequence five times.

Probe After Instruction

Generalization probes after instruction has ended reveal the extent of response maintenance. The question of how long probes should be conducted after instruction has ended should be answered by factors such as the sever- ity of the target behavior, the importance of the behavior change to the person’s quality of life, the strength and consistency of the response maintenance obtained by probes to date, and so forth. The need for long-term as- sessment of maintenance is especially critical for severe behavior problems. In some cases, maintenance probes over many months and several years may be indicated (e.g., Derby et al., 1997; Foxx, Bittle, & Faw, 1989; Wag- man, Miltenberger, & Woods, 1995).

If conducting systematic generalization and mainte- nance probes seems too difficult or too contrived, the practitioner should consider the relative importance of the target behavior to the student or client. If a behavior is important to target for intervention then assessing the

659

Generalization and Maintenance of Behavior Change

generalized outcomes of that intervention is worth what- ever effort that assessment requires.

Involve Significant Others All persons are potential teachers of all sorts of behavior changes. Just because you are designated as “teacher” or “behavior analyst” does not mean that you have an ex- clusive franchise on the ability to make deliberate be- havior changes. In fact, there is no possibility of such a franchise. Everyone in contact contributes to everyone else’s behavior, both to its changes and its maintenance.

— Donald M. Baer (1999, p. 12)

Teaching for generalized outcomes is a big job, and prac- titioners should try to get as much help as they can. People are almost always around where and when im- portant behaviors need to be prompted and reinforced; and when social behaviors are targeted, people are there by definition.

In virtually every behavior change program people other than the participant and the behavior analyst are in- volved, and their cooperation is crucial to the success of the program. Foxx (1996) stated that in programming success- ful behavior change interventions, “10% is knowing what to do; 90% is getting people to do it. . . . Many programs are unsuccessful because these percentages have been reversed” (p. 230). Although Foxx was referring to the challenge and importance of getting staff to implement programs with the consistency and fidelity needed for success, the same holds for the involvement of significant others.

Baer (1999) suggested identifying others who will or may be involved in a behavior change program as active sup- porters or tolerators. An active supporter is someone natu- rally present in the generalization setting who helps promote the generalization and maintenance of the target behavior by doing specific things. Active supporters help facilitate the desired generalized outcomes by arranging opportunities for the learner to use or practice the new skill, giving cues and response prompts for the behavior, and providing rein- forcement for performance of the target behavior.

Active supporters for a behavior change program to teach independent eating to a child with severe disabili- ties might include one or two key people in the school’s cafeteria, a volunteer or aide who works regularly with the child, the child’s parents, and an older sibling. These people are vital parts of the teaching team. If optimal generalization is to occur, the active supporters must see to it that the learner has many opportunities to use the new skill in the generalization environments they share, and the natural reinforcers they control in those environ- ments (e.g., praise, touch, smiles, companionship) must be used as consequences for the target behavior. It is not necessary, and probably not desirable, to limit an active

supporters list, to school staff, family members, and peers. They may not be regularly available in all the environ- ments and situations in which generalization is desired.

A tolerator is someone in the generalization setting who agrees not to behave in ways that would impede the generalization plan. Tolerators should be informed that the learner will be using a new skill in the generalization en- vironment and asked to be patient. In an independent eat- ing program a list of tolerators would likely include some members of the child’s family, school cafeteria staff, and peers. Beyond the home and school environments the be- havior analyst should consider the possible role of the gen- eral public with whom the child may find herself sharing a table or dining area in a public restaurant. The learner’s initial sloppiness and slowness (she may always be slop- pier and slower than most people) as she makes the first at- tempts beyond the familiar contingencies of home and school might result in various responses from strangers that could punish the new eating skills. Being stared at, laughed at, talked about, told to hurry up, or even offered assistance could reduce the possibility of generalization. Certainly, the behavior analyst can inform various school staff and family members of the ongoing eating program and request that they not interfere with the child’s attempts to eat independently. But the public at large is another issue. It is impossible to inform everyone of the program. However, by considering the types of intolerant behaviors that the learner may encounter in the generalization set- ting, the teaching program can be constructed to include practice under such intolerant conditions. Instructional tri- als might be contrived to reinforce the learner for ignoring rude remarks and continuing to eat independently.

Use the Least Intrusive, Least Costly Tactics Possible

Behavior analysts should use less intrusive and less costly tactics to promote generalization before using more in- trusive and costly tactics. As noted previously, simply reminding students to use their new skills in the general- ization setting is the easiest and least expensive of all methods that might promote generalization. Although practitioners should never assume that telling the learner to generalize will produce the desired outcomes, neither should they fail to include such a simple and cost-free tactic. Similarly, incorporating some of the most relevant features of the generalization setting into the instructional setting (i.e., programming common stimuli) often helps produce the needed generalization and is less costly than conducting instruction in the natural setting (e.g., Neef, Lensbower et al., 1990; van den Pol et al., 1981).

Not only is using less costly tactics good conserva- tion of the limited resources available for teaching, but

660

Generalization and Maintenance of Behavior Change

also less intrusive interventions with fewer moving parts are easier to withdraw. Systematic generalization probes will determine whether generalization has occurred and whether more elaborate and intrusive intervention and supports are needed.

Contrive Intervention Tactics as Needed to Achieve Important Generalized Outcomes

A practitioner should not be so concerned about intru- siveness that she fails to implement a potentially effective intervention or procedures that will achieve important outcomes for the learner. Therefore, if necessary, prac- tioners should disregard the previous guideline and con- trive as many instruction and generalization tactics as necessary to enable the learner to generalize and maintain critical knowledge and skills.

Rather than lament the lack of generalization or blame the learner for his inability to show generalized behavior changes, the behavior analyst should work to arrange whatever socially valid contingencies may be needed to extend and maintain the target behavior.

Some Final Words of Wisdom from Don Baer

The most difficult and important challenge facing behav- ioral practitioners is helping learners achieve generalized change in socially significant behaviors. A behavior

change—no matter how important initially—is of little value to the learner if it does not last over time, is not emit- ted in appropriate settings and situations, or occurs in re- stricted form when varied topographies are desired.

Research during the past 30 years has developed and advanced the knowledge base of what Stokes and Baer (1977) described as an “implicit technology of general- ization” into an increasingly explicit and effective set of strategies and tactics for promoting generalized behav- ior change. Knowledge of these methods, combined with knowledge of the basic principles and behavior change tactics described throughout this book, provides behav- ior analysts with a powerful approach for helping peo- ple enjoy healthy, happy, and productive lives.

We end this chapter with a dually wise observation by Don Baer (1999), in which he pointed out a funda- mental truth about the relation between a person’s expe- rience (in this case, the nature of a lesson) and what is learned or not learned from that experience. Like Skin- ner before him, Baer wisely reminded us not to blame the learner for not behaving as we think he should.

Learning one aspect of anything never means that you know the rest of it. Doing something skillfully now never means that you will always do it well. Resisting one temptation consistently never means that you now have character, strength, and discipline. Thus, it is not the learner who is dull, learning disabled, or immature, because all learners are alike in this regard: no one learns a generalized lesson unless a generalized lesson is taught. (p. 1) [emphasis in original]

Summary Generalized Behavior Change: Definitions and Key Concepts

1. Generalized behavior change has taken place if trained be- havior occurs at other times or in other places without hav- ing to be retrained completely in those time or places, or if functionally related behaviors occur that were not taught directly.

2. Response maintenance refers to the extent to which a learner continues to perform a behavior after a portion or all of the intervention responsible for the behavior’s initial appearance in the learner’s repertoire has been terminated.

3. Setting/situation generalization refers to the extent to which a learner emits the target behavior in settings or situations that are different from the instructional setting.

4. The instructional setting is the environment where instruc- tion occurs and encompasses all aspects of the environment,

planned or unplanned, that may influence the learner’s ac- quisition and generalization of the target behavior.

5. A generalization setting is any place or stimulus situation that differs from the instructional setting in some mean- ingful way and in which performance of the target behav- ior is desired.

6. Response generalization refers to the extent to which a learner emits untrained responses that are functionally equivalent to the trained response.

7. Some interventions yield significant and widespread gen- eralized effects across time, settings, and other behaviors; others produce circumscribed changes in behavior with limited endurance and spread.

8. Undesirable setting/situation generalization takes two common forms: overgeneralization, in which the behavior has come under control of a stimulus class that is too broad, and faulty stimulus control, in which the behavior

661

comes under the control of an irrelevant antecedent stimulus.

9. Undesired response generalization occurs when any of a learner’s untrained but functionally equivalent responses produce undesirable outcomes.

10. Other types of generalized outcomes (e.g., stimulus equiv- alence, contingency adduction, and generalization across subjects) do not fit easily into categories of response maintenance, setting/situation generalization, and re- sponse generalization.

11. The generalization map is a conceptual framework for combining and categorizing the various types of gener- alized behavior change (Drabman, Hammer, & Rosen- baum, 1979).

Planning for Generalized Behavior Change

12. The first step in promoting generalized behavior changes is to select target behaviors that will meet naturally exist- ing contingencies of reinforcement.

13. A naturally existing contingency is any contingency of re- inforcement (or punishment) that operates independent of the behavior analyst’s or practitioner’s efforts, including socially mediated contingencies contrived by other peo- ple and already in effect in the relevant setting.

14. A contrived contingency is any contingency of reinforce- ment (or punishment) designed and implemented by a be- havior analyst to achieve the acquisition, maintenance, and/or generalization of a targeted behavior change.

15. Planning for generalization includes identifying all the de- sired behavior changes and all the environments in which the learner should emit the target behavior(s) after direct training has ceased.

16. Benefits of developing the planning lists include a better understanding of the scope of the teaching task and an op- portunity to prioritize the most important behavior changes and settings for direct instruction.

Strategies and Tactics for Promoting Generalized Behavior Change

17. Researchers have developed and advanced what Stokes and Baer (1977) called an “implicit technology of gener- alization” into an increasingly explicit and effective set of methods for promoting generalized behavior change.

18. The strategy of teaching sufficient examples requires teaching a subset of all of the possible stimulus and re- sponse examples and then assessing the learner’s perfor- mance on untrained examples.

19. A generalization probe is any measurement of a learner’s performance of a target behavior in a setting and/or stimu- lus situation in which direct training has not been provided.

20. Teaching sufficient stimulus examples involves teaching the learner to respond correctly to more than one example

of an antecedent stimulus and probing for generalization to untaught stimulus examples.

21. As a general rule, the more examples the practitioner uses during instruction, the more likely the learner will be to re- spond correctly to untrained examples or situations.

22. Having the learner practice a variety of response topogra- phies helps ensure the acquisition of desired response forms and promotes response generalization. Often called multiple exemplar training, this tactic typically incorpo- rates numerous stimulus examples and response variations.

23. General case analysis is a systematic method for select- ing teaching examples that represent the full range of stim- ulus variations and response requirements in the generalization setting.

24. Negative, or “don’t do it,” teaching examples help learn- ers identify stimulus situations in which the target behav- ior should not be performed.

25. Minimum difference negative teaching examples, which share many characteristics with positive teaching exam- ples, help eliminate “generalization errors” due to over- generalization and faulty stimulus control.

26. The greater the similarity between the instructional set- ting and the generalization setting, the more likely the tar- get behavior will be emitted in the generalization setting.

27. Programming common stimuli means including in the in- structional setting stimulus features typically found in the generalization setting. Practitioners can identify possible stimuli to make common by direct observation in the gen- eralization setting and by asking people who are familiar with the generalization setting.

28. Teaching loosely—randomly varying noncritical aspects of the instructional setting within and across teaching ses- sions—(a) reduces the likelihood that a single or small group of noncritical stimuli will acquire exclusive con- trol over the target behavior and (b) makes it less likely that the learner’s performance will be impeded or “thrown off” by the presence of a “strange” stimulus in the gen- eralization setting.

29. A newly learned behavior may fail to contact an existing contingency of reinforcement because it has not been taught well enough. The solution for this kind of general- ization problem is to teach the learner to emit the target be- havior at the rate, accuracy, topography, latency, duration, and/or magnitude required by the naturally occurring con- tingencies of reinforcement.

30. The use of intermittent schedules of reinforcement and de- layed rewards can create indiscriminable contingencies, which promote generalized responding by making it dif- ficult for the learner to discriminate whether the next re- sponse will produce reinforcement.

31. Behavior traps are powerful contingencies of reinforce- ment with four defining features: (a) They are “baited”

Generalization and Maintenance of Behavior Change

662

Generalization and Maintenance of Behavior Change

with virtually irresistible reinforcers; (b) only a low-effort response already in the student’s repertoire is needed to enter the trap; (c) interrelated contingencies of reinforce- ment inside the trap motivate the student to acquire, ex- tend, and maintain targeted skills; and (d) they can remain effective for a long time.

32. One way to wake up an existing but inoperative contin- gency of reinforcement is to ask key people in the gener- alization setting to attend to and praise the learner’s performance of the target behavior.

33. Another tactic for waking up a natural contingency of re- inforcement is to teach the learner how to recruit rein- forcement in the generalization setting.

34. One tactic for mediating generalization is to bring the tar- get behavior under the control of a contrived stimulus in the instructional setting that will reliably prompt or aid the learner’s performance of the target behavior in the gen- eralization setting.

35. Teaching a learner self-management skills with which he can prompt and maintain targeted behavior changes in all relevant settings at all times is the most potentially effective approach to mediating generalized behavior changes.

36. The strategy of training to generalize is predicated on treat- ing “to generalize” as an operant response class that, like any other operant, is selected and maintained by contin- gencies of reinforcement.

37. One tactic for promoting response generalization is to re- inforce response variability. On a lag reinforcement sched- ule, reinforcement is contingent on a response being different in some defined way from the previous response (a Lag 1 schedule) or a specified number of previous re- sponses (Lag 2 or more).

38. The simplest and least expensive tactic for promoting gen- eralized behavior change is to tell the learner about the usefulness of generalization and then instruct him to do so.

Modifying and Terminating Successful Interventions

39. With most successful behavior change programs it is im- possible, impractical, or undesirable to continue the in- tervention indefinitely.

40. The shift from formal intervention procedures to a normal everyday environment can be accomplished by gradually withdrawing elements comprising the three components of the training program: (a) antecedents, prompts, or cue- related stimuli; (b) task modifications and criteria; and (c) consequences or reinforcement variables.

41. An important behavior change should not go unmade be- cause complete withdrawal of the intervention required to achieve it may never be possible. Some level of interven- tion may always be required to maintain certain behav- iors, in which case attempts must be made to continue necessary programming.

Guiding Principles for Promoting Generalized Outcomes

42. Efforts to promote generalized behavior change will be enhanced by adhering to five guiding principles:

• Minimize the need for generalization as much as possible.

• Conduct generalization probes before, during, and after instruction.

• Involve significant others whenever possible.

• Promote generalized behavior change with the least in- trusive, least costly tactics possible.

• Contrive intervention tactics as needed to achieve im- portant generalized outcomes.

663

Ethical Considerations for Applied Behavior Analysts

Key Terms

confidentiality conflict of interest

ethical codes of behavior ethics

informed consent

Content Area 1: Ethical Considerations

1-1 Solicit or otherwise influence clients only through the use of truthful and accurate representations of intervention efficacy and one’s professional competence in applied behavior analysis.

1-2 Practice within one’s limits of professional competence in applied behavior analysis; obtain consultation, supervision, or training; or make referrals as necessary.

1-3 Maintain competence by engaging in ongoing professional development activities.

1-4 Obtain informed consent within applicable legal and ethical standards.

1-5 Assist the client with identifying lifestyle or systems change goals and targets for behavior change that are consistent with the following:

(a) The applied dimension of applied behavior analysis (Baer, Wolf, & Risley 1968)

(b) Applicable laws

1-6 Initiate, continue, modify, or discontinue behavior analysis services only when the risk–benefit ratio of doing so is lower than the risk–benefit ratio for taking al- ternative actions.

1-7 Identify and reconcile contingencies that compromise the practitioner–client covenant, including relationships among the practitioner, the client, and other parties.

This chapter was written by José A. Martinez-Diaz, Thomas R. Freeman, Matthew Normand, and Timothy E. Heron.

664

Ethical dilemmas surface frequently in educa- tional and clinical practice. Consider these situations:

• A person residing in a private, rural, for-profit com- munity-based home for persons with developmental disabilities approaches the director and states that he wants to move to an apartment in a nearby town. Such a move would represent a loss of income to the agency, might generate additional transition costs (e.g., moving expenses, future on-site supervi- sion), and has the potential to be dangerous to the resident given the area of town that the person could afford. How could the director respond ethi- cally to the resident’s inquiry about moving without being biased by a conflict of interest?

• Julian, a student with severe disabilities, engages in frequent and severe self-injurious behavior (SIB) (e.g., head banging, eye gouging). Many positive and positive-reductive approaches have been attempted to reduce his SIB, but none have been successful. The support coordinator recom- mends the Self-Injurious Behavior Inhibition Sys- tem (SIBIS) as an option, but the parents object because they fear that electrical shock will hurt their son. Given that documented positive at- tempts have failed, is it an appropriate ethical course of action to recommend that SIBIS treat- ment be initiated?

• During the course of an annual individualized edu- cation program (IEP) meeting, Ms. Dougherty, a first-year teacher, perceives that a school district administrator is trying to “steer” the parents of a student with emotional disabilities into accepting a revised IEP without the provision of school-based applied behavior analysis services recommended by the majority of other team members. Ms. Dougherty hypothesizes that the administrator’s position is based on the added costs the financially strapped school district would bear if it provided these services. As a first-year teacher, Ms. Dougherty is concerned that if she speaks up, she might lose the favor of her principal and maybe her job. If she remains silent, the student might not re- ceive needed services. How might Ms. Dougherty serve as an advocate for the student, but not lose her position on the faculty?

Given each of these situations, how is the behavior analyst to respond ethically? This chapter addresses the issues surrounding and underlying these situations. We begin by discussing what ethics is and why it is impor- tant. Next, we present ethical behavior and behavior analysis standards for professional practice as a way for practitioners to navigate ethical dilemmas that arise in day-to-day practice. Finally, we discuss ethical issues within the context of client services (e.g., informed con- sent and conflict of interest).

Ethical Considerations for Applied Behavior Analysts

1-8 Use the most effective assessment and behavior change procedures within ap- plicable ethical standards, taking into consideration the guideline of minimal in- trusiveness of the procedure to the client.

1-9 Protect confidentiality.

1-10 Truthfully and accurately represent one’s contributions and those of others to the practice, discipline, and profession of applied behavior analysis.

1-11 Ensure that the dignity, health, and safety of one’s client are fully protected at all times.

1-12 Give preference to assessment and intervention methods that have been scien- tifically validated, and use scientific methods to evaluate those that have not yet been scientifically validated.

Content Area 8: Selecting Intervention Outcomes and Strategies

8-2 Make recommendations to the client regarding target outcomes based on such factors as: client preferences, task analysis, current repertoires, supporting environments, constraints, social validity, assessment results, and best available scientific evidence.

8-4 Make recommendations to the client regarding intervention strategies based on such factors as: client preferences, task analysis, current repertoires, supporting environments, constraints, social validity, assessment results and best available scientific evidence.

© 2006 The Behavior Analyst Certification Board, Inc.,® (BACB®) all rights reserved. A current version of this document may be found at www.bacb.com. Requests to reprint, copy, or distribute this document and questions about this document must be submitted directly to the BACB.

665

What Is Ethics and Why Is It Important? Definition of Ethics

Ethics refers to behaviors, practices, and decisions that address three basic and fundamental questions: What is the right thing to do? What is worth doing? What does it mean to be a good behavior analyst? (Reich, 1988; Smith, 1987, 1993). Guided by these questions, personal and professional practices are conducted for the principal pur- pose of helping others to improve their physical, social, psychological, familial, or personal condition. As Corey, Corey, and Callanan (1993) stated, “The basic purpose of practicing ethically is to further the welfare of the client” (p. 4).

What Is the Right Thing to Do?

Addressing the question of what is the right thing to do leads to an examination of several areas of influence: our personal histories of what is right and wrong, the con- text within which applied behavior analysis is practiced, including legal and illegal versus ethical and unethical issues, and codified ethical rules of conduct. Further, how to determine what is the right thing to do is derived from established principles, methods, and decisions that have been used successfully by other applied behavior ana- lysts and professionals primarily to ensure the welfare of those entrusted to our care, the well-being of the profes- sion, and ultimately the survival of the culture (Skinner, 1953, 1971).

At the outset, we would state that what is ethical (right) or unethical (wrong) behavior is ultimately related to cultural practices. Therefore, it is subject to differences across cultures and the passage of time. What might be acceptable in one culture may be unacceptable in other cultures. What might be acceptable at one time may be completely unacceptable 20 years later.

Personal Histories. When deciding how to pro- ceed with an assessment or intervention, all applied be- havior analysts are influenced by their personal histories of making decisions in similar situations. Pre- sumably, the analyst’s training and experiences will balance negative biases or predispositions that may carry over from his or her personal or cultural back- ground. For example, assume that a behavior analyst has a sibling who engages in self-injurious behavior (SIB). Further, assume that the practitioner faces a deci- sion on how to assist another family with a child who exhibits severe SIB. The practitioner may be influenced by his recollection of procedures his parents used (or

did not use) to help his sibling. Did they use punish- ment? Include the sibling in family gatherings? Seek competent programs and services? However, given the analyst’s training in a range of assessments and inter- ventions, the potential influence of such personal expe- rience is more likely to be overridden in favor of an unbiased exploration of clinically appropriate treatment alternatives.

Further, the practitioner’s cultural or religious up- bringing may influence decisions about the right course of action. Prior to formal training in behavior analysis, a practitioner reared in a family culture that supported “Spare the rod and spoil the child” may manage severe be- havior differently than if that same practitioner was reared in a family culture that supported a “He’ll grow out of it” philosophy. Subsequent to training, however, a more un- biased and informed set of principles is likely to surface.

Finally, a person’s professional training and experi- ences with other cases involving severe behavior chal- lenges will likely influence whether Method A (e.g., a differential reinforcement procedure) is favored over Method B (e.g., overcorrection). The behavior analyst must recognize that her personal history is likely to be a factor when making a decision. Still, the practitioner needs to be careful lest her personal background lead her to select inappropriate or ineffective interventions. To counteract this possibility and ensure that personal his- tory and background will not influence decision making over professional knowledge and experience, the practi- tioner can seek help from supervisors or colleagues, re- view the research literature, consult case studies to determine past courses of successful action, or excuse herself from the case.

The Context of Practice. Applied behavior ana- lysts work in schools, homes, community settings, job sites, and other natural environments. Rules within these environments cover a host of behaviors (e.g., at- tendance, use of sick time). Included in the rules are policy statements that are designed to help practitioners differentiate between legal issues and ethical issues. For example, some practices may be legal, but unethi- cal. Breaking a professional confidence, accepting val- ued family “heirlooms” in lieu of payment for services, or engaging in consensual sexual relations with a client over age 18 serve as examples of behaviors that are legal but unethical. Other behaviors are both illegal and unethical. For instance, misrepresenting personal skills or promised services; stealing the belongings of a client while rendering services; abusing a client physi- cally, emotionally, sexually, or socially; or engaging in consensual sexual interaction with persons under age 18 are all examples of behaviors that are both illegal and unethical (Greenspan & Negron, 1994). Behavior

Ethical Considerations for Applied Behavior Analysts

666

analysts who can discriminate between these legal and ethical distinctions are more likely to provide effective services, maintain a focused sensitivity toward clients, and not run afoul of the law or professional standards of conduct.

Ethical Codes of Behavior. All professional orga- nizations have generated or adopted ethical codes of behavior. These codes provide guidelines for associa- tion members to consider when deciding a course of action or conducting their professional duties. Also, these guidelines provide the standard by which gradu- ated sanctions can be imposed for deviating from the code (e.g., reprimand, censure, expulsion from the or- ganization). The Association for Behavior Analysis has adopted the American Psychological Association’s Code of Ethics. More discussion of this point is found later in the chapter.

What Is Worth Doing?

Questions related to what is worth doing directly address the goals and objectives of practice. What are we trying to accomplish? How are we trying to accomplish it? Clearly, social validity, cost–benefit ratio, and existing exigencies enter into decision making about what is worth doing.

Social Validity. Questions of social validity ask: Are the goals acceptable for the planned behavior change intervention? Are the procedures acceptable and are they aligned with best treatment practices (Peters & Heron, 1993)? Finally, do the results show meaningful, significant, and sustainable change (Wolf, 1978)? Most people would agree that teaching a child to read is a de- sirable goal. Using direct instruction, or another mea- surably effective instructional method, is procedurally sound, and outcomes that show improved reading are socially significant. In every sense, teaching a child or adult to read meets the ethical test for social validity. The new reading skill has a positive effect on the life of the individual. However, we cannot make the same claim for every skill in every situation. Does teaching street safety signs to a person who has mobility, sight, and hearing problems have true worth to the individual? How about teaching an Alzheimer’s patient to recall the serial order of U.S. presidents? Why teach an adult with developmental disabilities to play with coloring books? Does a first-grade student with autism need to spend 20 minutes of one-on-one discrete trial instruction per day to learn to discriminate among pictures of women wearing fall and spring fashions? Does a child with

quadriplegia really need to learn to use a standard pen- cil to write? In each of these cases, the present and fu- ture “worth” of the treatment with respect to goals, procedures, and outcomes is not clear. Given adequate programming and technological advances, teaching street safety signage to a person with mobility, sight, and hearing challenges, or the order of presidents to an Alzheimer’s patient, or coloring skills to an adult with developmental disabilities, seasonal fashions to a child with autism, or pencil-gripping skills to a person with quadriplegia could be accomplished. However, from an ethical standpoint, the questions are: Should it be ac- complished? And is it worth doing?

Cost–Benefit Ratio. Cost–benefit ratio decisions are contextual and involve a balance among planning, implementing, and evaluating a treatment or interven- tion (i.e., the cost side) and projecting future potential gain by the person (i.e., the benefit side). In other words, does the potential benefit to the individual jus- tify the short- and long-term cost for providing the ser- vice? For instance, is it worth the time, cost, and emotional resources to transport a student with learn- ing disabilities to an out-of-state private school with high-cost tuition when (a) comparable services can be provided free within the neighborhood public school system and (b) improved student learning and social behavior outcomes are uncertain at best? Is it worth the practitioner’s efforts (i.e., financial cost plus time) to continue to maintain an 11th grade student with devel- opment disabilities in an academic, college-oriented program when switching to a functional curriculum program would likely enhance employability, self- sufficiency, and independence? Ethically, it is difficult to support a decision to spend school district dollars for outside private services that otherwise should be provided through public support. Likewise, it is diffi- cult to defend a decision to subject a student to the continued reflexive establishing operations for nega- tive reinforcement generated by an academic program, when a functional curriculum model would likely meet more appropriate short-term and long-term goals. To address the thorny issue of cost versus benefit, Sprague and Horner (1991) suggested that decisions should be made by committee, and that the perspectives of those with the highest stake in the outcome be given the greatest consideration. They also recommended that a hierarchy of opinions and inputs be sought to gain the widest possible viewpoint.

Existing Exigencies. Some behaviors challenge practitioners to find effective solutions quickly. A child who engages in self-injurious behavior, a child with

Ethical Considerations for Applied Behavior Analysts

667

feeding problems, or a student who engages in severe disruptive behaviors present such cases. Most practi- tioners would agree that these behaviors are worth changing so that the potential for harm to the person or others is reduced. Further, most practitioners would agree that to not act might lead to an even worse condi- tion. So from an ethical standpoint, behaviors that are more serious warrant intervention consideration before behaviors that are less problematic. However, that such behaviors may necessitate a quick and expedited treat- ment is not an invitation to adopt a situational ethics perspective in which the promise of fast results in the short term is compromised by not fully considering long- term ramifications. In short, important questions rela- tive to effectiveness, intrusiveness, possible deleterious side effects of potential treatments, and independence also must be considered, even if that means delaying an intervention temporarily (Sprague, 1994).

What Does It Mean to Be a Good Behavior Analyst?

To be a good behavior analyst requires more than fol- lowing professional codes of conduct. Although adher- ence to the codes of the Association for Behavior Analysis and the Behavior Analyst Certification Board is neces- sary, it is not sufficient. Even keeping the client’s wel- fare at the forefront of decision making is not sufficient. Following the Golden Rule (Do unto others as you would have them do unto you), a code valued by virtually every culture and religion in the world, is also not sufficient (Maxwell, 2003). A good practitioner is self-regulating. That is, the ethical practitioner seeks ways to calibrate decisions over time to ensure that values, contingencies, and rights and responsibilities are integrated and an in- formed combination of these is considered (Smith, 1993).

Why Is Ethics Important?

Behavior analytic practitioners abide by ethical princi- ples to (a) produce “meaningful” behavior change of so- cial significance for the persons entrusted to their care (Hawkins, 1984), (b) reduce or eliminate harm (e.g., poor treatments, self-injury), and (c) conform to the ethical standards of learned societies and professional organiza- tions. Without an ethical compass, practitioners would be adrift when deciding whether a course of action is morally right or wrong, or whether they have slipped into the realm of situational ethics in which decisions to act (or not to act) are based on expediency, pressure, or mis- placed priorities (Maxwell, 2003). For instance, if the

teacher mentioned in our earlier scenario acquiesced to the administrator’s pressure to not support applied be- havior analysis services, when she might have taken a firmer stand if other teachers voiced their positions, sit- uational ethics may be at work.

Further, ethical practices are important because they increase the likelihood that appropriate services will be rendered to individuals. As a result, the individual and the culture will improve. Over time, practices that pro- duce such improvement survive and become codified into ethical rules of conduct. These codes are selected and change over time in response to new environmental is- sues, dilemmas, or questions.

Standards of Professional Practice for Applied Behavior Analysts What Are Professional Standards?

Professional standards are written guidelines or rules of practice that provide direction for conducting the prac- tices associated with an organization. Professional soci- eties and certification or licensing boards develop, refine, and revise the standards that govern their profession to provide members with parameters for appropriate be- havior in a dynamic and changing environment. In prac- tice, organizations initially form task forces to develop standards that are reviewed and approved by their re- spective boards of directors and members. In addition to prescribing rules of conduct, most professional organi- zations exact sanctions on members who do not follow the rules. Major code violations may result in expulsion from the organization, or revocation of certification or licensure.

Five complementary and interrelated documents de- scribe standards of professional conduct and ethical prac- tice for applied behavior analysts.

• Ethical Principles of Psychologists and Code of Con- duct (American Psychological Association, 2002)

• The Right to Effective Behavioral Treatment (Asso- ciation for Behavior Analysis, 1989)

• The Right to Effective Education (Association for Behavior Analysis, 1990)

• Guidelines for Responsible Conduct for Behav- ior Analysts (Behavior Analyst Certification Board, 2001)

• The BCBA and BCABA Behavior Analyst Task List—Third Edition (Behavior Analyst Certifica- tion Board, 2005)

Ethical Considerations for Applied Behavior Analysts

668

Ethical Considerations for Applied Behavior Analysts

Ethical Principles of Psychologists and Code of Conduct—American Psychological Association

The American Psychological Association first published its code of ethics in 1953 as Ethical Standards of Psy- chologists (American Psychological Association, 1953). Reflecting the changing nature of the field, eight revi- sions of the code were published between 1959 and 2002. In 1988 the Association for Behavior Analysis first adopted the American Psychological Association’s code of ethics (American Psychological Association, 2002) to guide professional practice. The five general principles and 10 areas of ethical standards on which the code is based are shown in Figure 1.

The Right to Effective Behavioral Treatment

The Association for Behavior Analysis has published two position papers describing clients’ rights. In 1986 the as- sociation established a task force to examine the rights of

people receiving behavioral treatment and how behavior analysts can ensure that clients are served appropriately. After two years of study, the task force outlined six basic client rights as the basis for directing the ethical and ap-

The Right to Effective Education

The Association for Behavior Analysis also adopted a position paper titled The Right to Effective Education (Association for Behavior Analysis, 1989). The full re- port of the task force (Barrett et al., 1991) was accepted by the ABA Executive Council. The abbreviated state- ment shown in Figure 3 was subsequently approved by majority vote of the general membership and now con- stitutes official ABA policy. This position statement es- sentially requires that assessment and educational interventions (a) rest on a foundation of solid research demonstrating effectiveness, (b) address functional re- lations between behavior and environmental events, and

Figure 2 The right to effective behavioral treatment.

1. An individual has a right to a therapeutic environment. 2. An individual has a right to services whose overriding goal is personal welfare. 3. An individual has a right to treatment by a competent behavior analyst. 4. An individual has a right to programs that teach functional skills. 5. An individual has a right to behavioral assessment and ongoing evaluation. 6. An individual has a right to the most effective treatment procedures available.

Adapted from “The Right to Effective Behavioral Treatment” by the Association for Behavior Analysis, 1989. Retrieved November 11, 2006, from www.abainternational.org/ABA/statements/treatment.asp. Copyright 1989 by the Association for Behavior Analysis. Adapted with permission.

Figure 1 Ethical principles of psychologists and code of conduct.

General Principles Ethical Standards

Principle A: Beneficence and 1. Resolving ethical Issues nonmaleficence 2. Competence

Principle B: Fidelity and responsibility 3. Human relations Principle C: Integrity 4. Privacy and confidentiality Principle D: Justice 5. Advertising and other public statements Principle E: Respect for people’s rights 6. Record keeping and fees

and dignity 7. Education and training 8. Research and publication 9. Assessment

10. Therapy

Adapted from “Ethical Principles of Psychologists and Code of Conduct,” by the American Psychological As- sociation, 2002. Retrieved November 11, 2003, from www.apa.org/ethics/code2002.html. Copyright 2002 by the American Psychological Association. Adapted with permission from the author.

propriate application of behavioral treatment (see Fig- ure 2) (Van Houten et al., 1988).

669

Figure 3 The right to effective education.

1. The student’s overall educational context should include: a. Social and physical school environments that encourage and maintain academic

achievement and progress, and discourage behavior inconsistent with those goals; b. Schools that treat students with care and individual attention, comparable to that

offered by a caring family; c. School programs that provide support and training for parents in parenting and

teaching skills; and d. Consequences and attention at home that encourage and maintain success at

school. 2. Curriculum and instructional objectives should:

a. Be based on empirically validated hierarchies or sequences of instructional objectives and measurable performance criteria that are demonstrated to promote cumulative mastery and that are of long-term value in the culture;

b. Specify mastery criteria that include both the accuracy and the speed dimensions of fluent performance;

c. Include objectives that specify both long-term and short-term personal and vocational success, and that, once mastered, will be maintained by natural consequences in everyday living; and

d. Include long-term retention and maintenance of skills and knowledge as explicitly measured instructional objectives.

3. Assessment and student placement should involve: a. Assessment and reporting methods that are sufficiently criterion-referenced to

promote useful decision-making based on actual levels of skills and knowledge rather than on categorical labels such as “emotionally disturbed” or “learning disabled,” and

b. Placement based on correspondence between measured entering skills and skills required as prerequisites for a given level in a hierarchically sequenced curriculum.

4. Instructional methods should: a. Allow students to master instructional objectives at their own pace and to respond as

rapidly and as frequently as they are able during at least some self-paced instructional session each day;

b. Provide sufficient practice opportunities to enable students to master skills and knowledge at each step in the curriculum;

c. Provide consequences designed to correct errors and/or to increase frequency of responding and that are adjusted to individual performance until they enable students to achieve desired outcomes;

d. Be sensitive to and adjust in response to measures of individual learning and performance, including use of individualized instruction when group instruction fails to produce desired outcomes;

e. Regularly employ the most advanced equipment to promote skill mastery via programs incorporating validated features described in this document; and

f. Be delivered by teachers who receive performance-based training, administrative and supervisory support, and evaluation in the use of measurably effective, scientifically validated instructional procedures, programs, and materials.

5. Measurement and summative evaluation should entail: a. Decision-making via objective curriculum-based measures of performance, and b. Reports of objectively measured individual achievement and progress rather than

subjective ratings, norm-referenced comparisons, or letter grading. 6. Responsibility for success should stipulate that:

a. Financial and operational consequences for school personnel depend on objective measures of student achievement;

b. Teachers, administrators, and the general educational program assume responsibility for student success, and change programs until students achieve their highest performance levels; and

c. Students and parents should be allowed and encouraged to change schools or school programs until their educational needs are met.

Adapted from “Students’ Right to Effective Education,” by the Association for Behavior Analysis, 1990. Retrieved November 11, 2006, from www.abainternational.org/ABA/statements/treatment.asp. Copyright 1990 by the As- sociation for Behavior Analysis. The Behavior Analyst, 1991, Volume 14(1), pages 79–82 show the full report. Adapted with permission.

Ethical Considerations for Applied Behavior Analysts

670

(c) are monitored and evaluated on a systematic and ongoing basis (Van Houten, 1994). Interventions are considered only when they are likely to be effective, based on empirical evidence and assessment results.

Guidelines for Responsible Conduct for Behavior Analysts

The Behavior Analyst Certification Board’s Guidelines for Responsible Conduct for Behavior Analysts (2004) describes specific expectations for professional practice and ethical conduct under 10 major areas (see Figure 4). The BACB’s guidelines are based on, and consistent with, a variety of other ethical guidelines, including The Bel- mont Report (produced by The National Commission for the Protection of Human Subjects of Biomedical and Be- havioral Research, 1979) and ethical codes developed and adopted by nine different professional organizations in behavior analysis and related fields (e.g., American Psychological Association, 2002; Florida Association for Behavior Analysis, 1988; National Association of School Psychologists, 2000; National Association of Social Workers, 1996).

BCBA and BCABA Behavior Analyst Task List

The BCBA and BCABA Behavior Analyst Task List de- scribes the knowledge, skills, and attributes expected of a certified behavior analyst. The task list describes 111 tasks (some have multiple subtasks) across 10 con- tent areas. Content Area 1, Ethical Considerations, con- sists of the 12 tasks listed at the beginning of this chapter.1 We will address these tasks in the remainder of this chapter.

Three of the tasks relate to professional competence (i.e., what behavior analysts need to do to become and remain professionally capable); the remaining nine tasks in Content Area 1 focus on ethics associated with the pro- vision of client services

Ensuring Professional Competence Professional competence in applied behavior analysis is achieved through academic training that involves formal coursework, supervised practica, and mentored profes- sional experience. Many excellent behavior analysts have been trained through university-based master’s and doc-

1The entire Task List is reprinted on the inside front and back covers of this book.

3Information on the certification requirements and process is available at the BACB Web site: www.BACB.com.

toral-level programs housed in psychology, education, social work, and other human service departments.2

ABA accreditation and certification bodies have specified minimum curriculum and supervised experi- ence requirements for behavior analysts. Both the Asso- ciation for Behavior Analysis (1993, 1997) and the Behavior Analyst Certification Board have set minimum standards for what constitutes training in behavior analy- sis. The Association for Behavior Analysis accredits uni- versity training programs; the BACB certifies individual practitioners. Practitioners must not only meet the certi- fication criteria, but also pass a certification examination. The BACB has conducted an extensive occupational analysis and developed the Behavior Analyst Task List (Behavior Analyst Certification Board, 2005), which specifies the minimum content that all behavior analysts should master (BACB, 2005; Shook, Johnston, & Mel- lichamp, 2004; cf. Martinez-Diaz, 2003).3

Obtaining Certification and Licensure

Potential consumers must be able to identify practicing behavior analysts who have demonstrated at least a min- imum standard of training and competence (Moore & Shook, 2001; Shook & Favell, 1996; Shook & Neisworth, 2005). In the past, most behavior analysts in private prac- tice were licensed in psychology, education, or clinical social work. The public had no way to determine whether the licensed professional had specific training in applied behavior analysis (Martinez-Diaz, 2003). In 1999 the BACB began credentialing behavior analysts in the United States and other countries with a certificate. The BACB certification program was based on Florida’s long-standing innovative certification program (Shook, 1993; Starin, Hemingway, & Hartsfield, 1993).

Practicing Within One’s Areas of Competence

Behavior analysts should practice within their areas of professional training, experience, and competence. For example, professionals with extensive experience with adults with developmental disabilities should restrict their services to this field. They should not suddenly begin working with young children diagnosed with autism- spectrum disorders. Likewise, analysts with extensive ex- perience working with adolescents and young adults

Ethical Considerations for Applied Behavior Analysts

2For a list of colleges and universities with graduate programs in behavior analysis, see the ABA Directory of Graduate Training in Behavior Analysis.

671

Figure 4 Behavior Analyst Certification Board® guidelines for responsible conduct for behavior analysts.

1.0 Responsible Conduct of a Behavior Analyst 1.01 Reliance on Scientific Knowledge 1.02 Competence and Professional Development 1.03 Competence 1.04 Professional Development 1.05 Integrity 1.06 Professional and Scientific Relationships 1.07 Dual Relationships 1.08 Exploitative Relationships

2.0 The Behavior Analyst’s Responsibility to Clients 2.01 Definition of Client 2.02 Responsibility 2.03 Consultation 2.04 Third-Party Requests for Services 2.05 Rights and Prerogatives of Clients 2.06 Maintaining Confidentiality 2.07 Maintaining Records 2.08 Disclosures 2.09 Treatment Efficacy 2.10 Documenting Professional and Scientific Work 2.11 Records and Data 2.12 Fees and Financial Arrangements 2.13 Accuracy in Reports to Those Who Pay for Services 2.14 Referrals and Fees 2.15 Interrupting or Terminating Services

3.0 Assessing Behavior 3.01 Environmental Conditions that Preclude Implementation 3.02 Environmental Conditions that Hamper Implementation 3.03 Functional Assessment 3.04 Accepting Clients 3.05 Consent—Client Records 3.06 Describing Program Objectives 3.07 Behavioral Assessment Approval 3.08 Describing Conditions for Program Success 3.09 Explaining Assessment Results

4.0 The Behavior Analyst and the Individual Behavior Change Program 4.01 Approving Interventions 4.02 Reinforcement/Punishment 4.03 Avoiding Harmful Reinforcers 4.04 Ongoing Data Collection 4.05 Program Modifications 4.06 Program Modifications Consent 4.07 Least Restrictive Procedures 4.08 Termination Criteria 4.09 Terminating Clients

5.0 The Behavior Analyst as a Teacher and/or Supervisor 5.01 Designing Competent Training Programs 5.02 Limitations on Training 5.03 Providing Course Objectives 5.04 Describing Course Requirements 5.05 Describing Evaluation Requirements 5.06 Providing Feedback to Students/Supervisees 5.07 Providing Behavior Analysis Principles in Teaching 5.08 Requirements of Supervisees 5.09 Training and Supervision 5.10 Feedback to Supervisees 5.11 Reinforcing Supervisee Behavior

Ethical Considerations for Applied Behavior Analysts

672

Figure 4 (continued)

6.0 The Behavior Analyst and the Workplace 6.01 Job Commitments 6.02 Assessing Employee Interactions 6.03 Preparing for Consultation 6.04 Employees Interventions 6.05 Employee Health and Well-Being 6.06 Conflicts with Organizations

7.0 The Behavior Analyst and Research 7.01 Scholarship and Research 7.02 Using Confidential Information for Didactic or Instructive Purposes 7.03 Conforming with Laws and Regulations 7.04 Informed Consent 7.05 Deception in Research 7.06 Informing of Future Use 7.07 Minimizing Interference 7.08 Commitments to Research Participants 7.09 Ensuring Participant Anonymity 7.10 Informing of Withdrawal 7.11 Debriefing 7.12 Answering Research Questions 7.13 Written Consent 7.14 Extra Credit 7.15 Acknowledging Contributions 7.16 Principal Authorship and Other Publication Credits 7.17 Paying Participants 7.18 Withholding Payment 7.19 Grant Reviews 7.20 Animal Research

8.0 The Behavior Analyst’s Ethical Responsibility to the Field of Behavior Analysis 8.01 Affirming Principles 8.02 Disseminating Behavior Analysis 8.03 Being Familiar with these Guidelines 8.04 Discouraging Misrepresentation by Non-Certified Individuals

9.0 The Behavior Analyst’s Responsibility to Colleagues 9.01 Ethical Violations by Colleagues 9.02 Accuracy of Data 9.03 Authorship and Findings 9.04 Publishing Data 9.05 Withholding Data

10.0 The Behavior Analyst’s Ethical Responsibility to Society 10.01 Promotion in Society 10.02 Scientific Inquiry 10.03 Public Statements 10.04 Statements by Others 10.05 Avoiding False or Deceptive Statements 10.06 Media Presentations 10.07 Testimonials 10.08 In-Person Solicitation

should not begin working with preschool clients, who would be outside of their expertise. A person whose en- tire career has been set in preschool or home-based set- tings with children should not start offering services in the area of organizational behavior management.

Ethical Considerations for Applied Behavior Analysts

Even within one’s competence area, practitioners who come upon a situation that exceeds their training or experience should make a referral to another behavior an- alyst or consultant. Where gaps in professional training exist, increased competence can be achieved by attending

673

workshops, seminars, classes, and other continuing edu- cation activities. When possible, analysts should work with mentors, supervisors, or colleagues who can provide enhanced training and professional development.

Maintaining and Expanding Professional Competence

Behavior analysts have an ethical responsibility to stay in- formed of advancements in the field. For example, con- ceptual advances and technical innovations in the 1990s in areas such as antecedent interventions, functional analysis, and motivating operations have had profound implications for clinical and educational practice. Be- havior analysts can maintain and expand their profes- sional competence by earning continuing education credits, attending and participating in professional con- ferences, reading professional literature, and presenting cases to peer review and oversight committees.

Continuing Education Units

Behavior analysts can expand their professional compe- tence and keep abreast of new developments by attend- ing training events offering continuing education unit (CEU) credits. The BACB requires a minimum number of CEUs per three-year renewal period to maintain cer- tification. CEU credits are available by attending work- shops at national and local conferences such as those of the Association for Behavior Analysis and its local af- filiates, or events sponsored by universities and other agencies approved as CEU providers by the BACB. Con- tinuing education credits demonstrate that the behavior analyst has added relevant awareness, knowledge, and/or skills to his or her repertoire.

Attending and Presenting at Conferences

Attending and participating in local, state, or national con- ferences enhance every behavior analyst’s skills. The axiom, “You never learn something so well as when you have to teach it” continues to have value. Hence, partici- pating in a conference helps practitioners refine their skills.

Professional Reading

Self-study is a fundamental way to maintain currency in an ever-changing field. In addition to routinely reading the Journal of Applied Behavior Analysis (JABA) and The Behavior Analyst, all behavior analysts should study other behavioral publications specific to their areas of expertise and interest.

Oversight and Peer Review Opportunities

When behavior analysts approach a difficult problem, they apply skills and techniques within their repertoire to ad- dress the problem. For example, in a case of a child with a history of severe and frequent face-slapping, an inter- vention that involved the child wearing a helmet or pro- tective device might be faded as differential reinforcement for appropriate object-holding becomes more effective. Fading the helmet presents no ethical problems as long as the technique remains conceptually systematic, is based in the basic principles of behavior, is tied to the published research literature in some way, and is effective in the pre- sent case. However, even the most technically proficient and professionally careful practitioner is subject to con- tingencies that can lead to treatment drift or error. This is where oversight and peer review come into play.

Many states have laws requiring that oversight be provided under specific circumstances, which are defined either by the type and severity of behaviors being ad- dressed or the restrictiveness of the procedure(s) being proposed. The presence or absence of laws in a particu- lar jurisdiction should not determine our dedication to a process of peer review and oversight, however these safe- guards protect not only the consumers of behavior ana- lytic services, but also behavior analysts themselves.

When findings are presented before a group of col- leagues outside of the behavior analytic community, clear demonstrations of procedures that comply with clinical and professional standards must apply. Moreover, such presen- tations set the occasion for behavior analysts to clarify be- havioral outcomes, present graphic displays that are easy to interpret, and explain in clear and rational language why various educational or clinical choices were made.

Making and Substantiating Professional Claims

Sometimes overzealous practicing behavior analysts are so certain of the superiority of operant and respondent principles that they make preemptive claims that are not realistic. For example, asserting, “I am certain I can help your son” borders on an unethical claim. A more ethical and appropriate statement may be, “I have had success working with other children who have profiles similar to your son’s.” Behavior analysts who are well acquainted with the professional literature on treatment effective- ness for the target behaviors, the functions that behav- iors serve (e.g., attention, escape), and specific client populations are less likely to make unsubstantiated and far-reaching claims.

A second aspect of this professional standard relates to presenting oneself as having certifications, licenses, educational experiences, or training that one does not have.

Ethical Considerations for Applied Behavior Analysts

674

Figure 5 Example of an informed consent form.

Informed Consent Form A.B.A. TECHNOLOGIES, INC.

CLIENT: ____________________________________________ DOB: _______________

STATEMENT OF AUTHORITY TO CONSENT: I certify that I have the authority to legally consent to assessment, release of information, and all legal issues involving the above- named client. Upon request, I will provide A.B.A. Technologies, Inc., with proper legal documentation to support this claim. I further hereby agree that if my status as legal guardian should change, I will immediately inform A.B.A. Technologies, Inc., of this change in status and will further immediately inform A.B.A. Technologies, Inc., of the name, address, and phone number of the person or persons who have assumed guardianship of the above-named client.

TREATMENT CONSENT: I consent for behavioral treatment to be provided for the above- named client by A.B.A. Technologies, Inc., and its staff. I understand that the procedures used will consist of manipulating antecedents and consequences to produce improvements in behavior. At the beginning of treatment behavior may get worse in the environment where the treatment is provided (e.g., “extinction burst”) or in other settings (e.g., “behavioral contrast”). As part of the behavioral treatment, physical prompting and manual guidance may be used. The actual treatment protocols that will be used have been explained to me.

I understand that I may revoke this consent at any time. However, I cannot revoke consent for action that has already been taken. A copy of this consent shall be as valid as the original.

PARENT/GUARDIAN: _________________________________ DATE: ______________

WITNESS: ___________________________________________ DATE: _______________

________________________________________________________________________

From ABA Technologies, Inc., by José A. Martinez-Diaz, Ph.D., BCBA, 129 W. Hibiscus Blvd., Melbourne, FL. Used with permission.

Falsely claiming to have applied behavior analysis expe- riences or valid credentials is always unethical and may be illegal in many states.

Ethical Issues in Client Services As stated earlier, although applied behavior analysis shares many ethical concerns with other disciplines, some ethical questions are specific to the choices posed when behavior analysis services are considered. For example, the decision to use an aversive procedure poses complex ethical questions that must be addressed prior to imple- mentation (Herr, O’Sullivan, & Dinerstein, 1999; Iwata, 1988; Repp & Singh, 1990).

Although our discussion is limited to the most com- mon areas of ethical concern related to the BACB stan- dards, we strongly advise that students and practitioners read the BACB’s guidelines and task-list objectives to maintain their ethical perspective and skills across a wide range of standards.

Informed Consent

Informed consent means that the potential recipient of services or participant in a research study gives his or her explicit permission before any assessment or treatment is provided. Informed consent requires more than ob- taining permission. Permission must come after full dis- closure and information is provided to the participant. Figure 5 shows an example of an informed consent let- ter in which such information is provided.

Three tests must be met before informed consent can be considered valid: (a) The person must demonstrate the capacity to decide, (b) the person’s decision must be vol- untary, and (c) the person must have adequate knowledge of all salient aspects of the treatment.

Capacity to Decide

To be considered capable of making an informed deci- sion, a participant must have (a) an adequate mental process or faculty by which he or she acquires knowl- edge, (b) the ability to select and express his or her

Ethical Considerations for Applied Behavior Analysts

675

Figure 6 Factors surrogates must consider when making informed consent decisions for persons who are incapacitated.

FOR A PERSON WHOSE WISHES MAY BE KNOWN OR INFERRED 1. The person’s current diagnosis and prognosis. 2. The person’s expressed preference regarding the treatment at issue. 3. The person’s relevant religious or personal beliefs. 4. The person’s behavior and attitude toward medical treatment. 5. The person’s attitude toward a similar treatment for another individual. 6. The person’s expressed concerns about the effects of his or her illness and treatment

on family and friends.

FOR A PERSON WHOSE WISHES ARE UNKNOWN AND PROBABLY UNKNOWABLE 1.The effects of the treatment on the physical, emotional, and mental functions of the

person. 2. The physical pain that the person would suffer from the treatment or from the

withholding or withdrawal of the treatment. 3. The humiliation, loss of dignity, and dependency the person is suffering as a result of

the condition or as a result of the treatment. 4. The effect the treatment would have on the life expectancy of the person. 5. The person’s potential for recovery, with the treatment and without the treatment. 6. The risks, side effects, and benefits of treatment.

Adapted From “Informed Consent for Health Care” by A. D. N. Hurley & J. L. O’Sullivan, 1999, in R. D. Diner- stein, S. S. Herr, & J. L. O’Sullivan (Eds.), A Guide to Consent, Washington DC, American Association on Mental Retardation, pp. 50–51. Used by permission.

choices, and (c) the ability to engage in a rational process of decision making. Concepts such as “mental process or faculty” are mentalistic to behavior analysts, and there are no agreed-upon evaluative tools for testing preinter- vention capacity. Capacity is questioned only if the per- son “has impaired or limited ability to reason, remember, make choices, see the consequences of actions, and plan for the future” (O’Sullivan, 1999, p. 13). A person is con- sidered mentally incapacitated if a disability affects his or her ability to understand the consequences of his or her actions (Turnbull & Turnbull, 1998).

According to Hurley and O’Sullivan (1999), “Capac- ity to give informed consent is a fluid concept and varies with each individual and proposed procedure” (p. 39). A person may have capacity to consent to participate in a pos- itive reinforcement program that poses little or no risk, but may not have the capacity to understand the complexities that might be involved in a more complex treatment such as overcorrection combined with response cost. Hence, practitioners should not assume that a client capable of in- formed consent at one level can provide informed consent if the complexity of the proposed treatment changes.

Capacity must be viewed within a legal context as well as a behavioral one. Courts have held that capacity requires that the person “rationally understand the nature of the procedure, its risk, and other relevant information” (Kaimowitz v. Michigan Department of Mental Health, 1973, as cited in Neef, Iwata, & Page, 1986, p. 237). The determination of capacity by persons with developmen-

tal disabilities poses specific challenges, and the behav- ior analyst would benefit from consulting a legal expert whenever a question of capacity arises for such persons.

When a person is deemed incapacitated, informed consent may be obtained either through a surrogate or a guardian.

Surrogate Consent. Surrogate consent is a legal process by which another individual—the surrogate—is authorized to make decisions for a person deemed incom- petent based on the knowledge of what the incapacitated person would have wanted. Family members or close friends most often serve as surrogates.

In most states, a surrogate’s authority is limited. A surrogate may not authorize treatment when the client is actively refusing treatment (e.g., when an adult with developmental disabilities refuses to sit in a dentist’s chair, a surrogate cannot consent to sedation); or for con- troversial medical procedures such as sterilization, abor- tion, or certain treatments for mental disorders (e.g., electroconvulsive therapy or psychotropic medication) (Hurley & O’Sullivan, 1999). The surrogate is required to consider specific information when making decisions. For a person who is incapacitated, the landmark case of Superintendent of Belchertown State School v. Saikewicz (1977) crystallized factors that surrogates must consider when making informed consent decisions for persons who cannot make their own decisions (Hurley & O’Sullivan, 1999). Figure 6 lists necessary information

Ethical Considerations for Applied Behavior Analysts

676

for the two primary areas of concern for surrogates: (a) decision making for a person who is incapacitated and whose wishes may be known or inferred, and (b) deci- sion making for a person whose wishes are unknown and probably unknowable (e.g., a person with profound disabilities).

Guardian Consent. Guardian consent is obtained through a guardian, a person whom a court appoints as a legal custodian of an individual. Guardianship is a complex legal issue that varies from state to state. Hence, only two main points will be made. First, guardianship may be sought when treatment is deemed necessary but a surrogate is inappropriate because the client without capacity refuses treatment. The greater the degree of guardianship, the less legal control a person has over his or her own life. Because helping the person become more independent is a goal of be- havior analysis, a decision to seek guardianship at any level should be considered only as a final option to be exercised in the most limited way that still resolves the issue.

Second, guardianship may be limited in any way that the court deems appropriate. In most states, a full guardian is responsible for essentially every important decision in a person’s life. The court, in protecting the rights of the individual, may determine that limited or temporary guardianship is more appropriate. The guardianship may apply only to financial or medical con- cerns, and may remain in effect only so long as a very specific concern is at issue (such as a need for surgery). In all guardianship cases, the court is the ultimate decision-making body and may take any action including revoking guardianship or determining who should serve as the guardian (O’Sullivan, 1999).

Voluntary Decision

Consent is considered voluntary when it is given in the absence of coercion, duress, or any undue influence and when it is issued with the understanding that it can be withdrawn at any time. As Yell (1998) stated, “Revoca- tion of consent has the same effects as an initial refusal to consent” (p. 274).

Family members, doctors, support staff, or others may exert strong influence over a person’s willingness to grant or deny consent (Hayes, Adams, & Rydeen, 1994). For example, individuals with developmental dis- abilities may be asked to make major decisions during interdisciplinary team meetings. The way questions are posed to the prospective client may suggest that consent is not entirely voluntary (e.g., “You want us to help you with this intervention, right?”).

Ethical Considerations for Applied Behavior Analysts

To ensure that a person’s consent is voluntary, a prac- titioner may want to discuss topics related to assessment and treatment privately and with an independent advo- cate present. Further, having discussions without severe time constraints reduces the pressure to evoke a response. Finally, anyone being asked to give consent should be given time to think, discuss, and review all options with a trusted confidant to ensure that consent is voluntary.

Knowledge of the Treatment

A person considering services or research participation must be provided information in clear, nontechnical lan- guage regarding (a) all important aspects of the planned treatment, (b) all potential risks and benefits of the planned procedure, (c) all potential alternative treatments, and (d) the right to refuse continued treatment at any time. The recipient must be able to answer questions correctly about the information he or she has been given, and should be able to describe the procedures in his or her own words. If, for example, a time-out procedure is part of the intervention package, the client should be able to describe all aspects of how time-out will operate and be able to say more than, “I’ll get in trouble if I hit another person.” Figure 7 outlines additional information that the potential client should have to ensure voluntary, informed consent (Cipani, Robinson, & Toro, 2003).

Treatment Without Consent

Most states have policies authorizing a course of action when a potential procedure is needed and informed con- sent by the individual is not possible. Typically, consent may be granted in the case of a life-threatening emer- gency, or when there is an imminent risk of serious harm. In educational instances in which the school district de- termines that services are needed or desired (e.g., special education programming), but the parents refuse, the dis- trict has progressive recourse through administrative re- view, mediation, and ultimately the court system (Turnbull & Turnbull, 1998). State statutes differ in in- stances in which consent is denied; thus, practitioners should consult current local or state statutes. Further, practitioners should recognize that rules and regulations associated with state laws may be subject to change and therefore require regular review.

Confidentiality

Professional relationships require confidentiality, mean- ing that any information regarding an individual receiv- ing or having received services may not be discussed with, or otherwise made available to any third party, un- less that individual has provided explicit authorization

677

Ethical Considerations for Applied Behavior Analysts

Figure 7 Information for clients to ensure informed consent.

Documentation Required: All forms indicating an understanding of information and agreement to various policies listed below must be signed and dated.

Issues Presented/Discussed: 1. Confidentiality and its limits—how information will be used, with whom it will be shared 2. Provider qualifications 3. Risks/benefits of treatment 4. Nature of procedures and alternatives 5. Logistics of service

• Monetary issues—Fee structure, billing, methods of payment, insurance issues, other fees

• Communication—ways to contact between appointments by phone, pager, etc. • Cancellations and missed appointments • Termination of services

6. Responsibilities of client and behavior analyst

Adapted from “Ethical and Risk Management Issues in the Practice of ABA” by E. Cipani, S. Robinson, and H. Toro (2003). Paper presented at the annual conference of the Florida Association for Behavior Analysis, St. Petersburg. Adapted with permission of the authors.

for release of that information. Although confidentiality is a professional ethical standard for behavior analysts, it is also a legal requirement in some states (Koocher & Keith-Spiegel, 1998). Figure 8 shows a standard release of information (ROI) form. Note that the ROI form clearly specifies what can be shared and released, and the expiration date for such releases.

Limits to Confidentiality

When service to a client begins, the limits on confiden- tiality must be fully explained to the client. For instance, confidentiality does not extend to abusive situations and when knowledge of impending harm to the individual or others is known. All professionals must report suspected abuse of children in all states, and suspected abuse of el- ders in most states. Figure 9 shows a form typical for in- forming the client of abuse-reporting requirements.

As stated previously, in circumstances in which the practicing behavior analyst becomes aware of the possi- bility of imminent, severe harm to the individual or an- other person, confidentiality no longer applies. In such cases, it is ethical to inform supervisors, administrators, or other caregivers of the pending injury so that appro- priate preventative action can be taken.

Breaches of Confidentiality

Breaches of confidentiality generally occur for two main reasons: (a) The breach is intentional to protect some- one from harm, or (b) the breach is unintentional and is the result of carelessness, neglect, or a misunderstand- ing of the nature of confidentiality. An intentional

breach of confidentiality is warranted when credible in- formation becomes available that imminent harm or danger is possible. For instance, if a reliable student ap- proached a teacher after learning that another student brought a gun to school, confidentiality can be breached to protect potential victims. Such a breach is intended to serve the greater good (i.e., protect others from im- minent harm). Unintentional breaches can occur when a teacher unwittingly provides confidential information to a parent about a child’s performance, but fails to con- firm that the parent who is requesting the information is the de facto legal guardian for the child. To avoid this second type of breach, practitioners must remain vigi- lant about information sharing throughout all phases of service delivery.

A good rule to follow with respect to confidentiality is: If uncertain whether confidentiality applies in a given situation, assume that it does. Some actions to protect confidentiality include locking file cabinets, requiring passwords to access computer files, avoiding the trans- mission of nonencrypted information across wireless sys- tems, and confirming an individual’s status as a surrogate or legal guardian before providing information concern- ing the client.

Protecting the Client’s Dignity, Health, and Safety

Dignity, health, and safety issues often center on the con- tingencies and physical structures present in the environ- ments in which people live and work. The behavior analyst should be acutely aware of these issues. Favell and McGim- sey (1993) provided a list of acceptable characteristics of

678

Ethical Considerations for Applied Behavior Analysts

Figure 9 Example of abuse-reporting protocol form.

A.B.A. Technologies, Inc. Confidentiality Act/Abuse-Reporting Protocol

Client: ___________________________

I understand that all information related to the above-named client’s assessment and treatment must be handled with strict confidentiality. No information related to the client, either verbal or written, will be released to other agencies or individuals without the express written consent of the client’s legal guardian. By law, the rules of confidentiality do not hold under the following conditions:

1. If abuse or neglect of a minor, disabled, or elderly person is reported or suspected, the professional involved is required to report it to the Department of Children and Families for investigation.

2. If, during the course of services, the professional involved receives information that someone’s life is in danger, that professional has a duty to warn the potential victim.

3. If our records, our subcontractor records or staff testimony are subpoenaed by court order, we are required to produce requested information or appear in court to answer questions regarding the client.

This consent expires 1 year after signature date below.

Parent/Legal Guardian Date

From ABA Technologies, Inc., by José A. Martinez-Diaz, Ph.D., BCBA, 129 W. Hibiscus Blvd., Melbourne, FL. Used with permission.

Figure 8 Release of information form.

A.B.A. TECHNOLOGIES, INC. RELEASE OF INFORMATION AND ASSESSMENT CONSENT FORM

CLIENT: _______________________________________________ DOB: ______________

PARENT/GUARDIAN NAME: __________________________________________________

I consent for the above-named client to participate in an assessment through A.B.A. Technologies, Inc. I consent to have the assessment with the above-named individual conducted at the following locations (Circle relevant ones)

Home School Other: ___________

I understand and consent to have the individuals responsible for care in the above-named locations involved in the assessment of the above-named client. In order to coordinate the assessment with these individuals, I authorize the release of the following confidential records to the individuals responsible for care in the above-named locations:

Evaluations/Assessments: ____________________________________________________

___________________________________________________________________________

IEP or Other Records: ________________________________________________________

Other: _____________________________________________________________________

I understand that these records may contain psychiatric and/or drug and alcohol information. I understand that these records may also contain references to blood-borne pathogens (e.g., HIV, AIDS). I understand that I may revoke this consent at any time; however, I cannot revoke consent for action that has already been taken. A copy of this release shall be as valid as the original. THIS CONSENT AUTOMATICALLY EXPIRES 30 DAYS AFTER TERMINATION OF SERVICES.

PARENT/GUARDIAN: _______________________________ DATE: _____________

From ABA Technologies, Inc., by José A. Martinez-Diaz, Ph.D., BCBA, 129 W. Hibiscus Blvd., Melbourne, FL. Used with permission.

679

Ethical Considerations for Applied Behavior Analysts

Figure 10 Defining an acceptable treatment environment.

1. The environment is engaging: Reinforcement is readily available; problem behaviors are reduced; reinforcement contingencies are effective; exploratory play and practice flourish; the environment is by definition humane.

2. Functional skills are taught and maintained: Environments are judged not by paperwork or records, but by observed evidence of training and progress; incidental teaching allows the natural environment to support skill acquisition and maintenance.

3. Behavior problems are ameliorated: A purely functional approach regarding both behaviors and procedures leads to effective interventions; arbitrary labels based on topographies are replaced by individualized definitions based on function.

4. The environment is the least restrictive alternative. Again, this is functionally defined, based on freedom of movement and activity engagement parameters; community settings may actually be more restrictive than institutional ones, depending on their behavioral effects.

5. The environment is stable: Changes in schedules, programs, peers, caregivers, etc., are minimized; skills are maintained; consistency and predictability.

6. The environment is safe: Physical safety is paramount; supervision and monitoring are adequate; peer review ensures that proper program procedures are based on function.

7. The client chooses to live there: Efforts are made to determine client choice; alternative environments may be sampled.

Adapted from “Defining an Acceptable Treatment Environment” by J. E. Favell and J. E. McGimsey, 1993. In R. Van Houten & S. Axelrod (Eds.), Behavior Analysis and Treatment (pp. 25–45). New York: Plenum Press. Copyright 1993 by Plenum Press. Used with permission.

treatment environments to ensure dignity, health, and safety (see Figure 10).

Dignity can be examined by addressing the follow- ing questions: Do I honor the person’s choices? Do I pro- vide adequate space for privacy? Do I look beyond the person’s disability and treat the person with respect? Be- havior analysts can help to ensure dignity for their clients by defining their own roles. By using operant principles of behavior, they teach skills that will enable learners to establish increasingly effective control over the contin- gencies in their own natural environments. Everyone has the right to say yes, no, or sometimes nothing at all (cf. Bannerman, Sheldon, Sherman, & Harchik, 1990).

Choice is a central principle in the delivery of ethi- cal behavioral services. In behavioral terms, the act of making a choice requires that both behavioral and stim- ulus alternatives be possible and available (Hayes et al., 1994). Practitioners must provide the client with behav- ioral alternatives, and the person must be capable of per- forming the actions required by the alternatives. To leave a room, a person must be physically capable of opening the door, and the door must not be blocked or locked. The term stimulus alternatives refers to the simultane- ous presence of more than one stimulus item from which to choose (e.g., choosing to eat an apple instead of an or- ange or pear). The point is that, to have a fair choice, a client must have alternatives, must be able to perform each alternative, and must be able to experience the nat- ural consequences of the chosen alternative.

Helping the Client Select Outcomes and Behavior Change Targets

The term outcomes refers to the lifestyle changes a client has identified as the ultimate goals of behavior analysis services. In part, it refers to a quality of life issue for the individual (Felce & Emerson, 2000; Risley, 1996). Ob- taining a job, establishing a loving relationship with a significant other, pursuing personal goals, engaging in community-based activities, and living independently are examples of outcomes. Ultimately, behaviors selected for change must benefit the individual, not the practitioner or caregiver. Peterson, Neef, Van Norman, and Ferreri (2005) expressed this point succinctly:

One reason often cited for providing opportunities for choices is that it is consistent with social values such as self-determination and empowerment. . . . We also main- tain that we can best “empower” students by arranging conditions to promote choices that benefit the student, and by helping students to recognize the factors that can influence their choices. (p. 126)

In the early days of the field, behavior analysts were criticized rightfully for targeting behaviors that mostly benefited staff (e.g., Winett & Winkler, 1972). For ex- ample, achieving a sheltered workshop of docile, com- pliant adults is not an appropriate goal if functional work-related or social skills are not addressed adequately. Compliance, although commonly identified in behavior

680

Ethical Considerations for Applied Behavior Analysts

programs, is not an appropriate goal by itself. Teaching compliant behavior becomes ethically justifiable when it facilitates the development of other functional or social skills that will, in turn, help the person achieve higher levels of independence.

To help people select appropriate outcomes and tar- get behaviors within an ethical framework, practitioners need to have a thorough understanding of variables that affect reinforcer assessment, the function of stimuli iden- tified as reinforcers in an assessment, and how these vari- ables interact when a choice is made. As Peterson and colleagues (2005) stated:

It may be that the act of choosing is effective only to the extent that it permits access to preferred reinforcers. We posit that assessment of the variables that affect prefer- ence contributes to efforts to promote choices that are beneficial. (p. 132)

Maintaining Records

In addition to data on behavior change, behavior analysts must keep records of their interactions with clients. As- sessment data, descriptions of interventions, and progress notes are part of a client record that must be considered confidential. The following points on maintaining records comply with the APA Ethical Principles of Psychologists and Code of Conduct (2002):

• A release must be obtained from the client or his or her guardian (unless ordered otherwise by a judge) before records may be shared with anyone.

• Records must be kept in a secure area.

• Well-maintained records facilitate the provision of future services, meet agency or institutional re- quirements, ensure accurate billing, allow for future research, and may comply with legal requirements.

• Records disposal should be complete (shredding is probably best).

• Electronic transmission of confidential records across any unsecured medium (e.g., fax lines in public areas, e-mail) is prohibited under the Health Insurance Portability and Accountability Act (HIPAA) regulations (1996).

Advocating for the Client Providing Necessary and Needed Services

Prior to initiating services, a behavior analyst has the re- sponsibility to validate that a referral warrants further ac- tion. This poses the first ethical challenge to the

practitioner: deciding whether to accept or reject the case. The decision to provide treatment may be divided into two sequential decisions: (1) Is the presenting problem amenable to behavioral intervention? and (2) is the pro- posed intervention likely to be successful?

Is the Problem Amenable to Behavior Treatment?

To determine whether behavioral intervention is neces- sary and appropriate, the behavior analyst should seek answers to the following questions:

1. Has the problem emerged suddenly?

a. Might the problem have a medical cause?

b. Has a medical evaluation been done?

2. Is the problem with the client or with someone else? (For example, a student who has done well through fourth grade suddenly exhibits “behavior problems” in fifth grade, although she remains well behaved at home. Perhaps a simple change of teachers will solve the problem.)

3. Have other interventions been tried?

4. Does the problem actually exist? (For example, a parent has serious concerns about a 3-year-old child who “refuses to eat” everything provided at every meal.)

5. Can the problem be solved simply or informally? (For example, the same 3-year-old “elopes” through a back door left wide open all day.)

6. Might the problem be better addressed by another discipline? (For example, a child with cerebral palsy may need adaptive equipment instead of be- havioral treatment.)

7. Is the behavioral problem considered an emergency?

Is the Proposed Intervention Likely to Be Successful?

Questions to ask when considering whether the inter- vention is likely to be successful include the following:

1. Is the client willing to participate?

2. Are the caregivers surrounding the client willing or able to participate?

3. Has the behavior been successfully treated in the research literature?

4. Is public support likely?

5. Does the behavior analyst have the appropriate ex- perience to deal with the problem?

681

Ethical Considerations for Applied Behavior Analysts

6. Will those most likely to be involved in imple- menting the program have adequate control of the critical environmental contingencies?

If the answer to all of these questions is yes, then the behavior analyst can act. If the answer to any of these questions is no, the behavior analyst should seriously consider declining to initiate treatment.

Embracing the Scientific Method

With respect to how the scientific method relates to ethics, practitioners use measurably effective, research-based methods and test emerging practices to assess their ef- fectiveness before implementing them. As space engi- neer James Oberg once stated, “In science keeping an open mind is a virtue, but not so open that your brains fall out” (cited in Sagan, 1996, p. 187). Embracing the scientific method is important as claims of the effective- ness of treatment regimens and intervention methods be- come more widespread and accepted by the public without critical examination (Shermer, 1997). In Sagan’s (1996) view, extraordinary claims require extraordinary evidence.

Heward and Silvestri (2005) elaborated further on the essential ingredients of extraordinary evidence.

What constitutes extraordinary evidence? In the strictest sense, and the sense that should be employed when eval- uating claims of educational effectiveness, evidence is the outcome of the application of the scientific method to test the effectiveness of a claim, a theory, or a prac- tice. Evidence becomes extraordinary when it is extraor- dinarily well tested. . . . In addition, extraordinary evidence requires replication. A single study, anecdote, or theoretical article, no matter how impressive the find- ings or how complicated the writing, is not a basis for practice. (p. 209)

Applied behavior analysts should base their practices on two primary sources: the scientific literature and direct and frequent measurements of behaviors. Literature- based data inform initial decisions about intervention. However, practicing ethically also means that consulta- tions with other professionals who have tackled similar problems is often necessary.4

Peer-reviewed scientific reports published in rep- utable outlets are a strong source of objective informa- tion concerning effective intervention strategies. When confronted with data that differ from well-established, peer-reviewed literature, practitioners are obliged to in- vestigate the situation further. By examining critical vari-

4Remember that in applied behavior analysis parlance, similarity means likeness in function (i.e., controlling variables) as well as form (i.e., topography).

ables or combinations of interactions, the practitioner would be likely to determine differences that would ben- efit the individual.

There are almost as many ideas about instructional strategies and behavioral treatments as there are teach- ers and people treating problem behaviors. Unfortunately, many ideas have not been empirically validated. Take, for example, the popular notion in education that repeated practice of a skill is harmful to the student because it de- stroys his or her motivation to learn and ultimately re- sults in unhappiness with school (e.g., Heron, Tincani, Peterson, & Miller, 2005; Kohn, 1997). Despite the pop- ular appeal of this philosophy, it is simply unsubstanti- ated. Decades of carefully controlled and replicated research have demonstrated that repeated practice is a necessary ingredient to skill mastery (Binder, 1996; Ericsson & Charness, 1994).

The area of autism treatment offers another example of untested claims permeating the mainstream culture. In the 1990s a technique called facilitated communication was hailed as an amazing breakthrough for people with autism. Reports surfaced claiming that children who pre- viously had no communication skills were now able to communicate, sometimes in very sophisticated language, with the aid of a facilitator. The facilitator guided the child’s hand over a keyboard as he or she typed messages. The results of numerous experiments showed that it was the facilitators who typed the messages, not the children. (e.g., Green & Shane, 1994; Szempruch & Jacobson, 1993; Wheeler, Jacobson, Paglieri, & Schwartz, 1993).

Carefully controlled scientific studies debunked what had become a popular perception among desperate par- ents and teachers. For behavior analysts, the cornerstone of ethical practice is providing effective services based on solid, replicated research evidence.

Evidenced-Based Best Practice and Least Restrictive Alternatives.

An essential component of ethically driven behavior analysis is that interventions and related practices be ev- idence based, and that the most powerful, but least in- trusive methods be used first. Further, intervention plans must be designed, implemented, and evaluated system- atically. If the individual fails to show progress, data sys- tems should be reviewed and if necessary interventions modified. If progress is made, the intervention should be phased out and assessed for generalization and mainte- nance. In all phases of an intervention, data and direct observations should drive treatment decisions.

682

Ethical Considerations for Applied Behavior Analysts

Conflict of Interest

A conflict of interest occurs when a principal party, alone or in connection with family, friends, or associates, has a vested interest in the outcome of the interaction. The most common form of conflict arises in the form of dual- role relationships. Conflicts arise when a person acting as a therapist enters into another type of relationship with the client, a family member, or a close associate of the client, or promises to enter into such a relationship in the fu- ture. These relationships may be financial, personal, pro- fessional (providing another service, for example), or otherwise beneficial to the therapist in some way.

Direct and frequent observations during assessment and intervention phases bring the behavior analyst into close contact with the client (and often family members, other professionals, and caregivers) in various natural set- tings. Personal relationships can be formed that cross pro- fessional boundaries. Family members may offer unsolicited gifts or make invitations to parties or other events. In every interaction, the behavior analyst must monitor her own behavior with vigilance and guard against crossing any personal or professional boundaries. This is especially true when treatment is provided within a private home. In service contexts, personal relation- ships of any kind can quickly develop ethical tangles, and are to be avoided.

Other professional conflicts of interest might arise as well. A teacher, for example, should not simultane- ously be a student’s employer outside of school because performance in one area can affect behavior in the other. A supervisee should be off limits as a potential romantic partner. A member of a peer review committee should not participate in reviews of his own work or the work of his supervisees. The general rule to follow with respect to conflict of interest is to avoid it. A practitioner in doubt

should consult with a supervisor or trusted and experi- enced confidant.

Conclusion Practicing ethically can be challenging and uncertain be- cause anticipating the full ramifications of a decision or a given course of action is not always possible. Further- more, it is hard work because practicing ethically requires vigilance, self-monitoring, and the application of dynamic guidelines and principles with virtually every case.

Returning to our scenarios at the beginning of the chapter, how might the director, the support coordinator, and the teacher proceed? Prescriptive, pat solutions are not possible. Yet these practitioners need not be at a loss as to what to do. In our view, ethical challenges, regard- less of their sources, are best addressed by revisiting the three questions posed early in the chapter: What is the right thing to do? What is worth doing? What does it mean to be a good practitioner? Behavior analysts who use these three questions as the focal points for their decision mak- ing will have a foundation for responding to such dilem- mas. Further, addressing these questions honestly, openly, and without prejudice will help practitioners avoid many potential ethical pitfalls and keep the client, student, or care recipient at the center of their efforts.

Finally, by practicing ethically, and by maintaining adherence to the scientific method and the principles, procedures, and dimensions of applied behavior analy- sis, practitioners will have a ready source of valid, accu- rate, reliable, and believable data to inform decision making. As a result, they will be more likely to achieve the original and continuing promise of applied behavior analysis: the application of the experimental analysis of behavior to problems of social importance.

Summary What Is Ethics and Why Is It Important?

1. Ethics describes behaviors, practices, and decisions that address three basic and fundamental questions: What is the right thing to do? What is worth doing? What does it mean to be a good behavior analyst?

2. Ethics is important because it helps practitioners decide whether a course of action is morally right or wrong, and it guides decisions to act irrespective of the demands of expediency, pressure, or priorities.

3. Ethical practices are derived from other behavior analysts and professionals to ensure the well-being of (a) clients,

(b) the profession itself, and (c) the culture as a whole. Over time, practices become codified into ethical rules of conduct that may change in response to a changing world.

4. Personal histories, including cultural and religious expe- riences, influence how practitioners decide a course of ac- tion in any given situation.

5. Practicing behavior analysts must be aware of the partic- ular settings in which they work and the specific rules and ethical standards applicable to those settings. Behavior might be illegal but ethical, legal but unethical, or illegal and unethical.

683

Standards of Professional Practice for Applied Behavior Analysts

6. Professional organizations adopt formalized statements of conduct, codes of professional behavior, or ethical stan- dards of behavior to guide their members in making deci- sions. Also, standards exact sanctions when deviations from the code occur.

7. The client’s welfare is at the forefront of ethical deci- sion making.

8. Behavior analysts must take steps to ensure that their pro- fessional conduct is self-regulating.

9. Professional standards, guidelines, or rules of practice are written statements that provide direction for conducting the practices associated with an organization.

10. Behavior analysts are guided by five documents with re- spect to ethical behavior: Ethical Principles of Psycholo- gists and Code of Conduct (APA, 2002), The Right to Effective Behavioral Treatment (ABA, 1989), The Right to Effective Education (ABA, 1990), Guidelines for Respon- sible Conduct for Behavior Analysts (BACB, 2001), and the Behavior Analyst Task List (BACB, 2005).

Ensuring Professional Competence

11. Professional competence in applied behavior analysis is achieved through formal academic training that involves coursework, supervised practica, and mentored profes- sional experience.

12. Behavior analysts must be truthful and accurate in all pro- fessional and personal interactions.

Ethical Issues in Client Services

13. Three tests must be met before informed consent can be considered valid: capacity to decide, voluntary decision, and knowledge of the treatment.

14. Confidentiality refers to the professional standard requir- ing that the behavior analyst not discuss or otherwise re- lease information regarding someone in his or her care. Information may be released only with formal permission of the individual or the individual’s guardian.

15. Practicing behavior analysts have a responsibility to pro- tect a client’s dignity, health, and safety. Various rights must be observed and protected, including the right to make choices, the right to privacy, the right to a thera- peutic treatment environment, and the right to refuse treatment.

16. The individual receiving services must be given the op- portunity to assist in selecting and approve of treatment goals. Outcomes must be selected that are primarily aimed at benefiting the individual receiving services.

17. Records of service must be maintained, preserved, and considered confidential.

18. In deciding to provide services, the behavior analyst must determine that services are needed, that medical causes have been ruled out, that the treatment environment will support service delivery, and that a reasonable expectation of success exists.

19. All sources of conflict of interest, and in particular dual re- lationships, are to be avoided.

Advocating for the Client

20. The choice to provide treatment may be divided into two sets of decision rules: (1) determining that the problem is amenable to behavioral intervention, and (2) evaluating the likely success of an intervention.

21. A conflict of interest occurs when a principal party, alone or in connection with family, friends, or associates, has a vested interest in the outcome of the interaction.

Ethical Considerations for Applied Behavior Analysts

684

Achenbach, T. M., & Edelbrock, C. S. (1991). Manual for the Child Behavior Checklist. Burlington: University of Vermont, Department of Psychiatry.

Adams, C., & Kelley, M. (1992). Managing sibling aggression: Overcorrection as an alternative to timeout. Behavior Modification, 23, 707–717.

Adelinis, J. D., Piazza, C. C., & Goh, H. L. (2001). Treatment of multiply con- trolled destructive behavior with food reinforcement. Journal of Applied Be- havior Analysis, 34, 97–100.

Adkins, V. K., & Mathews, R. M. (1997). Prompted voiding to reduce inconti- nence in community-dwelling older adults. Journal of Applied Behavior Analysis, 30, 153–156.

Adronis, P. T. (1983). Symbolic aggression by pigeons: Contingency coadduction. Unpublished doctoral dissertation, Uni- versity of Chicago, Department of Psy- chiatry and Behavior Analysis, Chicago.

Agran, M. (Ed.). (1997). Self-directed learn- ing: Teaching self-determination skills. Pacific Grove, CA: Brooks/Cole.

Ahearn, W. H. (2003). Using simultaneous presentation to increase vegetable con- sumption in a mildly selective child with autism. Journal of Applied Be- havior Analysis, 36, 361–365.

Ahearn, W. H., Clark, K. M., Gardenier, N. C., Chung, B. I., & Dube, W. V. (2003). Persistence of stereotypic behavior: Ex- amining the effects of external rein- forcers. Journal of Applied Behavior Analysis, 36, 439–448.

Ahearn, W. H., Kerwin, M. E., Eicher, P. S., Shantz, J., & Swearingin, W. (1996). An alternating treatments comparison of two intensive interventions for food refusal. Journal of Applied Behavior Analysis, 29, 321–332.

Alber, S. R., & Heward, W. L. (1996). “GOTCHA!” Twenty-five behavior traps guaranteed to extend your students’aca- demic and social skills. Intervention in School and Clinic, 31 (5), 285–289.

Alber, S. R., & Heward, W. L. (2000). Teaching students to recruit positive at- tention: A review and recommenda- tions. Journal of Behavioral Education, 10, 177–204.

Alber, S. R., Heward, W. L., & Hippler, B. J. (1999). Training middle school stu- dents with learning disabilities to re- cruit positive teacher attention. Exceptional Children, 65, 253–270.

Alber, S. R., Nelson, J. S., & Brennan, K. B. (2002). A comparative analysis of two homework study methods on elemen- tary and secondary students’ acquisi- tion and maintenance of social studies content. Education and Treatment of Children, 25, 172–196.

Alberto, P. A., & Troutman, A. C. (2006) Applied behavior analysis for teachers (7th ed.). Upper Saddle River, NJ: Mer- rill/Prentice Hall.

Alberto, P. A., Heflin, L. J., & Andrews, D. (2002). Use of the timeout ribbon pro- cedure during community-based in- struction. Behavior Modification, 26 (2), 297–311.

Albin, R. W., & Horner, R. H. (1988). Gen- eralization with precision. In R. H. Horner, G. Dunlap, & R. L. Koegel (Eds.), Generalization and mainte- nance: Life-style changes in applied set- tings (pp. 99–120). Baltimore: Brookes.

Alessi, G. (1992). Models of proximate and ultimate causation in psychology. American Psychologist, 48, 1359–1370.

Alexander, D. F. (1985). The effect of study skill training on learning disabled stu- dents’ retelling of expository material.

Journal of Applied Behavior Analysis, 18, 263–267.

Allen, K, E., Hart, B. M., Buell, J. S., Har- ris, F. R., & Wolf, M. M. (1964). Ef- fects of social reinforcement on isolate behavior of a nursery school child. Child Development, 35, 511–518.

Allen, K. D., & Evans, J. H. (2001). Expo- sure-based treatment to control excessive blood glucose monitoring. Journal of Applied Behavior Analysis, 34, 497–500.

Allen, L. D., Gottselig, M., & Boylan, S. (1982). A practical mechanism for using free time as a reinforcer in the classroom. Education and Treatment of Children, 5 (4), 347–353.

Allison, J. (1993). Response deprivation, re- inforcement, and economics. Journal of the Experimental Analysis of Behav- ior, 60, 129–140.

Altschuld, J. W., & Witkin, B. R. (2000). From needs assessment to action: Transforming needs into solution strategies. Thousand Oaks, CA: Sage.

Altus, D. E., Welsh, T. M., & Miller, L. K. (1991). A technology for program maintenance: Programming key re- searcher behaviors in a student hous- ing cooperative. Journal of Applied Behavior Analysis, 24, 667–675.

American Psychological Association. (1953). Ethical standards of psycholo- gists. Washington, DC: Author.

American Psychological Association. (2001). Publication Manual of the American Psychological Association (5th ed.). Washington, DC: Author

American Psychological Association. (2002). Ethical principles of psycholo- gists and code of conduct. Washington, DC: Author. Retrieved November 11, 2003, from www.apa.org/ethics/ code2002.html.

Bibliography

685

Bibliography

American Psychological Association. (2004). Ethical principles of psycholo- gists and code of conduct. Retrieved October 21, 2004, from www.apa .org/ethics.

Andersen, B. L., & Redd, W. H. (1980). Pro- gramming generalization through stim- ulus fading with children participating in a remedial reading program. Education and Treatment of Children, 3, 297–314.

Anderson, C. M., & Long, E. S. (2002). Use of a structured descriptive assessment methodology to identify variable affect- ing problem behavior. Journal of Ap- plied Behavior Analysis, 35, 137–154.

Andresen, J. T. (1991). Skinner and Chom- sky 30 years later OR: The return of the repressed. The Behavior Analyst, 14, 49–60.

Ardoin, S. P., Martens, B. K., & Wolfe, L. A. (1999). Using high-probability in- struction sequences with fading to in- crease student compliance during transitions. Journal of Applied Behav- ior Analysis, 32, 339–351.

Armendariz, F., & Umbreit, J. (1999). Using active responding to reduce disruptive behavior in a general education class- room. Journal of Positive Behavior In- terventions, 1, 152–158.

Arndorfer, R. E., Miltenberger, R. G., Woster, S. H., Rortvedt, A. K., & Gaffaney, T. (1994). Home-based de- scriptive and experimental analysis of problem behaviors in children. Topics in Early Childhood Special Educa- tion, 14, 64–87.

Arndorfer, R., & Miltenberger, R. (1993). Functional assessment and treatment of challenging behavior: A review with implications for early childhood. Topics in Early Childhood Special Education, 13, 82–105.

Arnesen, E. M. (2000). Reinforcement of ob- ject manipulation increases discovery. Unpublished bachelor’s thesis, Reed College, Portland, OR.

Arntzen, E., Halstadtrø, A., & Halstadtrø, M. (2003). Training play behavior in a 5-year-old boy with developmental dis- abilities. Journal of Applied Behavior Analysis, 36, 367–370.

Ashbaugh, R., & Peck, S. M. (1998). Treat- ment of sleep problems in a toddler: A replication of the faded bedtime with response cost protocol. Journal of Ap- plied Behavior Analysis, 31, 127–129.

Association for Behavior Analysis. (1989). The right to effective education. Kala- mazoo, MI: Author. Retrieved No- vember 11, 2006, from www. abainternational.org/ABA/statements/ treatment.asp.

Association for Behavior Analysis. (1990). Students’ right to effective education. Kalamazoo, MI: Author. Retrieved November 11, 2006, from www. abainternational.org/ABA/statements/ treatment.asp

Association for Behavior Analysis. (1993, 1997). Guidelines for the accreditation of programs in behavior analysis. Kala- mazoo, MI:Author. Retrieved December 2, 2003, from www. abainternational. org/sub/behaviorfield/education/ accreditation/index.asp.

Association for Persons with Severe Hand- icaps. (1987, May). Resolution on the cessation of intrusive interventions. TASH Newsletter, 5, 3.

Atwater, J. B., & Morris, E. K. (1988). Teachers’ instructions and children’s compliance in preschool classrooms: A descriptive analysis. Journal of Applied Behavior Analysis, 21, 157–167.

Axelrod, S. A. (1990). Myths that (mis)guide our profession. In A. C. Repp & N. N. Singh (Eds.), Per- spectives on the use of nonaversive and aversive interventions for persons with developmental disabilities (pp. 59–72). Sycamore, IL: Sycamore.

Axelrod, S., Hall, R. V., Weis, L., & Rohrer, S. (1971). Use of self-imposed contin- gencies to reduce the frequency of smoking behavior. Paper presented at the Fifth Annual Meeting of the Asso- ciation for the Advancement of Behav- ior Therapy, Washington, DC.

Ayllon, T., & Azrin, N. H. (1968). The token economy: A motivational system for therapy and rehabilitation. New York: Appleton-Century-Crofts.

Ayllon, T., & Michael, J. (1959). The psy- chiatric nurse as a behavioral engineer. Journal of the Experimental Analysis of Behavior, 2, 323–334.

Azrin, N. H. (1960). Sequential effects of punishment. Science, 131, 605–606.

Azrin, N. H., & Besalel, V. A. (1999). How to use positive practice, self-correction, and overcorrection (2nd ed.). Austin, TX: Pro-Ed.

Azrin, N. H., & Foxx, R. M. (1971). A rapid method of toilet training the institu-

tionalized retarded. Journal of Applied Behavior Analysis, 4, 89–99.

Azrin, N. H., & Holz, W. C. (1966). Pun- ishment. In W. K. Honig (Ed.), Operant behavior: Areas of research and application (pp. 380–447). New York: Appleton-Century-Crofts.

Azrin, N. H., & Nunn, R. G. (1973). Habit- reversal for habits and tics. Behavior Research and Therapy, 11, 619–628.

Azrin, N. H., & Powers, M. A. (1975). Elim- inating classroom disturbances of emo- tionally disturbed children by positive practice procedures. Behavior Therapy, 6, 525–534.

Azrin, N. H., & Wesolowski, M. D. (1974). Theft reversal: An overcorrection pro- cedure for eliminating stealing by re- tarded persons. Journal of Applied Behavior Analysis, 7, 577–581.

Azrin, N. H., Holz, W. C., & Hake, D. C. (1963). Fixed-ratio punishment by in- tense noise. Journal of the Experimen- tal Analysis of Behavior, 6, 141–148.

Azrin, N. H., Hutchinson, R. R., & Hake, D. C. (1963). Pain-induced fighting in the squirrel monkey. Journal of the Exper- imental Analysis of Behavior, 6, 620.

Azrin, N. H., Kaplan, S. J., & Foxx, R. M. (1973). Autism reversal: Eliminating stereotyped self-stimulation of retarded individuals. American Journal of Men- tal Deficiency, 78, 241–248.

Azrin, N. H., Nunn, R. G., & Frantz, S. E. (1980). Habit reversal vs. negative prac- tice treatment of nail biting. Behavior Research and Therapy, 18, 281–285.

Azrin, N. H., Rubin, H., O’Brien, F., Ayllon, T., & Roll, D. (1968). Behavioral engi- neering: Postural control by a portable operant apparatus. Journal of Applied Behavior Analysis, 1, 99–108.

Babyak, A. E., Luze, G. J., & Kamps, D. M. (2000). The good student game: Be- havior management for diverse class- rooms. Intervention in School and Clinic, 35 (4), 216–223.

Bacon-Prue, A., Blount, R., Pickering, D., & Drabman, R. (1980). An evaluation of three litter control procedures: Trash re- ceptacles, paid workers, and the marked item techniques. Journal of Applied Be- havior Analysis, 13, 165–170.

Baer, D. M. (1960). Escape and avoidance response of preschool children to two schedules of reinforcement withdrawal. Journal of the Experimental Analysis of Behavior, 3, 155–159.

686

Bibliography

Baer, D. M. (1961). Effect of withdrawal of positive reinforcement on an extin- guishing response in young children. Child Development, 32, 67–74.

Baer, D. M. (1962). Laboratory control of thumbsucking by withdrawal and rep- resentation of reinforcement. Journal of the Experimental Analysis of Behav- ior, 5, 525–528.

Baer, D. M. (1970). An age-irrelevant con- cept of development. Merrill-Palmer Quarterly, 16, 238–245.

Baer, D. M. (1971). Let’s take another look at punishment. Psychology Today, 5, 5–32.

Baer, D. M. (1975). In the beginning, there was the response. In E. Ramp & G. Semb (Eds.), Behavior analysis: Areas of research and application (pp. 16–30). Upper Saddle River, NJ: Prentice Hall.

Baer, D. M. (1977a). Reviewer’s comment: Just because it’s reliable doesn’t mean that you can use it. Journal of Applied Behavior Analysis, 10, 117–119.

Baer, D. M. (1977b). “Perhaps it would be better not to know everything.” Journal of Applied Behavior Analysis, 10, 167–172.

Baer, D. M. (1981). A hung jury and a Scot- tish verdict: “Not proven.” Analysis and Intervention in Developmental Dis- abilities, 1, 91–97.

Baer, D. M. (1982). Applied behavior analy- sis. In G. T. Wilson & C. M. Franks (Eds.), Contemporary behavior ther- apy: Conceptual and empirical foun- dations (pp. 277–309). New York: Guilford Press.

Baer, D. M. (1985). [Symposium discus- sant]. In C. E. Naumann (Chair), Dev- eloping response classes: Why re- invent the wheel? Symposium con- ducted at the Annual Conference of the Association for Behavior Analysis, Columbus, OH.

Baer, D. M. (1987). Weak contingencies, strong contingencies, and too many be- haviors to change. Journal of Applied Behavior Analysis, 20, 335–337.

Baer, D. M. (1991). Tacting “to a fault”. Journal of Applied Behavior Analysis, 24, 429–431.

Baer, D. M. (1999). How to plan for gener- alization (2nd ed.). Austin, TX: Pro-Ed.

Baer, D. M. (2005). Letters to a lawyer. In W. L. Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Car- tledge, R. Gardner, III, L. D. Peterson, S. B. Hersh, & J. C. Dardig (Eds.),

Focus on behavior analysis in educa- tion: Achievements, challenges, and op- portunities (pp. 3–30). Upper Saddle River, NJ: Merrill/Prentice Hall.

Baer, D. M., & Bushell, Jr., D. (1981). The future of behavior analysis in the schools? Consider its recent pact, and then ask a different question. School Psychology Review, 10(2), 259–270.

Baer, D. M., & Fowler, S. A. (1984). How should we measure the potential of self- control procedures for generalized ed- ucational outcomes? In W. L. Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behavior analysis in education (pp. 145–161). Columbus, OH: Charles E. Merrill.

Baer, D. M., & Richards, H. C. (1980). An interdependent group-oriented contin- gency system for improving academic performance. School Psychology Re- view, 9, 190–193.

Baer, D. M., & Schwartz, I. S. (1991). If re- liance on epidemiology were to be- come epidemic, we would need to assess its social validity. Journal of Ap- plied Behavior Analysis, 24, 321–234.

Baer, D. M., & Sherman, J. A. (1964). Re- inforcement control of generalized im- itation in young children. Journal of Experimental Child Psychology, 1, 37–49.

Baer, D. M., & Wolf, M. M. (1970a). Recent examples of behavior modification in preschool settings. In C. Neuringer & J. L. Michael (Eds.), Behavior mod- ification in clinical psychology (pp. 10–55). Upper Saddle River, NJ: Pren- tice Hall.

Baer, D. M., & Wolf, M. M. (1970b). The entry into natural communities of rein- forcement. In R. Ulrich, T. Stachnik, & J. Mabry (Eds.), Control of human be- havior (Vol. 2, pp. 319–324). Glenview, IL: Scott, Foresman.

Baer, D. M., Peterson, R. F., & Sherman, J. A. (1967). The development of imitation by reinforcing behavioral similarity of a model. Journal of the Experimental Analysis of Behavior, 10, 405–416.

Baer, D. M., Wolf, M. M., & Risley, T. R. (1968). Some current dimensions of ap- plied behavior analysis. Journal of Ap- plied Behavior Analysis, 1, 91–97.

Baer, D. M., Wolf, M. M., & Risley, T. (1987). Some still-current dimensions of applied behavior analysis. Journal of Applied Behavior Analysis, 20, 313–327.

Baer, R. A. (1987). Effects of caffeine on classroom behavior, sustained atten- tion, and a memory task in preschool children. Journal of Applied Behavior Analysis, 20, 225–234.

Baer, R. A., Blount, R., L., Detrich, R., & Stokes, T. F. (1987). Using intermittent reinforcement to program maintenance of verbal/nonverbal correspondence. Journal of Applied Behavior Analysis, 20, 179–184.

Baer, R. A., Tishelman, A. C., Degler, J. D., Osnes, P. G., & Stokes, T. F. (1992). Ef- fects of self- vs. experimenter-selection of rewards on classroom behavior in young children. Education and Treat- ment of Children, 15, 1–14.

Baer, R. A., Williams, J. A., Osnes, P. G., & Stokes, T. F. (1984). Delayed re- inforcement as an indiscriminable contingency in verbal/nonverbal corre- spondence training. Journal of Applied Behavior Analysis, 17, 429–440.

Bailey, D. B. (1984). Effects of lines of progress and semilogarithmic charts on ratings of charted data. Journal of Ap- plied Behavior Analysis, 17, 359–365.

Bailey, D. B., Jr., & Wolery, M. (1992). Teaching infants and preschoolers with disabilities (2nd ed). Upper Saddle River, NJ: Merrill/Prentice Hall.

Bailey, J. S. (2000). A futurist perspective for applied behavior analysis. In J. Austin & J. E. Carr (Eds.), Handbook of applied behavior analysis (pp. 473–488). Reno, NV: Context Press.

Bailey, J., & Meyerson, L. (1969). Vibration as a reinforcer with a profoundly re- tarded child. Journal of Applied Be- havior Analysis, 2, 135–137.

Bailey, J. S. & Pyles, D. A. M. (1989). Be- havioral diagnostics. In E. Cipani (Ed.), The treatment of severe behavior disor- ders: Behavior analysis approach (pp. 85–107). Washington, DC: American Association on Mental Retardation.

Bailey, S. L., & Lessen, E. I. (1984). An analysis of target behaviors in educa- tion: Applied but how useful? In W. L. Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behavior analysis in education (pp. 162–176). Columbus, OH: Charles E. Merrill.

Ballard, K. D., & Glynn, T. (1975). Behav- ioral self-management in story writ- ing with elementary school children. Journal of Applied Behavior Analysis, 8, 387–398.

687

Bibliography

Bandura, A. (1969). Principles of behavior modification. New York: Holt, Rine- hart & Winston.

Bandura, A. (1971). Vicarious and self-re- inforcement processes. In R. Glaser (Ed.), The nature of reinforcement. New York: Academic Press.

Bannerman, D. J., Sheldon, J. B., Sherman, J. A., & Harchik, A. E. (1990). Balanc- ing the rights to habilitation with the right to personal liberties: The rights of people with developmental disabilities to eat too many doughnuts and take a nap. Journal of Applied Behavior Analysis, 23, 79–89.

Barbetta, P. M., Heron, T. E., & Heward, W. L. (1993). Effects of active student re- sponse during error correction on the acquisition, maintenance, and general- ization of sight words by students with developmental disabilities. Journal of Applied Behavior Analysis, 26, 111–119.

Barker, M. R., Bailey, J. S., & Lee, N. (2004). The impact of verbal prompts on child safety-belt use in shopping carts. Journal of Applied Behavior Analysis, 37, 527–530.

Barkley, R., Copeland, A., & Sivage, C. (1980). A self-control classroom for hy- peractive children. Journal of Autism and Developmental Disorders, 10, 75–89.

Barlow, D. H., & Hayes, S. C. (1979). Al- ternating treatments design: One strat- egy for comparing the effects of two treatments in a single behavior. Journal of Applied Behavior Analysis, 12, 199–210.

Baron, A., & Galizio, M. (2005). Positive and negative reinforcement: Should the distinction be preserved? The Behavior Analyst, 28, 85–98.

Baron, A., & Galizio, M. (2006). The dis- tinction between positive and negative reinforcement: Use with care. The Be- havior Analyst, 29, 141–151.

Barrett, B. H., Beck, R., Binder, C., Cook, D. A., Engelmann, S., Greer, R. D., Kyrklund, S. J., Johnson, K. R., Mal- oney, M., McCorkle, N., Vargas, J. S., & Watkins, C. L. (1991). The right to effective education. The Behavior An- alyst, 14 (1), 79–82.

Barrish, H. H., Saunders, M., & Wolf, M. M. (1969). Good behavior game: Effects of individual contingencies for group consequences on disruptive be-

havior in a classroom. Journal of Ap- plied Behavior Analysis, 2, 119–124.

Barry, A. K. (1998). English grammar: Lan- guage as human behavior. Upper Sad- dle River, NJ: Prentice Hall.

Barton, E. S., Guess, D., Garcia, E., & Baer, D. M. (1970). Improvement of retar- dates’ mealtime behaviors by timeout procedures using multiple baseline techniques. Journal of Applied Behav- ior Analysis, 3, 77–84.

Barton, L. E., Brulle, A. R., & Repp, A. C. (1986). Maintenance of therapeutic change by momentary DRO. Journal of Applied Behavior Analysis, 19, 277–282.

Barton-Arwood, S. M., Wehby, J. H., Gunter, P. L., & Lane, K. L. (2003). Functional behavior assessment rating scales: In- trarater reliability with students with emotional or behavioral disorders. Behavior Disorders, 28, 386–400.

Baum, W. M. (1994). Understanding be- haviorism: Science, behavior, and cul- ture. New York: Harper Collins.

Baum, W. M. (2005). Understanding be- haviorism: Science, behavior, and cul- ture (2nd ed.). Malden, MA: Blackwell Publishing.

Bay-Hinitz, A. K., Peterson, R. F., & Quil- itch, H. R. (1994). Cooerrative games: A way to modify aggressive and coop- erative behaviors in young children. Journal of Applied Behavior Analysis, 27, 435–446.

Becker, W. C., & Engelmann, S. E. (1978). Systems for basic instruction: Theory and applications. In A. Catania & T. Brigham (Eds.), Handbook of applied behavior analysis: Social and instruc- tional processes. New York: Irvington.

Becker, W. C., Engelmann, S., & Thomas, D. R. (1975). Teaching 2: Cognitive learning and instruction. Chicago: Sci- ence Research Associates.

Behavior Analyst Certification Board. (2001). Guidelines for responsible con- duct for behavior analysts. Tallahassee, FL: Author. Retrieved November 11, 2003, from http://bacb.com/consum _frame.html.

Behavior Analyst Certification Board. (2005). Behavior analyst task list, third edition. Tallahassee, FL: Author. Re- trieved November 11, 2003, from http:// bacb.com/consum_frame.html.

Belfiore, P. J., Skinner, C. H., & Ferkis, M. A. (1995). Effects of response and

trial repetition on sight-word training for students with learning disabilities. Journal of Applied Behavior Analysis, 28, 347–348.

Bell, K. E., Young, K. R., Salzberg, C. L., & West, R. P. (1991). High school driver education using peer tutors, direct in- struction, and precision teaching. Jour- nal of Applied Behavior Analysis, 24, 45–51.

Bellack, A. S., & Hersen, M. (1977). Be- havior modification: An introductory textbook. New York: Oxford Univer- sity Press.

Bellack, A. S., & Schwartz, J. S. (1976). As- sessment for self-control programs. In M. Hersen & A. S. Bellack (Eds.), Be- havioral assessment: A practical hand- book (pp. 111–142). New York: Pergamon Press.

Bellamy, G. T., Horner, R. H., & Inman, D. P. (1979). Vocational habilitation of severely retarded adults. Austin, TX: Pro-Ed.

Bender, W. N., & Mathes, M. Y. (1995). Stu- dents with ADHD in the inclusive classroom: A hierarchical approach to strategy selection. Intervention in School & Clinic, 30 (4), 226–234.

Bennett, K., & Cavanaugh, R. A. (1998). Ef- fects of immediate self-correction, de- layed self-correction, and no correction on the acquisition and maintenance of multiplication facts by a fourth-grade stu- dent with learning disabilities. Journal of Applied BehaviorAnalysis, 31, 303–306.

Bicard, D. F. & Neef, N. A. (2002). Effects of strategic versus tactical instructions on adaptation to changing contingencies in children with ADHD. Journal of Ap-

Bijou, S. W. (1955). A systematic approach to an experimental analysis of young chil- dren. Child Development, 26, 161–168.

Bijou, S. W. (1957). Patterns of reinforcement and resistance to extinction in young children. Child Development, 28, 47–54.

Bijou, S. W. (1958). Operant extinction after fixed-interval schedules with young children. Journal of the Experimental Analysis of Behavior, 1, 25–29.

Bijou, S. W., & Baer, D. M. (1961). Child development: Vol. 1. A systematic and empirical theory. New York: Appleton- Century-Crofts.

Bijou, S. W., & Baer, D. M. (1965). Child development: Vol. 2. Universal stage of

plied Behavior Analysis, 35, 375–389.

688

Bibliography

infancy. New York: Appleton-Century- Crofts.

Bijou, S. W., Birnbrauer, J. S., Kidder, J. D., & Tague, C. (1966). Programmed in- struction as an approach to teaching of reading, writing, and arithmetic to re- tarded children. The Psychological Record, 16, 505–522.

Bijou, S. W., Peterson, R. F., & Ault, M. H. (1968). A method to integrate descrip- tive and experimental field studies at the level of data and empirical con- cepts. Journal of Applied Behavior Analysis, 1, 175–191.

Billings, D. C., & Wasik, B. H. (1985). Self- instructional training with preschool- ers: An attempt to replicate. Journal of Applied Behavior Analysis, 18, 61–67.

Billingsley, F., White, D. R., & Munson, R. (1980). Procedural reliability: A ration- ale and an example. Behavioral As- sessment, 2, 247–256.

Binder, C. (1996). Behavioral fluency: Evo- lution of a new paradigm. The Behav- ior Analyst, 19, 163–197.

Binder, L. M., Dixon, M. R., & Ghezzi, P. M. (2000). A procedure to teach self- control to children with attention deficit hyperactivity disorder. Journal of Ap- plied Behavior Analysis, 33, 233–237.

Birnbrauer, J. S. (1979). Applied behavior analysis, service, and the acquisition of knowledge. The Behavior Analyst, 2, 15–21.

Birnbrauer, J. S. (1981). External validity and experimental investigation of indi- vidual behavior. Analysis and Inter- vention in Developmental Disabilities, 1, 117–132.

Birnbrauer, J. S., Wolf, M. M., Kidder, J. D., & Tague, C. E. (1965). Classroom be- havior of retarded pupils with token re- inforcement. Journal of Experimental Child Psychology, 2, 219–235.

Bishop, B. R., & Stumphauzer, J. S. (1973). Behavior therapy of thumb sucking in children: A punishment (time out) and generalization effect—what’s a mother to do? Psychological Reports, 33, 939–944.

Bjork, D. W. (1997). B. F. Skinner: A life. Washington, DC: American Psycho- logical Association.

Blew, P. A., Schwartz, I. S., & Luce, S. C. (1985). Teaching functional commu- nity skills to autistic children using nonhandicapped peer tutors. Journal

of Applied Behavior Analysis, 18, 337–342.

Blick, D. W., & Test, D. W. (1987). Effects of self-recording on high school stu- dents’ on-task behavior. Learning Dis- ability Quarterly, 10, 203–213.

Bloom, L. (1970). Language development: Form and function in emerging gram- mars. Cambridge, MA: MIT Press.

Bloom, M., Fischer, J., & Orme, J. G. (2003). Evaluating practice: Guide- lines for the accountable professional (4th ed.). Boston: Allyn & Bacon.

Bolin, E. P., & Goldberg, G. M. (1979). Be- havioral psychology and the Bible: General and specific considerations. Journal of Psychology and Theology, 7, 167–175.

Bolstad, O., & Johnson, S. (1972). Self- regulation in the modification of disruptive classroom behavior. Jour- nal of Applied Behavior Analysis, 5, 443–454.

Bondy, A., & Frost, L. (2002). The Picture Ex- change Communication System. Newark, DE: Pyramid Educational Products.

Boring, E. G. (1941). Statistical frequencies as dynamic equilibria. Psychological Review, 48, 279–301.

Bornstein, P. H., & Quevillon, R. P. (1976). The effects of a self-instructional pack- age on overactive preschool boys. Journal of Applied Behavior Analysis, 9, 179–188.

Bosch, S., & Fuqua, W. R. (2001). Behav- ioral cusps: A model for selecting tar- get behaviors. Journal of Applied Behavior Analysis, 34, 123–125.

Bourret, J., Vollmer, T. R., & Rapp, J. T. (2004). Evaluation of a vocal mand as- sessment and vocal mand procedures. Journal of Applied Behavior Analysis, 37, 129–144.

Bowers, F. E., Woods, D. W., Carlyon, W. D., & Friman, P. C. (2000). Using positive peer reporting to improve the social interactions and acceptance of socially isolated adolescents in resi- dential care: A systematic replication. Journal of Applied Behavior Analysis, 33, 239–242.

Bowman, L. G., Piazza, C. C., Fisher, W., Hagopian, L. P., & Kogan, J. S. (1997). Assessment of preference for varied versus constant reinforcement. Journal of Applied Behavior Analysis, 30, 451–458.

Boyajian, A. E., DuPaul, G. J., Wartel Han- dler, M., Eckert, T. L., & McGoey, K. E. (2001). The use of classroom- based brief functional analyses with preschoolers at risk for attention deficit hyperactivity disorder. School Psy- chology Review, 30, 278–293.

Boyce, T. E., & Geller, E. S. (2001). A tech- nology to measure multiple driving behaviors without self-report or partic- ipant reactivity. Journal of Applied Be- havior Analysis, 34, 39–55.

Boyle, J. R., & Hughes, C. A. (1994). Ef- fects of self-monitoring and subsequent fading of external prompts on the on- task behavior and task productivity of elementary students with moderate mental retardation. Journal of Behav- ioral Education, 4, 439–457.

Braam, S. J., & Poling, A. (1982). Develop- ment of intraverbal behavior in men- tally retarded individuals through transfer of stimulus control procedures: Classification of verbal responses. Applied Research in Mental Retarda- tion, 4, 279–302.

Braine, M. D. S. (1963). The ontogeny of English phrase structure: The first phrase. Language, 39, 1–13.

Brame, P. B. (2001). Making sustained silent reading (SSR) more effective: Effects of a story fact recall game on students’ off-task behavior during SSR and re- tention of story facts. Unpublished doc- toral dissertation, The Ohio State University, Columbus, OH.

Brame, P., Bicard, S. C., Heward, W. L., & Greulich, H. (2007). Using an indis- criminable group contingency to “wake up” sustained silent reading: Effects on off-task behavior and recall of story facts. Manuscript submitted for publi- cation review.

Brantley, D. C., & Webster, R. E. (1993). Use of an independent group contin- gency management system in a regular classroom setting. Psychology in the Schools, 30, 60–66.

Brantner, J. P., & Doherty, M. A. (1983). A review of timeout: A conceptual and methodological analysis. In S. Axelrod & J. Apsche (Eds.), The effects of pun- ishment on human behavior (pp. 87–132). New York: Academic Press.

Brethower, D. C., & Reynolds, G. S. (1962). A facilitative effect of punish- ment on unpunished behavior. Journal

689

Bibliography

of the Experimental Analysis of Be- havior, 5, 191–199.

Briggs, A., Alberto, P., Sharpton, W., Berlin, K., McKinley, C., & Ritts, C. (1990). Generalized use of a self-operated audio prompt system. Education and Train- ing in Mental Retardation, 25, 381–389.

Brigham, T. A. (1980). Self-control revis- ited: Or why doesn’t anyone read Skinner anymore? The Behavior Ana- lyst, 3, 25–33.

Brigham, T. A. (1983). Self-management: A radical behavioral perspective. In P. Karoly & F. H. Kanfer (Eds.), Self- management and behavior change: From theory to practice (pp. 32–59). New York: Pergamon Press.

Brobst, B., & Ward, P. (2002). Effects of public posting, goal setting, and oral feedback on the skills of female soccer players. Journal of Applied Behavior Analysis, 27, 247–257.

Broden, M., Hall, R. V., & Mitts, B. (1971). The effect of self-recording on the classroom behavior of two eighth-grade students. Journal of Applied Behavior Analysis, 4, 191–199.

Brothers, K. J., Krantz, P. J., & McClan- nahan, L. E. (1994). Office paper re- cycling: A function of container proximity. Journal of Applied Behav- ior Analysis, 27, 153–160.

Browder, D. M. (2001). Curriculum and as- sessment for students with moderate and severe disabilities. New York: Guilford Press.

Brown, K. A., Wacker, D. P., Derby, K. M., Peck, S. M., Richman, D. M., Sasso, G. M., Knutson, C. L., & Harding, J. W. (2000). Evaluating the effects of functional communication training in the presence and absence of establish- ing operations. Journal of Applied Be- havior Analysis, 33, 53–71.

Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.

Brown, R., Cazden, C., & Bellugi, U. (1969). The child’s grammar from I to III (pp. 28–73). In J. P. Hill (Ed.), The 1967 symposium on child psychology. Min- neapolis: University of Minnesota Press.

Brown, S. A., Dunne, J. D, & Cooper, J. O. (1996) Immediate retelling’s effect on student retention. Education and Treat- ment of Children, 19, 387–407.

Browning, R. M. (1967). A same-subject de- sign for simultaneous comparison of

three reinforcement contingencies. Behavior Research and Therapy, 5, 237–243.

Budd, K. S., & Baer, D. M. (1976). Behavior modification and the law: Implications of recent judicial decisions. Journal of Psychiatry and Law, 4, 171–244.

Burgio, L. D., Whitman, T. L., & Johnson, M. R. (1980). A self-instructional pack- age for increasing attending behavior in educable mentally retarded children. Journal of Applied Behavior Analysis, 13, 443–459.

Bushell, D., Jr., & Baer, D. M. (1994). Mea- surably superior instruction means close, continual contact with the rele- vant outcome data. Revolutionary! In R. Gardner, III, D. M. Sainato, J. O. Cooper, T. E. Heron, W. L. Heward, J. Eshleman, & T. A. Grossi (Eds.), Behavior analysis in education: Focus on measurably superior instruction (pp. 3–10). Pacific Grove, CA: Brooks/Cole.

Byrd, M. R., Richards, D. F., Hove, G., & Friman, P. C. (2002). Treatment of early onset hair pulling as a simple habit. Behavior Modification, 26 (3), 400–411.

Byrne, T., LeSage, M. G., & Poling, A. (1997). Effects of chlorpromazine on rats’ acquisition of lever-press re- sponding with immediate and delayed reinforcement. Pharmacology Bio- chemistry and Behavior, 58, 31–35.

Caldwell, N. K., Wolery, M., Werts, M. G., & Caldwell, Y. (1996). Embedding in- structive feedback into teacher-student interactions during independent seat- work. Journal of Behavioral Educa- tion, 6, 459–480.

Cameron, J. (2005). The detrimental effects of reward hypothesis: Persistence of a view in the face of disconfirming evi- dence. In W. L. Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner, III, L. D. Peterson, S. B. Hersh, & J. C. Dardig (Eds.), Focus on behavior analysis in education: Achievements, challenges, and opportunities (pp. 304–315). Upper Saddle River, NJ: Merrill/Prentice Hall.

Cammilleri, A. P., & Hanley, G. P. (2005). Use of a lag differential reinforcement contingency to increase varied selec- tions of classroom activities. Jour- nal of Applied Behavior Analysis, 38, 111–115.

Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.

Campbell, R. C., & Stremel-Campbell, K. (1982). Programming “loose training” as a strategy to facilitate language gen- eralization. Journal of Applied Behav- ior Analysis, 15, 295–301.

Carr, E. G., & Durand, V. M. (1985). Re- ducing behavior problems through functional communication training. Journal of Applied Behavior Analysis, 18, 111–126.

Carr, E. G., & Kologinsky, E. (1983). Ac- quisition of sign language by autistic children II: Spontaneity and general- ization effects. Journal of Applied Be- havior Analysis, 16, 297–314.

Carr, E. G., & Lovaas, I. O. (1983). Contin- gent electric shock as a treatment for severe behavior problems. In S. Axel- rod & J. Apsche (Eds.), The effects of punishment on human behavior (pp. 221–245). New York: Academic Press.

Carr, J. E., & Burkholder, E. O. (1998). Cre- ating single-subject design graphs with Microsoft Excel. Journal of Applied Behavior Analysis, 31 (2), 245–251.

Carr, J. E., Kellum, K. K., & Chong, I. M. (2001). The reductive effects of non- contingent reinforcement: Fixed-time versus variable-time schedules. Jour- nal of Applied Behavior Analysis, 34, 505–509.

Carr, J. E., Nicolson, A. C., & Higbee, T. S. (2000). Evaluation of a brief multiple- stimulus preference assessment in a naturalistic context. Journal of Applied Behavior Analysis, 33, 353–357.

Carroll, R. J., & Hesse, B. E. (1987). The ef- fects of alternating mand and tact train- ing on the acquisition of tacts. The Analysis of Verbal Behavior, 5, 55–65.

Carter, J. F. (1993). Self-management: Edu- cation’s ultimate goal. Teaching Ex- ceptional Children, 25(3), 28–32.

Carter, M., & Grunsell, J. (2001). The be- havior chain interruption strategy: A re- view of research and discussion of future directions. Journal of the Asso- ciation for Persons with Severe Hand- icaps, 26 (1), 37–49.

Carton, J. S., & Schweitzer, J. B. (1996). Use of token economy to increase compliance during hemodialysis. Journal of Applied Behavior Analysis, 29, 111–113.

690

Bibliography

Catania, A. C. (1972). Chomsky’s formal analysis of natural languages: A behav- ioral translation. Behaviorism, 1, 1–15.

Catania, A. C. (1975). The myth of self-re- inforcement. Behaviorism, 3, 192–199.

Catania, A. C. (1976). Self-reinforcement revisited. Behaviorism, 4, 157–162.

Catania, A. C. (1992). B.F. Skinner, Or- ganism. American Psychologist, 48, 1521–1530.

Catania, A. C. (1998). Learning (4th ed.). Upper Saddle River, NJ: Prentice Hall.

Catania, A. C., & Harnad, S. (Eds.). (1988). The selection of behavior: The operant behaviorism of B. F. Skinner: Com- ments and controversies. New York: Cambridge University Press.

Catania, A. C., & Hineline, P. N. (Eds.). (1996). Variations and selections: An anthology of reviews from the Journal of the Experimental Analysis of Behavior. Bloomington, IN: Society for the Experimental Analysis of Behavior.

Cautela, J. R. (1971). Covert conditioning. In A. Jacobs & L. B. Sachs (Eds.), The psychology of private events: Perspec- tive on covert response systems (pp. 109–130). New York: Academic Press.

Cavalier, A., Ferretti, R., & Hodges, A. (1997). Self-management within a classroom token economy for students with learning disabilities. Research in Developmental Disabilities, 18 (3), 167–178.

Cavanaugh, R. A., Heward, W. L., & Donel- son, F. (1996). Effects of response cards during lesson closure on the aca- demic performance of secondary stu- dents in an earth science course. Journal of Applied Behavior Analysis, 29, 403–406.

Chadsey-Rusch, J., Drasgow, E., Reinoehl, B., Halle, J., & Collet-Klingenberg, L. (1993). Using general-case instruction to teach spontaneous and generalized requests for assistance to learners with severe disabilities. Journal of the As- sociation for Persons with Severe Handicaps, 18, 177–187.

Charlop, M. H., Burgio, L. D., Iwata, B. A., & Ivancic, M. T. (1988). Stimulus vari- ation as a means of enhancing punish- ment effects. Journal of Applied Behavior Analysis, 21, 89–95.

Charlop-Christy, M. H., & Carpenter, M. H. (2000). Modified incidental teaching sessions: A procedure for parents to in- crease spontaneous speech in their chil-

dren with autism. Journal of Positive Behavioral Interventions, 2, 98–112.

Charlop-Christy, M. H., & Haymes, L. K. (1998). Using objects of obsession as token reinforcers for children with autism. Journal of Autism and Devel- opmental Disorders, 28 (3), 189–198.

Chase, P. N. (2006). Teaching the distinc- tion between positive and negative re- inforcement. The Behavior Analyst, 29, 113–115.

Chase, P. N., & Danforth, J. S. (1991). The role of rules in concept learning. In L. J. Hayes and P. N. Chase (Eds.), Dia- logues on verbal behavior (pp. 205–225). Reno, NV: Context Press.

Chase, P., Johnson, K., & Sulzer-Azaroff, B. (1985). Verbal relations within instruc- tion: Are there subclasses of the in- traverbal? Journal of the Experimental Analysis of Behavior, 43, 301–314.

Chiang, S. J., Iwata, B. A., & Dorsey, M. F. (1979). Elimination of disruptive bus riding behavior via token reinforce- ment on a “distance-based” schedule. Education and Treatment of Children, 2, 101–109.

Chiesa, M. (1994). Radical behaviorism: The philosophy and the science. Boston: Authors Cooperative.

Chomsky, N. (1957). Syntactic structures. The Hague: Mouton and Company.

Chomsky, N. (1959). Review of B. F. Skin- ner’s Verbal behavior. Language, 35, 26–58.

Chomsky, N. (1965). Aspects of a theory of syntax. Cambridge, MA: MIT Press.

Christian, L., & Poling, A. (1997). Using self-management procedures to im- prove the productivity of adults with developmental disabilities in a com- petitive employment setting. Journal of Applied Behavior Analysis, 30, 169–172.

Christle, C. A., & Schuster, J. W. (2003). The effects of using response cards on stu- dent participation, academic achieve- ment, and on-task behavior during whole-class, math instruction. Journal of Behavioral Education, 12, 147–165.

Ciccone, F. J., Graff, R. B., & Ahearn, W. H. (2006). Stimulus preference assessments and the utility of a moderate category. Behavioral Intervention, 21, 59–63.

Cipani, E. C., & Spooner, F. (1994). Cur- ricular and instructional approaches for persons with severe disabilities. Boston: Allyn & Bacon.

Cipani, E., Brendlinger, J., McDowell, L., & Usher, S. (1991). Continuous vs. in- termittent punishment: A case study. Journal of Developmental and Physi- cal Disabilities, 3, 147–156.

Cipani, E., Robinson, S., & Toro, H. (2003). Ethical and risk management issues in the practice of ABA. Paper presented at annual conference of the Florida As- sociation for Behavior Analysis, St. Petersburg.

Clark, H. B., Rowbury, T., Baer, A., & Baer, D. M. (1973). Time out as a punishing stimulus in continuous and intermittent schedules. Journal of Applied Behav- ior Analysis, 6, 443–455.

Codding, R. S., Feinberg, A. B., Dunn, E. K., & Pace, G. M. (2005). Effects of im- mediate performance feedback on im- plementation of behavior support plans. Journal of Applied Behavior Analysis, 38, 205–219.

Cohen, J. A. (1960). A coefficient of agree- ment for nominal scales. Educational and Psychological Measurement, 20, 37–46.

Cohen-Almeida, D., Graff, R. B., & Ahearn, W. H. (2000). A comparison of verbal and tangible stimulus preference as- sessments. Journal of Applied Behav- ior Analysis, 33, 329–334.

Cole, G. A., Montgomery, R. W., Wilson, K. M., & Milan, M. A. (2000). Para- metric analysis of overcorrection du- ration effects: Is longer really better than shorter? Behavior Modification, 24, 359–378.

Coleman-Martin, M. B., & Wolff Heller, K. (2004). Using a modified constant prompt-delay procedure to teach spelling to students with physical dis- abilities. Journal of Applied Behavior Analysis, 37, 469–480.

Conaghan, B. P., Singh, N. N., Moe, T. L., Landrum, T. J., & Ellis, C. R. (1992). Acquisition and generalization of man- ual signs by hearing-impaired adults with mental retardation. Journal of Be- havioral Education, 2, 175–203.

Connell, M. C., Carta, J. J., & Baer, D. M. (1993). Programming generalization of in-class transition skills: Teaching preschoolers with developmental de- lays to self-assess and recruit contin- gent teacher praise. Journal of Applied Behavior Analysis, 26, 345–352.

Conners, J., Iwata, B. A., Kahng, S. W., Han- ley, G. P, Worsdell, A. S., & Thompson,

691

Bibliography

R. H. (2000). Differential responding in the presence and absence of discrimina- tive stimuli during multi-element func- tional analyses. Journal of Applied Behavior Analysis, 33, 299–308.

Conroy, M. A., Fox, J. J., Bucklin, A., & Good, W. (1996). An analysis of the re- liability and stability of the Motivation Assessment Scale in assessing the chal- lenging behaviors of persons with de- velopmental disabilities. Education and Training in Mental Retardation and De- velopmental Disabilities, 31, 243–250.

Cooke, N. L. (1984). Misrepresentations of the behavioral model in preservice teacher education textbooks. In W. L. Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behavior analysis in education (pp. 197–217). Columbus, OH: Charles E. Merrill.

Cooper, J. O. (1981). Measuring behavior (2nd ed.). Columbus, OH: Charles E. Merrill.

Cooper, J. O. (2005). Applied research: The separation of applied behavior analysis and precision teaching. In W. L. Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner, III, L. D. Peterson, S. B. Hersh, & J. C. Dardig (Eds.), Focus on behavior analysis in education: Achievements, challenges, and oppor- tunities (pp. 295–303). Upper Saddle River, NJ: Prentice Hall/Merrill.

Cooper, J. O., Kubina, R., & Malanga, P. (1998). Six procedures for showing standard celeration charts. Journal of Precision Teaching & Celeration, 15 (2), 58–76.

Cooper, K. J., & Browder, D. M. (1997). The use of a personal trainer to enhance par- ticipation of older adults with severe disabilities in a community water exer- cise class. Journal of Behavioral Edu- cation, 7, 421–434.

Cooper, L. J., Wacker, D. P., McComas, J. J., Brown, K., Peck, S. M., Richman, D., Drew, J., Frischmeyer, P., & Mil- lard, T. (1995). Use of component analysis to identify active variables in treatment packages for children with feeding disorders. Journal of Applied Behavior Analysis, 28, 139–153.

Cooper, L. J., Wacker, D. P., Thursby, D., Plagmann, L. A., Harding, J., Millard, T., & Derby, M. (1992). Analysis of the effects of task preferences, task de- mands, and adult attention on child be-

havior in outpatient and classroom set- tings. Journal of Applied Behavior Analysis, 25, 823–840.

Copeland, R. E., Brown, R. E., & Hall, R. V. (1974). The effects of principal- implemented techniques on the behav- ior of pupils. Journal of Applied Behavior Analysis, 7, 77–86.

Corey, G., Corey, M. S., & Callanan, P. (1993). Issues and ethics in the helping professions (4th ed.). Pacific Grove, CA: Brooks/Cole.

Costenbader, V., & Reading-Brown, M. (1995). Isolation timeout used with stu- dents with emotional disturbance. Exc- eptional Children, 61 (4), 353–364.

Cowdery, G., Iwata, B. A., & Pace, G. M. (1990). Effects and side effects of DRO as treatment for self-injurious behav- ior. Journal of Applied Behavior Analy- sis, 23, 497–506.

Cox, B. S., Cox, A. B., & Cox, D. J. (2000). Motivating signage prompts safety belt use among drivers exiting senior com- munities. Journal of Applied Behavior Analysis, 33, 635–638.

Craft, M. A., Alber, S. R., & Heward, W. L. (1998). Teaching elementary students with developmental disabilities to re- cruit teacher attention in a general ed- ucation classroom: Effects on teacher praise and academic productivity. Journal of Applied Behavior Analysis, 31, 399–415.

Crawford, J., Brockel, B., Schauss, S., & Miltenberger, R. G. (1992). A compar- ison of methods for the functional as- sessment of stereotypic behavior. Journal of the Association for Persons with Severe Handicaps, 17, 77–86.

Critchfield, T. S. (1993). Behavioral pharma- cology and verbal behavior: Diazepam effects on verbal self-reports. The Analy- sis of Verbal Behavior, 11, 43–54.

Critchfield, T. S. (1999). An unexpected ef- fect of recording frequency in reactive self-monitoring. Journal of Applied Behavior Analysis, 32, 389–391.

Critchfield, T. S., & Kollins, S. H. (2001). Temporal discounting: Basic research and the analysis of socially important behavior. Journal of Applied Behavior Analysis, 34, 101–122.

Critchfield, T. S., & Lattal, K. A. (1993). Ac- quisition of a spatially defined operant with delayed reinforcement. Journal of the Experimental Analysis of Behavior, 59, 373–387.

Critchfield, T. S., & Vargas, E. A. (1991). Self-recording, instructions, and public self-graphing: Effects on swimming in the absence of coach verbal interaction. Behavior Modification, 15, 95–112.

Critchfield, T. S., Tucker, J. A., & Vuchinich, R. E. (1998). Self-report methods. In K. A. Lattal & M. Perone (Eds.), Handbook of research methods in human operant behavior (pp. 435–470). New York: Plenum.

Cromwell, O. (1650, August 3). Letter to the general assembly of the Church of Scotland. Available online: http://en .wikiquote.org/wiki/Oliver_Cromwell

Crosbie, J. (1999). Statistical inference in behavior analysis: Useful friend. The Behavior Analyst, 22, 105–108.

Cushing, L. S., & Kennedy, C. H. (1997). Academic effects of providing peer support in general education class- rooms on students without disabilities. Journal of Applied Behavior Analysis, 30, 139–151.

Cuvo, A. J. (1979). Multiple-baseline design in instructional research: Pitfalls of measurement and procedural advan- tages. American Journal of Mental De- ficiency, 84, 219–228.

Cuvo,A. J. (2000). Development and function of consequence classes in operant be- havior. The Behavior Analyst, 23, 57–68.

Cuvo, A. J. (2003). On stimulus generaliza- tion and stimulus classes. Journal of Behavioral Education, 12, 77–83.

Cuvo, A. J., Lerch, L. J., Leurquin, D. A., Gaffaney, T. J., & Poppen, R. L. (1998). Response allocation to concurrent fixed-ratio reinforcement schedules with work requirements by adults with mental retardation and typical pre- school children. Journal of Applied Be- havior Analysis, 31, 43–63.

Dalton, T., Martella, R., & Marchand- Martella, N. E. (1999). The effects of a self-management program in reduc- ing off-task behavior. Journal of Be- havioral Education, 9, 157–176.

Daly, P. M., & Ranalli, P. (2003). Using countoons to teach self-monitoring skills. Teaching Exceptional Children, 35 (5), 30–35.

Dams, P-C. (2002). A little night music. In R. W. Malott & H. Harrison, I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using behavior analysis to get your act together (pp. 7–3–7-4).

692

Bibliography

Kalamazoo, MI: Department of Psy- chology, Western Michigan University.

Dardig, J. C., & Heward, W. L. (1976). Sign here: A contracting book for children and their families. Kalamazoo, MI: Behaviordelia.

Dardig, J. C., & Heward, W. L. (1981a). A systematic procedure for prioritizing IEP goals. The Directive Teacher, 3, 6–8.

Dardig, J. C., & Heward, W. L. (1981b). Sign here: A contracting book for chil- dren and their parents (2nd ed.). Bridgewater, NJ: Fournies.

Darwin, C. (1872/1958). The origin of species (6th ed.). New York: Mentor. (Original work published 1872)

Davis, C. A., & Reichle, J. (1996). Variant and invariant high-probability requests: Increasing appropriate behaviors in children with emotional-behavioral dis- orders. Journal of Applied Behavior Analysis, 29, 471–482.

Davis, C. A., Brady, M. P., Williams, R. E., & Burta, M. (1992). The effects of self- operated auditory prompting tapes on the performance fluency of persons with severe mental retardation. Edu- cation and Training in Mental Retar- dation, 27, 39–50.

Davis, L. L., & O’Neill, R. E. (2004). Use of response cards with a group of students with learning disabilities including those for whom English is a second language. Journal of Applied Behavior Analysis, 37, 219–222.

Davis, P. K., & Chittum, R. (1994). A group- oriented contingency to increase leisure activities of adults with traumatic brain injury. Journal of Applied Behavior Analysis, 27, 553–554.

Davison, M. (1999). Statistical inference in behavior analysis: Having my cake and eating it too. The Behavior Ana- lyst, 22, 99-103.

Dawson, J. E., Piazza, C. C., Sevin, B. M., Gulotta, C. S., Lerman, D. & Kelley, M. L. (2003). Use of the high-proba- bility instructional sequence and escape extinction in a child with food refusal. Journal of Applied Behavior Analysis, 36, 105–108.

De Luca, R. B., & Holborn, S. W. (1990). Effects of fixed-interval and fixed-ratio schedules of token reinforcement on exercise with obese and nonobese boys. Psychological Record, 40, 67–82.

De Luca, R. B., & Holborn, S. W. (1992). Effects of a variable-ratio reinforce-

ment schedule with changing criteria on exercise in obese and nonobese boys. Journal of Applied Behavior Analysis, 25, 671–679.

De Martini-Scully, D., Bray, M. A., & Kehle, T. J. (2000). A packaged intervention to reduce disruptive behaviors in gen- eral education students. Psychology in the Schools, 37 (2), 149–156.

de Zubicaray, G., & Clair, A. (1998). An evaluation of differential reinforcement of other behavior, differential rein- forcement of incompatible behaviors, and restitution for the management of aggressive behaviors. Behavioral In- terventions, 13, 157–168.

Deaver, C. M., Miltenberger, R. G., & Stricker, J. M. (2001). Functional analysis and treatment of hair twirling in a young child. Journal of Applied Behavior Analysis, 34, 535–538.

DeCatanzaro, D., & Baldwin, G. (1978). Ef- fective treatment of self-injurious be- havior through a forced arm exercise. Journal of Applied Behavior Analysis, 1, 433–439.

DeHaas-Warner, S. (1992). The utility of self-monitoring for preschool on-task behavior. Topics in Early Childhood Special Education, 12, 478–495.

Deitz, D. E. D. (1977). An analysis of pro- gramming DRL schedules in educa- tional settings. Behavior Research and Therapy, 15, 103–111.

Deitz, D. E. D., & Repp, A. C. (1983). Re- ducing behavior through reinforcement. Exceptional Education Quarterly, 3, 34–46.

Deitz, S. M. (1977). An analysis of pro- gramming DRL schedules in educa- tional settings. Behavior Research and Therapy, 15, 103–111.

Deitz, S. M., & Repp, A. C. (1973). De- creasing classroom misbehavior through the use of DRL schedules of reinforcement. Journal of Applied Be- havior Analysis, 6, 457–463.

Deitz, S. M. (1982). Defining applied be- havior analysis: An historical analogy. The Behavior Analyst, 5, 53–64.

Deitz, S. M., & Repp, A. C. (1983). Reduc- ing behavior through reinforcement. Exceptional Education Quarterly, 3, 34–46.

Deitz, S. M., Slack, D. J., Schwarzmueller, E. B., Wilander, A. P., Weatherly, T. J., & Hilliard, G. (1978). Reducing inap- propriate behavior in special class-

rooms by reinforcing average interre- sponse times: Interval DRL. Behavior Therapy, 9, 37–46.

DeLeon, I. G., & Iwata, B. A. (1996). Eval- uation of a multiple-stimulus presenta- tion format for assessing reinforcer preferences. Journal of Applied Be- havior Analysis, 29, 519–533.

Deleon, I. G., Anders, B. M., Rodriguez- Catter, V., & Neidert, P. L. (2000). The effects of noncontingent access to sin- gle-versus multiple-stimulus sets on self-injurious behavior. Journal of Ap- plied Behavior Analysis, 33, 623–626.

DeLeon, I. G., Fisher, W. W., Rodriguez-Cat- ter, V., Maglieri, K., Herman, K., & Marhefka, J. M. (2001). Examination of relative reinforcement effects of stimuli identified through pretreatment and daily brief preference assessments. Journal of Applied Behavior Analysis, 34, 463–473.

DeLeon, I. G., Iwata, B. A., Conners, J., & Wallace, M. D. (1999). Examination of ambiguous stimulus preferences with duration-based measures. Journal of Ap- plied Behavior Analysis, 32, 111–114.

DeLeon, I. G., Iwata, B. A., Goh, H., & Worsdell, A. S. (1997). Emergence of reinforcer preference as a function of schedule requirements and stimulus similarity. Journal of Applied Behavior Analysis, 30, 439–449.

DeLissovoy, V. (1963). Head banging in early childhood: A suggested cause. Journal of Genetic Psychology, 102, 109–114.

Delprato, D. J. (2002). Countercontrol in be- havior analysis. The Behavior Analyst, 25, 191–200.

Delprato, D. J., & Midgley, B. D. (1992). Some fundamentals of B. F. Skinner’s behaviorism. American Psychologist, 48, 1507–1520.

DeLuca, R. V., & Holborn, S. W. (1992). Ef- fects of a variable-ratio reinforcement schedule with changing criteria on ex- ercise in obese and nonobese boys. Journal of Applied Behavior Analysis, 25, 671–679.

DeMyer, M. K., & Ferster, C. B. (1962). Teaching new social behavior to schiz- ophrenic children. Journal of the American Academy of Child Psychia- try, 1, 443–461.

Derby, K. M., Wacker, D. P., Berg, W., De- Raad, A., Ulrich, S., Asmus, J., Hard- ing, J., Prouty, A., Laffey, P., & Stoner, E. A. (1997). The long-term effects of functional communication training in

693

Bibliography

home settings. Journal of Applied Be- havior Analysis, 30, 507–531.

Derby, K. M., Wacker, D. P., Sasso, G., Steege, M., Northup, J., Cigrand, K., & Asmus, J. (1992). Brief functional as- sessment techniques to evaluate aber- rant behavior in an outpatient setting: A summary of 79 cases. Journal of Ap- plied Behavior Analysis, 25, 713–721.

DeVries, J. E., Burnette, M. M., & Redmon, W. K. (1991). AIDS prevention: Im- proving nurses’ compliance with glove wearing through performance feed- back. Journal of Applied Behavior Analysis, 24, 705–711.

Dewey, J. (1939). Experience and education. New York: Macmillan.

Dickerson, E. A., & Creedon, C. F. (1981). Self-selection of standards by children: The relative effectiveness of pupil- selected and teacher-selected standards of performance. Journal of Applied Be- havior Analysis, 14, 425–433.

Didden, R., Prinsen, H., & Sigafoos, J. (2000). The blocking effect of pictor- ial prompts on sight-word reading. Journal of Applied Behavior Analysis, 33, 317–320.

Dinsmoor, J. A. (1952). A discrimination based on punishment. Quarterly Journal of Experimental Psychology, 4, 27–45.

Dinsmoor, J. A. (1995a). Stimulus control: Part I. The Behavior Analyst, 18, 51–68.

Dinsmoor, J. A. (1995b). Stimulus control: Part II. The Behavior Analyst, 18, 253–269.

Dinsmoor, J. A. (2003). Experimental. The Behavior Analyst, 26, 151–153.

Dixon, M. R., & Cummins, A. (2001), Self- control in children with autism: Re- sponse allocation during delays to reinforcement. Journal of Applied Be- havior Analysis, 34, 491–495.

Dixon, M. R., & Falcomata, T. S. (2004). Preference for progressive delays and concurrent physical therapy exercise in an adult with acquired brain injury. Journal of Applied Behavior Analysis, 37, 101–105.

Dixon, M. R., & Holcomb, S. (2000). Teach- ing self-control to small groups of du- ally diagnosed adults. Journal of Applied Behavior Analysis, 33, 611–614.

Dixon, M. R., Benedict, H., & Larson, T. (2001). Functional analysis and treat- ment of inappropriate verbal behavior. Journal of Applied Behavior Analysis, 34, 361–363.

Dixon, M. R., Hayes, L. J., Binder, L. M., Manthey, S., Sigman, C., & Zdanow- ski, D. M. (1998). Using a self-control training procedure to increase appro- priate behavior. Journal of Applied Be- havior Analysis, 31, 203–210.

Dixon, M. R., Rehfeldt, R. A., & Randich, L. (2003). Enhancing tolerance to delayed reinforcers: The role of intervening ac- tivities. Journal of Applied Behavior Analysis, 36, 263–266.

Doke, L. A., & Risley, T. R. (1972). The or- ganization of day care environments: Required vs. optional activities. Journal of Applied Behavior Analysis, 5, 453–454.

Donahoe, J. W., & Palmer, D. C. (1994). Learning and complex behavior. Boston: Allyn and Bacon.

Dorigo, M., & Colombetti, M. (1998). Robot shaping: An experiment in behavior en- gineering. Cambridge, MA: MIT Press.

Dorow, L. G., & Boyle, M. E. (1998). In- structor feedback for college writing assignments in introductory classes. Journal of Behavioral Education, 8, 115–129.

Downing, J. A. (1990). Contingency con- tracting: A step-by-step format. Teach- ing Exceptional Children, 26 (2), 111–113.

Drabman, R. S., Hammer, D., & Rosen- baum, M. S. (1979). Assessing gener- alization in behavior modification with children: The generalization map. Be- havioral Assessment, 1, 203–219.

Drabman, R. S., Spitalnik, R., & O’Leary, K. D. (1973). Teaching self-control to disruptive children. Journal of Abnor- mal Psychology, 82, 10–16.

Drasgow, E., Halle, J., & Ostrosky, M. M. (1998). Effects of differential rein- forcement on the generalization of a re- placement mand in three children with severe language delays. Journal of Ap- plied Behavior Analysis, 31, 357–374.

Drash, P. W., High, R. L., & Tudor, R. M. (1999). Using mand training to estab- lish an echoic repertoire in young chil- dren with autism. The Analysis of Verbal Behavior, 16, 29–44.

Ducharme, D. W., & Holborn, S. W. (1997). Programming generalization of social skills in preschool children with hear- ing impairments. Journal of Applied Behavior Analysis, 30, 639–651.

Ducharme, J. M., & Rushford, N. (2001). Proximal and distal effects of play on

child compliance with a brain-injured parent. Journal of Applied Behavior Analysis, 34, 221–224.

Duker, P. C., & Seys, D. M. (1996). Long- term use of electrical aversion treat- ment with self-injurious behavior. Research in Developmental Disabili- ties, 17, 293–301.

Duker, P. C., & van Lent, C. (1991). Induc- ing variability in communicative ges- tures used by severely retarded individuals. Journal of Applied Behav- ior Analysis, 24, 379–386.

Dunlap, G., & Johnson, J. (1985). Increasing the independent responding of autistic children with unpredictable supervi- sion. Journal of Applied Behavior Analysis, 18, 227–236.

Dunlap, G., de Perczel, M., Clarke, S., Wil- son, D., Wright, S., White, R., & Gomez, A. (1994). Choice making to promote adaptive behavior for students with emotional and behavioral chal- lenges. Journal of Applied Behavior Analysis, 27, 505–518.

Dunlap, G., Kern-Dunlap, L., Clarke, S., & Robbins, F. R. (1991). Functional as- sessment, curricular revision, and severe behavior problems. Journal of Applied Behavior Analysis, 24, 387–397.

Dunlap, G., Koegel, R. L., Johnson, J., & O’Neill, R. E. (1987). Maintaining performance of autistic clients in com- munity settings with delayed contin- gencies. Journal of Applied Behavior Analysis, 20, 185–191.

Dunlap, L. K., & Dunlap, G. (1989). A self- monitoring package for teaching sub- traction with regrouping to students with learning disabilities. Journal of Applied Behavior Analysis, 22, 309–314.

Dunlap, L. K., Dunlap, G., Koegel, L. K., & Koegel, R. L. (1991). Using self- monitoring to increase independence. Teaching Exceptional Children, 23(3), 17–22.

Dunn, L. M., & Dunn, L. M. (1997). Pea- body Picture Vocabulary Test—III. Cir- cle Pines, MN: American Guidance Service.

Durand, V. M. (1999). Functional commu- nication training using assistive de- vices: Recruiting natural communities of reinforcement. Journal of Applied Behavior Analysis, 32, 247–267.

Durand, V. M., & Carr, E. G. (1987). Social influences on “self-stimulatory” behavior: Analysis and treatment

694

Bibliography

application. Journal of Applied Behav- ior Analysis, 20, 119–132.

Durand, V. M., & Carr, E. G. (1992). An analysis of maintenance following functional communication training. Journal of Applied Behavior Analysis, 25, 777–794.

Durand, V. M., & Crimmins, D. (1992). The Motivation Assessment Scale. Topeka, KS: Monaco & Associates.

Durand, V. M., Crimmins, D. B., Caufield, M., & Taylor, J. (1989). Reinforcer as- sessment I: Using problem behavior to select reinforcers. Journal of the Asso- ciation for Persons with Severe Hand- icaps, 14, 113–126.

Duvinsky, J. D., & Poppen, R. (1982). Human performance on conjunctive fixed-interval fixed-ratio schedules. Journal of the Experimental Analysis of Behavior, 37, 243–250.

Dyer, K., Schwartz, I., & Luce, S. C. (1984). A supervision program for increasing functional activities for severely hand- icapped students in a residential setting. Journal of Applied Behavior Analysis, 17, 249–259.

Ebanks, M. E., & Fisher, W. W. (2003). Al- tering the timing of academic prompts to treat destructive behavior maintained by escape. Journal of Applied Behav- ior Analysis, 36, 355–359.

Eckert, T. L., Ardoin, S., P., Daly, III, E. J., & Martens, B. K. (2002). Improving oral reading fluency: A brief experi- mental analysis of combining an antecedent intervention with conse- quences. Journal of Applied Behavior Analysis, 35, 271–281.

Ecott, C. L., Foate, B. A. L., Taylor, B., & Critchfield, T. S. (1999). Further evalu- ation of reinforcer magnitude effects in noncontingent schedules. Journal of Ap- plied Behavior Analysis, 32, 529–532.

Edwards, K. J., & Christophersen, E. R. (1993). Automated data acquisition through time-lapse videotape record- ing. Journal of Applied Behavior Analysis, 24, 503–504.

Egel, A. L. (1981). Reinforcer variation: Implications for motivating develop- mentally disabled children. Journal of Applied Behavior Analysis, 14, 345–350.

Egel, A. L. (1982). Programming the gen- eralization and maintenance of treat- ment gains. In R. L. Koegal, A. Rincover, & A. L. Egel (Eds.), Edu-

cating and understanding autistic chil- dren (pp. 281–299). San Diego, CA: College-Hill Press.

Elliot, S. N., Busse, R. T., & Shapiro, E. S. (1999). Intervention techniques for academic problems. In C. R. Reynolds & T. B. Gutkin (Eds.), The handbook of school psychology (3rd ed., pp. 664–685). New York: John Wiley & Sons.

Ellis, E. S., Worthington, L. A., & Larkin, M. J. (2002). Executive summary of the research synthesis on effective teach- ing principles and the design of quality tools for educators. [available on- line: http://idea.uoregon.edu/~ncite/ documents/techrep/tech06.html]

Emerson, E., Reever, D. J., & Felce, D. (2000). Palmtop computer technologies for behavioral observation research. In T. Thompson, D. Felce, & F. J. Symons (Eds.), Behavioral observation: Tech- nology and applications in develop- mental disabilities (pp. 47–59). Baltimore: Paul H. Brookes.

Engelmann, S. (1975). Your child can suc- ceed. New York: Simon & Shuster.

Engelmann, S., & Carnine, D. (1982). Theory of instruction: Principles and applications. New York: Irvington.

Engelmann, S., & Colvin, D. (1983). Generalized compliance training: A direct-instruction program for man- aging severe behavior problems. Austin, TX: Pro-Ed.

Epling, W. F., & Pierce, W. D. (1983). Ap- plied behavior analysis: New directions from the laboratory. The Behavior An- alyst, 6, 27–37.

Epstein, L. H., Beck, B., Figueroa, J., Farkas, G., Kazdin, A., Daneman, D., & Becker, D. (1981). The effects of targeting im- provement in urine glucose on meta- bolic control in children with insulin dependent diabetes. Journal of Applied Behavior Analysis, 14, 365–375.

Epstein, R. (1982). Skinner for the class- room. Champaign, IL: Research Press.

Epstein, R. (1990). Generativity theory and creativity. In M. A. Runco & R. S. Al- bert (Eds.), Theories of creativity (pp. 116–140). Newbury Park, CA: Sage.

Epstein, R. (1991). Skinner, creativity, and the problem of spontaneous behavior. Psychological Science, 2, 362–370.

Epstein, R. (1996). Cognition, creativity, and behavior: Selected essays. Westport, CT: Praeger.

Epstein, R. (1997). Skinner as self-manager. Journal of Applied Behavior Analysis, 30, 545–568.

Ericsson, K. A., & Charness, N. (1994). Ex- pert performance. Its structure and acquisition. American Psychologist, 49 (8), 725–747.

Ervin, R., Radford, P., Bertsch, K., Piper, A., Ehrhardt, K., & Poling, A. (2001). A descriptive analysis and critique of the empirical literature on school-based functional assessment. School Psy- chology Review, 30, 193–210.

Eshleman, J. W. (1991). Quantified trends in the history of verbal behavior research. The Analysis of Verbal Be- havior, 9, 61–80.

Eshleman, J. W. (2004, May 31). Celeration analysis of verbal behavior: Research papers presented at ABA 1975–present. Paper presented at the 30th Annual Convention of the Association for Be- havior Analysis, Boston, MA.

Falcomata, T. S., Roane, H. S., Hovanetz, A. N., Kettering, T. L., & Keeney, K. M. (2004). An evaluation of response cost in the treatment of inappropriate vo- calizations maintained by automatic reinforcement. Journal of Applied Be- havior Analysis, 37, 83–87.

Falk, J. L. (1961). Production of polydipsia in normal rats by an intermittent food schedule. Science, 133, 195–196.

Falk, J. L. (1971). The nature and determi- nants of adjunctive behavior. Physiology and Behavior, 6, 577–588.

Fantuzzo, J. W., & Clement, P. W. (1981). Generalization of the effects of teacher- and self-administered token reinforcers to nontreated students. Journal of Applied Behavior Analysis, 14, 435–447.

Fantuzzo, J. W., Rohrbeck, C. A., Hightower, A. D., & Work, W. C. (1991). Teacher’s use and children’s preferences of re- wards in elementary school. Psychology in the Schools, 28, 175–181.

Farrell, A. D. (1991). Computers and be- havioral assessment: Current applica- tions, future possibilities, and obstacles to routine use. Behavioral Assessment, 13, 159–179.

Favell, J. E., & McGimsey, J. E. (1993). Defining an acceptable treatment envi- ronment. In R. Van Houten & S. Axel- rod (Eds.), Behavior analysis and treatment (pp. 25–45). New York: Plenum Press.

695

Bibliography

Favell, J. E., Azrin, N. H., Baumeister, A. A., Carr, E. G., Dorsey, M. F., Fore- hand, R., Foxx, R. M., Lovaas, I. O., Rincover, A., Risley, T. R., Romanczyk, R. G., Russo, D. C., Schroeder, S. R., & Solnick, J. V. (1982). The treatment of self-injurious behavior. Behavior Ther- apy, 13, 529–554.

Favell, J. E., McGimsey, J. F., & Jones, M. L. (1980). Rapid eating in the retarded: Reduction by nonaversive procedures. Behavior Modification, 4, 481–492.

Fawcett, S. B. (1991). Social validity: A note on methodology. Journal of Applied Behavior Analysis, 24, 235–239.

Felce, D., & Emerson, E. (2000). Obser- vational methods in assessment of quality of life. In T. Thompson, D. Felce, & F. J. Symons (Eds.), Behav- ioral observation: Technology and applications in developmental dis- abilities (pp. 159–174). Baltimore: Paul H. Brookes.

Felixbrod, J. J., & O’Leary, K. D. (1973). Effects of reinforcement on children’s academic behavior as a function of self- determined and externally imposed sys- tems. Journal of Applied Behavior Analysis, 6, 241–250.

Felixbrod, J. J., & O’Leary, K. D. (1974). Self-determination of academic stan- dards by children: Toward freedom from external control. Journal of Edu- cational Psychology, 66, 845–850.

Ferguson, D. L., & Rosales-Ruiz, J. (2001). Loading the problem loader: The ef- fects of target training and shaping on trailer-loading behavior of horses. Journal of Applied Behavior Analysis, 34, 409–424.

Ferrari, M., & Harris, S. (1981). The limits and motivational potential of sensory stimuli as reinforcers for autistic chil- dren. Journal of Applied Behavior Analysis, 14, 339–343.

Ferreri, S. J., Allen, N., Hessler, T., Nobel, M., Musti-Rao, S., & Salmon, M. (2006). Battling procrastination: Self- managing studying and writing for candidacy exams and dissertation de- fenses. Symposium at 32nd Annual Convention of the Association for Be- havior Analysis, Atlanta, GA.

Ferreri, S. J., Neef, N. A., & Wait, T. A. (2006). The assessment of impulsive choice as a function of the point of re- inforcer delay. Manuscript submitted for publication review.

Ferster, C. B., & DeMyer, M. K. (1961). The development of performances in autis- tic children in an automatically con- trolled environment. Journal of Chronic Diseases, 13, 312–345.

Ferster, C. B., & DeMyer, M. K. (1962). A method for the experimental analysis of the behavior of autistic children. American Journal of Orthopsychiatry, 32, 89–98.

Ferster, C. B., & Skinner, B. F. (1957). Schedules of reinforcement. Englewood Cliffs, NJ: Prentice Hall.

Fink, W. T., & Carnine, D. W. (1975). Con- trol of arithmetic errors using informa- tional feedback and graphing. Journal of Applied Behavior Analysis, 8, 461.

Finney, J. W., Putnam, D. E., & Boyd, C. M. (1998). Improving the accuracy of self- reports of adherence. Journal of Ap- plied Behavior Analysis, 31, 485–488.

Fisher, R. (1956). Statistical methods and statistical inference. London: Oliver & Boyd.

Fisher, W. W., & Mazur, J. E. (1997). Basic and applied research on choice re- sponding. Journal of Applied Behavior Analysis, 30, 387–410.

Fisher, W. W., Kelley, M. E., & Lomas, J. E. (2003). Visual aids and structured cri- teria for improving visual inspection and interpretation of single-case de- signs. Journal of Applied Behavior Analysis, 36, 387–406.

Fisher, W. W., Kuhn, D. E., & Thompson, R. H. (1998). Establishing discrimina- tive control of responding using func- tional and alternative reinforcers during functional communication training. Journal of Applied Behavior Analysis, 31, 543–560.

Fisher, W. W., Lindauer, S. E., Alterson, C. J., & Thompson, R. H. (1998). As- sessment and treatment of destructive behavior maintained by stereotypic ob- ject manipulation. Journal of Applied Behavior Analysis, 31, 513–527.

Fisher, W. W., Piazza, C. C., Bowman, L. G., & Almari, A. (1996). Integrating care- giver report with a systematic choice assessment to enhance reinforcer iden- tification. American Journal on Men- tal Retardation, 101, 15–25.

Fisher, W. W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., Owens, J. C., & Slevin, I. (1992). A comparison of two approaches for identifying reinforcers for persons with severe and profound

disabilities. Journal of Applied Behav- ior Analysis, 25, 491–498.

Fisher, W. W., Piazza, C. C., Bowman, L. G., Kurtz, P. F., Sherer, M. R., & Lachman, S. R. (1994). A preliminary evaluation of empirically derived consequences for the treatment of pica. Journal of Applied Behavior Analysis, 27, 447–457.

Fisher, W. W., Piazza, C. C., Cataldo, M. E., Harrell, R., Jefferson, G., & Conner, R. (1993). Functional communication training with and without extinction and punishment. Journal of Applied Behavior Analysis, 26, 23–36.

Flaute, A. J., Peterson, S. M., Van Norman, R. K., Riffle, T., & Eakins, A. (2005). Motivate me! 20 tips for using a Moti- vAider® to improve your classroom. Teaching Exceptional Children Plus, 2 (2) Article 3. Retrieved March 1, 2006, from http://escholarship.bc.edu/ education/tecplus/vol2/iss2/art3.

Fleece, L., Gross, A., O’Brien, T., Kistner, J., Rothblum, E., & Drabman, R. (1981). Elevation of voice volume in young developmentally delayed chil- dren via an operant shaping procedure. Journal of Applied Behavior Analysis, 14, 351–355.

Flood, W. A., & Wilder, D. A. (2002). An- tecedent assessment and assessment- based treatment of off-task behavior in a child diagnosed with Attention Deficit- Hyperactivity Disorder (ADHD). Edu- cation and Treatment of Children, 25 (3), 331–338.

Flora, S. R. (2004). The power of reinforce- ment. Albany: State University of New York Press.

Florida Association for Behavior Analysis. (1988). The behavior analyst’s code of ethics. Tallahassee, FL: Author.

Forthman, D. L., & Ogden, J. J. (1992). The role of applied behavior analysis in zoo management: Today and tomorrow. Journal of Applied Behavior Analysis, 25, 647–652

Foster, W. S. (1978). Adjunctive behavior: An underreported phenomenon in applied behavior analysis? Journal of Applied Behavioral Analysis, 11, 545–546.

Fowler, S. A., & Baer, D. M. (1981). ”Do I have to be good all day?” The timing of delayed reinforcement as a factor in generalization. Journal of Applied Be- havior Analysis, 14, 13–24.

696

Bibliography

Fox, D. K., Hopkins, B. L., & Anger, A. K. (1987). The long-term effects of a token economy on safety performance in open-pit mining. Journal of Applied Behavior Analysis, 20, 215–224.

Foxx, R. M. (1982). Decreasing behaviors of persons with severe retardation and autism. Champaign, IL: Research Press.

Foxx, R. M. (1996). Twenty years of applied behavior analysis in treating the most se- vere problem behavior: Lessons learned. The Behavior Analyst, 19, 225–235.

Foxx, R. M., & Azrin, N. H. (1972). Resti- tution: A method of eliminating ag- gressive-disruptive behavior of retarded and brain damaged patients. Behavior Research and Therapy, 10, 15–27.

Foxx, R. M., & Azrin, N. H. (1973). The elimination of autistic self-stimulatory behavior by overcorrection. Journal of Applied Behavior Analysis, 6, 1–14.

Foxx, R. M., & Bechtel, D. R. (1983). Over- correction: A review and analysis. In S. Axelrod & J. Apsche (Eds.), The effects of punishment on human behavior (pp. 133–220). New York: Academic Press.

Foxx, R. M., & Rubinoff, A. (1979). Be- havioral treatment of caffeinism: Re- ducing excessive coffee drinking. Journal of Applied Behavior Analysis, 12, 335–344.

Foxx, R. M., & Shapiro, S. T. (1978). The timeout ribbon: A non-exclusionary timeout procedure. Journal of Applied Behavior Analysis, 11, 125–143.

Foxx, R. M., Bittle, R. G., & Faw, G. D. (1989). A maintenance strategy for dis- continuing aversive procedures: A 52- month follow-up of the treatment of aggression. American Journal on Men- tal Retardation, 94, 27–36.

Freeland, J. T., & Noell, G. H. (1999). Main- taining accurate math responses in el- ementary school students: The effects of delayed intermittent reinforcement and programming common stimuli. Journal of Applied Behavior Analysis, 32, 211–215.

Freeland, J. T., & Noell, G. H. (2002). Pro- gramming for maintenance: An in- vestigation of delayed intermittent reinforcement and common stimuli to create indiscriminable contingencies. Journal of Behavioral Education, 11, 5–18.

Friedling, C., & O’Leary, S. G. (1979). Ef- fects of self-instructional training on sec- ond- and third-grade hyperactive

children:A failure to replicate. Journal of Applied Behavior Analysis, 12, 211–219.

Friman, P. C. (1990). Nonaversive treat- ment of high-rate disruptions: Child and provider effects. Exceptional Chil- dren, 57, 64–69.

Friman, P. C. (2004). Up with this I shall not put: 10 reasons why I disagree with Branch and Vollmer on behavior used as a count noun. The Behavior Analyst, 27, 99–106.

Friman, P. C., & Poling, A. (1995). Making life easier with effort: Basic findings and applied research on response effort. Journal of Applied Behavior Analysis, 28, 538–590.

Friman, P. C., Hayes, S. C., & Wilson, K. G. (1998). Why behavior analysts should study emotion: The example of anxi- ety. Journal of Applied Behavior Analy- sis, 31, 137–156.

Fuller, P. R. (1949). Operant conditioning of a vegetative organism. American Jour- nal of Psychology, 62, 587–590.

Fuqua, R. W., & Schwade, J. (1986). Social validation of applied research: A selec- tive review and critique. In A. Poling & R. W. Fuqua (Eds.), Research meth- ods in applied behavior analysis (pp. 265–292). New York: Plenum Press.

Gable, R. A., Arllen, N. L., & Hendrickson, J. M. (1994). Use of students with emo- tional/behavioral disorders as behavior change agents. Education and Treat- ment of Children, 17 (3), 267–276.

Galbicka, G. (1994). Shaping in the 21st century: Moving percentile schedules into applied settings. Journal of Ap- plied Behavior Analysis, 27, 739–760.

Gallagher, S. M., & Keenan, M (2000). In- dependent use of activity materials by the elderly in a residential setting. Journal of Applied Behavior Analysis, 33, 325–328.

Gambrill, E. (2003). Science and its use and neglect in the human services. In K. S. Budd & T. Stokes (Eds.), A small matter of proof: The legacy of Donald M. Baer (pp. 63–76). Reno, NV: Con- text Press.

Gambrill, E. D. (1977). Behavior modifica- tion: Handbook of assessment, inter- vention, and evaluation. San Francisco: Jossey-Bass.

Garcia, E. E. (1976). The development and generalization of delayed imitation. Journal of Applied Behavior Analy- sis, 9, 499.

Garcia, E. E., & Batista-Wallace, M. (1977). Parental training of the plural mor- pheme in normal toddlers. Journal of Applied Behavior Analysis, 10, 505.

Gardner, III, R., Heward, W. L., & Grossi, T. A. (1994). Effects of response cards on student participation and academic achievement: A systematic replication with inner-city students during whole- class science instruction. Journal of Ap- plied Behavior Analysis, 27, 63–71.

Garfinkle, A. N., & Schwartz, I. S. (2002). Peer imitation: Increasing social in- teractions in children with autism and other developmental disabilities in in- clusive preschool classrooms. Topics in Early Childhood Special Educa- tion, 22, 26–38.

Garner, K. (2002). Case study: The consci- entious kid. In R. W. Malott & H. Har- rison, I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using be- havior analysis to get your act to- gether (p. 3-13). Kalamazoo, MI: Department of Psychology, Western Michigan University.

Gast, D. L., Jacobs, H. A., Logan, K. R., Murray, A. S., Holloway, A., & Long, L. (2000). Pre-session assessment of preferences for students with profound multiple disabilities. Education and Training in Mental Retardation and De- velopmental Disabilities, 35, 393–405.

Gaylord-Ross, R. (1980). A decision model for the treatment of aberrant behavior in applied settings. In W. Sailor, B. Wilcox, & L. Brown (Eds.), Methods of instruction for severely handicapped students (pp. 135–158). Baltimore: Paul H. Brookes.

Gaylord-Ross, R. J., Haring, T. G., Breen, C., & Pitts-Conway, V. (1984). The training and generalization of social in- teraction skills with autistic youth. Journal of Applied Behavior Analysis, 17, 229–247.

Gee, K., Graham, N., Goetz, L., Oshima, G., & Yoshioka, K. (1991). Teaching stu- dents to request the continuation of rou- tine activities by using time delay and decreasing physical assistance in the context of chain interruption. Journal of the Association for Persons with Se- vere Handicaps, 10, 154–167.

Geller, E. S., Paterson, L., & Talbott, E. (1982). A behavioral analysis of incen- tive prompts for motivating seat belt

697

Bibliography

use. Journal of Applied Behavior Analysis, 15, 403–415.

Gena, A., Krantz, P. J., McClannahan, L. E., & Poulson, C. L. (1996). Training and generalization of affective behavior displayed by youth with autism. Journal of Applied Behavior Analysis, 29, 291–304.

Gentile, J. R., Roden, A. H., & Klein, R. D. (1972). An analysis-of-variance model for the intrasubject replication design. Journal of Applied Behavior Analysis, 5, 193–198.

Gewirtz, J. L., & Baer, D. M. (1958). The effect of brief social deprivation on behaviors for a social reinforcer. Journal of Abnormal Social Psychol- ogy, 56, 49–56.

Gewirtz, J. L., & Pelaez-Nogueras, M. (2000). Infant emotions under the pos- itive-reinforcer control of caregiver at- tention and touch. In J. C. Leslie & D. Blackman (Eds.), Issues in experimen- tal and applied analyses of human behavior (pp. 271–291). Reno, NV: Context Press.

Glenn, S. S. (2004). Individual behavior, cul- ture, and social change. The Behavior Analyst, 27, 133–151.

Glenn, S. S., Ellis, J., & Greenspoon, J. (1992). On the revolutionary nature of the oper- ant as a unit of behavioral selection. American Psychologist, 47, 1329–1336.

Glynn, E. L. (1970). Classroom applications of self-determined reinforcement. Journal of Applied Behavior Analysis, 3, 123–132.

Glynn, E. L., Thomas, J. D., & Shee, S. M. (1973). Behavioral self-control of on- task behavior in an elementary class- room. Journal of Applied Behavior Analysis, 6, 105–114.

Glynn, S. M. (1990). Token economy: Ap- proaches for psychiatric patients: Progress and pitfalls over 25 years. Behavior Modification, 14 (4), 383–407.

Goetz, E. M., & Baer, D. M. (1973). Social control of form diversity and the emer- gence of new forms in children’s block- building. Journal of Applied Behavior Analysis, 6, 209–217.

Goetz, L., Gee, K., & Sailor, W. (1985). Using a behavior chain interruption strategy to teach communication skills to students with severe disabilities. Journal of the Association for Persons with Severe Handicaps, 10, 21–30.

Goh, H-L., & Iwata, B. A. (1994). Behav- ioral persistence and variability during

extinction of self-injury maintained by escape. Journal of Applied Behavior Analysis, 27, 173–174.

Goldiamond, I. (1965). Self-control proce- dures in personal behavior problems. Psychological Reports, 17, 851–868.

Goldiamond, I. (1974). Toward a construc- tional approach to social problems: Ethical and constitutional issues raised by applied behavior analysis. Behavi- orism, 2, 1–85.

Goldiamond, I. (1976a). Self-reinforcement. Journal of Applied Behavior Analysis, 9, 509–514.

Goldiamond, I. (1976b). Fables, armadyllics, and self-reinforcement. Journal of Ap- plied Behavior Analysis, 9, 521–525.

Gottschalk, J. M., Libby, M. E., & Graff, R. B. (2000). The effects of establish- ing operations on preference assess- ment outcomes. Journal of Applied Behavior Analysis, 33, 85–88.

Grace, N. C., Kahng, S., & Fisher, W. W. (1994). Balancing social acceptability with treatment effectiveness of an in- trusive procedure: A case report. Journal of Applied Behavior Analysis, 27, 171–172.

Graf, S. A., & Lindsley, O. R. (2002). Standard Celeration Charting 2002. Poland, OH: Graf Implements.

Gray, J. A. (1979). Ivan Pavlov. New York: Penguin Books.

Green, C. W., & Reid, D. H. (1996). Defin- ing, validating, and increasing indices of happiness among people with pro- found multiple disabilities. Journal of Applied Behavior Analysis, 29, 67–78.

Green, C. W., Gardner, S. M., & Reid, D. H. (1997). Increasing indices of happiness among people with profound multiple disabilities: A program replication and component analysis. Journal of Applied Behavior Analysis, 30, 217–228.

Green, G., & Shane, H. C. (1994). Science, reason, and facilitated communication. Journal of the Association for Persons with Severe Handicaps, 19, 151–172.

Green, L., & Freed, D. W. (1993). The sub- stitutability of reinforcers. Journal of the Experimental Analysis of Behavior, 60, 141–158.

Greene, B. F., Bailey, J. S., & Barber, F. (1981). An analysis and reduction of disruptive behavior on school buses. Journal of Applied Behavior Analysis, 14, 177–192.

Greenspan, S., & Negron, E. (1994). Ethi- cal obligations of special services per-

sonnel. Special Services in the Schools, 8 (2), 185–209.

Greenwood, C. R., & Maheady, L. (1997). Measurable change in student perfor- mance: Forgotten standard in teacher preparation? Teacher Education and Special Education, 20, 265–275.

Greenwood, C. R., Delquadri, J. C., & Hall, R. V. (1984). Opportunity to re- spond and student academic achieve- ment. In W. L. Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behavior analysis in educa- tion (pp. 58–88). Columbus, OH: Merrill.

Greer, R. D. (1983). Contingencies of the science and technology of teaching and pre-behavioristic research prac- tices in education. Educational Re- searcher, 12, 3–9.

Gresham, F. M. (1983). Use of a home-based dependent group contingency system in controlling destructive behavior: A case study. School Psychology Review, 12 (2), 195–199.

Gresham, F. M., Gansle, K. A., & Noell, G. H. (1993). Treatment integrity in applied behavior analysis with children. Journal of Applied Behavior Analysis, 26, 257–263.

Griffen, A. K., Wolery, M., & Schuster J. W. (1992). Triadic instruction of chained food preparation responses: Acquisition and observational learn- ing. Journal of Applied Behavior Analysis, 25, 193–204.

Griffith, R. G. (1983). The administrative issues: An ethical and legal perspec- tive. In S. Axelrod & J. Apsche (Eds.), The effects of punishment on human behavior (pp. 317–338). New York: Academic Press.

Grossi, T. A. (1998). Using a self-operated auditory prompting system to improve the work performance of two employ- ees with severe disabilities. Journal of The Association for Persons with Se- vere Handicaps, 23, 149–154.

Grossi, T. A., & Heward, W. L. (1998). Using self-evaluation to improve the work productivity of trainees in a community-based restaurant training program. Education and Training in Mental Retardation and Developmental Disabilities, 33, 248–263.

Grossi, T. A., Kimball, J. W., & Heward, W. L. (1994). What did you say? Using review of tape-recorded interactions to increase social acknowledgments by

698

Bibliography

trainees in a community-based voca- tional program. Research in Develop- mental Disabilities, 15, 457–472.

Grunsell, J., & Carter, M. (2002). The be- havior change interruption strategy: Generalization to out-of-routine con- texts. Education and Training in Men- tal Retardation and Developmental Disabilities, 37 (4), 378–390.

Guilford, J. P. (1965). Fundamental statis- tics in psychology and education. New York: McGraw-Hill.

Gumpel, T. P., & Shlomit, D. (2000). Ex- ploring the efficacy of self-regulatory training as a possible alternative to so- cial skills training. Behavioral Disor- ders, 25, 131–141.

Gunter, P. L., Venn, M. L., Patrick, J., Miller, K. A., & Kelly, L. (2003). Efficacy of using momentary time samples to de- termine on-task behavior of students with emotional/behavioral disorders. Education and Treatment of Children, 26, 400–412.

Gutowski, S. J., & Stromer, R. (2003). De- layed matching to two-picture samples by individuals with and without dis- abilities: An analysis of the role of naming. Journal of Applied Behavior Analysis, 36, 487–505.

Guttman, N., & Kalish, H. (1956). Dis- criminability and generalization. Jour- nal of Experimental Psychology, 51, 79–88.

Haagbloom, S. J., Warnick, R., Warnick J. E., Jones, V. K.,Yarbrough, G. L., Rus- sell, T. M., et al. (2002). The 100 most eminent psychologists of the 20th cen- tury. Review of General Psychology, 6, 139–152.

Hackenberg, T. D. (1995). Jacques Loeb, B. F. Skinner, and the legacy of pre- diction and control. The Behavior An- alyst, 18, 225–236.

Hackenberg, T. D., & Axtell, S. A. M. (1993). Humans’ choices in situations of time-based diminishing returns. Journal of the Experimental Analysis of Behavior, 59, 445–470.

Hagopian, L. P., & Adelinis, J. D. (2001) Re- sponse blocking with and without redi- rection for the treatment of pica. Journal of Applied Behavior Analysis, 34, 527–530.

Hagopian, L. P., & Thompson, R. H. (1999). Reinforcement of compliance with res- piratory treatment in a child with cystic fibrosis. Journal of Applied Be- havior Analysis, 32, 233–236.

Hagopian, L. P., Farrell, D. A., & Amari, A. (1996). Treating total liquid refusal with backward chaining and fading. Journal of Applied Behavior Analysis, 29, 573–575.

Hagopian, L. P., Fisher, W. W., Sullivan, M. T., Acquisto, J., & Leblanc, L. A. (1998). Effectiveness of functional communication training with and with- out extinction and punishment: A sum- mary of 21 inpatient cases. Journal of Applied Behavior Analysis, 31, 211–235.

Hagopian, L. P., Rush, K. S., Lewin, A. B., & Long, E. S. (2001). Evaluating the predictive validity of a single stimulus engagement preference assessment. Journal of Applied Behavior Analysis, 34, 475–485.

Hake, D. F., & Azrin, N. H. (1965). Condi- tioned punishment. Journal of the Ex- perimental Analysis of Behavior, 6, 297–298.

Hake, D. F., Azrin, N. H., & Oxford, R. (1967). The effects of punishment in- tensity on squirrel monkeys. Journal of the Experimental Analysis of Behavior, 10, 95–107.

Hall, G. A., & Sundberg, M. L. (1987). Teaching mands by manipulating con- ditioned establishing operations. The Analysis of Verbal Behavior, 5, 41–53.

Hall, R. V., & Fox, R. G. (1977). Changing- criterion designs: An alternative applied behavior analysis procedure. In B. C. Etzel, J. M. LeBlanc, & D. M. Baer (Eds.), New developments in behav- ioral research: Theory, method, and ap- plication (pp. 151–166). Hillsdale, NJ: Erlbaum.

Hall, R. V., Cristler, C., Cranston, S. S., & Tucker, B. (1970). Teachers and par- ents as researchers using multiple base- line designs. Journal of Applied Behavior Analysis, 3, 247–255.

Hall, R. V., Delquadri, J. C., & Harris, J. (1977, May). Opportunity to respond: A new focus in the field of applied be- havior analysis. Paper presented at the Midwest Association for Behavior Analysis, Chicago, IL.

Hall, R. V., Lund, D., & Jackson, D. (1968). Effects of teacher attention on study be- havior. Journal of Applied Behavior Analysis, 1, 1–12.

Hall, R. V., Panyan, M., Rabon, D., & Bro- den. D. (1968). Instructing beginning teachers in reinforcement procedures which improve classroom control.

Journal of Applied Behavior Analysis, 1, 315–322.

Hall, R., Axelrod, S., Foundopoulos, M., Shellman, J., Campbell, R. A., & Cranston, S. S. (1971). The effective use of punishment to modify behavior in the classroom. Educational Technol- ogy, 11 (4), 24–26.

Hall, R.V., Fox, R., Williard, D., Goldsmith, L., Emerson, M., Owen, M., Porcia, E., & Davis, R. (1970). Modification of disrupting and talking-out behavior with the teacher as observer and ex- perimenter. Paper presented at the American Educational Research Asso- ciation Convention, Minneapolis, MN.

Hallahan, D. P., Lloyd, J. W., Kosiewicz, M. M., Kauffman, J. M., & Graves, A. W. (1979). Self-monitoring of atten- tion as a treatment for a learning dis- abled boy’s off-task behavior. Learning Disability Quarterly, 2, 24–32.

Hamblin, R. L., Hathaway, C., & Wodarski, J. S. (1971). Group contingencies, peer tutoring and accelerating academic achievement. In E. A. Ramp & B. L. Hopkins (Eds.), A new direction for ed- ucation: Behavior analysis (Vol. 1, pp. 41–53). Lawrence: University of Kansas.

Hamlet, C. C., Axelrod, S., & Kuerschner, S. (1984). Eye contact as an antecedent to compliant behavior. Journal of Ap- plied Behavior Analysis, 17, 553–557.

Hammer-Kehoe, J. (2002). Yoga. In R. W. Malott & H. Harrison (Eds.), I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using behavior analy- sis to get your act together (p. 10-5). Kalamazoo, MI: Department of Psy- chology, Western Michigan University.

Hammill, D., & Newcomer, P. L. (1997). Test of language development—3. Austin, TX: Pro-Ed.

Hanley, G. P., Iwata, B. A., & Thompson, R. H. (2001). Reinforcement schedule thinning following treatment with func- tional communication training. Journal of Applied Behavior Analysis, 34, 17–38.

Hanley, G. P., Iwata, B. A., Thompson, R. H., & Lindberg, J. S. (2000). A component analysis of “stereotypy and reinforce- ment” for alternative behavior. Journal of Applied Behavior Analysis, 33, 299–308.

Hanley, G. P., Piazza, C. C., Fisher, W. W., Contrucci, S. A., & Maglieri, K. A. (1997). Evaluation of client preference of function-based treatment packages.

699

Bibliography

Journal of Applied Behavior Analysis, 30, 459–473.

Hanley, G. P., Piazza, C. C., Keeney, K. M., Blakeley-Smith, A. B., & Worsdell, A. S. (1998). Effects of wrist weights on self-injurious and adaptive behav- iors. Journal of Applied Behavior Analysis, 31, 307–310.

Haring, T. G., & Kennedy, C. H. (1990). Contextual control of problem behav- ior. Journal of Applied Behavior Analy- sis, 23, 235–243.

Harlow, H. R. (1959). Learning set and error factor theory. In S. Kock (Ed.), Psychology: A study of science (Vol. 2, pp. 492–537). New York: McGraw-Hill.

Harrell, J. P. (2002). Case study: Taking time to relax. In R. W. Malott & H. Harrison, I’ll stop procrastinating when I get around to it: Plus other cool ways to suc- ceed in school and life using behavior analysis to get your act together (p. 10-2). Kalamazoo, MI: Department of Psy- chology, Western Michigan University.

Harris, F. R., Johnston, M. K., Kelly, C. S., & Wolf, M. M. (1964). Effects of positive social reinforcement on regressed crawl- ing of a nursery school child. Journal of Educational Psychology, 55, 35–41.

Harris, K. R. (1986). Self-monitoring of at- tentional behavior versus self-monitor- ing of productivity: Effects on on-task behavior and academic response rate among learning disabled children. Journal of Applied Behavior Analysis, 19, 417–424.

Hart, B., & Risley, T. R. (1975). Incidental teaching of language in the preschool. Journal of Applied Behavior Analysis, 8, 411–420.

Hart, B. M., Allen, K. E., Buell, J. S., Har- ris, F. R., & Wolf, M. M. (1964). Ef- fects of social reinforcement on operant crying. Journal of Experimental Child Psychology, 1, 145–153.

Hartmann, D. P. (1974). Forcing square pegs into round holes: Some comments on “an analysis-of-variance model for the intrasubject replication design”. Journal of Applied Behavior Analysis, 7, 635–638.

Hartmann, D. P. (1977). Considerations in the choice of interobserver reliability estimates. Journal of Applied Behavior Analysis, 10, 103–116.

Hartmann, D. P., & Hall, R. V. (1976). The changing criterion design. Journal of Applied Behavior Analysis, 9, 527–532.

Hartmann, D. P., Gottman, J. M., Jones, R. R., Gardner, W., Kazdin, A. E., & Vaught, R. S. (1980). Interrupted time- series analysis and its application to behavioral data. Journal of Applied Be- havior Analysis, 13, 543–559.

Harvey, M. T., May, M. E., & Kennedy, C. H. (2004). Nonconcurrent multiple base- line designs and the evaluation of edu- cational systems. Journal of Behavioral Education, 13, 267–276.

Hawkins, R. P. (1975). Who decided that was the problem? Two stages of responsi- bility for applied behavior analysts. In W. S. Wood (Ed.), Issues in evaluating behavior modification (pp. 195–214). Champaign, IL: Research Press.

Hawkins, R. P. (1979). The functions of as- sessment. Journal of Applied Behavior Analysis, 12, 501–516.

Hawkins, R. P. (1984). What is “meaning- ful” behavior change in a severely/ profoundly retarded learner: The view of a behavior analytic parent. In W. L Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behav- ior analysis in education (pp. 282–286). Upper Saddle River, NJ: Prentice-Hall/Merrill.

Hawkins, R. P. (1986). Selection of target behaviors. In R. O. Nelson & S. C. Hayes (Eds.), Conceptual foundations of behavioral assessment (pp. 331–385). New York: Guilford Press.

Hawkins, R. P. (1991). Is social validity what we are interested in? Journal of Applied Behavior Analysis, 24, 205–213.

Hawkins, R. P., & Anderson, C. M. (2002). On the distinction between science and practice: A reply to Thyer and Adkins. The Behavior Analyst, 26, 115–119.

Hawkins, R. P., & Dobes, R. W. (1977). Be- havioral definitions in applied behav- ior analysis: Explicit or implicit? In B. C. Etzel, J. M. LeBlanc, & D. M. Baer (Eds.), New developments in be- havioral research: Theory, method, and application (pp. 167–188). Hillsdale, NJ: Erlbaum.

Hawkins, R. P., & Dotson, V. A. (1975). Re- liability scores that delude: An Alice in Wonderland trip through the mislead- ing characteristics of interobserver agreement scores in interval recording. In E. Ramp & G. Semp (Eds.), Be- havior analysis: Areas of research and application (pp. 359–376). Upper Sad- dle River, NJ: Prentice Hall.

Hawkins, R. P., & Hursh, D. E. (1992). Lev- els of research for clinical practice: It isn’t as hard as you think. West Virginia Journal of Psychological Research and Practice, 1, 61–71.

Hawkins, R. P., Mathews, J. R., & Hamdan, L. (1999). Measuring behavioral health outcomes: A practical guide. New York: Kluwer Academic/Plenum.

Hayes, L. J., Adams, M. A., & Rydeen, K. L. (1994). Ethics, choice, and value. In L. J. Hayes et al. (Eds.), Ethical issues in developmental disabilities (pp. 11–39). Reno, NV: Context Press.

Hayes, S. C. (1991). The limits of techno- logical talk. Journal of Applied Behav- ior Analysis, 24, 417–420.

Hayes, S. C. (Ed.). (1989). Rule-governed behavior: Cognition, contingencies, and instructional control. Reno, NV: Context Press.

Hayes, S. C., & Cavior, N. (1977). Multiple tracking and the reactivity of self- monitoring: I. Negative behaviors. Behavior Therapy, 8, 819–831.

Hayes, S. C., & Cavior, N. (1980). Multiple tracking and the reactivity of self- monitoring: II. Positive behaviors. Be- havioral Assessment, 2, 238–296.

Hayes, S. C., & Hayes, L. J. (1993). Applied implications of current JEAB research on derived relations and delayed rein- forcement. Journal of Applied Behavior Analysis, 26, 507–511.

Hayes, S. C., Rincover, A., & Solnick, J. V. (1980). The technical drift of applied behavior analysis. Journal of Applied Behavior Analysis, 13, 275–285.

Hayes, S. C., Rosenfarb, I., Wulfert, E., Munt, E. D., Korn, D., & Zettle, R. D. (1985). Self-reinforcement effects: An artifact of social standard setting? Journal of Applied Behavior Analysis, 18, 201–214.

Hayes, S. C., Zettle, R. D., & Rosenfarb, I. (1989). Rule-following. In S. C. Hayes (Ed.), Rule-governed behavior: Cogni- tion, contingencies, and instructional control (pp. 191–220). New York: Plenum Press.

Haynes, S. N., & Horn, W. F. (1982). Reac- tivity in behavioral observation: A methodological and conceptual critique. Behavioral Assessment, 4, 369–385.

Health Insurance Portability and Account- ability Act (HIPAA). (1996). Wash- ington, DC: Office for Civil Rights. Retrieved December 14, 2003, from

700

Bibliography

http://aspe.hhs.gov/admnsimp/ pl104191.htm>http://aspe.hhs.gov/ admnsimp/pl104191.htm.

Heckaman, K. A., Alber, S. R., Hooper, S., & Heward, W. L. (1998). A compari- son of least-to-most prompts and pro- gressive time delay on the disruptive behavior of students with autism. Journal of Behavior Education, 8, 171–201.

Heflin, L. J., & Simpson, R. L. (2002). Un- derstanding intervention controver- sies. In B. Scheuermann & J. Weber (Eds.), Autism: Teaching does make a difference (pp. 248–277). Belmont, CA: Wadsworth.

Helwig, J. (1973). Effects of manipulating an antecedent event on mathematics re- sponse rate. Unpublished manuscript, Ohio State University, Columbus, OH.

Heron, T. E., & Harris, K. C. (2001). The educational consultant: Helping pro- fessionals, parents, and students in in- clusive classrooms (4th ed.). Austin, TX: Pro-Ed.

Heron, T. E., & Heward, W. L. (1988). Ecological assessment: Implications for teachers of learning disabled stu- dents. Learning Disability Quarterly, 11, 224–232.

Heron, T. E., Heward, W. L., Cooke, N. L., & Hill, D. S. (1983). Evaluation of a class- wide peer tutoring system: First graders teach each other sight words. Education and Treatment of Children, 6, 137–152.

Heron, T. E., Hippler, B. J., & Tincani, M. J. (2003). How to help students complete classwork and homework assignments. Austin, TX: Pro-Ed.

Heron, T. E., Tincani, M. J., Peterson, S. M., & Miller, A. D. (2005). Plato’s allegory of the cave revisited. Disciples of the light appeal to the pied pipers and prisoners in the darkness. In W. L. Heward, T. E. Heron, N.A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner, III, L.D.Peterson,S.B.Hersh,&J.C.Dardig (Eds.), Focus on behavior analysis in ed- ucation: Achievements, challenges, and opportunities (pp. 267–282), Upper Sad- dle River, NJ: Merrill/Prentice Hall.

Herr, S. S., O’Sullivan, J. L., & Dinerstein, R. D. (1999). Consent to extraordinary interventions. In R. D. Dinerstein, S. S. Herr, & J. L. O’Sullivan (Eds.), A guide to consent (pp. 111–122). Washington DC: American Association on Mental Retardation.

Herrnstein, R. J. (1961). Relative and ab- solute strength of a response as a func- tion of frequency of reinforcement. Journal of the Experimental Analysis of Behavior 4, 267–272.

Herrnstein, R. J. (1970). On the law of ef- fect. Journal of the Experimental Analysis of Behavior 13, 243–266.

Hersen, M., & Barlow, D. H. (1976). Single case experimental designs: Strategies for studying behavior change. New York: Pergamon Press.

Heward, W. L. (1978, May). The delayed multiple baseline design. Paper pre- sented at the Fourth Annual Conven- tion of the Association for Behavior Analysis, Chicago.

Heward, W. L. (1980). A formula for indi- vidualizing initial criteria for reinforce- ment. Exceptional Teacher, 1 (9), 7, 9.

Heward, W. L. (1994). Three “low-tech” strategies for increasing the frequency of active student response during group in- struction. In R. Gardner, D. M. Sainato, J. O. Cooper, T. E. Heron, W. L. Heward, J. Eshleman, & T. A. Grossi (Eds.), Behavior analysis in education: Focus on measurably superior instruction (pp. 283–320). Monterey, CA: Brooks/Cole.

Heward, W. L. (2003). Ten faulty notions about teaching and learning that hinder the effectiveness of special education. The Journal of Special Education, 36 (4), 186–205.

Heward, W. L. (2005). Reasons applied be- havior analysis is good for education and why those reasons have been in- sufficient. In W. L. Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner, III, L. D. Peterson, S. B. Hersh, & J. C. Dardig (Eds.), Focus on behavior analysis in education: Achievements, challenges, and opportunities (pp. 316–348). Upper Saddle River, NJ: Merrill/Prentice Hall.

Heward, W. L. (2006). Exceptional children: An introduction to special education (8th ed.). Upper Saddle River, NJ: Mer- rill/Prentice Hall.

Heward, W. L., & Cooper, J. O. (1992). Rad- ical behaviorism: A productive and needed philosophy for education. Journal of Behavioral Education, 2, 345–365.

Heward, W. L., & Eachus, H. T. (1979). Ac- quisition of adjectives and adverbs in sentences written by hearing impaired

and aphasic children. Journal of Ap- plied Behavior Analysis, 12, 391–400.

Heward, W. L., & Silvestri, S. M. (2005a). The neutralization of special education. In J. W. Jacobson, J. A. Mulick, & R. M. Foxx (Eds.), Fads: Dubious and im- probable treatments for developmental disabilities. Mahwah NJ: Erlbaum.

Heward, W. L., & Silvestri, S. M. (2005b). Antecedent. In G. Sugai & R. Horner (Eds.), Encyclopedia of behavior modi- fication and cognitive behavior therapy, Vol. 3: Educational applications (pp. 1135–1137). Thousand Oaks, CA: Sage.

Heward, W. L., Dardig, J. C., & Rossett, A. (1979). Working with parents of hand- icapped children. Upper Saddle River, NJ: Merrill/Prentice Hall.

Heward, W. L., Heron, T. E., Gardner, III, R., & Prayzer, R. (1991). Two strate- gies for improving students’ writing skills. In G. Stoner, M. R. Shinn, & H. M. Walker (Eds.), A school psy- chologist’s interventions for regular education (pp. 379–398). Washington, DC: National Association of School Psychologists.

Heward, W. L., Heron, T. E., Neef, N. A., Pe- terson, S. M., Sainato, D. M., Cartledge, G., Gardner, III, R., Peterson, L. D., Hersh, S. B., & Dardig, J. C. (Eds.). (2005). Focus on behavior analysis in education: Achievements, challenges, and opportunities. Upper Saddle River, NJ: Prentice Hall/Merrill.

Hewett, F. M. (1968). The emotionally dis- turbed child in the classroom. New York: McGraw-Hill.

Higbee, T. S., Carr, J. E., & Harrison, C. D. (1999). The effects of pictorial versus tangible stimuli in stimulus preference assessments. Research in Developmen- tal Disabilities, 20, 63–72.

Higbee, T. S., Carr, J. E., & Harrison, C. D. (2000). Further evaluation of the multiple-stimulus preference assess- ment. Research in Developmental Dis- abilities, 21, 61–73.

Higgins, J. W., Williams, R. L., & McLaugh- lin, T. F. (2001). The effects of a token economy employing instructional con- sequences for a third-grade student with learning disabilities: A data-based case study. Education and Treatment of Children, 24 (1), 99–106.

Himle, M. B., Miltenberger, R. G., Flessner, C., & Gatheridge, B. (2004). Teaching safety skills to children to prevent gun

701

Bibliography

play. Journal of Applied Behavior Analysis, 1, 1–9.

Himle, M. B., Miltenberger, R. G., Gath- eridge, B., & Flessner, C., (2004). An evaluation of two procedures for train- ing skills to prevent gun play in chil- dren. Pediatrics, 113, 70–77.

Hineline, P. N. (1977). Negative reinforce- ment and avoidance. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 364–414). Upper Saddle River, NJ: Prentice Hall.

Hineline, P. N. (1992). A self-interpretive behavior analysis. American Psychol- ogist, 47, 1274–1286.

Hobbs, T. R., & Holt, M. M. (1976). The ef- fects of token reinforcement on the behavior of delinquents in cottage set- tings. Journal of Applied Behavior Analysis, 9, 189–198.

Hoch, H., McComas, J. J., Johnson, L, Faranda, N., & Guenther, S. L. (2002). The effects of magnitude and quality of reinforcement on choice responding during play activities. Journal of Ap- plied Behavior Analysis, 35, 171–181.

Hoch, H., McComas, J. J., Thompson, A. L., & Paone, D. (2002). Concurrent rein- forcement schedules: Behavior change and maintenance without extinction. Journal of Applied Behavior Analysis, 35, 155–169.

Holcombe, A., Wolery, M., & Snyder, E. (1994). Effects of two levels of proce- dural fidelity with constant time delay with children’s learning. Journal of Be- havioral Education, 4, 49–73.

Holcombe, A., Wolery, M., Werts, M. G., & Hrenkevich, P. (1993). Effects of instructive feedback on future learn- ing. Journal of Behavioral Education, 3, 259–285.

Holland, J. G. (1978). Behaviorism: Part of the problem or part of the solution? Journal of Applied Behavior Analysis, 11, 163–174.

Holland, J. G., & Skinner, B. F. (1961). The analysis of behavior: A program for self- instruction. New York: McGraw-Hill.

Holman, J., Goetz, E. M., & Baer, D. M. (1977). The training of creativity as an operant and an examination of its gen- eralization characteristics. In B. C. Etzel, J. M. LeBlanc, & D. M. Baer (Eds.), New developments in behav- ioral research: Theory, method, and practice (pp. 441–471). Hillsdale, NJ: Erlbaum.

Holmes, G., Cautela, J., Simpson, M., Motes, P., & Gold, J. (1998). Factor structure of the School Reinforcement Survey Schedule: School is more than grades. Journal of Behavioral Educa- tion, 8, 131–140.

Holth, P. (2003), Generalized imitation and generalized matching to sample. The Behavior Analyst, 26, 155–158.

Holz, W. C., & Azrin, N. H. (1961). Dis- criminative properties of punishment. Journal of the Experimental Analysis of Behavior, 4, 225–232.

Holz, W. C., & Azrin, N. H. (1962). Recov- ery during punishment by intense noise. Journal of the Experimental Analysis of Behavior, 6, 407–412.

Holz, W. C., Azrin, N. H., & Ayllon, T. (1963). Elimination of behavior of mental patients by response-produced extinction. Journal of the Experimen- tal Analysis of Behavior, 6, 407–412.

Homme, L., Csanyi, A. P., Gonzales, M. A., & Rechs, J. R. (1970). How to use con- tingency contracting in the classroom. Champaign, IL: Research Press.

Honig, W. K. (Ed.). (1966). Operant behav- ior: Areas of research and application. New York: Appleton-Century-Crofts.

Hoover, H. D., Hieronymus,A. N., Dunbar, S. B. Frisbie, D. A., & Switch (1996). Iowa Tests of Basic Skills. Chicago: Riverside.

Hopkins, B. L. (1995). Applied behavior analysis and statistical process control? Journal of Applied Behavior Analysis, 28, 379–386.

Horner, R. D., & Baer, D. M. (1978). Mul- tiple-probe technique: A variation on the multiple baseline design. Journal of Applied Behavior Analysis, 11, 189–196.

Horner, R. D., & Keilitz, I. (1975). Training mentally retarded adolescents to brush their teeth. Journal of Applied Behavior Analysis, 8, 301–309.

Horner, R. H. (2002). On the status of knowledge for using punishment: A commentary. Journal of Applied Be- havior Analysis, 35, 465–467.

Horner, R. H., & McDonald, R. S. (1982). Comparison of single instance and gen- eral case instruction in teaching of a generalized vocational skill. Journal of the Association for Persons with Severe Handicaps, 7 (3), 7–20.

Horner, R. H., Carr, E. G., Halle, J., McGee, G., Odom, S., & Wolery, M. (2005). The use of single-subject research to

identify evidence-based practice in spe- cial education. Exceptional Children, 71, 165–179.

Horner, R. H., Day, M., Sprague, J., O’Brien, M., & Heathfield, L. (1991). Inter- spersed requests: A nonaversive pro- cedure for reducing aggression and self-injury during instruction. Journal of Applied Behavior Analysis, 24, 265–278.

Horner, R. H., Dunlap, G., & Koegel, R. L. (1988). Generalization and mainte- nance: Life-style changes in applied settings. Baltimore: Paul H. Brookes.

Horner, R. H., Eberhard, J. M., & Sheehan, M. R. (1986). Teaching generalized table bussing: The importance of neg- ative teaching examples. Behavior Modification, 10, 457–471.

Horner, R. H., Sprague, J., & Wilcox, B. (1982). Constructing general case pro- grams for community activities. In B. Wilcox & G. T. Bellamy (Eds.), Design of high school programs for severely handicapped students (pp. 61–98). Bal- timore: Paul H. Brookes.

Horner, R. H., Williams, J. A., & Steveley, J. D. (1987). Acquisition of generalized telephone use by students with moder- ate and severe mental retardation. Research in Developmental Disabili- ties, 8, 229–247.

Houten, R. V., & Rolider, A. (1990). The use of color mediation techniques to teach number identification and single digit multiplication problems to children with learning problems. Education and Treatment of Children, 13, 216–225.

Howell, K. W. (1998). Curriculum-based evaluation: Teaching and decision making (3rd ed.). Monterey, CA: Brooks/Cole.

Hughes, C. (1992). Teaching self-instruc- tion using multiple exemplars to pro- duce generalized problem-solving by individuals with severe mental retar- dation. Journal on Mental Retardation, 97, 302–314.

Hughes, C. (1997). Self-instruction. In M. Agran (Ed.), Self-directed learning: Teaching self-determination skills (pp. 144–170). Pacific Grove, CA: Brooks/ Cole.

Hughes, C., & Lloyd, J. W. (1993). An analysis of self-management. Journal of Behavioral Education, 3, 405–424.

Hughes, C., & Rusch, F. R. (1989). Teaching supported employees with severe men- tal retardation to solve problems.

702

Bibliography

Journal of Applied Behavior Analysis, 22, 365–372.

Hughes, C., Harmer, M. L., Killian, D. J., & Niarhos, F. (1995). The effects of multiple-exemplar self-instructional training on high school students’ gen- eralized conversational interactions. Journal of Applied Behavior Analysis, 28, 201–218.

Hume, K. M., & Crossman, J. (1992). Mu- sical reinforcement of practice behav- iors among competitive swimmers. Journal of Applied Behavior Analysis, 25, 665–670.

Humphrey, L. L., Karoly, P., & Kirschen- baum, D. S. (1978). Self-management in the classroom: Self-imposed re- sponse cost versus self-reward. Behav- ior Therapy, 9, 592–601.

Hundert, J., & Bucher, B. (1978). Pupils’ self-scored arithmetic performance: A practical procedure for maintaining ac- curacy. Journal of Applied Behavior Analysis, 11, 304.

Hunt, P., & Goetz, L. (1988). Teaching spon- taneous communication in natural set- tings using interrupted behavior chains. Topics in Language Disorders, 9, 58–71.

Hurley, A. D. N., & O’Sullivan, J. L. (1999). Informed consent for health care. In R. D. Dinerstein, S. S. Herr, & J. L. O’Sullivan (Eds.), A guide to consent (pp. 39–55). Washington DC: Ameri- can Association on Mental Retardation.

Hutchinson, R. R. (1977). By-products of aversive control In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of oper- ant behavior (pp. 415–431). Upper Sad- dle River, NJ: Prentice Hall.

Huybers, S., Van Houten, R., & Malenfant, J. E. L. (2004). Reducing conflicts between motor vehicles and pedestrians: The sep- arate and combined effects of pavement markings and a sign prompt. Journal of Applied Behavior Analysis, 37, 445–456.

Individuals with Disabilities Education Act of 1997, P. L. 105–17, 20 U.S.C. para. 1400 et seq.

Irvin, D. S., Thompson, T. J., Turner, W. D., & Williams, D. E. (1998). Utilizing re- sponse effort to reduce chronic hand mouthing. Journal of Applied Behav- ior Analysis, 31, 375–385.

Isaacs, W., Thomas, I., & Goldiamond, I. (1960). Application of operant condi- tioning to reinstate verbal behavior in psychotics. Journal of Speech and Hearing Disorders, 25, 8–12.

Iwata, B. A. (1987). Negative reinforcement in applied behavior analysis: An emerg- ing technology. Journal of Applied Be- havior Analysis, 20, 361–378.

Iwata, B. A. (1988). The development and adoption of controversial default tech- nologies. The Behavior Analyst, 11, 149–157.

Iwata, B. A. (1991). Applied behavior analysis as technological science. Journal of Applied Behavior Analysis, 24, 421–424.

Iwata, B. A. (2006). On the distinction be- tween positive and negative reinforce- ment. The Behavior Analyst, 29, 121–123.

Iwata, B. A., & DeLeon, I. (1996). The func- tional analysis screening tool. Gaines- ville, FL: The Florida Center on Self-Injury, The University of Florida.

Iwata, B. A., & Michael, J. L. (1994). Ap- plied implications of theory and re- search on the nature of reinforcement. Journal of Applied Behavior Analysis, 27, 183–193.

Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bau- man, K. E., & Richman, G. S. (1994). Toward a functional analysis of self-in- jury. Journal of Applied Behavior Analysis, 27, 197–209. (Reprinted from Analysis and Intervention in Develop- mental Disabilities, 2, 3–20, 1982).

Iwata, B. A., Pace, G. M., Cowdery, G. E., & Miltenberger, R. G. (1994). What makes extinction work: An analysis of procedural form and function. Journal of Applied Behavior Analysis, 27, 131–144.

Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R., Vollmer, T. R., Smith, R. G., Rodgers, T. A., Lerman, D. C., Shore, B. A., Mazaleski, J. L., Goh, H., Cowdery, G. E., Kalsher, M. J., & Willis, K. D. (1994). The functions of self-injurious behavior: An experimental-epidemiological analy- sis. Journal of Applied Behavior Analysis, 27, 215–240.

Iwata, B. A., Pace, G. M., Kissel, R. C., Nau, P. A., & Farber, J. M. (1990). The Self- Injury Trauma (SIT) Scale: A method for quantifying surface tissue damage caused by self-injurious behavior. Journal of Applied Behavior Analysis, 23, 99–110.

Iwata, B. A., Smith, R. G., & Michael, J. (2000). Current research on the influ- ence of establishing operations on be-

havior in applied settings. Journal of Ap- plied Behavior Analysis, 33, 411–418.

Iwata, B. A., Wallace, M. D., Kahng, S., Lindberg, J. S., Roscoe, E. M., Con- ners, J., Hanley, G. P., Thompson, R. H., & Worsdell, A. S. (2000). Skill acqui- sition in the implementation of func- tional analysis methodology. Journal of Applied Behavior Analysis, 33, 181–194.

Jacobs, H. E., Fairbanks, D., Poche, C. E., & Bailey, J. S. (1982). Multiple incentives in encouraging car pool formation on a university campus. Journal of Applied Behavior Analysis, 15, 141–149.

Jacobson, J. M., Bushell, D., & Risley, T. (1969). Switching requirements in a Head Start classroom. Journal of Ap- plied Behavior Analysis, 2, 43–47.

Jacobson, J. W., Foxx, R. M., & Mulick, J. A. (Eds.). (2005). Controversial ther- apies for developmental disabilities: Fads, fashion, and science in profes- sional practice. Hillsdale, NJ: Erlbaum.

Jason, L. A., & Liotta, R. F. (1982). Reduc- tion of cigarette smoking in a univer- sity cafeteria. Journal of Applied Behavior Analysis, 15, 573–577.

Johnson, B. M., Miltenberger, R. G., Egemo- Helm, K. R., Jostad, C., Flessner, C. A., & Gatheridge, B. (2005). Evaluation of behavior skills training for teaching ab- duction prevention skills to young chil- dren. Journal of Applied Behavior Analysis, 38, 67–78.

Johnson, B. M., Miltenberger, R. G., Knud- son, P., Egemo-Helm, K., Kelso, P., Jostad, C., & Langley, L. (2006). A pre- liminary evaluation of two behavioral skills training procedures for teaching abduction-prevention skills to school children. Journal of Applied Behavior Analysis, 39, 25–34.

Johnson, K. R., & Layng, T. V. J. (1992). Breaking the structuralist barrier: Lit- eracy and numeracy with fluency. American Psychologist, 47, 1475–1490.

Johnson, K. R., & Layng, T. V. J. (1994). The Morningside model of generative instruction. In R. Gardner, D. M., III, Sainato, J. O., Cooper, T. E., Heron, W. L., Heward, J., Eshleman, & T. A. Grossi (Eds.), Behavior analysis in ed- ucation: Focus on measurably superior instruction (pp. 173–197). Pacific Grove, CA: Brooks/Cole.

Johnson, T. (1973). Addition and subtraction math program with stimulus shaping

703

Bibliography

and stimulus fading. Produced pursuant to a grant from the Ohio Department of Education, BEH Act, P.L. 91-203, Title VI-G; OE G-0-714438(604). J. E. Fisher & J. O. Cooper, project codirectors.

Johnston, J. M. (1979). On the relation be- tween generalization and generality. The Behavior Analyst, 2, 1–6.

Johnston, J. M. (1991). We need a new model of technology. Journal of Ap- plied Behavior Analysis, 24, 425–427.

Johnston, J. M., & Pennypacker, H. S. (1980). Strategies and tactics for Human Be- havioral Research. Hillsdale, NJ: Erl- baum.

Johnston, J. M., & Pennypacker, H. S. (1993a). Strategies and tactics for human behavioral research (2nd ed.). Hillsdale, NJ: Erlbaum.

Johnston, J. M., & Pennypacker, H. S. (1993b). Readings for Strategies and tactics of behavioral research (2nd ed.). Hillsdale, NJ: Erlbaum.

Johnston, M. K., Kelly, C. S., Harris, F. R., & Wolf, M. M. (1966). An application of reinforcement principles to the de- velopment of motor skills of a young child. Child Development, 37, 370–387.

Johnston, R. J., & McLaughlin, T. F. (1982). The effects of free time on assignment completion and accuracy in arithmetic: A case study. Education and Treatment of Children, 5, 33–40.

Jones, F. H., & Miller, W. H. (1974). The ef- fective use of negative attention for re- ducing group disruption in special elementary school classrooms. Psyc- hological Record, 24, 435–448.

Jones, F. H., Fremouw, W., & Carples, S. (1977). Pyramid training of elementary school teachers to use a classroom man- agement “skill package.” Journal of Ap- plied Behavior Analysis, 10, 239–254.

Jones, R. R., Vaught, R. S., & Weinrott, M. (1977). Time-series analysis in operant research. Journal of Applied Behavior Analysis, 10, 151–166.

Journal of Applied Behavior Analysis. (2000). Manuscript preparation check- list. Journal of Applied Behavior Analy- sis, 33, 399. Lawrence, KS: Society for the Experimental Analysis of Behavior.

Kachanoff, R., Leveille, R., McLelland, H., & Wayner, M. J. (1973). Schedule in- duced behavior in humans. Physiology and Behavior, 11, 395–398.

Kadushin, A. (1972). The social work in- terview. New York: Columbia Univer- sity Press.

Kahng, S. W., & Iwata, B. A. (1998). Com- puterized systems for collecting real- time observational data. Journal of Applied Behavior Analysis, 31, 253–261.

Kahng, S. W., & Iwata, B. A. (2000). Com- puter systems for collecting real-time observational data. In T. Thompson, D. Felce, & F. J. Symons (Eds.), Behav- ioral observation: Technology and ap- plications in developmental disabilities (pp. 35–45). Baltimore: Paul H. Brookes.

Kahng, S. W., Iwata, B. A., DeLeon, I. G., & Wallace, M. D. (2000). A comparison of procedures for programming non- contingent reinforcement schedules. Journal of Applied Behavior Analysis, 33, 223–231.

Kahng, S. W., Iwata, B. A., Fischer, S. M., Page, T. J., Treadwell, K. R. H., Williams, D. E., & Smith, R. G. (1998). Temporal distributions of problems be- havior based on scatter plot analysis. Journal of Applied Behavior Analysis, 31, 593–604.

Kahng, S. W., Iwata, B. A., Thompson, R. H., & Hanley, G. P. (2000). A method for identifying satiation versus extinction effects under noncontingent reinforce- ment schedules. Journal of Applied Be- havior Analysis, 33, 419–432.

Kahng, S. W., Tarbox, J., & Wilke, A. (2001). Use of multicomponent treat- ment for food refusal. Journal of Ap- plied Behavior Analysis, 34, 93–96.

Kahng, S., & Iwata, B. A. (1999). Corre- spondence between outcomes of brief and extended functional analyses. Journal of Applied Behavior Analysis, 32, 149–159.

Kahng, S., Iwata, B. A., Fischer, S. M., Page, T. J., Treadwell, K. R. H., Williams, D. E., & Smith, R. G. (1998). Temporal distributions of prob- lem behavior based on scatter plot analysis. Journal of Applied Behavior Analysis, 31, 593–604.

Kahng, S., Iwata, B. A., & Lewin, A. B. (2002). Behavioral treatment of self-in- jury, 1964 to 2000. American Journal of Mental Retardation, 107 (3), 212–221.

Kame’enui, E. (2002, September). Begin- ning reading failure and the quantifi- cation of risk: Behavior as the supreme index. Presentation to the Focus on Be- havior Analysis in Education Confer- ence, Columbus, OH.

Kanfer, F. H. (1976). The many faces of self- control, or behavior modification changes its focus. Paper presented at

the Fifth International Banff Confer- ence, Banff, Alberta, Canada.

Kanfer, F. H., & Karoly P. (1972). Self- control: A behavioristic excursion into the lion’s den. Behavior Therapy, 3, 398–416.

Kantor, J. R. (1959). Interbehavioral psy- chology. Granville, OH: Principia Press.

Katzenberg, A. C. (1975). How to draw graphs. Kalamazoo, MI: Behaviordelia.

Kauffman, J. M. (2005). Characteristics of emotional and behavioral disorders of children and youth (8th ed.). Upper Saddle River, NJ: Merrill/Prentice Hall.

Kaufman, K. F., & O’Leary, K. D. (1972). Reward, cost, and self-evaluation pro- cedures for disruptive adolescents in a psychiatric hospital school. Journal of Applied Behavior Analysis, 5, 293–309.

Kazdin, A. E. (1973). The effects of vicari- ous reinforcement on attentive behavior in the classroom. Journal of Applied Behavior Analysis, 6, 77–78.

Kazdin, A. E. (1977). Artifact, bias, and complexity of assessment: The ABCs of reliability. Journal of Applied Be- havior Analysis, 10, 141–150.

Kazdin, A. E. (1978). History of behavior modification. Austin: TX: Pro-Ed.

Kazdin, A. E. (1979). Unobtrusive measures in behavioral assessment. Journal of Ap- plied Behavior Analysis, 12, 713–724.

Kazdin, A. E. (1980). Acceptability of alter- native treatments for deviant child be- havior. Journal of Applied Behavior Analysis, 13, 259–273.

Kazdin, A. E. (1982). Single case research designs: Methods for clinical and ap- plied settings. Boston: Allyn and Bacon.

Kazdin, A. E. (1982). Observer effects: Re- activity of direct observation. New Di- rections for Methodology of Social and Behavioral Science, 14, 5–19.

Kazdin, A. E. (2001). Behavior modification in applied settings (6th ed.). Belmont, CA: Wadsworth.

Kazdin, A. E., & Hartmann, D. P. (1978). The simultaneous-treatment design. Behavior Therapy, 9, 912–922.

Kee, M., Hill, S. M., & Weist, M. D. (1999). School-based behavior management of cursing, hitting, and spitting in a girl with profound retardation. Education and Treatment of Children, 22 (2), 171–178.

Kehle, T. J., Bray, M. A., Theodore, L. A., Jenson, W. R., & Clark, E. (2000). A multi-component intervention de- signed to reduce disruptive classroom

704

Bibliography

behavior. Psychology in the Schools, 37 (5), 475–481.

Keller, C. L., Brady, M. P., & Taylor, R. L. (2005). Using self-evaluation to im- prove student teacher interns’ use of specific praise. Education and Train- ing in Mental Retardation and Devel- opmental Disabilities, 40, 368–376.

Keller, F. S. (1941). Light aversion in the white rat. Psychological Record, 4, 235–250.

Keller, F. S. (1982). Pedagogue’s progress. Lawrence, KS: TRI Publications.

Keller, F. S. (1990). Burrhus Frederic Skin- ner (1904–1990) (a thank you). Jour- nal of Applied Behavior Analysis, 23, 404–407.

Keller, F. S., & Schoenfeld, W. M. (1950/1995). Principles of psychology. Acton, MA: Copley Publishing Group.

Keller, F. S., & Schoenfeld, W. N. (1995). Principles of psychology. Acton, MA: Copley Publishing Group. (Reprinted from Principles of psychology: A sys- tematic text in the science of behav- ior. New York: Appleton-Century- Crofts, 1950)

Kelley, M. E., Piazza, C. C., Fisher, W. W., & Oberdorff, A. J. (2003). Acquisition of cup drinking using previously re- fused foods as positive and negative re- inforcement. Journal of Applied Behavior Analysis, 36, 89–93.

Kelley, M. L., Jarvie, G. J., Middlebrook, J. L., McNeer, M. F., & Drabman, R. S. (1984). Decreasing burned children’s pain behavior: Impacting the trauma of hydrotherapy. Journal of Applied Behavior Analysis, 17, 147–158.

Kellum, K. K., Carr, J. E., & Dozier, C. L. (2001). Response-card instruction and student learning in a college class- room. Teaching of Psychology, 28 (2), 101–104.

Kelshaw-Levering, K., Sterling-Turner, H., Henry, J. R., & Skinner, C. H. (2000). Randomized interdependent group con- tingencies: Group reinforcement with a twist. Psychology in the Schools, 37 (6), 523–533.

Kennedy, C. H. (1994). Automatic rein- forcement: Oxymoron or hypothetical construct. Journal of Behavioral Edu- cation, 4 (4), 387–395.

Kennedy, C. H. (2005). Single-case designs for educational research. Boston: Allyn and Bacon.

Kennedy, C. H., & Haring, T. G. (1993). Teaching choice making during social interactions to students with profound

multiple disabilities. Journal of Applied Behavior Analysis, 26, 63–76.

Kennedy, C. H., & Sousa, G. (1995). Func- tional analysis and treatment of eye poking. Journal of Applied Behavior Analysis, 28, 27–37.

Kennedy, C. H., Meyer, K. A., Knowles, T., & Shukla, S. (2000). Analyzing the multiple functions of stereotypical be- havior for students with autism: Impli- cations for assessment and treatment. Journal of Applied Behavior Analysis, 33, 559–571.

Kennedy, G. H., Itkonen, T., & Lindquist, K. (1994). Nodality effects during equivalence class formation: An exten- sion to sight-word reading and concept development. Journal of Applied Be- havior Analysis, 27, 673–683.

Kern, L., Dunlap, G., Clarke, S., & Childs, K. E. (1995). Student-assisted func- tional assessment interview. Diagnost- ique, 19, 29–39.

Kern, L., Koegel, R., & Dunlap, G. (1984). The influence of vigorous versus mild exercise on autistic stereotyped behav- iors. Journal of Autism and Develop- mental Disorders, 14, 57–67.

Kerr, M. M., & Nelson, C. M. (2002). Strategies for addressing behavior problems in the classroom (4th ed.). Upper Saddle River, NJ: Merrill/ Prentice Hall.

Killu, K., Sainato, D. M., Davis, C. A., Os- pelt, H., & Paul, J. N. (1998). Effects of high-probability request sequences on preschoolers’ compliance and dis- ruptive behavior. Journal of Behavioral Education, 8, 347–368.

Kimball, J. W. (2002). Behavior-analytic in- struction for children with autism: Phi- losophy matters. Focus on Autism and Other Developmental Disabilities, 17 (2), 66–75.

Kimball, J. W., & Heward, W. L. (1993). A synthesis of contemplation, prediction, and control. American Psychologist, 48, 587–588.

Kirby, K. C., & Bickel, W. K. (1988). Toward an explicit analysis of generalization: A stimulus control interpretation. The Be- havior Analyst, 11, 115–129.

Kirby, K. C., Fowler, S. A., & Baer, D. M. (1991). Reactivity in self-recording: Obtrusiveness of recording procedure and peer comments. Journal of Applied Behavior Analysis, 24, 487–498.

Kladopoulos, C. N., & McComas, J. J. (2001). The effects of form training on

foul-shooting performance in members of a women’s college basketball team. Journal of Applied Behavior Analysis, 34, 329–332.

Klatt, K. P., Sherman, J. A., & Sheldon, J. B. (2000). Effects of deprivation on en- gagement in preferred activities by persons with developmental disabili- ties. Journal of Applied Behavior Analy- sis, 33, 495–506.

Kneedler, R. D., & Hallahan, D. P. (1981). Self-monitoring of on-task behavior with learning disabled children: Cur- rent studies and directions. Exceptional Education Quarterly, 2 (3), 73–82.

Knittel, D. (2002). Case study: A profes- sional guitarist on comeback road. In R. W. Malott & H. Harrison (Eds.), I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using behavior analysis to get your act together (pp. 8-5–8-6). Kalamazoo, MI: Department of Psy- chology, Western Michigan University.

Kodak, T., Grow, L., & Northrup, J. (2004). Functional analysis and treatment of elopement for a child with attention deficit hyperactivity disorder. Journal of Applied Behavior Analysis, 37, 229–232.

Kodak, T., Miltenberger, R. G., & Roma- niuk, C. (2003). The effects of differ- ential negative reinforcement of other behavior and noncontingent escape on compliance. Journal of Applied Be- havior Analysis, 36, 379–382.

Koegal, R. L., & Rincover, A. (1977). Re- search on the differences between generalization and maintenance in extra-therapy responding. Journal of Applied Behavior Analysis, 10, 1–12.

Koegel, L. K., Carter, C. M., & Koegel, R. L. (2003). Teaching children with autism self-initiations as a pivotal re- sponse. Topics in Language Disorders, 23 (2), 134–145.

Koegel, L. K., Koegel, R. L., Hurley, C., & Frea, W. D. (1992). Improving social skills and disruptive behavior in chil- dren with autism through self-manage- ment. Journal of Applied Behavior Analysis, 25, 341–353.

Koegel, R. L., & Frea, W. (1993). Treatment of social behavior in autism through the modification of pivotal social skills. Journal of Applied Behavior Analysis, 26, 369–377.

Koegel, R. L., & Koegel, L. K. (1988). Gen- eralized responsivity and pivotal be- haviors. In R. H. Horner, G. Dunlap, &

705

Bibliography

R. L. Koegel (Eds.), Generalization and maintenance: Life-style changes in applied settings (pp. 41–66). Balti- more; Paul H. Brookes.

Koegel, R. L., & Koegel, L. K. (1990). Ex- tended reductions in stereoptypic be- havior of students with autism through a self-management treatment package. Journal of Applied Behavior Analysis, 23, 119–127.

Koegel, R. L., & Williams, J. A. (1980). Di- rect versus indirect response-reinforcer relationships in teaching autistic chil- dren. Journal of Abnormal Child Psy- chology, 8, 537–547.

Koegel, R. L., Koegel, L. K., & Schreibman, L. (1991). Assessing and training parents in teaching pivotal behaviors. In R. J. Prinz (Ed.), Advances in behavioral as- sessment of children and families (Vol. 5, pp. 65–82). London: Jessica Kingsley.

Koehler, L. J., Iwata, B. A., Roscoe, E. M., Rolider, N. U., & O’Steen, L. E. (2005). Effects of stimulus variation on the reinforcing capability of nonpre- ferred stimuli. Journal of Applied Be- havior Analysis, 38, 469–484.

Koenig, C. H., & Kunzelmann, H. P. (1980). Classroom learning screening. Colum- bus, OH: Charles E. Merrill.

Kohler, F. W., & Greenwood, C. R. (1986). Toward a technology of generalization: The identification of natural contin- gencies of reinforcement. The Behav- ior Analyst, 9, 19–26.

Kohler, F. W., Strain, P. S., Maretsky, S., & Decesare, L. (1990). Promoting posi- tive and supportive interactions be- tween preschoolers: An analysis of group-oriented contingencies. Journal of Early Intervention, 14 (4), 327–341.

Kohn, A. (1997). Students don’t “work”— they learn. Education Week, Septem- ber 3, 1997. Retrieved January 1, 2004, from www.alfiekohn.org/teaching/ edweek/sdwtl.htm.

Komaki, J. L. (1998). When performance improvement is the goal: A new set of criteria for criteria. Journal of Applied Behavior Analysis, 31, 263–280.

Konarski, E. A., Jr., Crowell, C. R., & Dug- gan, L. M. (1985). The use of response deprivation to increase the academic per- formance of EMR students. Applied Re- search in Mental Retardation, 6, 15–31.

Konarski, E. A., Jr., Crowell, C. R., John- son, M. R., & Whitman T. L. (1982). Response deprivation, reinforcement, and instrumental academic perfor-

mance in an EMR classroom. Behavior Therapy, 13, 94–102.

Konarski, E. A., Jr., Johnson, M. R., Crow- ell, C. R., & Whitman T. L. (1980). Re- sponse deprivation and reinforcement in applied settings: A preliminary re- port. Journal of Applied Behavior Analysis, 13, 595–609.

Koocher, G. P., & Keith-Spiegel, P. (1998). Ethics in psychology: Professional standards and cases. New York, Ox- ford: Oxford University Press.

Kosiewicz, M. M., Hallahan, D. P., Lloyd, J. W., & Graves, A. W. (1982). Effects of self-instruction and self-correction procedures on handwriting perfor- mance. Learning Disability Quarterly, 5 (1), 71–78.

Kostewicz, D. E., Kubina, R. M., Jr., & Cooper, J. O. (2000). Managing ag- gressive thoughts and feelings with daily counts of non-aggressive thoughts and feelings: A self-experiment. Journal of Behavior Therapy and Ex- perimental Psychiatry, 31, 177–187.

Kounin, J. (1970). Discipline and group management in classrooms. New York: Holt, Rinehart & Winston.

Kozloff, M. A. (2005). Fads in general ed- ucation: Fad, fraud, and folly. In J. W. Jacobson, R. M. Foxx, & J. A. Mulick (Eds.), Controversial therapies in de- velopmental disabilities: Fads, fash- ion, and science in professional practice (pp. 159–174). Hillsdale, NJ: Erlbaum.

Krantz, P. J., & McClannahan, L. E. (1993). Teaching children with autism to initi- ate to peers: Effects of a script-fading procedure. Journal of Applied Behav- ior Analysis, 26, 121–132.

Krantz, P. J., & McClannahan, L. E. (1998). Social interaction skills for children with autism: A script-fading procedure for beginning readers. Journal of Ap- plied Behavior Analysis, 31, 191–202.

Krasner, L. A., & Ullmann, L. P. (Eds.). (1965). Research in behavior modifi- cation: New developments and impli- cations. New York: Holt, Rinehart & Winston.

Kubina, R. M., Jr. (2005). The relations among fluency, rate building, and prac- tice: A response to Doughty, Chase, and O’Shields (2004). The Behavior Ana- lyst, 28, 73–76.

Kubina, R. M., & Cooper, J. O. (2001). Changing learning channels: An effi- cient strategy to facilitate instruction

and learning. Intervention in School and Clinic, 35, 161–166.

Kuhn, S. A. C., Lerman, D. C., & Vorndran, C. M. (2003). Pyramidal training for families of children with problem be- havior. Journal of Applied Behavior Analysis, 36, 77–88.

La Greca, A. M., & Schuman, W. B. (1995). Adherence to prescribed medical regi- mens. In M. C. Roberts (Ed.), Hand- book of pediatric psychology (2nd ed., pp. 55–83). New York: Guildford.

LaBlanc, L. A., Coates, A. M., Daneshvar, S. Charlop-Christy, Morris, C., & Lan- caster, B. M. (2003). Journal of Applied Behavior Analysis, 36, 253–257.

Lalli, J. S., Browder, D. M., Mace, F. C., & Brown, D. K. (1993). Teacher use of descriptive analysis data to implement interventions to decrease students’ problem behaviors. Journal of Applied Behavior Analysis, 26, 227–238.

Lalli, J. S., Casey, S., & Kates, K. (1995). Reducing escape behavior and increas- ing task completion with functional communication training, extinction, and response chaining. Journal of Ap- plied Behavior Analysis, 28, 261–268.

Lalli, J. S., Livezey, K., & Kates, D. (1996). Functional analysis and treatment of eye poking with response blocking. Journal of Applied Behavior Analysis, 29, 129–132.

Lalli, J. S., Mace, F. C., Livezey, K., & Kates, K. (1998). Assessment of stimulus gen- eralization gradients in the treatment of self-injurious behavior. Journal of Ap- plied Behavior Analysis, 31, 479–483.

Lalli, J. S., Zanolli, K., & Wohn, T. (1994). Using extinction to promote response variability. Journal of Applied Behavior Analysis, 27, 735–736.

Lambert, M. C., Cartledge, G., Lo, Y., & Heward, W. L. (2006). Effects of re- sponse cards on disruptive behavior and participation by fourth-grade students during math lessons in an urban school. Journal of Positive Behavioral Inter- ventions, 8, 88–99.

Lambert, N., Nihira, K., & Leland, H. (1993). Adaptive Behavior Scale— School (2nd ed.). Austin, TX: Pro-Ed.

Landry, L., & McGreevy, P. (1984). The paper clip counter (PCC): An inexpen-

Kubina, R. M., Haertel, M. W., & Cooper, J. O. (1994). Reducing negative inner behavior of senior citizens: The one- minute counting procedure. Journal of Precision Teaching, 11 (2), 28–35.

706

Bibliography

sive and reliable device for collecting behavior frequencies. Journal of Pre- cision Teaching, 5, 11–13.

Lane, S. D., & Critchfield, T. S. (1998). Clas- sification of vowels and consonants by individuals with moderate mental re- tardation: Development of arbitrary re- lations via match-to-sample training with compound stimuli. Journal of Ap- plied Behavior Analysis, 31, 21–41.

Laraway, S., Snycerski, S., Michael, J., & Poling, A. (2001). The abative effect: A new term to describe the action of antecedents that reduce operant re- sponding. The Analysis of Verbal Be- havior, 18, 101–104.

Larowe, L. N., Tucker, R. D., & McGuire, J. M. (1980). Lunchroom noise con- trol using feedback and group re- inforcement. Journal of School Psychology, 18, 51–57.

Lasiter, P. S. (1979). Influence of contingent responding on schedule-induced activ- ity in human subjects. Physiology and Behavior, 22, 239–243.

Lassman, K. A., Jolivette, K., & Wehby, J. H. (1999). “My teacher said I did good work today!”: Using collaborative be- havioral contracting. Teaching Excep- tional Children, 31 (4), 12–18.

Lattal, K. A. (1969). Contingency manage- ment of toothbrushing behavior in a summer camp for children. Journal of Applied Behavior Analysis, 2, 195–198.

Lattal, K. A. (1995). Contingency and be- havior analysis. The Behavior Analyst, 18, 209–224.

Lattal, K. A. (Ed.). (1992). Special issue: Reflections on B. F. Skinner and psy- chology. American Psychologist, 47, 1269–1533.

Lattal, K. A., & Griffin, M. A. (1972). Pun- ishment contrast during free operant re- sponding. Journal of the Experimental Analysis of Behavior, 18, 509–516.

Lattal, K. A., & Lattal, A. D. (2006). And yet . . .: Further comments on distin- guishing positive and negative reinforce- ment. The Behavior Analyst, 29, 129–134.

Lattal, K. A., & Neef, N. A. (1996). Recent reinforcement-schedule research and ap- plied behavior analysis. Journal of Ap- plied Behavior Analysis, 29, 213–220.

Lattal, K. A., & Shahan, T. A. (1997). Dif- fering views of contingencies: How contiguous. The Behavior Analyst, 20, 149–154.

LaVigna, G. W., & Donnellen, A. M. (1986). Alternatives to punishment: Solving be-

havior problems with nonaversive strategies. New York: Irvington.

Layng, T. V. J., & Andronis, P. T. (1984). To- ward a functional analysis of delusional speech and hallucinatory behavior. The Behavior Analyst, 7, 139–156.

Lee, C., & Tindal, G. A. (1994). Self-record- ing and goal-setting: Task and math productivity of low-achieving Korean elementary school students. Journal of Behavioral Education, 4, 459–479.

Lee, R., McComas, J. J., & Jawor, J. (2002). The effects of differential and lag rein- forcement schedules on varied verbal responding by individuals with autism. Journal of Applied Behavior Analysis, 35, 391–402.

Lee, V. L. (1988). Beyond behaviorism. Hillsdale, NJ: Erlbaum.

Leitenberg, H. (1973). The use of single- case methodology in psychotherapy research. Journal of Abnormal Psy- chology, 82, 87–101.

Lenz, M., Singh, N., & Hewett, A. (1991). Overcorrection as an academic reme- diation procedure. Behavior Modifica- tion, 15, 64–73.

Lerman, D. C. (2003). From the laboratory to community application: Transla- tional research in behavior analysis. Journal of Applied Behavior Analysis, 36, 415–419.

Lerman, D. C., & Iwata, B. A. (1993). De- scriptive and experimental analysis of variables maintaining self-injurious be- havior. Journal of Applied Behavior Analysis, 26, 293–319.

Lerman, D. C., & Iwata, B. A. (1995). Prevalence of the extinction burst and its attenuation during treatment. Jour- nal of Applied Behavior Analysis, 28, 93–94.

Lerman, D. C., & Iwata, B. A. (1996a). De- veloping a technology for the use of op- erant extinction in clinical settings: An examination of basic and applied re- search. Journal of Applied Behavior Analysis, 29, 345–382.

Lerman, D. C., & Iwata, B. A. (1996b). A methodology for distinguishing between extinction and punishment effects associated with response block- ing. Journal of Applied Behavior Analy- sis, 29, 231–234.

Lerman, D. C., & Vorndran, C. M. (2002). On the status of knowledge for using punishment: Implications for treating behavior disorders. Journal of Applied Behavior Analysis, 35, 431–464.

Lerman, D. C., Iwata, B. A., & Wallace, M. D. (1999). Side effects of extinction: Prevalence of bursting and aggression during the treatment of self-injurious behavior. Journal of Applied Behavior Analysis, 32, 1–8.

Lerman, D. C., Iwata, B. A., Smith, R. G., Zarcone, J. R., & Vollmer, T. R. (1994). Transfer of behavioral function as a contributing factor in treatment relapse. Journal of Applied Behavior Analysis, 27, 357–370.

Lerman, D. C., Iwata, B. A., Zarcone, J. R., & Ringdahl, J. (1994). Assessment of stereotypic and self-injurious behavior as adjunctive responses. Journal of Ap- plied Behavior Analysis, 27, 715–728.

Lerman, D. C., Kelley, M. E., Vorndran, C. M., & Van Camp, C. M. (2003). Col- lateral effects of response blocking dur- ing the treatment of stereotypic behavior. Journal of Applied Behavior Analysis, 36, 119–123.

Lerman, D. C., Kelley, M. E., Vorndran, C. M., Kuhn, S. A. C., & LaRue, Jr., R. H. (2002). Reinforcement magnitude and responding during treatment with differential reinforcement. Journal of Applied Behavior Analysis, 35, 29–48.

Levondoski, L. S., & Cartledge, G. (2000). Self-monitoring for elementary school children with serious emotional distur- bances: Classroom applications for increased academic responding. Be- havioral Disorders, 25, 211–224.

Lewis, T. J., Powers, L. J., Kelk, M. J., & Newcomer, L. L. (2002). Reducing problem behaviors on the playground: An intervention of the application of school-wide positive behavior sup- ports. Psychology in the Schools, 39 (2), 181–190.

Lewis, T., Scott, T. & Sugai, G. (1994). The problem behavior questionnaire: A teacher-based instrument to develop functional hypotheses of problem

Lerman, D. C., Iwata, B. A., Shore, B. A., & DeLeon, I. G. (1997). Effects of in- termittent punishment on self-injurious behavior: An evaluation of schedule thinning. Journal of Applied Behavior Analysis, 30, 187–201.

Lerman, D. C., Iwata, B. A., Shore, B. A., & Kahng, S. W. (1996). Responding maintained by intermittent reinforce- ment: Implications for the use of ex- tinction with problem behavior in clinical settings. Journal of Applied Behavior Analysis, 29, 153–171.

707

Bibliography

behavior in general education class- rooms. Diagnostique, 19, 103–115.

Lindberg, J. S., Iwata, B. A., Kahng, S. W., & DeLeon, I. G. (1999). DRO contin- gencies: Analysis of variable-momen- tary schedules. Journal of Applied Behavior Analysis, 32, 123–136.

Lindberg, J. S., Iwata, B. A., Roscoe, E. M., Worsdell, A. S., & Hanley, G. P. (2003). Treatment efficacy of noncontingent re- inforcement during brief and extended application. Journal of Applied Behav- ior Analysis, 36, 1–19.

Lindsley, O. R. (1956). Operant conditioning methods applied to research in chronic schizophrenia. Psychiatric Research Reports, 5, 118–139.

Lindsley, O. R. (1960). Characteristics of the behavior of chronic psychotics as re- vealed by free-operant conditioning methods. Diseases of the Nervous Sys- tem (Monograph Supplement), 21, 66–78.

Lindsley, O. R. (1968). A reliable wrist counter for recording behavior rates. Journal of Applied Behavior Analy- sis, 1, 77–78.

Lindsley, O. R. (1971). An interview. Teaching Exceptional Children, 3, 114–119.

Lindsley, O. R. (1985). Quantified trends in the results of behavior analysis. Presi- dential address at the Eleventh Annual Convention of the Association for Be- havior Analysis, Columbus, OH.

Lindsley, O. R. (1990). Precision teaching: By teachers for children. Teaching Ex- ceptional Children, 22, 10–15.

Lindsley, O. R. (1992). Precision teaching: Discoveries and effects. Journal of Ap- plied Behavior Analysis, 25, 51–57.

Lindsley, O. R. (1996). The four free-oper- ant freedoms.The Behavior Analyst, 19, 199–210.

Linehan, M. (1977). Issues in behavioral in- terviewing. In J. D. Cone & R. P. Hawkins (Eds.), Behavioral assessment: New directions in clinical psychology (pp. 30–51). New York: Bruner/Mazel.

Linscheid, T. R, Iwata, B. A., Ricketts, R. W., Williams, D. E., & Griffin, J. C. (1990). Clinical evaluation of the self-injurious behavior inhibiting system (SIBIS). Journal of Applied Behavior Analysis, 23, 53–78.

Linscheid, T. R., & Meinhold, P. (1990). The controversy over aversives: Basic oper- ant research and the side effects of pun- ishment. In A. C. Repp & N. N. Singh (Eds.), Perspectives on the use of non-

aversive and aversive interventions for persons with developmental disabilities (pp. 59–72). Sycamore, IL: Sycamore.

Linscheid, T. R., & Reichenbach, H. (2002). Multiple factors in the long-term ef- fectiveness of contingent electric shock treatment for self-injurious behavior: A case example. Research in Develop- mental Disabilities, 23, 161–177.

Linscheid, T. R., Iwata, B. A., Ricketts, R. W., Williams, D. E., & Griffin, J. C. (1990). Clinical evaluation of SIBIS: The self- injurious behavior inhibiting system. Journal of Applied Behavior Analysis, 23, 53–78.

Linscheid, T. R., Pejeau C., Cohen S., & Footo-Lenz, M. (1994). Positive side effects in the treatment of SIB using the Self-Injurious Behavior Inhibiting Sys- tem (SIBIS): Implications for operant and biochemical explanations of SIB. Research in Developmental Disabili- ties, 15, 81–90.

Lipinski, D. P., Black, J. L., Nelson, R. O., & Ciminero, A. R. (1975). Influence of motivational variables on the reactivity and reliability of self-recording. Journal of Consulting and Clinical Psychology, 43, 637–646.

Litow, L., & Pumroy, D. K. (1975). A brief review of classroom group-oriented contingencies. Journal of Applied Be- havior Analysis, 3, 341–347.

Lloyd, J. W., Bateman, D. F., Landrum, T. J., & Hallahan, D. P. (1989). Self- recording of attention versus produc- tivity. Journal of Applied Behavior Analysis. 22, 315–323.

Lloyd, J. W., Eberhardt, M. J., Drake, G. P., Jr. (1996). Group versus individual re- inforcement contingencies with the context of group study conditions. Journal of Applied Behavior Analysis, 29, 189–200.

Lo, Y. (2003). Functional assessment and individualized intervention plans: In- creasing the behavior adjustment of urban learners in general and special education settings. Unpublished doc- toral dissertation. Columbus, OH: The Ohio State University.

Logan, K. R., & Gast, D. L. (2001). Con- ducting preference assessments and re- inforcer testing for individuals with profound multiple disabilities: Issues and procedures. Exceptionality, 9 (3), 123–134.

Logan, K. R., Jacobs, H. A., Gast, D. L., Smith, P. D., Daniel, J., & Rawls, J.

(2001). Preferences and reinforcers for students with profound multiple dis- abilities: Can we identify them? Journal of Developmental and Physi- cal Disabilities, 13, 97–122.

Long, E. S., Miltenberger, R. G., Ellingson, S. A., & Ott, S. M. (1999). Augmenting simplified habit reversal in the treat- ment of oral-digital habits exhibited by individuals with mental retardation. Journal of Applied Behavior Analysis, 32, 353–365.

Lovaas, O. I. (1977). The autistic child: Lan- guage development through behavior modification. New York: Irvington.

Lovitt, T. C. (1973). Self-management pro- jects with children with behavioral disabilities. Journal of Learning Dis- abilities, 6, 138–150.

Lovitt, T. C., & Curtiss, K. A. (1969). Aca- demic response rates as a function of teacher- and self-imposed contingen- cies. Journal of Applied Behavior Analysis, 2, 49–53.

Lowenkron, B. (2004). Meaning: A verbal behavior account. The Analysis of Ver- bal Behavior, 20, 77–97.

Luce, S. C., & Hall, R. V. (1981). Contin- gent exercise: A procedure used with differential reinforcement to reduce bizarre verbal behavior. Education and Treatment of Children, 4, 309–327.

Luce, S. C., Delquadri, J., & Hall, R. V. (1980). Contingent exercise: A mild but powerful procedure for suppressing in- appropriate verbal and aggressive be- havior. Journal of Applied Behavior Analysis, 13, 583–594.

Luciano, C. (1986). Acquisition, mainte- nance, and generalization of produc- tive intraverbal behavior through transfer of stimulus control procedures. Applied Research in Mental Retarda- tion, 7, 1–20.

Ludwig, R. L. (2004). Smiley faces and spin- ners: Effects of self-monitoring of productivity with an indiscriminable contingency of reinforcement on the on-task behavior and academic produc- tivity by kindergarteners during inde- pendent seatwork. Unpublished master’s thesis, The Ohio State University.

Luiselli, J. K. (1984). Controlling disruptive behaviors of an autistic child: Parent- mediated contingency management in the home setting. Education and Treat- ment of Children, 3, 195–203.

Lynch, D. C., & Cuvo, A. J. (1995). Stimu- lus equivalence instruction of fraction-

708

Bibliography

decimal realations. Journal of Applied Behavior Analysis, 28, 115–126.

Lyon, C. S., & Lagarde, R. (1997). Tokens for success: Using the graduated rein- forcement system. Teaching Excep- tional Children, 29 (6), 52–57.

Maag, J. W., Reid, R., & DiGangi, S. A. (1993). Differential effects of self- monitoring attention, accuracy, and productivity. Journal of Applied Be- havior Analysis, 26, 329–344.

Mabry, J. H., (1994). Review of R. A. Har- ris’ Linguistic wars. The Analysis of Verbal Behavior, 12, 79–86.

Mabry, J. H., (1995). Review of Pinker’s The language instinct. The Analysis of Ver- bal Behavior, 12, 87–96.

MacCorquodale, K. (1970). On Chomsky’s review of Skinner’s Verbal behavior. Journal of the Experimental Analysis of Behavior, 13, 83–99.

MacCorquodale, K., & Meehl, P. (1948). On a distinction between hypothetical con- structs and intervening variables. Psyc- hological Record, 55, 95–107.

MacDuff, G.S. Krantz, P. J., & McClannahan, L. E. (1993). Teaching children with autism to use photographic activity schedules: Maintenance and generaliza- tion of complex response chains. Journal of Applied Behavior Analysis, 26, 89–97.

Mace, F. C. (1996). In pursuit of general be- havioral relations. Journal of Applied Behavior Analysis, 29, 557–563.

Mace, F. C., & Belfiore, P. (1990). Behavioral momentum in the treatment of escape- motivated stereotypy. Journal of Applied Behavior Analysis, 23, 507–514.

Mace, F. C., Hock, M. L., Lalli, J. S., West, B. J., Belfiore, P., Pinter, E., & Brown, D. K. (1988). Behavioral momentum in the treatment of noncompliance. Journal of Applied Behavior Analysis, 21, 123–141.

Mace, F. C., Page, T. J., Ivancic, M. T., & O’Brien, S. (1986). Effectiveness of brief time-out with and without con- tingency delay: A comparative analy- sis. Journal of Applied Behavior Analysis, 19, 79–86.

MacNeil, J., & Thomas, M. R. (1976). Treatment of obsessive-compulsive hairpulling (trichotillomania) by be- havioral and cognitive contingency manipulation. Journal of Behavior Therapy and Experimental Psychiatry, 7, 391–392.

Madden, G. J., Chase, P. N., & Joyce, J. H. (1998). Making sense of sensitivity in

the human operant literature. The Be- havior Analyst, 21, 1–12.

Madsen, C. H., Becker, W. C., Thomas, D. R., Koser, L., & Plager, E. (1970). An analy- sis of the reinforcing function of “sit down” commands. In R. K. Parker (Ed.), Readings in educational psychology (pp. 71–82). Boston: Allyn & Bacon.

Maglieri, K. A., DeLeon, I. G., Rodriguez- Catter, V., & Sevin, B. M. (2000). Treatment of covert food stealing in an individual with Prader-Willi Syndrome. Journal of Applied Behavior Analysis, 33, 615–618.

Maheady, L., Mallete, B., Harper, G. F., & Saca, K. (1991). Heads together: A peer-mediated option for improving the academic achievement of heteroge- neous learning groups. Remedial and Special Education, 12 (2), 25–33.

Mahoney, M. J. (1971). The self-manage- ment of covert behavior: A case study. Behavior Therapy, 2, 575–578.

Mahoney, M. J. (1976). Terminal terminol- ogy. Journal of Applied Behavior Analysis, 9, 515–517.

Malesky, B. M. (1974). Behavior record- ing as treatment. Behavior Therapy, 5, 107–111.

Maloney, K. B., & Hopkins, B. L. (1973). The modification of sentence structure and its relationship to subjective judgments of creativity in writing. Journal of Ap- plied Behavior Analysis, 6, 425–433.

Malott, R. W. (1981). Notes from a radical behaviorist. Kalamazoo, MI: Author.

Malott, R. W. (1984). Rule-governed be- havior, self-management, and the de- velopmentally disabled: A theoretical analysis. Analysis and Intervention in Developmental Disabilities, 6, 53–68.

Malott, R. W. (1988). Rule-governed be- havior and behavioral anthropology. The Behavior Analyst, 11, 181–203.

Malott, R. W. (1989). The achievement of evasive goals: Control by rules de- scribing contingencies that are not di- rect acting. In S. C. Hayes (Ed.), Rule- governed behavior: Cognition, contin- gencies, and instructional control (pp. 269–322). Reno, NV: Context Press.

Malott, R. W. (2005a). Self-management. In M. Hersen & J. Rosqvist, (Eds.), En- cyclopedia of behavior modification and cognitive behavior therapy (Vol- ume I: Adult Clinical Applications) (pp. 516–521). Newbury Park, CA: Sage.

Malott, R. W. (2005b). Behavioral systems analysis and higher education. In W. L.

Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner III, L. D. Peterson, S. B. Hersh, & J. C. Dardig (Eds.), Focus on behavior analysis in education: Achievements, challenges, and oppor- tunities (pp. 211–236). Upper Saddle River, NJ: Merrill/Prentice Hall.

Malott, R. W., & Garcia, M. E. (1991). The role of private events in rule-governed behavior. In L. J. Hayes & P. Chase (Eds.), Dialogues on verbal behavior (pp. 237–254). Reno, NV: Context Press.

Malott, R. W., & Harrison, H. (2002). I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using behavior analy- sis to get your act together. Kalamazoo, MI: Department of Psychology, West- ern Michigan University.

Malott, R. W., & Suarez, E. A. (2004). Ele- mentary principles of behavior (5th ed.). Upper Saddle River, NJ: Prentice Hall.

Malott, R. W., & Trojan Suarez, E. A. (2004). Elementary principles of be- havior (5th ed.). Upper Saddle River, NJ: Prentice Hall.

Malott, R. W., General, D. A., & Snapper, V. B. (1973). Issues in the analysis of be- havior. Kalamazoo, MI: Behaviordelia.

Malott, R. W., Tillema, M., & Glenn, S. (1978). Behavior analysis and behavior modification: An introduction. Kala- mazoo, MI: Behaviordelia.

Mank, D. M., & Horner, R. H. (1987). Self- recruited feedback: A cost-effective procedure for maintaining behavior. Research in Developmental Disabili- ties, 8, 91–112.

March, R., Horner, R. H., Lewis-Palmer, T., Brown, D., Crone, D., Todd, A. W. et al. (2000). Functional Assessment Checklist for Teachers and Staff (FACTS). Eugene, OR: University of Oregon, Department of Educational and Community Supports.

Marckel, J. M., Neef, N. A., & Ferreri, S. J. (2006). A preliminary analysis of teaching improvisation with the picture exchange communication system to children with autism. Journal of Ap- plied Behavior Analysis, 39, 109–115.

Marcus, B. A., & Vollmer, T. R. (1995). Ef- fects of differential negative reinforce- ment on disruption and compliance. Journal of Applied Behavior Analysis, 28, 229–230.

Marholin, D., II, & Steinman, W. (1977). Stimulus control in the classroom as a

709

Bibliography

function of the behavior reinforced. Journal of Applied Behavior Analysis, 10, 465–478.

Marholin, D., Touchette, P. E., & Stuart, R. M. (1979). Withdrawal of chronic chlorpromazine medication: An exper- imental analysis. Journal of Applied Behavior Analysis, 12, 150–171.

Markle, S. M. (1962). Good frames and bad: A grammar of frame writing (2nd ed.). New York: Wiley.

Markwardt, F. C. (2005). Peabody Individual Achievement Test. Circle Pines, MN: American Guidance Service.

Marmolejo, E. K., Wilder, D. A., & Bradley, L. (2004). A preliminary analysis of the effects of response cards on student performance and participation in an upper division university course. Journal of Applied Behavior Analysis, 37, 405–410.

Marr, J. (2003). Empiricism. In K. A. Lattal & P. C. Chase (Eds.), Behavior theory and philosophy (pp. 63–82). New York: Kluwer/Plenum.

Marshall, A. E., & Heward, W. L. (1979). Teaching self-management to incar- cerated youth. Behavioral Disorders, 4, 215–226.

Marshall, K. J., Lloyd, J. W., & Hallahan, D. P. (1993). Effects of training to in- crease self-monitoring accuracy. Journal of Behavioral Education, 3, 445–459.

Martella, R., Leonard, I. J., Marchand- Martella, N. E., & Agran, M. (1993). Self-monitoring negative statements. Journal of Behavioral Education, 3, 77–86.

Martens, B. K., Lochner, D. G., & Kelly, S. Q. (1992). The effects of variable-interval reinforcement on academic engagement: A demonstration of matching theory. Journal of Applied Behavior Analysis, 25, 143–151.

Martens, B. K., Witt, J. C., Elliott, S. N., & Darveaux, D. (1985). Teacher judg- ments concerning the acceptability of school-based interventions. Professional Psychology: Research and Practice, 16, 191–198.

Martin, G., & Pear, J. (2003). Behavior mod- ification: What it is and how to do it (7th ed.). Upper Saddle River, NJ: Prentice Hall.

Martinez-Diaz, J. A. (2003). Raising the bar. Presidential address presented at the an- nual conference of the Florida Associ- ation for Behavior Analysis, St. Petersburg.

Mastellone, M. (1974). Aversion therapy: A new use of the old rubberband. Journal of Behavior Therapy and Experimen- tal Psychiatry, 5, 311–312.

Matson, J. L., & Taras, M. E., (1989). A 20- year review of punishment and alter- native methods to treat problem behaviors in developmentally delayed persons. Research in Developmental Disabilities, 10, 85–104.

Mattaini, M. A. (1995). Contingency dia- grams as teaching tools. The Behavior Analyst, 18, 93–98.

Maurice, C. (1993). Let me hear your voice: A family’s triumph over autism. New York: Fawcett Columbine.

Maurice, C. (2006). The autism wars. In W. L. Heward (Ed.), Exceptional children: An introduction to special education (8th ed., pp. 291–293). Upper Saddle River, NJ: Merrill/Prentice Hall.

Maxwell, J. C. (2003). There’s no such thing as “business” ethics: There’s only one rule for decision making. New York: Warner Business Books: A Time Warner Company.

Mayer, G. R., Sulzer, B., & Cody, J. J. (1968). The use of punishment in mod- ifying student behavior. Journal of Spe- cial Education, 2, 323–328.

Mayfield, K. H., & Chase, P. N. (2002). The effects of cumulative practice on math- ematics problem solving. Journal of Ap- plied Behavior Analysis, 35, 105–123.

Mayhew, G., & Harris, F. (1979). Decreas- ing self-injurious behavior. Behavior Modification, 3, 322–326.

Mazaleski, J. L., Iwata, B. A., Rodgers, T. A., Vollmer, T. R., & Zarcone, J. R. (1994). Protective equipments as treatment for stereotypic hand mouthing: Sensory ex- tinction or punishment effects? Journal of Applied Behavior Analysis, 27, 345–355.

McAllister, L. W., Stachowiak, J. G., Baer, D. M., & Conderman, L. (1969). The application of operant conditioning techniques in a secondary school class- room. Journal of Applied Behavior Analysis, 2, 277–285.

McCain, L. J., & McCleary, R. (1979). The statistical analysis of the simple inter- rupted time series quasi-experiment. In T. D. Cook & D. T. Campbell (Eds.),

Quasi-experimentation: Design and analysis issues for field settings. Chicago: Rand McNally.

McClannahan, L. E., & Krantz, P. J. (1999). Activity schedules for children with autism: Teaching independent behav- ior. Bethesda, MD: Woodbine House.

McClannahan, L. E., McGee, G. G., Mac- Duff, G. S., & Krantz, P. J. (1990). As- sessing and improving child care: A person appearance index for children with autism. Journal of Applied Be- havior Analysis, 23, 469–482.

McConnell, M. E. (1999). Self-monitoring, cueing, recording, and managing: Teaching students to manage their own behavior. Teaching Exceptional Chil- dren, 32 (2), 14–21.

McCord, B. E., Iwata, B. A., Galensky, T. L., Ellingson, S. A., & Thomson, R. J. (2001). Functional analysis and treat- ment of problems behavior evoked by noise. Journal of Applied Behavior Analysis, 34, 447–462.

McCullough, J. P., Cornell, J. E., McDaniel, M. H., & Mueller, R. K. (1974). Uti- lization of the simultaneous treatment design to improve student behavior in a first-grade classroom. Journal of Consulting and Clinical Psychology, 42, 288–292.

McEntee, J. E., & Saunders, R. R. (1997). A response-restriction analysis of stereotypy in adolescents with mental retardation: Implications for applied behavior analysis. Journal of Applied Behavior Analysis, 30, 485–506.

McEvoy, M. A., & Brady, M. P. (1988). Contingent access to play materials as an academic motivator for autistic and behavior disordered children. Edu- cation and Treatment of Children, 11, 5–18.

McFall, R. M. (1977). Parameters of self- monitoring. In R. B. Stuart (Ed.), Be- havioral self-management (pp. 196–214). New York: Bruner/Mazel.

McGee, G. G., Krantz, P. J., & McClan- nahan, L. E. (1985). The facilitative effects of incidental teaching on preposition use by autistic children. Journal of Applied Behavior Analy- sis, 18, 17–31.

McGee, G. G., Morrier, M., & Daly, T. (1999). An incidental teaching ap- proach to early intervention for tod- dlers with autism. Journal of the Association for Persons with Severe Handicaps, 24, 133–146.

Martens, B. K., Hiralall, A. S., & Bradley, T. A. (1997). A note to teacher: Im- proving student behavior through goal setting and feedback. School Psychol- ogy Quarterly, 12, 33–41.

710

Bibliography

McGill, P. (1999). Establishing operations: Implications for assessment, treatment, and prevention of problem behavior. Journal of Applied Behavior Analysis, 32, 393–418.

McGinnis, E. (1984). Teaching social skills to behaviorally disordered youth. In J. K. Grosenick, S. L. Huntze, E. McGinnis, & C. R. Smith (Eds.), Social/affective interventions in behav- iorally disordered youth (pp. 87–112). De Moines, IA: Department of Public Instruction.

McGonigle, J. J., Rojahn, J., Dixon, J., & Strain, P. S. (1987). Multiple treatment interference in the alternating treatments design as a function of the intercompo- nent interval length. Journal of Applied Behavior Analysis, 20, 171–178.

McGuffin, M. E., Martz, S. A., & Heron, T. E. (1997). The effects of self- correction versus traditional spelling on the spelling performance and mainte- nance of third grade students. Journal of Behavioral Education, 7, 463–476.

McGuire, M. T., Wing, R. R., Klem, M. L., & Hill, J. O. (1999). Behavioral strate- gies of individuals who have main- tained long-term weight losses. Obesity Research, 7, 334–341.

McIlvane, W. J., & Dube, W. V. (1992). Stimulus control shaping and stimulus control topographies. The Behavior An- alyst, 15, 89–94.

McIlvane, W. J., Dube, W. V., Green, G., & Serna, R. W. (1993). Programming conceptual and communication skill development: A methodological stim- ulus class analysis. In A. P. Kaiser & D. B. Gray (Eds.), Enhancing chil- dren’s communication (Vol. 2, pp. 243–285). Baltimore: Brookes.

McKerchar, P. M., & Thompson, R. H. (2004). A descriptive analysis of po- tential reinforcement contingencies in the preschool classroom. Journal of Applied Behavior Analysis, 21, 157.

McLaughlin, T., & Malaby, J. (1972). Re- ducing and measuring inappropriate verbalizations in a token classroom. Journal of Applied Behavior Analysis, 5, 329–333.

McNeish, J., Heron, T. E., & Okyere, B. (1992). Effects of self-correction on the spelling performance of junior high stu- dents with learning disabilities. Journal of Behavioral Education, 2, 17–27.

McPherson, A., Bonem, M., Green, G., & Osborne, J. G. (1984). A citation analy-

sis of the influence on research of Skin- ner’s Verbal behavior. The Behavior Analyst, 7, 157–167.

McWilliams, R., Nietupski, J., & Hamre- Nietupski, S. (1990). Teaching complex activities to students with moderate handicaps through the forward chain- ing of shorter total cycle response se- quences. Education and Training in Mental Retardation, 25 (3), 292–298.

Mechling, L. C., & Gast, D. L. (1997). Com- bination audio/visual self-prompting system for teaching chained tasks to students with intellectual disabilities. Education and Training in Mental Re- tardation and Developmental Disabil- ities, 32, 138–153.

Meichenbaum, D., & Goodman, J. (1971). The developmental control of operant motor responding by verbal operants. Journal of Experimental Child Psy- chology, 7, 553–565.

Mercatoris, M., & Craighead, W. E. (1974). Effects of nonparticipant observation on teacher and pupil classroom behav- ior. Journal of Educational Psychology, 66, 512–519.

Meyer, L. H., & Evans, I. M. (1989). Nonaversive intervention for behavior problems: A manual for home and com- munity. Baltimore: Paul H. Brookes.

Michael, J. (1974). Statistical inference for individual organism research: Mixed blessing or curse? Journal of Applied Behavior Analysis, 7, 647–653.

Michael, J. (1975). Positive and negative re- inforcement, a distinction that is no longer necessary; or a better way to talk about bad things. Behaviorism, 3, 33–44.

Michael, J. (1980). Flight from behavior analysis. The Behavior Analyst, 3, 1–22.

Michael, J. (1982). Distinguishing between discriminative and motivational func- tions of stimuli. Journal of the Experi- mental Analysis of Behavior, 37, 149–155.

Michael, J. (1982). Skinner’s elementary ver- bal relations: Some new categories. The Analysis of Verbal Behavior, 1, 1–4.

Michael, J. (1984). Verbal behavior. Journal of the Experimental Analysis of Behav- ior, 42, 363–376.

Michael, J. (1988). Establishing operations and the mand. The Analysis of Verbal Behavior, 6, 3–9.

Michael, J. (1991). Verbal behavior: Ob- jectives, exams, and exam answers. Kalamazoo, MI: Western Michigan University.

Michael, J. (1992). Introduction I. In Verbal behavior by B. F. Skinner (Reprinted edition). Cambridge, MA: B. F. Skinner Foundation.

Michael, J. (1993). Concepts and principles of behavior analysis. Kalamazoo, MI: Society for the Advancement of Be- havior Analysis.

Michael, J. (1993). Establishing operations. The Behavior Analyst, 16, 191–206.

Michael, J. (1995). What every student of behavior analysis ought to learn: A sys- tem for classifying the multiple effects of behavioral variables. The Behavior Analyst, 18, 273–284.

Michael, J. (2000). Implications and refine- ments of the establishing operation concept. Journal of Applied Behavior Analysis, 33, 401–410.

Michael, J. (2003). The multiple control of verbal behavior. Invited tutorial pre- sented at the 29th Annual Convention of the Association for Behavior Analy- sis, San Francisco, CA.

Michael, J. (2004). Concepts and principles of behavior analysis (rev. ed.) Kalama- zoo, MI: Society for the Advancement of Behavior Analysis.

Michael, J. (2006). Comment on Baron and Galizio. The Behavior Analyst, 29, 117–119.

Michael, J., & Shafer, E. (1995). State nota- tion for teaching about behavioral pro- cedures. The Behavior Analyst, 18, 123–140.

Michael, J., & Sundberg, M. L. (2003, May 23). Skinner’s analysis of verbal behav- ior: Beyond the elementary verbal oper- ants. Workshop conducted at the 29th Annual Convention of the Association for Behavior Analysis, San Francisco, CA.

Miguel, C. F., Carr, J. E., & Michael, J. (2002). Effects of stimulus-stimulus pairing procedure on the vocal behav- ior of children diagnosed with autism. The Analysis of Verbal Behavior, 18, 3–13.

Millenson, J. R. (1967). Principles of behav- ioral analysis. New York: Macmillan.

Miller, A. D., Hall, S. W., & Heward, W. L. (1995). Effects of sequential 1-minute time trials with and without inter-trial feedback and self-correction on general and special education students’ fluency with math facts. Journal of Behavioral Education, 5, 319–345.

Miller, D. L., & Kelley, M. L. (1994). The use of goal setting and contingency contracting for improving children’s

711

Bibliography

homework performance. Journal of Ap- plied Behavior Analysis, 27, 73–84.

Miller, D. L., & Stark, L. J. (1994). Contin- gency contracting for improving ad- herence in pediatric populations. Journal of the American Medical As- sociation, 271 (1), 81–83.

Miller, N., & Dollard, J. (1941). Social learning and imitation. New Haven, CT: Yale University Press.

Miller, N., & Neuringer, A. (2000). Rein- forcing variability in adolescents with autism. Journal of Applied Behavior Analysis, 33, 151–165.

Miltenberger, R. (2004). Behavior modifi- cation: Principles and procedures (3rd ed.). Belmont, CA: Wadsworth/Thom- son Learning.

Miltenberger, R. G. (2001). Behavior mod- ification: Principles and procedures (2nd ed.). Belmont, CA: Wadsworth/ Thomson Learning.

Miltenberger, R. G., & Fuqua, R. W. (1981). Overcorrection: A review and critical analysis. The Behavioral Ana- lyst, 4, 123–141.

Miltenberger, R. G., Flessner, C., Gath- eridge, B., Johnson, B., Satterlund, M., & Egemo, K. (2004). Evaluation of be- havior skills training to prevent gun play in children. Journal of Applied Be- havior Analysis, 37, 513–516.

Miltenberger, R. G., Fuqua, R. W., & Woods, D. W. (1998). Applying behavior analy- sis to clinical problems: Review and analysis of habit reversal. Journal of Ap- plied Behavior Analysis, 31, 447–469.

Miltenberger, R. G., Gatheridge, B., Satter- lund, M., Egemo-Helm, K. R., Johnson, B. M., Jostad, C., Kelso, P., & Flessner, C. A. (2005). Teaching safety skills to children to prevent gun play: An evalu- ation of in situ training. Journal of Ap- plied Behavior Analysis, 38, 395–398.

Miltenberger, R. G., Rapp, J., & Long, E. (1999). A low-tech method for conduct- ing real time recording. Journal of Ap- plied Behavior Analysis, 32, 119–120.

Mineka, S. (1975). Some new perspectives on conditioned hunger. Journal of Ex- perimental Psychology: Animal Be- havior Processes, 104, 143–148.

Mischel, H. N., Ebbesen, E. B., & Zeiss, A. R. (1972). Cognitive and attentional mechanisms in delay of gratification. Journal of Personality and Social Psy- chology, 16, 204–218.

Mischel, W., & Gilligan, C. (1964). Delay of gratification, motivation for the pro-

hibited gratification, and responses to temptation. Journal of Abnormal and Social Psychology, 69, 411–417.

Mitchell, R. J., Schuster, J. W., Collis, B. C., & Gassaway, L. J. (2000). Teaching vo- cational skills with a faded auditory prompting system. Education and Training in Mental Retardation and De- velopmental Disabilities, 35, 415–427.

Mitchem, K. J., & Young, K. R. (2001). Adapting self-management programs for classwide use: Acceptability, feasi- bility, and effectiveness. Remedial and Special Education, 22, 75–88.

Mitchem, K. J.,Young, K. R., West, R. P., & Benyo, J. (2001). CWPASM: A class- wide peer-assisted self-management program for general education class- rooms. Education and Treatment of Children, 24, 3–14.

Molè, P. (2003). Ockham’s razor cuts both ways: The uses and abuses of sim- plicity in scientific theories. Skeptic, 10 (1), 40–47.

Moore, J. (1980). On behaviorism and pri- vate events. Psychological Record, 30, 459–475.

Moore, J. (1984). On behaviorism, knowl- edge, and causal explanation. Psycho- logical Record, 34, 73–97.

Moore, J. (1985). Some historical and con- ceptual relations among logical posi- tivism, operationism, and behaviorism. The Behavior Analyst, 8, 53–63.

Moore, J. (1995). Radical behaviorism and the subjective-objective distinction. The Behavior Analyst, 18, 33–49.

Moore, J. (2000). Thinking about thinking and feeling about feeling. The Behavior Analyst, 23 (1), 45–56.

Moore, J. (2003). Behavior analysis, men- talism, and the path to social justice. The Behavior Analyst, 26, 181–193.

Moore, J. W., Mueller, M. M., Dubard, M., Roberts, D. S., & Sterling-Turner, H. E. (2002). The influence of therapist at- tention on self-injury during a tangible condition. Journal of Applied Behavior Analysis, 35, 283–286.

Moore, J., & Cooper, J. O. (2003). Some proposed relations among the domains of behavior analysis. The Behavior An- alyst, 26, 69–84.

Moore, J., & Shook, G. L. (2001). Certifi- cation, accreditation and quality con- trol in behavior analysis. The Behavior Analyst, 24, 45–55.

Moore, R., & Goldiamond, I. (1964). Error- less establishment of visual discrimi-

nation using fading procedures. Journal of the Experimental Analysis of Behav- ior, 7, 269–272.

Morales v. Turman, 364 F. Supp. 166 (E.D. Tx. 1973).

Morgan, D., Young, K. R., & Goldstein, S. (1983). Teaching behaviorally disor- dered students to increase teacher at- tention and praise in mainstreamed classrooms. Behavioral Disorders, 8, 265–273.

Morgan, Q. E. (1978). Comparison of two “Good Behavior Game” group contin- gencies on the spelling accuracy of fourth-grade students. Unpublished master’s thesis, The Ohio State Uni- versity, Columbus.

Morris, E. K. (1991). Deconstructing “tech- nological to a fault”. Journal of Applied Behavior Analysis, 24, 411–416.

Morris, E. K., & Smith, N. G. (2003). Bib- liographic processes and products, and a bibliography of the published primary- source works of B. F. Skinner. The Be- havior Analyst, 26, 41–67.

Morris, R. J. (1985). Behavior modifica- tion with exceptional children: Prin- ciples and practices. Glenview, IL: Scott, Foresman.

Morse, W. H., & Kelleher, R. T. (1977). De- terminants of reinforcement and pun- ishment. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 174–200). Upper Saddle River, NJ: Prentice Hall.

Morton, W. L., Heward, W. L., & Alber, S. R. (1998). When to self-correct? A com- parison of two procedures on spelling performance. Journal of Behavioral Education, 8, 321–335.

Mowrer, O. H. (1950). Learning theory and personality dynamics. New York: The Ronald Press Company.

Moxley, R. A. (1990). On the relationship between speech and writing with im- plications for behavioral approaches to teaching literacy. The Analysis of Ver- bal Behavior, 8, 127–140.

Moxley, R. A. (1998). Treatment-only designs and student self-recording strategies for public school teachers. Education and Treatment of Children, 21, 37–61.

Moxley, R. A. (2004). Pragmatic selection- ism: The philosophy of behavior analysis. The Behavior Analyst Today, 5, 108–125.

Moxley, R. A., Lutz, P. A., Ahlborn, P., Boley, N., & Armstrong, L. (1995). Self-recording word counts of freewrit-

712

Bibliography

ing in grades 1-4. Education and Treat- ment of Children, 18, 138–157.

Mudford, O. C. (1995). Review of the gen- tle teaching data. American Journal on Mental Retardation, 99, 345–355.

Mueller, M. M., Moore, J. W., Doggett, R. A., & Tingstrom, D. H. (2000). The effectiveness of contingency-specific prompts in controlling bathroom graf- fiti. Journal of Applied Behavior Analy- sis, 33, 89–92.

Mueller, M. M., Piazza, C. C., Moore, J. W., Kelley, M. E., Bethke, S. A., Pruett, A. E., Oberdorff, A. J., & Layer, S. A. (2003). Training parents to implement pediatric feeding proto- cols. Journal of Applied Behavior Analysis, 36, 545–562.

Mueller, M., Moore, J., Doggett, R. A., & Tingstrom, D. (2000). The effectiveness of contingency-specific and contin- gency nonspecific prompts in control- ling bathroom graffiti. Journal of Applied Behavior Analysis, 33, 89–92.

Murphy, E. S., McSweeny, F. K., Smith, R. G., & McComas, J. J. (2003). Dy- namic changes in reinforcer effective- ness: Theoretical, methodological, and practical implications for applied re- search. Journal of Applied Behavior Analysis, 36, 421–438.

Murphy, R. J., Ruprecht, M. J., Baggio, P., & Nunes, D. L. (1979). The use of mild punishment in combination with rein- forcement of alternate behaviors to re- duce the self-injurious behavior of a profoundly retarded individual. AAESPH Review, 4, 187–195.

Musser, E. H., Bray, M. A., Kehle, T. J., & Jenson, W. R. (2001). Reducing dis- ruptive behaviors in students with seri- ous emotional disturbance. School Psychology Review, 30 (2), 294–304.

Myer, J. S. (1971). Some effects of non- contingent aversive stimulation. In R. F. Brush (Ed.), Aversive condition- ing and learning (pp. 469–536). NY: Academic Press.

Myles, B. S., Moran, M. R., Ormsbee, C. K., & Downing, J. A. (1992). Guidelines for establishing and maintaining token economies. Intervention in School and Clinic, 27 (3), 164–169.

Nakano,Y. (2004). Toward the establishment of behavioral ethics: Ethical principles of behavior analysis in the era of em- pirically supported treatment (EST). Japanese Journal of Behavior Analy- sis, 19 (1), 18–51.

Narayan, J. S., Heward, W. L., Gardner R., III, Courson, F. H., & Omness, C. (1990). Using response cards to in- crease student participation in an ele- mentary classroom. Journal of Applied Behavior Analysis, 23, 483–490.

National Association of School Psycholo- gists. (2000). Professional conduct manual: Principles for professional ethics and guidelines for the provision of school psychological services. Bethesda, MD: NASP Publications.

National Association of Social Workers. (1996). The NASW code of ethics. Washington, DC: Author.

National Commission for the Protection of Human Subjects of Biomedical and Be- havioral Research. (1979). The Belmont Report: Ethical principles and guide- lines for the protection of human sub- jects of research. Washington, DC: Department of Health, Education, and Welfare. Retrieved November 11, 2003, from http://ohsr.od.nih.gov/mpa/bel mont.php3.

National Reading Panel (2000). Teaching chil- dren to read: An evidence-based assess- ment of the scientific research literature on reading and its implications for read- ing instruction: Reports of the subgroups. (NIH Pub No. 00-4754). Bethesda, MD: National Institute of Child Health and Human Development. [Available at: www.nichd.nih.gov.publications/nrp. report/htm]

National Reading Panel. www.nationalread ingpanel.org. Retrieved November 29, 2005, from www.nationalreadingpanel. org

Neef, N. A., Bicard, D. F., & Endo, S. (2001). Assessment of impulsivity and the development of self-control by stu- dents with attention deficit hyperac- tivity disorder. Journal of Applied Behavior Analysis, 34, 397–408.

Neef, N. A., Bicard, D. F., Endo, S., Coury, D. L., & Aman, M. G. (2005). Evalua- tion of pharmacological treatment of im- pulsivity by students with attention deficit hyperactivity disorder. Journal of Applied Behavior Analysis, 38, 135–146.

Neef, N. A., Iwata, B. A., & Page, T. J. (1980). The effects of interspersal training versus high density reinforcement on spelling acquisition and retention. Journal of Ap- plied Behavior Analysis, 13, 153–158.

Neef, N. A., Lensbower, J., Hockersmith, I., DePalma, V., & Gray, K. (1990). In vivo versus stimulation training: An in-

teractional analysis of range and type of training exemplars. Journal of Ap- plied Behavior Analysis, 23, 447–458.

Neef, N. A., Mace, F. C., & Shade, D. (1993). Impulsivity in students with se- rious emotional disturbance: The inter- active effects of reinforcer rate, delay, and quality. Journal of Applied Behav- ior Analysis, 26, 37–52.

Neef, N. A., Mace, F. C., Shea, M. C., & Shade, D. (1992). Effects of rein- forcer rate and reinforcer quality on time allocation: Extensions of match- ing theory to educational settings. Journal of Applied Behavior Analy- sis, 25, 691–699.

Neef, N. A., Markel, J., Ferreri, S., Bicard, D. F., Endo, S., Aman, M. G., Miller, K. M., Jung, S., Nist, L., & Armstrong, N. (2005). Effects of modeling versus instructions on sensitivity to reinforce- ment schedules. Journal of Applied Be- havior Analysis, 38, 23–37.

Neef, N. A., Parrish, J. M., Hannigan, K. F., Page, T. J., & Iwata, B. A. (1990). Teaching self-catheterization skills to children with neurogenic bladder com- plications. Journal of Applied Behav- ior Analysis, 22, 237–243.

Neisser, U. (1976). Cognition and reality. San Francisco: Freeman.

Nelson, R. O., & Hayes, S. C. (1981). The- oretical explanations for reactivity in self-monitoring. Behavior Modifica- tion, 5, 3–14.

Neuringer, A. (1993). Reinforced variation and selection. Animal Learning and Behavior, 21, 83–91.

Neuringer, A. (2004). Reinforced variabil- ity in animals and people: Implication for adaptive action. American Psychol- ogist, 59, 891–906.

Nevin, J. A. (1998). Choice and momentum. In W. O’Donohue (Ed.), Learning and behavior therapy (pp. 230–251). Boston: Allyn and Bacon.

Newman, B., Buffington, D. M., Hemmes, N. S., & Rosen, D. (1996). Answering objections to self-management and re- lated concepts. Behavior and Social Is- sues, 6, 85–95.

Newman, B., Reinecke, D. R., & Meinberg, D. (2000). Self-management of varied responding in children with autism. Behavioral Interventions, 15, 145–151.

Newman, R. S., & Golding, L. (1990). Chil- dren’s reluctance to seek help with school work. Journal of Educational Psychology, 82, 92–100.

713

Bibliography

Newstrom, J., McLaughlin, T. F., & Sweeney, W. J. (1999). The effects of contingency contracting to improve the mechanics of written language with a middle school student with behavior disorders. Child & Family Behavior Therapy, 21 (1), 39–48.

Newton, J. T., & Sturmey, P. (1991). The Motivation Assessment Scale: Inter- rater reliability and internal consistency in a British sample. Journal of Mental Deficiency Research, 35, 472–474.

Nihira, K., Leland, H., & Lambert, N. K. (1993). Adaptive Behavior Scale—Res- idential and Community (2nd ed.). Austin, TX: Pro-Ed.

Ninness, H. A. C., Fuerst, J., & Rutherford, R. D. (1991). Effects of self-manage- ment training and reinforcement on the transfer of improved conduct in the ab- sence of supervision. Journal of Ap- plied Behavior Analysis, 24, 499–508.

Noell, G. H., VanDerHeyden, A. M., Gatti, S. L., & Whitmarsh, E. L. (2001). Functional assessment of the effects of escape and attention on students’ com- pliance during instruction. School Psy- chology Quarterly, 16, 253–269.

Nolan, J. D. (1968). Self-control procedures in the modification of smoking behav- ior. Journal of Consulting and Clinical Psychology, 32, 92–93.

North, S. T., & Iwata, B. A. (2005). Moti- vational influences on performance maintained by food reinforcement. Journal of Applied Behavior Analysis, 38, 317–333.

Northup, J. (2000). Further evaluation of the accuracy of reinforcer surveys: A sys- tematic replication. Journal of Applied Behavior Analysis, 33, 335–338.

Northup, J., George, T., Jones, K., Brous- sard, C., & Vollmer, T. R. (1996). A comparison of reinforcer assessment methods: The utility of verbal and pictorial choice procedures. Journal of Applied Behavior Analysis, 29, 201–212.

Northup, J., Vollmer, T. R., & Serrett, K. (1993). Publication trends in 25 years of the Journal of Applied Behavior Analysis. Journal of Applied Behavior Analysis, 26, 527–537.

Northup, J., Wacker, D., Sasso, G., Steege, M., Cigrand, K., Cook, J., & DeRaad, A. (1991). A brief functional analysis of aggressive and alternative behavior in an outclinic setting. Journal of Ap- plied Behavior Analysis, 24, 509–522.

Novak, G. (1996). Developmental psychol- ogy: Dynamical systems and behavior analysis. Reno, NV: Context Press.

O’Brien, F. (1968). Sequential contrast ef- fects with human subjects. Journal of the Experimental Analysis of Behavior, 11, 537–542.

O’Brien, S., & Karsh, K. G. (1990). Treatment acceptability, consumer, therapist, and society. In A. C. Repp & N. N. Singh (Eds.), Perspectives on the use of non- aversive and aversive interventions for persons with developmental disabilities (pp. 503–516). Sycamore, IL: Sycamore.

O’Donnell, J. (2001). The discriminative stimulus for punishment or SDp. The Behavior Analyst, 24, 261–262.

O’Leary, K. D. (1977). Teaching self-man- agement skills to children. In D. Upper (Ed.), Perspectives in behavior therapy. Kalamazoo, MI: Behaviordelia.

O’Leary, K. D., & O’Leary, S. G. (Eds.). (1972). Classroom management: The successful use of behavior modifica- tion. New York: Pergamon.

O’Leary, K. D., Kaufman, K. F., Kass, R. E., & Drabman, R. S. (1970). The effects of loud and soft reprimands on the behavior of disruptive students. Exceptional Children, 37, 145–155.

O’Leary, S. G., & Dubey, D. R. (1979). Ap- plications of self-control procedures by children: A review. Journal of Applied Behavior Analysis, 12, 449–465.

O’Neill, R. E. Horner, R. H., Albin, R. W., Sprague, J. R., Storey, K., & Newton, J. S. (1997). Functional assessment for problem behavior: A practical hand- book (2nd ed.). Pacific Grove, CA: Brooks/Cole.

O’Reilly, M. F. (1995). Functional analysis and treatment of escape-maintained ag- gression correlated with sleep depriva- tion. Journal of Applied Behavior Analysis, 28, 225–226.

O’Reilly, M., Green, G., & Braunling- McMorrow, D. (1990). Self-administered written prompts to teach home accident prevention skills to adults with brain injuries. Journal of Applied Behavior Analysis, 23, 431–446.

O’Sullivan, J. L. (1999). Adult guardianship and alternatives. In R. D. Dinerstein, S. S. Herr, & J. L. O’Sullivan (Eds.), A guide to consent (pp. 7–37). Washing- ton DC: American Association on Men- tal Retardation.

Odom, S. L., Hoyson, M., Jamieson, B., & Strain, P. S. (1985). Increasing handi-

capped preschoolers’ peer social inter- actions: Cross-setting and component analysis. Journal of Applied Behavior Analysis, 18, 3–16.

Oliver, C. O., Oxener, G., Hearn, M., & Hall, S. (2001). Effects of social proximity on multiple aggressive behaviors. Journal of Applied Behavior Analysis, 34, 85–88.

Ollendick, T. H., Matson, J. L., Esvelt- Dawson, K., & Shapiro, E. S. (1980). Increasing spelling achievement: An analysis of treatment procedures utiliz- ing an alternating treatments design. Journal of Applied Behavior Analysis, 13, 645–654.

Ollendick, T., Matson, J., Esveldt-Dawson, K., & Shapiro, E. (1980). An initial investigation into the parameters of overcorrection. Psychological Reports, 39, 1139–1142.

Olympia, D. W., Sheridan, S. M., Jenson, W. R., & Andrews, D. (1994). Using student-managed interventions to increase homework completion and accuracy. Journal of Applied Behavior Analysis, 27, 85–99.

Ortiz, K. R., & Carr, J. E. (2000). Multiple- stimulus preference assessments: A comparison of free-operant and re- stricted-operant formats. Behavioral In- terventions, 15, 345–353.

Osborne, J. G. (1969). Free-time as a rein- forcer in the management of classroom behavior. Journal of Applied Behavior Analysis, 2, 113–118.

Osgood, C. E. (1953). Method and theory in experimental psychology. New York: Oxford University Press.

Osnes, P. G., Guevremont, D. C., & Stokes, T. F. (1984). If I say I’ll talk more, then I will: Correspondence training to in- crease peer-directed talk by socially withdrawn children. Behavior Modifi- cation, 10, 287–299.

Osnes, P. G., & Lieblein, T. (2003). An ex- plicit technology of generalization. The Behavior Analyst Today, 3, 364–374.

Overton, T. (2006). Assessing learners with special needs: An applied approach (5th ed.). Upper Saddle River, NJ: Prentice Hall.

Owens, R. E. (2001). Language develop- ment: An introduction (5th ed.). Boston: Allyn & Bacon.

Pace, G. M., & Troyer, E. A. (2000). The ef- fects of a vitamin supplement on the pica of a child with severe mental re- tardation. Journal of Applied Behavior Analysis, 33, 619–622.

714

Bibliography

Pace, G. M., Ivancic, M. T., Edwards, G. L., Iwata, B. A., & Page, T. J. (1985). As- sessment of stimulus preference and re- inforcer value with profoundly retarded individuals. Journal of Applied Behav- ior Analysis, 18, 249–255.

Paclawskyj, T. R., & Vollmer, T. R. (1995). Reinforcer assessment for children with developmental disabilities and visual impairments. Journal of Applied Be- havior Analysis, 28, 219–224.

Paclawskyj, T., Matson, J., Rush, K., Smalls, Y., & Vollmer, T. (2000). Questions about behavioral function (QABF): Behavioral checklist for functional assessment of aberrant behavior. Research in Develop- mental Disabilities, 21, 223–229.

Page, T. J., & Iwata, B. A. (1986). Interob- server agreement: History, theory, and current methods. In A. Poling & R. W. Fuqua (Eds.), Research methods in ap- plied behavior analysis (pp. 92–126). New York: Plenum Press.

Palmer, D. C. (1991). A behavioral inter- pretation of memory. In L. J. Hayes & P. N. Chase (Eds.), Dialogues on verbal behavior (pp. 261–279). Reno NV: Context Press.

Palmer, D. C. (1996). Achieving parity: The role of automatic reinforcement. Journal of the Experimental Analysis of Behavior, 65, 289–290.

Palmer, D. C. (1998). On Skinner’s rejection of S-R psychology. The Behavior Ana- lyst, 21, 93–96.

Palmer, D. C. (1998). The speaker as lis- tener: The interpretations of structural regularities in verbal behavior. The Analysis of Verbal Behavior, 15, 3–16.

Panyan, M., Boozer, H., & Morris, N. (1970). Feedback to attendants as a re- inforcer for applying operant tech- niques. Journal of Applied Behavior Analysis, 3, 1–4.

Parker, L. H., Cataldo, M. F., Bourland, G., Emurian, C. S., Corbin, R. J., & Page, J. M. (1984). Operant treatment of oro- facial dysfunction in neuromuscular disorders. Journal of Applied Behavior Analysis, 17, 413–427.

Parrott, L. J. (1984). Listening and under- standing. The Behavior Analyst, 7, 29–39.

Parsons, M. B., Reid, D. H., Reynolds, J., & Bumgarner, M. (1990). Effects of cho- sen versus assigned jobs on the work performance of persons with severe handicaps. Journal of Applied Behav- ior Analysis, 23, 253–258.

Parsonson, B. S. (2003). Visual analysis of graphs: Seeing is believing. In K. S. Budd & T. Stokes (Eds.), A small matter of proof: The legacy of Donald M. Baer (pp. 35–51). Reno, NV: Context Press.

Parsonson, B. S., & Baer, D. M. (1978). The analysis and presentation of graphic data. In T. R. Kratochwill (Ed.), Single subject research: Strategies for evalu- ating change (pp. 101–165). New York: Academic Press.

Parsonson, B. S., & Baer, D. M. (1986). The graphic analysis of data. In A. Poling & R. W. Fuqua (Eds.), Research meth- ods in applied behavior analysis (pp. 157–186). New York: Plenum Press.

Parsonson, B. S., & Baer, D. M. (1992). The visual analysis of graphic data, and cur- rent research into the stimuli control- ling it. In T. R. Kratochwill & J. R. Levin (Eds.), Single subject research design and analysis: New directions for psychology and education (pp. 15–40). New York: Academic Press.

Partington, J. W., & Bailey, J. S. (1993). Teaching intraverbal behavior to pre- school children. The Analysis of Verbal Behavior, 11, 9–18.

Partington, J. W., & Sundberg, M. L. (1998). The assessment of basic language and learning skills (The ABLLS). Pleasant Hill, CA: Behavior Analysts, Inc.

Patel, M. R., Piazza, C. C., Kelly, M. L., Ochsner, C. A., & Santana, C. M. (2001). Using a fading procedure to in- crease fluid consumption in a child with feeding problems. Journal of Ap- plied Behavior Analysis, 34, 357–360.

Patel, M. R., Piazza, C. C., Martinez, C. J., Volkert, V. M., & Santana, C. M. (2002). An evaluation of two differen- tial reinforcement procedures with es- cape extinction to treat food refusal. Journal of Applied Behavior Analysis, 35, 363–374.

Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia.

Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Vol. 4: A social interactional approach. Eugene, OR: Castalia.

Pavlov, I. P. (1927). Conditioned reflexes: An investigation of the physiological activity of the cerebral cortex (W. H. Grant, Trans.). London: Oxford Uni- versity Press.

Pavlov, I. P. (1927/1960). Conditioned re- flexes (G. V. Anrep, Trans.). New York: Dover.

Pelios, L., Morren, J., Tesch, D., & Axelrod, S. (1999). The impact of functional analysis methodology on treatment choice for self-injurious and aggressive behavior. Journal of Applied Behavior Analysis, 32, 185–195.

Pennypacker, H. S. (1981). On behavioral analysis. The Behavior Analyst, 3, 159–161.

Pennypacker, H. S. (1994). A selectionist view of the future of behavior analysis in education. In R. Gardner, III, D. M. Sainato, J. O. Cooper, T. E. Heron, W. L. Heward, J. Eshleman, & T. A. Grossi (Eds.), Behavior analysis in education: Focus on measurably superior instruc- tion (pp. 11–18). Monterey, CA: Brooks/Cole.

Pennypacker, H. S., & Hench, L. L. (1997). Making behavioral technology transfer- able. The Behavior Analyst, 20, 97–108.

Pennypacker, H. S., Gutierrez, A., & Linds- ley, O. R. (2003). Handbook of the Stan- dard Celeration Chart. Gainesville, FL: Xerographics.

Pennypacker, H. S., Koenig, C., & Linds- ley, O. (1972). Handbook of the Stan- dard Behavior Chart. Kansas City: Precision Media.

Peters, M., & Heron, T. E. (1993). When the best is not good enough: An examina- tion of best practice. Journal of Special Education, 26 (4), 371–385.

Peters, R., & Davies, K. (1981). Effects of self-instructional training on cognitive impulsivity of mentally retarded ado- lescents. American Journal of Mental Deficiency, 85, 377–382.

Peterson, I., Homer, A. L., & Wonderlich, S. A. (1982). The integrity of indepen- dent variables in behavior analysis. Journal of Applied Behavior Analysis, 15, 477–492.

Peterson, N. (1978). An introduction to ver- bal behavior. Grand Rapids, MI: Be- havior Associates, Inc.

Peterson, S. M., Neef, N. A., Van Norman, R., & Ferreri, S. J. (2005). Choice mak- ing in educational settings. In W. L. Heward, T. E. Heron, N. A. Neef, S. M. Peterson, D. M. Sainato, G. Cartledge, R. Gardner, III, L. D. Peterson, S. B.

Pelaez-Nogueras, M., Gewirtz, J. L., Field, T., Cigales, M., Malphurs, J., Clasky, S., & Sanchez, A. (1996). Infants’ preference for touch stimulation in face-to-face interactions. Journal of Applied Developmental Psychology, 17, 199–213.

715

Bibliography

Hersh, & J. C. Dardig (Eds.), Focus on behavior analysis in education: Achievements, challenges, and oppor- tunities (pp. 125–136). Upper Saddle River, NJ: Merrill/Prentice Hall.

Pfadt, A., & Wheeler, D. J. (1995). Using statistical process control to make data- based clinical decisions. Journal of Ap- plied Behavior Analysis, 28, 349–370.

Phillips, E. L., Phillips, E. A., Fixen, D. L., & Wolf, M. M. (1971). Achievement Place: Modification of the behaviors of predelinquent boys with a token econ- omy. Journal of Applied Behavior Analysis, 4, 45–59.

Piaget, J. (1952). The origins of intelligence in children. (M. Cook, Trans.). New York: International University Press.

Piazza, C. C., & Fisher, W. (1991). A faded bedtime with response cost protocol for treatment of multiple-sleep problems in children. Journal of Applied Behav- ior Analysis, 24, 129–140.

Piazza, C. C., Bowman, L. G., Contrucci, S. A., Delia, M. D., Adelinis, J. D., & Goh, H-L. (1999). An evaluation of the properties of attention as reinforcement for destructive and appropriate behav- ior. Journal of Applied Behavior Analy- sis, 32, 437–449.

Piazza, C. C., Fisher, W. W., Hagopian, L. P., Bowman, L. G., & Toole, L. (1996). Using a choice assessment to predict reinforcer effectiveness. Journal of Applied Behavior Analy- sis, 29, 1–9.

Piazza, C. C., Moses, D. R., & Fisher, W. W. (1996). Differential reinforcement of al- ternative behavior and demand fading in the treatment of escape maintained destructive behavior. Journal of Applied Behavior Analysis, 29, 569–572.

Piazza, C. C., Roane, H. S., Kenney, K. M., Boney, B. R., & Abt, K. A. (2002). Varying response effort in the treatment of pica maintained by automatic rein- forcement. Journal of Applied Behavior Analysis, 35, 233–246.

Pierce, K. L., & Schreibman, L. (1994). Teaching daily living skills to children with autism in unsupervised settings through pictorial self-management. Journal of Applied Behavior Analysis, 27, 471–481.

Pierce, W. D., & Epling, W. F. (1999). Behavior analysis and learning (2nd ed.). Upper Saddle River, NJ: Prentice Hall/Merrill.

Pindiprolu, S. S., Peterson, S. M. P., Rule, S., & Lignuaris/Kraft, B. (2003). Using web-mediated experiential case-based instruction to teach functional behav- ioral assessment skills. Teacher Edu- cation in Special Education, 26, 1–16.

Pinker, S. (1994). The language instinct. New York: Harper Perennial.

Pinkston, E. M., Reese, N. M., Leblanc, J. M., & Baer, D. M. (1973). Inde- pendent control of a preschool child’s aggression and peer interaction by contingent teacher attention. Journal of Applied Behavior Analysis, 6, 115–124.

Plazza, C. C., Bowman, L. G., Contrucci, S. A., Delia, M. D., Adelinis, J. D., & Goh, H.-L. (1999). An evaluation of the properties of attention as reinforcement for destructive and appropriate behav- ior. Journal of Applied Behavioral Analysis, 32, 437–449.

Poche, C., Brouwer, R., & Swearingen, M. (1981). Teaching self-protection to young children. Journal of Applied Be- havior Analysis, 14, 169–176.

Poling, A., & Normand, M. (1999). Noncon- tingent reinforcement: An inappropriate description of time-based schedules that reduce behavior. Journal of Applied Be- havior Analysis, 32, 237–238.

Poling, A., & Ryan, C. (1982). Differential- reinforcement-of-other-behavior sched- ules: Therapeutic applications. Behavior Modification, 6, 3–21.

Poling, A., Methot, L. L., & LeSage, M. G. (Eds.). (1995). Fundamentals of be- havior analytic research. New York: Plenum Press.

Poplin, J., & Skinner, C. (2003). Enhancing academic performance in a classroom serving students with serious emotional disturbance: Interdependent group con- tingencies with randomly selected components. School Psychology Re- view, 32 (2), 282–296.

Post, M., Storey, K., & Karabin, M. (2002). Cool headphones for effective prompts: Supporting students and adults in work and community environments. Teaching Exceptional Children, 34, 60–65.

Potts, L., Eshleman, J. W., & Cooper, J. O. (1993). Ogden R. Lindsley and the his- torical development of Precision Teach- ing. The Behavior Analyst, 16 (2), 177–189.

Poulson, C. L. (1983). Differential rein- forcement of other-than-vocalization

as a control procedure in the condi- tioning of infant vocalization rate. Journal of Experimental Child Psy- chology, 36, 471–489.

Powell, J., & Azrin, N. (1968). Behavioral engineering: Postural control by a portable operant apparatus. Journal of Applied Behavior Analysis, 1, 63–71.

Powell, J., Martindale, B., & Kulp, S. (1975). An evaluation of time-sample measures of behavior. Journal of Ap- plied Behavior Analysis, 8, 463–469.

Powell, J., Martindale, B., Kulp, S., Martin- dale, A., & Bauman, R. (1977). Taking a closer look: Time sampling and mea- surement error. Journal of Applied Be- havior Analysis, 10, 325–332.

Powell, T. H., & Powell, I. Q. (1982). The use and abuse of using the timeout procedure for disruptive pupils. The Pointer, 26, 18–22.

Powers, R. B., Osborne, J. G., & Anderson, E. G. (1973). Positive reinforcement of litter removal in the natural environ- ment. Journal of Applied Behavior Analysis, 6, 579–586.

Premack, D. (1959). Toward empirical be- havioral laws: I. Positive reinforcement. Psychological Review, 66, 219–233.

President’s Council on Physical Fitness. www.fitness.gov.

Progar, P. R., North, S. T., Bruce, S. S., Di- novi, B. J., Nau, P. A., Eberman, E. M., Bailey, J. R., Jr., & Nussbaum, C. N. (2001). Putative behavioral history ef- fects and aggression maintained by es- cape from therapists. Journal of Applied Behavioral Analysis, 34, 69–72.

Pryor, K. (1999). Don’t shoot the dog! The new art of teaching and training (rev. ed.). New York: Bantam Books.

Pryor, K. (2005). Clicker trained flight in- struction. Retrieved April 10, 2005, from http://clickertraining.com/training/ humans/job/flight_training.

Pryor, K., Haag, R., & O’Reilly, J. (1969). The creative porpoise: Training for novel behavior. Journal of the Experimental Analysis of Behavior, 12, 653–661.

Pryor, K., & Norris, K. S. (1991). Dolphin societies: Discoveries and puzzles. Berkeley: University of California Press.

Rachlin, H. (1970). The science of self- control. Cambridge: Harvard Univer- sity Press.

Rachlin, H. (1974). Self-control. Behavior- ism, 2, 94–107.

716

Bibliography

Rachlin, H. (1977). Introduction to modern behaviorism (2nd ed.). San Francisco: W. H. Freeman.

Rachlin, H. (1995). Self-control. Cambridge: Harvard University Press.

Rapp, J. T., Miltenberger, R. G., & Long, E. S. (1998). Augmenting simplified habit reversal with an awareness en- hancement device. Journal of Applied Behavior Analysis, 31, 665–668.

Rapp, J. T., Miltenberger, R. G., Galensky, T. L., Ellingson, S. A., & Long, E. S. (1999). A functional analysis of hair pulling. Journal of Applied Behavior Analysis, 32, 329–337.

Rasey, H. W., & Iversen, I. H. (1993). An experimental acquisition of maladap- tive behaviors by shaping. Journal of Behavior Therapy and Experimental Psychiatry, 24, 37–43.

Readdick, C. A., & Chapman, P. L. (2000). Young children’s perceptions of time out. Journal of Research in Childhood Education, 15 (1), 81–87.

Reese, E. P. (1966). The analysis of human operant behavior. Dubuque, IA: Brown.

Rehfeldt, R. A., & Chambers, M. C. (2003). Functional analysis and treatment of verbal perseverations displayed by an adult with autism. Journal of Applied Behavior Analysis, 36, 259–261.

Reich, W. T. (1988). Experiential ethics as a foundation for dialogue between health communications and health-care ethics. Journal of Applied Communi- cation Research, 16, 16–28.

Reid, D. H., Parsons, M. B., Green, C. W., & Browning, L. B. (2001). Increasing one aspect of self-determination among adults with severe multiple disabilities in supported work. Jou- rnal of Applied Behavioral Analysis, 34, 341–344.

Reid, D. H., Parsons, M. B., Phillips, J. F., & Green, C. W. (1993). Reduction of self- injurious hand mouthing using re- sponse blocking. Journal of Applied Behavior Analysis, 26, 139–140.

Reid, R., & Harris, K. R. (1993). Self- monitoring attention versus self- monitoring of performance: Effects on attention and academic performance. Exceptional Children, 60, 29–40.

Reimers, T. M., & Wacker, D. P. (1988). Par- ents’ ratings of the acceptability of be- havior treatment recommendations made in an outpatient clinic: A pre- liminary analysis of the influence of

treatment effectiveness. Behavioral Disorders, 14, 7–15.

Reitman, D., & Drabman, R. S. (1999). Multifaceted uses of a simple timeout record in the treatment of a noncom- pliant 8-year-old boy. Education and Treatment of Children, 22 (2), 136–145.

Reitman, D., & Gross, A. M. (1996). De- layed outcomes and rule-governed be- havior among “noncompliant” and “compliant” boys: A replication and ex- tension. The Analysis of Verbal Behav- ior, 13, 65–77.

Repp, A. C., & Horner, R. H. (Eds.). (1999). Functional analysis of prob- lem behavior: From effective assess- ment to effective support. Belmont, CA: Wadsworth.

Repp, A. C., & Karsh, K. G. (1994). Laptop computer system for data recording and contextual analyses. In T. Thompson & D. B. Gray (Eds.), Destructive behav- ior in developmental disabilities: Di- agnosis and treatment (pp. 83–101). Thousand Oaks, CA: Sage.

Repp, A. C., & Singh, N. N. (Eds.). (1990). Perspectives on the use of nonaversive and aversive interventions for persons with developmental disabilities. Syca- more, IL: Sycamore.

Repp, A. C., Barton, L. E., & Brulle, A. R. (1983). A comparison of two proce- dures for programming the differential reinforcement of other behaviors. Journal of Applied Behavior Analysis, 16, 435–445.

Repp, A. C., Dietz, D. E. D., Boles, S. M., Dietz, S. M., & Repp, C. F. (1976). Dif- ferences among common methods for calculating interobserver agreement. Journal of Applied Behavior Analysis, 9, 109–113.

Repp, A. C., Harman, M. L., Felce, D., Vanacker, R., & Karsh, K. L. (1989). Conducting behavioral assessments on computer collected data. Behavioral Assessment, 2, 249–268.

Repp, A. C., Karsh, K. G., Johnson, J. W., & VanLaarhoven, T. (1994). A compari- son of multiple versus single examples of the correct stimulus on task acquisi- tion and generalization by persons with developmental disabiliteis. Journal of Behavioral Education, 6, 213–230.

Rescorla, R. (1988). Pavlovian conditioning: It’s not what you think it is. American Psychologist, 43, 151–160.

Reynolds, G. S. (1961). Behavioral contrast. Journal of the Experimental Analysis of Behavior, 4, 57–71.

Reynolds, G. S. (1968). A primer of operant conditioning. Glenview, IL: Scott, Foresman.

Reynolds, G. S. (1975). A primer of operant conditioning (Rev. ed.). Glenview, IL: Scott, Foresman.

Reynolds, N. J., & Risley, T. R. (1968). The role of social and material reinforcers in increasing talking of a disadvantaged preschool child. Journal of Applied Be- havior Analysis, 1, 253–262.

Rhode, G., Morgan, D. P., & Young, K. R. (1983). Generalization and mainte- nance of treatment gains of behav- iorally handicapped students from resource rooms to regular classrooms using self-evaluation procedures. Journal of Applied Behavior Analysis, 16, 171–188.

Richman, D. M., Berg, W. K., Wacker, D. P., Stephens, T., Rankin, B., & Kilroy, J. (1997). Using pretreatment assess- ments to enhance and evaluate existing treatment packages. Journal of Applied Behavior Analysis, 30, 709–712.

Richman, D. M., Wacker, D. P., Asmus, J. M., Casey, S. D., & Andelman, M. (1999). Further analysis of problem behavior in response class hierarchies. Journal of Behavior Analysis, 32, 269–283.

Ricketts, R. W., Goza, A. B., & Matese, M. (1993). A 4-year follow-up of treatment of self-injury. Journal of Behavior Therapy and Experimental Psychiatry, 24 (1), 57–62.

Rincover, A. (1978). Sensory extinction: A procedure for eliminating self-stimula- tory behavior in psychotic children. Journal of Abnormal Child Psychology, 6, 299–310.

Rincover, A. (1981). How to use sensory ex- tinction. Austin, TX: Pro-Ed.

Rincover, A., & Koegel, R. L. (1975). Set- ting generality and stimulus control in autistic children. Journal of Applied Behavior Analysis, 8, 235–246.

Rincover, A., & Newsom, C. D. (1985). The relative motivational properties of sen- sory reinforcement with psychotic chil- dren. Journal of Experimental Child Psychology, 24, 312–323.

Rincover, A., Cook, R., Peoples, A., & Packard, D. (1979). Sensory extinction and sensory reinforcement principles for programming multiple adaptive

717

Bibliography

behavior change. Journal of Applied Behavior Analysis, 12, 221–233.

Rindfuss, J. B., Al-Attrash, M., Morrison, H., & Heward, W. L. (1998, May). Using guided notes and response cards to im- prove quiz and exam scores in an eighth grade American history class. Paper presented at 24th Annual Convention of the Association for Behavior Analysis, Orlando, FL.

Ringdahl, J. E., Kitsukawa, K., Andelman, M. S., Call, N., Winborn, L. C., Bar- retto, A., & Reed, G. K. (2002). Differ- ential reinforcement with and without instructional fading. Journal of Applied Behavior Analysis, 35, 291–294.

Ringdahl, J. E., Vollmer, T. R., Borrero, J. C., & Connell, J. E. (2001). Fixed- time schedule effects as a function of baseline reinforcement rate. Journal of Applied Behavior Analysis, 34, 1–15.

Ringdahl, J. E., Vollmer, T. R., Marcus, B. A., & Roane, H. S (1997). An ana- logue evaluation of environmental enrichment: The role of stimulus pref- erence. Journal of Applied Behavior Analysis, 30, 203–216.

Riordan, M. M., Iwata, B. A., Finney, J. W., Wohl, M. K., & Stanley, A. E. (1984). Behavioral assessment and treatment of chronic food refusal in handicapped children. Journal of Applied Behavior Analysis, 17, 327–341.

Risley, T. (1996). Get a life! In L. Kern Koegel, R. L. Koegel, & G. Dunlap (Eds.), Positive behavioral support (pp. 425–437). Baltimore: Paul H. Brookes.

Risley, T. (2005). Montrose M. Wolf (1935–2004). Journal of Applied Be- havior Analysis, 38, 279–287.

Risley, T. R. (1968). The effects and side ef- fects of punishing the autistic behav- iors of a deviant child. Journal of Applied Behavior Analysis, 1, 21–34.

Risley, T. R. (1969, April). Behavior modi- fication: An experimental-therapeutic endeavor. Paper presented at the Banff International Conference on Behavior Modification, Banff, Alberta, Canada.

Risley, T. R. (1997). Montrose M. Wolf: The origin of the dimensions of applied be- havior analysis. Journal of Applied Be- havior Analysis, 30, 377–381.

Risley, T. R. (2005). Montrose M. Wolf (1935–2004). Journal of Applied Be- havior Analysis, 38, 279–287.

Risley, T. R., & Hart, B. (1968). Developing correspondence between the non-verbal

and verbal behavior of preschool chil- dren. Journal of Applied Behavior Analysis, 1, 267–281.

Roane, H. S., Fisher, W. W., & McDonough, E. M. (2003). Progressing from pro- grammatic to discovery research: A case example with the overjustification effect. Journal of Applied Behavior Analysis, 36, 23–36.

Roane, H. S., Kelly, M. L., & Fisher, W. W. (2003). The effects of noncontingent access to food on the rate of object mouthing across three settings. Journal of Applied Behavior Analysis, 36, 579–582.

Roane, H. S., Lerman, D. C., & Vorndran, C. M. (2001). Assessing reinforcers under progressive schedule require- ments. Journal of Applied Behavior Analysis, 34, 145–167.

Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Evaluation of a brief stimulus preference assessment. Journal of Applied Behavior Analysis, 31, 605–620.

Roberts-Pennell, D., & Sigafoos, J. (1999). Teaching young children with devel- opmental disabilities to request more play using the behavior chain inter- ruption strategy. Journal of Applied Research in Intellectual Disabilities, 12, 100–112.

Robin, A. L., Armel, S., & O’Leary, K. D., (1975). The effects of self-instruction on writing deficiencies. Behavior Ther- apy, 6, 178–187.

Robin, A., Schneider, M., & Dolnick, M., (1976). The turtle technique: An ex- tended case study of self-control in the classroom. Psychology in the Schools, 13, 449–453.

Robinson, P. W., Newby, T. J., & Gansell, S. L. (1981). A token system for a class of underachieving hyperactive children. Journal of Applied Behavior Analysis, 14, 307–315.

Rodgers, T. A., & Iwata, B. A. (1991). An analysis of error-correction procedures during discrimination training. Journal of Applied Behavior Analysis, 24, 775–781.

Rohn, D. (2002). Case study: Improving guitar skills. In R. W. Malott & H. Har- rison, I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using behavior analysis to get your act together (p. 8-4). Kalamazoo, MI: De-

partment of Psychology, Western Michigan University.

Rolider, A., & Van Houten, R. (1984). The effects of DRO alone and DRO plus reprimands on the undesirable behav- ior of three children in home settings. Education and Treatment of Children, 7, 17–31.

Rolider, A., & Van Houten, R. (1984). Train- ing parents to use extinction to elimi- nate nighttime crying by gradually increasing the criteria for ignoring cry- ing. Education and Treatment of Chil- dren, 7, 119–124.

Rolider, A., & Van Houten, R. (1985). Sup- pressing tantrum behavior in public places through the use of delayed pun- ishment mediated by audio recordings. Behavior Therapy, 16, 181–194.

Rolider, A. , & Van Houten, R. (1993). The interpersonal treatment model. In R. Van Houten & S. Axelrod (Eds.), Behavior analysis and treatment (pp. 127–168). New York: Plenum Press.

Romanczyk, R. G. (1977). Intermittent pun- ishment of self-stimulation: Effective- ness during application and extinction. Journal of Consulting and Clinical Psychology, 45, 53–60.

Romaniuk, C., Miltenberger, R., Conyers, C., Jenner, N., Jurgens, M., & Ringen- berg, C. (2002). The influence of ac- tivity choice on problem behaviors maintained by escape versus attention. Journal of Applied Behavioral Analy- sis, 35, 349–362.

Romaniuk, C., Miltenberger, R., Conyers, C., Jenner, N., Roscoe, E. M., Iwata, B. A., & Goh, H.-L. (1998). A com- parison of noncontingent reinforcement and sensory extinction as treatments for self-injurious behavior. Journal of Ap- plied Behavior Analysis, 31, 635–646.

Romeo, F. F. (1998). The negative effects of using a group contingency system of classroom management. Journal of Instructional Psychology, 25 (2), 130–133.

Romer, L. T., Cullinan, T., & Schoenberg, B. (1994). General case training of re- questing: A demonstration and analy- sis. Education and Training in Mental Retardation, 29, 57–68.

Rosales-Ruiz, J., & Baer, D. M. (1997). Be- havioral cusps: A developmental and pragmatic concept for behavior analy- sis. Journal of Applied Behavior Analy- sis, 30, 533–544.

718

Bibliography

Roscoe, E. M., Iwata, B. A., & Goh, H.-L. (1998). A comparison of noncontingent reinforcement and sensory extinction as treatments for self-injurious behav- ior. Journal of Applied Behavior Analy- sis, 31, 635–646.

Roscoe, E. M., Iwata, B. A., & Kahng, S. (1999). Relative versus absolute rein- forcement effects: Implications for preference assessments. Journal of Ap- plied Behavior Analysis, 32, 479–493.

Rose, J. C., De Souza, D. G., & Hanna, E. S. (1996). Teaching reading and spelling: Exclusion and stimulus equivalence. Journal of Applied Behavior Analysis, 29, 451–469.

Rose, T. L. (1978). The functional relation- ship between artificial food colors and hyperactivity. Journal of Applied Be- havior Analysis, 11, 439–446.

Ross, C., & Neuringer, A. (2002). Reinforce- ment of variations and repetitions along three independent response dimensions. Behavioral Processes, 57, 199–209.

Rozensky, R. H. (1974). The effect of tim- ing of self-monitoring behavior on re- ducing cigarette consumption. Journal of Consulting and Clinical Psychology, 5, 301–307.

Rusch, F. R., & Kazdin, A. E. (1981). To- ward a methodology of withdrawal de- signs for the assessment of response maintenance. Journal of Applied Be- havior Analysis, 14, 131–140.

Russell, B., & Whitehead A. N. (1910–1913). Principia mathematica. Cambridge, MA: University Press.

Ruth, W. J. (1996). Goal setting and behav- ioral contracting for students with emo- tional and behavioral difficulties: Analysis of daily, weekly, and total goal attainment. Psychology in the Schools, 33, 153–158.

Ryan, C. S., & Hemmes, N. S. (2005). Ef- fects of the contingency for homework submission on homework submission and quiz performance in a college course. Journal of Applied Behavior Analysis, 38, 79–88.

Ryan, S., Ormond, T., Imwold, C., & Ro- tunda, R. J. (2002). The effects of a public address system on the off-task behavior of elementary physical edu- cation students. Journal of Applied Be- havior Analysis, 35, 305–308.

Sagan, C. (1996). The demon-haunted world: Science as a candle in the dark. New York: Ballantine.

Saigh, P. A., & Umar, A. M. (1983). The ef- fects of a good behavior game on the disruptive behavior of Sudanese ele- mentary school students. Journal of Ap- plied Behavior Analysis, 16, 339–344.

Sainato, D. M., Strain, P. S., Lefebvre, D., & Rapp, N. (1990). Effects of self-eval- uation on the independent work skills of preschool children with disabilities. Exceptional Children, 56, 540–549.

Sajwaj, T., Culver, P., Hall, C., & Lehr, L. (1972). Three simple punishment tech- niques for the control of classroom dis- ruptions. In G. Semb (Ed.), Behavior analysis and education. Lawrence: University of Kansas.

Saksida, L. M., Raymond, S. M., & Touret- zky, D.S. (1997). Shaping robot behav- ior using principles from instrumental conditioning. Robotics and Autonomous Systems, 22, 231–249.

Salend, S. J. (1984b). Integrity of treatment in special education research. Mental Retardation, 22, 309–315.

Salend, S. J., Ellis, L. L., & Reynolds, C. J. (1989). Using self-instruction to teach vocational skills to individuals who are severely retarded. Education and Training of the Mentally Retarded, 24, 248–254.

Salvy, S.-J., Mulick, J. A., Butter, E., Bartlett, R. K., & Linscheid, T. R. (2004). Con- tingent electric shock (SIBIS) and a conditioned punisher eliminate severe head banging in a preschool child. Be- havioral Interventions, 19, 59–72.

Salzinger, K. (1978). Language behavior. In A. C. Catania & T. A. Brigham (Eds.), Handbook of applied behavior analy- sis: Social and instructional processes (pp. 275–321). New York: Irvington.

Santogrossi, D. A., O’Leary, K. D., Ro- manczyk, R. G., & Kaufman, K. F. (1973). Self-evaluation by adolescents in a psychiatric hospital school token program. Journal of Applied Behavior Analysis, 6, 277–287.

Sarakoff, R. A., & Strumey, P. (2004). The effects of behavioral skills training on staff implementation of discrete-trial teaching. Journal of Applied Behavior Analysis, 37, 535–538.

Saraokoff, R. A., Taylor, B. A., & Poulson, C. L. (2001). Teaching children with autism to engage in conversational ex- changes: Script fading with embedded textual stimuli. Journal of Applied Be- havior Analysis, 34, 81–84.

Sasso, G. M., Reimers, T. M., Cooper, L. J., Wacker, D., Berg, W., Steege, M., Kelly, L., & Allaire, A. (1992). Use of descriptive and experimental analysis to identify the functional properties of aberrant behavior in school settings. Journal of Applied Behavior Analysis, 25, 809–821.

Saudargas, R. A., & Bunn, R. D. (1989). A hand-held computer system for class- room observation. Journal of Special Education, 9, 200–206.

Saudargas, R. A., & Zanolli, K. (1990). Mo- mentary time sampling as an estimate of percentage time: A field validation. Journal of Applied Behavior Analysis, 23, 533–537.

Saunders, M. D., Saunders, J. L., & Saun- ders, R. R. (1994). Data collection with bar code technology. In T. Thompson & D. B. Gray (Eds.), Destructive be- havior in developmental disabilities: Diagnosis and treatment (pp. 102–116). Thousand Oaks, CA: Sage.

Savage, T. (1998). Shaping: The link be- tween rats and robots. Connection Sci- ence, 10 (3/4), 321–340.

Savage-Rumbaugh, E. S. (1984). Verbal be- havior at the procedural level in the chimpanzee. Journal of the Experimen- tal Analysis of Behavior, 41, 223–250.

Saville, B. K., Beal, S. A., & Buskist, W. (2002). Essential readings for graduate students in behavior analysis: A survey of the JEAB and JABA Boards of Edi- tors. The Behavior Analyst, 25, 29–35.

Schepis, M. M., Reid, D. H., Behrmann, M. M., & Sutton, K. A. (1998). In- creasing communicative interactions of young children with autism using a voice output communication aid and naturalistic teaching. Journal of Ap- plied Behavior Analysis, 31, 561–578.

Scheuermann, B., & Webber, J. (1996). Level systems: Problems and solutions. Beyond Behavior, 7, 12–17.

Schleien, S. J., Wehman, P., & Kiernan, J. (1981). Teaching leisure skills to severely handicapped adults: An age-appropriate darts game. Journal of Applied Behav- ior Analysis, 14, 513–519.

Schlinger, H., & Blakely, E. (1987). Function-altering effects of contingency- specifying stimuli. The Behavior Ana- lyst, 10, 41–45.

Schoenberger, T. (1990). Understanding and the listener: Conflicting views. The Analysis of Verbal Behavior, 8, 141–150.

719

Bibliography

Schoenberger, T. (1991). Verbal under- standing: Integrating the conceptual analyses of Skinner, Ryle, and Wittgen- stein. The Analysis of Verbal Behavior, 9, 145–151.

Schoenfeld, W. N. (1995). “Reinforcement” in behavior theory. The Behavior Ana- lyst, 18, 173–185.

Schumm, J. S., Vaughn, D., Haager, D., Mc- Dowell, J., Rothlein, L., & Saumell, L. (1995). General education teacher plan- ning: What can students with learning disabilities expect? Exceptional Chil- dren, 61, 335–352.

Schuster, J. W., Griffen, A. K., & Wolery, M. (1992). Comparison of simultaneous prompting and constant time delay pro- cedures in teaching sight words to ele- mentary students with moderate mental retardation. Journal of Behavioral Ed- ucation, 7, 305–325.

Schwartz, B. (1974). On going back to na- ture: A review of Seligman and Hager’s Biological Boundaries of Learning. Journal of the Experimental Analysis of Behavior, 21, 183–198.

Schwartz, I. S., & Baer, D. M. (1991). Social validity assessments: Is current prac- tice state of the art? Journal of Applied Behavior Analysis, 24, 189–204.

Schwarz, M. L., & Hawkins, R. P. (1970). Application of delayed reinforcement procedures to the behavior of an ele- mentary school child. Journal of Ap- plied Behavior Analysis, 3, 85–96.

Schweitzer, J. B., & Sulzer-Azaroff, B. (1988). Self-control: Teaching toler- ance for delay in impulsive children. Journal of the Experimental Analysis of Behavior, 50, 173–186.

Scott, D., Scott, L. M., & Goldwater, B. (1997). A performance improvement program for an international-level track and field athlete. Journal of Applied Be- havior Analysis, 30, 573–575.

Seymour, F. W., & Stokes, T. F. (1976). Self- recording in training girls to increase work rate and evoke staff praise in an institution for offenders. Journal of Ap- plied Behavior Analysis, 9, 41–54.

Seymour, M. A. (2002). Case study: A re- tired athlete runs down comeback road. In R. W. Malott & H. Harrison, I’ll stop procrastinating when I get around to it: Plus other cool ways to succeed in school and life using be- havior analysis to get your act to-

gether (p. 7-12). Kalamazoo, MI: De- partment of Psychology, Western Michigan University.

Shahan, T. A., & Chase, P. N. (2002). Novelty, stimulus control, and operant variabil- ity. The Behavior Analyst, 25, 175–190.

Shermer, S. (1997). Why people believe weird things. New York: W. H. Freeman.

Shimamune, S., & Jitsumori, M. (1999). Ef- fects of grammar instruction and flu- ency training on the learning of the and a by native speakers of Japanese. The Analysis of Verbal Behavior, 16, 3–16.

Shimmel, S. (1977). Anger and its control in Greco-Roman and modern psychol- ogy. Psychiatry, 42, 320–327.

Shimmel, S. (1979). Free will, guilt, and self-control in rabbinic Judaism and contemporary psychology. Judaism, 26, 418–429.

Shimoff, E., & Catania, A. C. (1995). Using computers to teach behavior analysis. The Behavior Analyst, 18, 307–316.

Shirley, M. J., Iwata, B. A., Kahng, S. W., Mazaleski, J. L., & Lerman, D. C. (1997). Does functional communica- tion training compete with ongoing contingencies of reinforcement? An analysis during response acquisition and maintenance. Journal of Applied Behavior Analysis, 30, 93–104.

Shook, G. L. (1993). The professional cre- dential in behavior analysis. The Be- havior Analyst, 16, 87–101.

Shook, G. L., & Favell, J. E. (1996). Identi- fying qualified professionals in behav- ior analysis. In C. Maurice, G. Green, & S. C. Luce (Eds.), Behavioral inter- vention for young children with autism: A manual for parents and profession- als (pp. 221–229). Austin, TX: Pro-Ed.

Shook, G. L., & Neisworth, J. (2005). En- suring appropriate qualifications for ap- plied behavior analyst professionals: The Behavior Analyst Certification Board. Exceptionality 13 (1), 3–10.

Shook, G. L., Johnston, J. M., & Mel- lichamp, F. (2004). Determining es- sential content for applied behavior analyst practitioners. The Behavior An- alyst, 27, 67–94.

Shook, G. L., Rosales, S. A., & Glenn, S. (2002). Certification and training of be- havior analyst professionals. Behavior Modification, 26 (1), 27–48.

Shore, B. A., Iwata, B. A., DeLeon, I. G., Kahng, S., & Smith, R. G. (1997). An

analysis of reinforcer substitutability using object manipulation and self- injury as competing responses. Jour- nal of Applied Behavior Analysis, 30, 439–449.

Sideridis, G. D., & Greenwood, C. R. (1996). Evaluating treatment effects in single-subject behavioral experiments using quality-control charts. Journal of Behavioral Education, 6, 203–211.

Sidman, M. (1960). Tactics of scientific re- search. New York: Basic Books.

Sidman, M. (1960/1988). Tactics of scien- tific research: Evaluating experimental data in psychology. New York: Basic Books/Boston: Authors Cooperative (reprinted).

Sidman, M. (1971). Reading and auditory- visual equivalences. Journal of Speech and Hearing Research, 14, 5–13.

Sidman, M. (1994). Equivalence relations and behavior: A research story. Boston: Author’s Cooperative.

Sidman, M. (2000). Applied behavior analy- sis: Back to basics. Behaviorology, 5 (1), 15–37.

Sidman, M. (2002). Notes from the beginning of time. The Behavior Analyst, 25, 3–13.

Sidman, M. (2006). The distinction between positive and negative reinforcement: Some additional considerations. The Behavior Analyst, 29, 135–139.

Sidman, M., & Cresson, O., Jr. (1973). Reading and crossmodal transfer of stimulus equivalences in severe retar- dation. American Journal of Mental Deficiency, 77, 515–523.

Sidman, M., & Stoddard, L. T. (1967). The effectiveness of fading in programming a simultaneous form discrimination for retarded children. Journal of the Exper- imental Analysis of Behavior, 10, 3–15.

Sidman, M., & Tailby, W. (1982). Condi- tional discrimination vs. matching-to- sample: An expansion of the testing paradigm. Journal of the Experimental Analysis of Behavior, 37, 5–22.

Sigafoos, J., Doss, S., & Reichle, J. (1989). Developing mand and tact repertoires with persons with severe developmen- tal disabilities with graphic symbols. Research in Developmental Disabili- ties, 11, 165–176.

Sigafoos, J., Kerr, M., & Roberts, D. (1994). Interrater reliability of the Mo- tivation Assessment Scale: Failure to replicate with aggressive behavior.

720

Bibliography

Research in Developmental Disabili- ties, 15, 333–342.

Silvestri, S. M. (2004). The effects of self- scoring on teachers’positive statements during classroom instruction. Unpub- lished doctoral dissertation. Columbus, OH: The Ohio State University.

Silvestri, S. M. (2005). How to make a graph using Microsoft Excel. Unpublished manuscript. Columbus, OH: The Ohio State University.

Simek, T. C., O’Brien, R. M., & Figlerski, L. B. (1994). Contracting and chaining to improve the performance of a col- lege golf team: Improvement and dete- rioration. Perceptual and Motor Skills, 78 (3), 1099.

Simpson, M. J. A., & Simpson, A. E. (1977). One-zero and scan method for sampling behavior. Animal Behavior, 25, 726–731.

Singer, G., Singer, J., & Horner, R. (1987). Using pretask requests to increase the probability of compliance for students with severe disabilities. Journal of the Association for Persons with Severe Handicaps, 12, 287–291.

Singh, J., & Singh, N. N. (1985). Compari- son of word-supply and word-analysis error-correction procedures on oral reading by mentally retarded children. American Journal of Mental Defi- ciency, 90, 64–70.

Singh, N. N. (1990). Effects of two error- correction procedures on oral read- ing errors. Behavior Modification, 11, 165–181.

Singh, N. N., & Katz, R. C. (1985). On the modification of acceptability ratings for alternative child treatments. Behavior Modification, 9, 375–386.

Singh, N. N., & Singh, J. (1984). Antecedent control of oral reading errors and self- corrections by mentally retarded chil- dren. Journal of Applied Behavior Analysis, 17, 111–119.

Singh, N. N., & Singh, J. (1986). Increasing oral reading proficiency: A comparative analyst off drill and positive practice overcorrection procedures. Behavior Modification, 10, 115–130.

Singh, N. N., & Winton, A. S. (1985). Con- trolling pica by components of an over- correction procedure. American Journal of Mental Deficiency, 90, 40–45.

Singh, N. N., Dawson, M. J., & Manning, P. (1981). Effects of spaced responding

DRI on the stereotyped behavior of pro- foundly retarded persons. Journal of Ap- plied Behavior Analysis, 14, 521–526.

Singh, N. N., Dawson, M. J., & Manning, P. (1981). The effects of physical re- straint on self-injurious behavior. Journal of Mental Deficiency Re- search, 25, 207–216.

Singh, N. N., Singh, J., & Winton, A. S. (1984). Positive practice overcorrection of oral reading errors. Behavior Modi- fication, 8, 23–37.

Skiba, R., & Raison, J. (1990). Relationship between the use of timeout and acade- mic achievement. Exceptional Chil- dren, 57 (1), 36–46.

Skinner, B. F. (1938). The behavior of or- ganisms. New York: Appleton-Century- Crofts.

Skinner, B. F. (1938/1966). The behavior of organisms: An experimental analy- sis. New York: Appleton-Century. (Copyright renewed in 1966 by the B. F. Skinner Foundation, Cambridge, MA.)

Skinner, B. F. (1948). Walden two. New York: Macmillan.

Skinner, B. F. (1948). Superstition in the pi- geon. Journal of Experimental Psy- chology, 38, 168–172.

Skinner, B. F. (1953). Science and human behavior. New York: MacMillan.

Skinner, B. F. (1956). A case history in sci- entific method. American Psychologist, 11, 221–233.

Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.

Skinner, B. F. (1966). Operant behavior. In W. K. Honig (Ed.), Operant behavior: Areas of research and application (pp. 12–32). New York: Appleton-Century- Crofts.

Skinner, B. F. (1967). B. F. Skinner: An au- tobiography. In E. G. Boring & G. Lindzey (Eds.), A history of psychol- ogy in autobiography (Vol. 5, pp. 387–413). New York: Irvington.

Skinner, B. F. (1969). Contingencies of reinforcement: A theoretical analysis. New York: Appleton-Century-Crofts.

Skinner, B. F. (1971). Beyond freedom and dignity. New York: Knopf.

Skinner, B. F. (1974). About behaviorism. New York: Knopf.

Skinner, B. F. (1976). Particulars of my life. Washington Square, NY: New York University Press.

Skinner, B. F. (1978). Reflections on behav- iorism and society. Upper Saddle River, NJ: Prentice Hall.

Skinner, B. F. (1979). The shaping of a be- haviorist. Washington Square, NY: New York University Press.

Skinner, B. F. (1981a). Selection by conse- quences. Science, 213, 501–504.

Skinner, B. F. (1981b). How to discover what you have to say—A talk to stu- dents. The Behavior Analyst, 4, 1–7.

Skinner, B. F. (1982). Contrived reinforce- ment. The Behavior Analyst, 5, 3–8.

Skinner, B. F. (1983a). A matter of conse- quences. Washington Square, NY: New York University Press.

Skinner, B. F. (1983b). Intellectual self- management in old age. American Psy- chologist, 38, 239–244.

Skinner, B. F. (1989). Recent issues in the analysis of behavior. Columbus, OH: Merrill.

Skinner, B. F., & Vaughan, M. E. (1983). Enjoy old age: A program of self- management. New York: Norton.

Skinner, C. H., Cashwell, T. H., & Skinner, A. L (2000). Increasing tootling: The effects of a peer-mediated group con- tingency program on students’ reports of peers’prosocial behavior. Psychology in the Schools, 37 (3), 263–270.

Skinner, C. H., Fletcher, P. A., Wildmon, M., & Belfiore, P. J. (1996). Improving as- signment preference through inter- spersing additional problems: Brief versus easy problems. Journal of Be- havioral Education, 6, 427–436.

Skinner, C. H., Skinner, C. F., Skinner, A. L., & Cashwell, T. H. (1999). Using interdependent contingencies with groups of students: Why the principal kissed a pig. Educational Administration Quarterly, 35 (Suppl.), 806–820.

Smith, B. W., & Sugai, G. (2000). A self- management functional assessment- based behavior support plan for a middle school student with EBD. Journal of Positive Behavior Inter- ventions, 2, 208–217.

Smith, D. H. (1987). Telling stories as a way of doing ethics. Journal of the Florida Medical Association, 74, 581–588.

Smith, D. H. (1993). Stories, values, and pa- tient care decisions. In C. Conrad (Ed.), Ethical nexus (pp. 123–148). Norwood, NJ: Ablex.

721

Bibliography

Smith, L. D. (1992). On prediction and con- trol: B. F. Skinner and the technologi- cal ideal of science. American Psychologist, 47, 216–223.

Smith, R. G., & Iwata, B. A. (1997). An- tecedent influences of behavior disor- ders. Journal of Applied Behavior Analysis, 30, 343–375.

Smith, R. G., Iwata, B. A., & Shore, B. A. (1995). Effects of subject-versus ex- perimenter-selected reinforcers on the behavior of individuals with profound developmental disabilities. Journal of Applied Behavior Analysis, 28, 61–71.

Smith, R. G., Iwata, B. A., Goh, H., & Shore, B. A. (1995). Analysis of establishing operations for self-injury maintained by escape. Journal of Applied Behav- ior Analysis, 28, 515–535.

Smith, R. G., Russo, L., & Le, D. D. (1999). Distinguishing between extinction and punishment effects of response block- ing: A replication. Journal of Applied Behavior Analysis, 32, 367–370.

Smith, R., Michael, J., & Sundberg, M. L. (1996). Automatic reinforcement and automatic punishment in infant vocal behavior. The Analysis of Verbal Be- havior, 13, 39–48.

Smith, S., & Farrell, D. (1993). Level system use in special education: Classroom in- tervention with prima facie appeal. Behavioral Disorders, 18 (4), 251–264.

Snell, M. E., & Brown, F. (2000). Instruction of students with severe disabilities (5th ed.). Upper Saddle River, NJ: Merrill Prentice Hall.

Snell, M. E., & Brown, F. (2006). Instruc- tion of students with severe disabilities (6th ed.). Upper Saddle River, NJ: Prentice Hall.

Solanto, M. V., Jacobson, M. S., Heller, L., Golden, N. H., & Hertz, S. (1994). Rate of weight gain of inpatients with anorexia nervosa under two behavioral contracts. Pediatrics, 93 (6), 989.

Solomon, R. L. (1964). Punishment. Amer- ican Psychologist, 19, 239–253.

Spies, R. A., & Plake, B. S. (Eds.). (2005). Sixteenth mental measurements year- book. Lincoln, NE: Buros Institute of Mental Measurements.

Spooner, F., Spooner, D., & Ulicny, G. R. (1986). Comparisons of modified backward chaining: Backward chain- ing with leaps ahead and reverse chain- ing with leaps ahead. Education and Treatment of Children, 9 (2), 122–134.

Spradlin, J. E. (1966). Environmental fac- tors and the language development of retarded children. In S. Rosenberg (Ed.), Developments in applied psy- cholinguist research (pp. 261–290). Riverside, NJ: MacMillian.

Spradlin, J. E. (1996). Comments on Ler- man and Iwata (1996). Journal of Ap- plied Behavior Analysis, 29, 383–385.

Spradlin, J. E. (2002). Punishment: A pri- mary response. Journal of Applied Be- havior Analysis, 35, 475–477.

Spradlin, J. E., Cotter, V. W., & Baxley, N. (1973). Establishing a conditional dis- crimination without direct training: A study of transfer with retarded adoles- cents. American Journal of Mental De- ficiency, 77, 556–566.

Sprague, J. R., & Horner, R. H. (1984). The effects of single instance, multiple in- stance, and general case training on generalized vending machine used by moderately and severely handicapped students. Journal of Applied Behavior Analysis, 17, 273–278.

Sprague, J. R., & Horner, R. H. (1990). Easy does it: Preventing challenging behaviors. Teaching Exceptional Chil- dren, 23, 13–15.

Sprague, J. R., & Horner, R. H. (1991). De- termining the acceptability of behavior support plans. In M. Wang, H. Walberg, & M. Reynolds (Eds.), Handbook of special education (pp. 125–142). Ox- ford, London: Pergamon Press.

Sprague, J. R., & Horner, R. H. (1992). Co- variation within functional response classes: Implications for treatment of severe problem behavior. Journal of Applied Behavior Analysis, 25, 735–745.

Sprague, J., & Walker, H. (2000). Early iden- tification and intervention for youth with antisocial and violent behavior. Exceptional Children, 66, 367–379.

Spreat, S., & Connelly, L. (1996). Reliabil- ity analysis of the Motivation Assess- ment Scale. American Journal on Mental Retardation, 100, 528–532.

Staats, A. W., & Staats, C. K. (1963). Complex human behavior: A system- atic extension of learning principles. New York: Holt, Rinehart and Winston.

Stack, L. Z., & Milan, M. A. (1993). Im- proving dietary practices of elderly in- dividuals: The power of prompting, feedback, and social reinforcement. Journal of Applied Behavior Analysis, 26, 379–387.

Staddon, J. E. R. (1977). Schedule-induced behavior. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 125–152). Upper Saddle River, NJ: Prentice Hall.

Stage, S. A., & Quiroz, D. R. (1997). A meta-analysis of interventions to de- crease disruptive classroom behavior in public education settings. School Psy- chology Review, 26, 333–368.

Starin, S., Hemingway, M., & Hartsfield, F. (1993). Credentialing behavior analysts and the Florida behavior analysis certi- fication program. The Behavior Ana- lyst, 16, 153–166.

Steege, M. W., Wacker, D. P., Cigrand, K. C., Berg, W. K., Novak, C. G., Reimers, T. M., Sasso, G. M., & DeRaad, A. (1990). Use of negative reinforcement in the treatment of self-injurious be- havior. Journal of Applied Behavior Analysis, 23, 459–467.

Stephens, K. R., & Hutchison, W. R. (1992). Behavioral personal digital as- sistants: The seventh generation of computing. The Analysis of Verbal Be- havior, 10, 149–156.

Steuart, W. (1993). Effectiveness of arousal and arousal plus overcorrection to reduce nocturnal bruxism. Journal of Behavior Therapy & Experimental Psychiatry, 24, 181–185.

Stevenson, H. C., & Fantuzzo, J. W. (1984). Application of the “generalization map” to a self-control intervention with school-aged children. Journal of Ap- plied Behavior Analysis, 17, 203–212.

Stewart, C. A., & Singh, N. N. (1986). Over- correction of spelling deficits in men- tally retarded persons. Behavior Modification, 10, 355–365.

Stitzer, M. L., Bigelow, G. E., Liebson, I. A., & Hawthorne, J. W. (1982). Contingent reinforcement for benzodiazepine-free urines: Evaluation of a drug abuse treat- ment intervention. Journal of Applied Behavior Analysis, 15, 493–503.

Stokes, T. (2003). A genealogy of applied behavior analysis. In K. S. Budd & T. Stokes (Eds.), A small matter of proof: The legacy of Donald M. Baer (pp. 257–272). Reno, NV: Context Press.

Stokes, T. F., & Baer, D. M. (1976). Preschool peers as mutual generaliza- tion-facilitating agents. Behavior Ther- apy, 7, 599–610.

Stokes, T. F., & Baer, D. M. (1977). An im- plicit technology of generalization.

722

Bibliography

Journal of Applied Behavior Analysis, 10, 349–367.

Stokes, T. F., & Osnes, P. G. (1982). Pro- gramming the generalization of chil- dren’s social behavior. In P. S. Strain, M. J. Guralnick, & H. M. Walker (Eds.), Children’s social behavior: De- velopment, assessment, and modifi- cation (pp. 407–443) Orlando, FL: Academic Press.

Stokes, T. F., & Osnes, P. G. (1989). An op- erant pursuit of generalization. Behavior Therapy, 20, 337–355.

Stokes, T. F., Baer, D. M., & Jackson, R. L. (1974). Programming the generaliza- tion of a greeting response in four re- tarded children. Journal of Applied Behavior Analysis, 7, 599–610.

Stokes, T. F., Fowler, S. A., & Baer, D. M. (1978). Training preschool children to recruit natural communities of rein- forcement. Journal of Applied Behavior Analysis, 11, 285–303.

Stolz, S. B. (1978). Ethical issues in be- havior modification. San Francisco: Jossey-Bass.

Strain, P. S., & Joseph, G. E. (2004). A not so good job with “Good job.” Journal of Positive Behavior Interventions, 6 (1), 55–59.

Strain, P. S., McConnell, S. R., Carta, J. J., Fowler, S. A., Neisworth, J. T., & Wol- ery, M. (1992). Behaviorism in early intervention. Topics in Early Childhood Speciàl Education, 12, 121–142.

Strain, P. S., Shores, R. E., & Kerr, M. M. (1976). An experimental analysis of “spillover” effects on the social inter- action of behaviorally handicapped preschool children. Journal of Applied Behavior Analysis, 9, 31–40.

Striefel, S. (1974). Behavior modification: Teaching a child to imitate. Austin, TX: Pro-Ed.

Stromer, R. (2000). Integrating basic and ap- plied research and the utility of Lattal and Perone’s Handbook of Research Methods in Human Operant Behavior. Journal of Applied Behavior Analysis, 33, 119–136.

Stromer, R., McComas, J. J., & Rehfeldt, R. A. (2000). Designing interventions that include delayed reinforcement: Im- plications of recent laboratory research. Journal of Applied Behavior Analysis, 33, 359–371.

Sugai, G. M., & Tindal, G. A. (1993). Effective school consultation: An in-

teractive approach. Pacific Grove, CA: Brooks/Cole.

Sulzer-Azaroff, B., & Mayer, G. R. (1977). Applying behavior-analysis procedures with children and youth. New York: Holt, Rinehart & Winston.

Sundberg, M. L. (1983). Language. In J. L. Matson, & S. E. Breuning (Eds.), Assessing the mentally retarded (pp. 285–310). New York: Grune & Stratton.

Sundberg, M. L. (1991). 301 research top- ics from Skinner’s book Verbal be- havior. The Analysis of Verbal Behavior, 9, 81–96.

Sundberg, M. L. (1993). The application of establishing operations. The Behavior Analyst, 16, 211–214.

Sundberg, M. L. (1998). Realizing the po- tential of Skinner’s analysis of verbal behavior. The Analysis of Verbal Be- havior, 15, 143–147.

Sundberg, M. L. (2004). A behavioral analy- sis of motivation and its relation to mand training. In L. W. Williams (Ed.), Developmental disabilities: Etiology, assessment, intervention, and integra- tion. Reno, NV: Context Press.

Sundberg, M. L., & Michael, J. (2001). The value of Skinner’s analysis of verbal be- havior for teaching children with autism. Behavior Modification, 25, 698–724.

Sundberg, M. L., & Partington, J. W. (1998). Teaching language to children with autism or other developmental disabil- ities. Pleasant Hill, CA: Behavior An- alysts, Inc.

Sundberg, M. L., Endicott, K., & Eigenheer, P. (2000). Using intraverbal prompts to establish tacts for children with autism. The Analysis of Verbal Behav- ior, 17, 89–104.

Sundberg, M. L., Loeb, M., Hale, L., & Eigenheer, P. (2002). Contriving estab- lishing operations to teach mands for information. The Analysis of Verbal Be- havior, 18, 14–28.

Sundberg, M. L., Michael, J., Partington, J. W., & Sundberg, C. A. (1996). The role of automatic reinforcement in early language acquisition. The Analysis of Verbal Behavior, 13, 21–37.

Sundberg, M. L., San Juan, B., Dawdy, M., & Arguelles, M. (1990). The acquisition of tacts, mands, and intraverbals by indi- viduals with traumatic brain injury. The Analysis of Verbal Behavior, 8, 83–99.

Surratt, P.R., Ulrich, R.E., & Hawkins, R. P. (1969). An elementary student as a be-

havioral engineer. Journal of Applied Behavior Analysis. 85–92.

Sutherland, K. S., Wehby, J. H., & Yoder, P. J. (2002). An examination of the rela- tion between teacher praise and students with emotional/behavioral disorders’op- portunities to respond to academic re- quests. Journal of Emotional and Behavioral Disorders, 10, 5–13.

Swanson, H. L., & Sachse-Lee, C. (2000). A meta-analysis of single-subject- design intervention research for stu- dents with LD. Journal of Learning Disabilities, 38, 114–136.

Sweeney, W. J., Salva, E., Cooper, J. O., & Talbert-Johnson, C. (1993). Using self-evaluation to improve difficult-to- read handwriting of secondary stu- dents. Journal of Behavioral Education, 3, 427–443.

Symons, F. J., Hoch, J., Dahl, N. A., & Mc- Comas, J. J. (2003). Sequential and matching analyses of self-injurious be- havior: A case of overmatching in the natural environment. Journal of Ap- plied Behavior Analysis, 36, 267–270.

Symons, F. J., McDonald, L. M., & Wehby, J. H. (1998). Functional assessment and teacher collected data. Education and Treatment of Children, 21 (2), 135–159.

Szempruch, J., & Jacobson, J. W. (1993). Evaluating the facilitated communica- tions of people with developmental dis- abilities. Research in Developmental Disabilities, 14, 253–264.

Tang, J., Kennedy, C. H., Koppekin, A., & Caruso, M. (2002). Functional analy- sis of stereotypical ear covering in a child with autism. Journal of Applied Behavior Analysis, 35, 95–98.

Tapp, J. T., & Walden, T. A. (2000). A sys- tem for collecting and analysis of ob- servational data from videotape. In T. Thompson, D. Felce, & F. J. Symons (Eds.), Behavioral observation: Tech- nology and applications in develop- mental disabilities (pp. 61–70). Baltimore: Paul H. Brookes.

Tapp, J. T., & Wehby, J. H. (2000). Obser- vational software for laptop computers and optical bar code readers. In T. Thompson, D. Felce, & F. J. Symons (Eds.), Behavioral observation: Tech- nology and applications in develop- mental disabilities (pp. 71–81). Baltimore: Paul H. Brookes.

Tapp, J. T., Wehby, J. H., & Ellis, D. M. (1995). A multiple option observation

723

Bibliography

system for experimental studies: MOOSES. Behavior Research Methods Instruments & Computers, 27, 25–31.

Tarbox, R. S. F., Wallace, M. D., & Williams, L. (2003). Assessment and treatment of elopement: A replication and extension. Journal of Applied Be- havior Analysis, 36, 239–244.

Tarbox, R. S. F., Williams, W. L., & Friman, P. C. (2004). Extended diaper wearing: Effects on continence in and out of the diaper. Journal of Applied Behavior Analysis, 37, 101–105.

Tawney, J., & Gast, D. (1984). Single subject research in special education. Colum- bus, OH: Charles E. Merrill.

Taylor, L. K., & Alber, S. R. (2003). The ef- fects of classwide peer tutoring on the spelling achievement of first graders with disabilities. The Behavior Analyst Today, 4, 181–189.

Taylor, R. L. (2006). Assessment of excep- tional students: Educational and psy- chological procedures (7th ed.). Boston: Pearson/Allyn and Bacon.

Terrace, H. S. (1963a). Discrimination learn- ing with and without “errors.” Journal of the Experimental Analysis of Behav- ior, 6, 1–27.

Terrace, H. S. (1963b). Errorless transfer of a discrimination across two continua. Journal of the Experimental Analysis of Behavior, 6, 223–232.

Terris, W., & Barnes, M. (1969). Learned resistance to punishment and subse- quent responsiveness to the same and novel punishers. Psychonomic Sci- ence, 15, 49–50.

Test, D. W., & Heward, W. L. (1983). Teach- ing road signs and traffic laws to learn- ing disabled students. Science Education, 64, 129–139.

Test, D. W., & Heward, W. L. (1984). Accu- racy of momentary time sampling: A comparison of fixed- and variable- interval observation schedules. In W. L. Heward, T. E. Heron, D. S. Hill, and J. Trap-Porter (Eds.), Focus on behavior analysis in education (pp. 177–194). Columbus, OH: Charles E. Merrill.

Test, D. W., Spooner, F., Keul, P. K., & Grossi, T. (1990). Teaching adoles- cents with severe disability to use the public telephone. Behavior Modifica- tion, 14, 157–171.

Thompson, R. H., & Iwata, B. A. (2000). Response acquisition under direct and indirect contingencies of reinforce-

ment. Journal of Applied Behavior Analysis, 33, 1–11.

Thompson, R. H., & Iwata, B. A. (2001). A descriptive analysis of social conse- quences following problem behavior. Journal of Applied Behavior Analysis, 34, 169–178.

Thompson, R. H., & Iwata, B. A. (2003). A review of reinforcement control proce- dures. Journal of Applied Behavior Analysis, 38, 257–278.

Thompson, R. H., & Iwata, B. A. (2005). A review of reinforcement control proce- dures. Journal of Applied Behavior Analysis, 38, 257–278.

Thompson, R. H., Fisher, W. W., Piazza, C. C., & Kuhn D. E. (1998). The eval- uation and treatment of aggression maintained by attention and automatic reinforcement. Journal of Applied Be- havior Analysis, 31, 103–116.

Thompson, R. H., Iwata, B. A., Conners, J., & Roscoe, E. M. (1999). Effects of re- inforcement for alternative behavior during punishment of self-injury. Journal of Applied Behavior Analysis, 32, 317–328.

Thompson, T. J., Braam, S. J., & Fuqua, R. W. (1982). Training and generaliza- tion of laundry skills: A multiple probe evaluation with handicapped persons. Journal of Applied Behavior Analysis, 15, 177–182.

Thompson, T., Symons, F. J., & Felce, D. (2000). Principles of behavioral obser- vation. In T. Thompson, F. J. Symons, & D. Felce (Eds.), Behavioral obser- vation: Technology and applications in developmental disabilities (pp. 3–16). Baltimore: Paul H. Brookes.

Thomson, C., Holmber, M., & Baer, D. M. (1974). A brief report on a comparison of time sampling procedures. Journal of Applied Behavior Analysis, 7, 623–626.

Thoresen, C. E., & Mahoney, M. J. (1974). Behavioral self-control. New York: Holt, Rinehart & Winston.

Timberlake, W., & Allison, J. (1974). Re- sponse deprivation: An empirical ap- proach to instrumental performance. Psychological Review, 81, 146–164.

Tincani, M. (2004). Comparing the Picture Exchange Communication System and sign language training for children with autism. Focus on Autism and Other De- velopmental Disabilities, 19, 152–163.

Todd, A. W., Horner, R. W., & Sugai, G. (1999). Self-monitoring and self-

recruited praise: Effects on problem behavior, academic engagement, and work completion in a typical class- room. Journal of Positive Behavior In- terventions, 1, 66–76, 122.

Todd, J. T., & Morris, E. K. (1983). Mis- conception and miseducation: Presen- tations of radical behaviorism in psychology textbooks. The Behavior Analyst, 6, 153–160.

Todd, J. T., & Morris, E. K. (1992). Case histories in the great power of steady misrepresentation. American Psychol- ogist, 47, 1441–1453.

Todd, J. T., & Morris, E. K. (1993). Change and be ready to change again. American Psychologist, 48, 1158–1159.

Todd, J. T., & Morris, E. K. (Eds.). (1994). Modern perspectives on John B. Wat- son and classical behaviorism. West- port, CT: Greenwood Press.

Touchette, P. E., MacDonald, R. F., & Langer, S. N. (1985). A scatter plot for identifying stimulus control of problem behavior. Journal of Applied Behavior Analysis, 18, 343–351.

Trammel, D. L., Schloss, P. J., & Alper, S. (1994). Using self-recording, evalua- tion, and graphing to increase comple- tion of homework assignments. Journal of Learning Disabilities, 27, 75–81.

Trap, J. J., Milner-Davis, P., Joseph, S., & Cooper, J. O. (1978). The effects of feed- back and consequences on transitional cursive letter formation. Journal of Ap- plied Behavior Analysis, 11, 381–393.

Trask-Tyler, S. A., Grossi, T. A., & Heward, W. L. (1994). Teaching young adults with developmental disabilities and vi- sual impairments to use tape-recorded recipes: Acquisition, generalization, and maintenance of cooking skills. Journal of Behavioral Education, 4, 283–311.

Tucker, D. J., & Berry, G. W. (1980). Teaching severely multihandicapped students to put on their own hearing aids. Journal of Applied Behavior Analysis, 13, 65–75.

Tufte, E. R. (1983). The visual display of quantitative information. Chesire, CT: Graphics Press.

Tufte, E. R. (1990). Envisioning informa- tion. Chesire, CT: Graphics Press.

Turkewitz, H., O’Leary, K. D., & Ironsmith, M. (1975). Generalization and mainte- nance of appropriate behavior through self-control. Journal of Consulting and Clinical Psychology, 43, 577–583.

724

Bibliography

Turnbull, H. R., III., & Turnbull, A. P. (1998). Free appropriate public edu- cation: The law and children with dis- abilities. Denver: Love.

Tustin, R. D. (1994). Preference for re- inforcers under varying schedule arrangements: A behavioral economic analysis. Journal of Applied Behavior Analysis, 28, 61–71.

Twohig, M. P., & Woods, D. W. (2001). Evaluating the duration of the compet- ing response in habit reversal: A para- metric analysis. Journal of Applied Behavior Analysis, 34, 517–520.

Twyman, J., Johnson, H., Buie, J., & Nel- son, C. M. (1994). The use of a warn- ing procedure to signal a more intrusive timeout contingency. Behavioral Dis- orders, 19 (4), 243–253.

U. S. Department of Education. (2003). Proven methods: Questions and an- swers on No Child Left Behind. Wash- ington, DC: Author. Retrieved October 24, 2005, from www.ed.gov/nclb/ methods/whatworks/doing.html.

Ulman, J. D., & Sulzer-Azaroff, B. (1975). Multi-element baseline design in edu- cational research. In E. Ramp & G. Semb (Eds.), Behavior analysis: Areas of research and application (pp. 371–391). Upper Saddle River, NJ: Prentice Hall.

Ulrich, R. E., & Azrin, N. H. (1962). Reflexive fighting in response to aver- sive stimulation. Journal of the Ex- perimental Analysis of Behavior, 5, 511–520.

Ulrich, R. E., Stachnik, T. & Mabry, J. (Eds.). (1974). Control of human be- havior (Vol. 3), Behavior modification in education. Glenview, IL: Scott, Foresman and Company.

Ulrich, R. E., Wolff, P. C., & Azrin, N. H. (1962). Shock as an elicitor of intra- and inter-species fighting behavior. Animal Behavior, 12, 14–15.

Umbreit, J., Lane, K., & Dejud, C. (2004). Improving classroom behavior by mod- ifying task difficulty: Effects of in- creasing the difficulty of too-easy tasks. Journal of Positive Behavioral Inter- ventions, 6, 13–20.

Valk, J. E. (2003). The effects of embedded instruction within the context of a small group on the acquisition of imitation skills of young children with disabili- ties. Unpublished doctoral dissertation, The Ohio State University.

Van Acker, R., Grant, S. H., & Henry, D. (1996). Teacher and student behavior as a function of risk for aggression. Education and Treatment of Children, 19, 316–334.

Van Camp, C. M., Lerman, D. C., Kelley, M. E., Contrucci, S. A., & Vorndran, C. M. (2000). Variable-time reinforce- ment schedules in the treatment of so- cially maintained problem behavior. Journal of Applied Behavior Analysis, 33, 545–557.

van den Pol, R. A., Iwata, B. A., Ivancic, M. T., Page, T. J., Neef, N. A., & Whit- ley, F. P. (1981). Teaching the handi- capped to eat in public places: Acquisition, generalization and main- tenance of restaurant skills. Journal of Applied Behavior Analysis, 14, 61–69.

Van Houten, R. (1979). Social validation: The evolution of standards of competency for target behaviors. Journal of Applied Behavior Analysis, 12, 581–591.

Van Houten, R. (1993). Use of wrist weights to reduce self-injury maintained by sensory reinforcement. Journal of Ap- plied Behavior Analysis, 26, 197–203.

Van Houten, R. (1994). The right to effec- tive behavioral treatment. In L. J. Hayes, G. J. Hayes, S. C. Moore, & P. M. Gjezzi (Eds.), Ethical issues in developmental disabilities (pp. 103–118). Reno, NV: Context Press.

Van Houten, R., Axelrod, S., Bailey, J. S., Favell, J. E., Foxx, R. M., Iwata, B. A., & Lovaas, O. I. (1988). The right to ef- fective behavioral treatment. The Be- havior Analyst, 11, 111–114.

Van Houten, R., & Doleys, D. M. (1983). Are social reprimands effective? In S. Axelrod & J. Apsche (Eds.). The effects of punishment on human behavior (pp. 45–70). New York: Academic Press.

Van Houten, R., & Malenfant, J. E. L. (2004). Effects of a driver enforcement program on yielding to pedestrians. Journal of Applied Behavior Analysis, 37, 351–363.

Van Houten, R., Malenfant, J. E., Austin, J., & Lebbon, A. (2005). The effects of a seatbelt-gearshift delay prompt on the seatbelt use of motorists who do not regularly wear seatbelts. Journal of Applied Behavior Analysis, 38, 195–203.

Van Houten, R., Malenfant, L., & Rolider, A. (1985). Increasing driver yielding and pedestrian signaling with prompting,

feedback, and enforcement. Journal of Applied Behavior Analysis, 18, 103–110.

Van Houten, R., & Nau, P. A. (1981). A com- parison of the effects of posted feed- back and increased police surveillance on highway speeding. Journal of Ap- plied Behavior Analysis, 14, 261–271.

Van Houten, R., & Nau, P. A. (1983). Feed- back interventions and driving speed: A parametric and comparative analy- sis. Journal of Applied Behavior Analy- sis, 17, 253–281.

Van Houten, R., Nau, P. A., Mackenzie- Keating, S. E., Sameoto, D., & Col- avecchia, B. (1982). An analysis of some variables influencing the effec- tiveness of reprimands. Journal of Ap- plied Behavior Analysis, 15, 65–83.

Van Houten, R., Nau, P. A., & Marini, Z. (1980). An analysis of public posting in reducing speeding behavior on an urban highway. Journal of Applied Be- havior Analysis, 13, 383–395.

Van Houten, R., & Retting, R. A. (2001). In- creasing motorist compliance and cau- tion at stop signs. Journal of Applied Behavior Analysis, 434, 185–193.

Van Houten, R., & Retting, R. A. (2005). In- creasing motorist compliance and cau- tion at stop signs. Journal of Applied Behavior Analysis, 34, 185–193.

Van Houten, R., & Rolider, A. (1988). Re- creating the scene: An effective way to provide delayed punishment for in- appropriate motor behavior. Journal of Applied Behavior Analysis, 21, 187–192.

Van Houten, R., & Rolider, A. (1990). The role of reinforcement in reducing inap- propriate behavior: Some myths and misconceptions. In A. C. Repp & N. N. Singh (Eds.), Perspectives on the use of nonaversive and aversive interven- tions for persons with developmental disabilities (pp. 119–127). Sycamore, IL: Sycamore.

Van Norman, R. K. (2005). The effects of functional communication training, choice making, and an adjusting work schedule on problem behavior main- tained by negative reinforcement. Un- published doctoral dissertation, The Ohio State University, Columbus.

Vargas, E. A. (1986). Intraverbal behavior. In P. N. Chase & L. J. Parrott (Eds.), Psychological aspects of language (pp. 128–151). Springfield, IL: Charles C. Thomas.

725

Bibliography

Vargas, J. S. (1978). A behavioral approach to the teaching of composition. The Be- havior Analyst, 1, 16–24.

Vargas, J. S. (1984). What are your exercises teaching? An analysis of stimulus con- trol in instructional materials. In W. L. Heward, T. E. Heron, D. S. Hill, & J. Trap-Porter (Eds.), Focus on behavior analysis in education (pp. 126–141). Columbus, OH: Merrill.

Vargas, J. S. (1990). B. F. Skinner–The last few days. Applied Behavior Analysis, 23, 409–410.

Vaughan, M. (1989). Rule-governed behav- ior in behavior analysis: A theoretical and experimental history. In S. C. Hayes (Ed.), Rule-governed behavior: Cognition, contingencies, and instruc- tional control (pp. 97–118). New York: Plenum Press.

Vaughan, M. E., & Michael, J. L. (1982). Automatic reinforcement: An impor- tant but ignored concept. Behaviorism, 10, 217–227.

Vollmer, T. R. (1994). The concept of auto- matic reinforcement: Implications for behavioral research in developmental disabilities. Research in Developmen- tal Disabilities, 15 (3), 187–207.

Vollmer, T. R. (2002). Punishment happens: Some comments on Lerman and Vorn- dran’s review. Journal of Applied Be- havior Analysis, 35, 469–473.

Vollmer, T. R. (2006, May). On the utility of automatic reinforcement in applied behavior analysis. Paper presented at the 32nd annual meeting of the As- sociation for Behavior Analysis, At- lanta, GA.

Vollmer, T. R., & Hackenberg, T. D. (2001). Reinforcement contingencies and so- cial reinforcement: Some reciprocal re- lations between basic and applied research. Journal of Applied Behavior Analysis, 34, 241–253.

Vollmer, T. R., & Iwata, B. A. (1991). Es- tablishing operations and reinforcement effects. Journal of Applied Behavior Analysis, 24, 279–291.

Vollmer, T. R., & Iwata, B. A. (1992). Dif- ferential reinforcement as treatment for behavior disorders: Procedural and functional variations. Research in De- velopmental Disabilities, 13, 393–417.

Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Roane, H. S. (1995). Progress- ing from brief assessments to extended

experimental analyses in the evaluation of aberrant behavior. Journal of Applied Behavior Analysis, 28, 561–576.

Vollmer, T. R., Progar, P. R., Lalli, J. S., Van Camp, C. M., Sierp, B. J., Wright, C. S., Nastasi, J., & Eisenschink, K. J. (1998). Fixed-time schedules attenuate extinction-induced phenomena in the treatment of severe aberrant behavior. Journal of Applied Behavior Analysis, 31, 529–542.

Vollmer, T. R., Roane, H. S., Ringdahl, J. E., & Marcus, B. A. (1999). Evalu- ating treatment challenges with dif- ferential reinforcement of alternative behavior. Journal of Applied Behavior Analysis, 32, 9–23.

Wacker, D. P., & Berg, W. K. (1983). Effects of picture prompts on the acquisition of complex vocational tasks by mentally retarded adolescents. Journal of Ap- plied Behavior Analysis, 16, 417–433.

Wacker, D. P., Berg, W. K., Wiggins, B., Muldoon, M., & Cavanaugh, J. (1985). Evaluation of reinforcer preferences for profoundly handicapped students. Journal of Applied Behavior Analysis, 18, 173–178.

Wacker, D., Steege, M., Northup, J., Reimers, T., Berg, W., & Sasso, G. (1990). Use of functional analysis and acceptability measures to assess and treat severe behavior problems: An out- patient clinic model. In A. C. Repp & N. N. Singh (Eds.), Perspectives on the use of nonaversive and aversive inter- ventions for persons with develop- mentaly disabilities (pp. 349–359). Sycamore, IL: Sycamore.

Wagman, J. R., Miltenberger, R. G., & Arn- dorfer, R. E. (1993). Analysis of a sim- plified treatment for stuttering. Journal of Applied Behavior Analysis, 26, 53–61.

Wagman, J. R., Miltenberger, R. G., & Woods, D. W. (1995). Long-term follow-up of a behavioral treatment for stuttering. Journal of Applied Behavior Analysis, 28, 233–234.

Wahler, R. G., & Fox, J. J. (1980). Solitary toy play and time out: A family treat- ment package for children with aggres- sive and oppositional behavior. Journal of Applied Behavior Analysis, 13, 23–39.

Walker, H. M. (1983). Application of response cost in school settings: Outcomes, issues and recommendations. Exceptional Ed- ucation Quarterly, 3, 46–55.

Walker, H. M. (1997). The acting out child: Coping with classroom disruption (2nd ed.). Longmont, CO: Sopris West.

Wallace, I. (1977). Self-control techniques of famous novelists. Journal of Applied Behavior Analysis, 10, 515–525.

Ward, P., & Carnes, M. (2002). Effects of posting self-set goals on collegiate football players’ skill execution during practice and games. Journal of Applied Behavior Analysis, 35, 1–12.

Warner, S. F. (1992). Facilitating basic vo- cabulary acquisition with milieu teach- ing procedures. Journal of Early Intervention, 16, 235–251.

Warren, S. F., Rogers-Warren, A., & Baer, D. M. (1976). The role of offer rates in controlling sharing by young children. Journal of Applied Behavior Analysis, 9, 491–497.

Watkins, C. L., Pack-Teixteira, L., & Howard, J.S. (1989). Teaching in- traverbal behavior to severely retarded children. The Analysis of Verbal Be- havior, 7, 69–81.

Watson, D. L., & Tharp, R. G. (2007). Self- directed behavior: Self-modification for personal adjustment (9th ed.). Belmont, CA: Wadsworth/Thomson Learning.

Watson, J. B. (1913). Psychology as the be- haviorist views it. Psychological Re- view, 20, 158–177.

Watson, J. B. (1924). Behaviorism. New York: W. W. Norton.

Watson, P. J., & Workman, E. A. (1981). The nonconcurrent multiple baseline across-individuals design: An exten- sion of the traditional multiple base- line design. Journal of Behavior Therapy and Experimental Psychiatry, 12, 257–259.

Watson, T. S. (1996). A prompt plus delayed contingency procedure for reducing bathroom graffiti. Journal of Applied Behavior Analysis, 29, 121–124.

Webber, J., & Scheuermann, B. (1991). Ac- centuate the positive . . . eliminate the negative. Teaching Exceptional Chil- dren, 24 (1), 13–19.

Weber, L. H. (2002). The cumulative record as a management tool. Behavioral Technology Today, 2, 1–8.

Weeks, M., & Gaylord-Ross, R. (1981). Task difficulty and aberrant behavior in severely handicapped students. Journal of Applied Behavior Analysis, 14, 449–463.

726

Bibliography

Wehby, J. H., & Hollahan, M. S. (2000). Ef- fects of high-probability requests on the latency to initiate academic tasks. Journal of Applied Behavior Analysis, 33, 259–262.

Weiher, R. G., & Harman, R. E. (1975). The use of omission training to reduce self- injurious behavior in a retarded child. Behavior Therapy, 6, 261–268.

Weiner, H. (1962). Some effects of response cost upon human operant behavior. Journal of Experimental Analysis of Behavior, 5, 201–208.

Wenrich, W. W., Dawley, H. H., & General, D. A. (1976). Self-directed systematic desensitization: A guide for the student, client, and therapist. Kalamazoo, MI: Behaviordelia.

Werts, M. G., Caldwell, N. K., & Wolery, M. (1996). Peer modeling of response chains: Observational learning by stu- dents with disabilities. Journal of Ap- plied Behavior Analysis, 29, 53–66.

West, R. P., & Smith, T. G. (2002, Septem- ber 21). Managing the behavior of groups of students in public schools: Clocklights and group contingencies. Paper presented at The Ohio State University Third Focus on Behavior Analysis in Education Conference, Columbus, OH.

West, R. P., Young, K. R., & Spooner, F. (1990). Precision teaching: An intro- duction. Teaching Exceptional Chil- dren, 22, 4–9.

Wetherington, C. L. (1982). Is adjunctive be- havior a third class of behavior? Neur- oscience and Biobehavioral Reviews, 6, 329–350.

Whaley, D. L., & Malott, R. W. (1971). Ele- mentary principles of behavior. Engle- wood Cliffs, NJ: Prentice Hall.

Whaley, D. L., & Surratt, S. L. (1968). Attitudes of science. Kalamazoo, MI: Behaviordelia.

Wheeler, D. L., Jacobson, J. W., Paglieri, R. A., & Schwartz, A. A. (1993). An experimental assessment of facilitated communication. Mental Retardation, 31, 49–60.

White, A., & Bailey, J. (1990). Reducing dis- ruptive behaviors of elementary physi- cal education students with sit and watch. Journal of Applied Behavior Analysis, 23, 353–359.

White, D. M. (1991). Use of guided notes to promote generalized notetaking behav-

ior of high school students with learn- ing disabilities. Unpublished master’s thesis. Columbus, OH: The Ohio State University.

White, G. D. (1977). The effects of observer presence on the activity level of fami- lies. Journal of Applied Behavior Analysis, 10, 734.

White, M. A. (1975). Natural rates of teacher approval and disapproval in the class- room. Journal of Applied Behavior Analysis, 8, 367–372.

White, O. (2005) Trend lines. In G. Sugai & R. Horner (Eds.), Encyclopedia of be- havior modification and cognitive be- havior therapy, Volume 3: Educational applications. Pacific Grove, CA; Sage Publications.

White, O. R. (1971). The “split-middle”: A “quickie” method of trend estimation (working paper No. 1). Eugene: Uni- versity of Oregon, Regional Center for Handicapped Children.

White, O. R., & Haring, N. G. (1980). Exc- eptional teaching (2nd ed.). Columbus, OH: Charles E. Merrill.

Wieseler, N. A., Hanson, R. H., Chamber- lain, T. P., & Thompson, T. (1985). Functional taxonomy of stereotypic and self-injurious behavior. Mental Re- tardation, 23, 230–234.

Wilder, D. A., & Carr, J. E. (1998). Recent advances in the modification of establishing operations to reduce aber- rant behavior. Behavioral Interven- tions, 13, 43–59.

Wilder, D. A., Masuda, A., O’Conner, C., & Baham, M. (2001). Brief functional analysis and treatment of bizarre vo- calizations in an adult with schizo- phrenia. Journal of Applied Behavior Analysis, 34, 65–68.

Wilkenfeld, J., Nickel, M., Blakely, E., & Poling, A. (1992). Acquisition of lever press responding in rats with delayed reinforcement: A comparison of three procedures. Journal of the Experimen- tal Analysis of Behavior, 51, 431–443.

Wilkinson, G. S. (1994). Wide Range Achievement Test—3. Austin, TX: Pro-Ed.

Wilkinson, L. A. (2003). Using behavioral consultation to reduce challenging be- havior in the classroom. Preventing School Failure, 47 (3), 100–105.

Williams, C. D. (1959). The elimination of tantrum behavior by extinction proce-

dures. Journal of Abnormal and Social Psychology, 59, 269.

Williams, D. E., Kirkpatrick-Sanchez, S., & Iwata, B. A. (1993). A comparison of shock intensity in the treatment of long- standing and severe self-injurious be- havior. Research in Developmental Disabilities, 14, 207–219.

Williams, G., Donley, C. R., & Keller, J. W. (2000). Teaching children with autism to ask questions about hidden objects. Journal of Applied Behavior Analysis, 33, 627–630.

Williams, J. A., Koegel, R. L., & Egel, A. L. (1981). Response-reinforcer relation- ships and improved learning in autistic children. Journal of Applied Behavior Analysis, 14, 53–60.

Williams, J. L. (1973). Operant learning: Procedures for changing behavior. Monterey, CA: Brooks/Cole.

Windsor, J., Piche, L. M., & Locke, P. A. (1994). Preference testing: A compari- son of two presentation methods. Research in Developmental Disabili- ties, 15, 439–455.

Winett, R. A., & Winkler, R. C. (1972). Cur- rent behavior modification in the class- room: Be still, be quiet, be docile. Journal of Applied Behavior Analysis, 5, 499–504.

Winett, R. A., Moore, J. F., & Anderson, E. S. (1991). Extending the concept of social validity: Behavior analysis for disease prevention and health promo- tion. Journal of Applied Behavior Analysis, 24, 215–230.

Winett, R. A., Neale, M. S., & Grier, H. C. (1979). Effects of self-monitoring on residential electricity consumption. Journal of Applied Behavior Analysis, 12, 173–184.

Witt, J. C., Noell, G. H., LaFleur, L. H., & Mortenson, B. P. (1997). Teacher use of interventions in general education set- tings: Measurement and analysis of the independent variable. Journal of Ap- plied Behavior Analysis, 30, 693–696.

Wittgenstein, L. (1953). Philosophical in- vestigations. New York: Macmillan.

Wolery, M. (1994). Procedural fidelity: A re- minder of its functions. Journal of Be- havioral Education, 4, 381–386.

Wolery, M., & Gast, D. L. (1984). Effective and efficient procedures for the trans- fer of stimulus control. Topics in Early Childhood Special Education, 4, 52–77.

727

Bibliography

Wolery, M., & Schuster, J. W. (1997). In- structional methods with students who have significant disabilities. Journal of Special Education, 31, 61–79.

Wolery, M., Ault, M. J., Gast, D. L., Doyle, P.M., & Griffen, A. K. (1991). Teach- ing chained tasks in dyads: Acquisition of target and observational behaviors. The Journal of Special Education, 25 (2), 198–220.

Wolf, M. M. (1978). Social validity: The case for subjective measurement or how applied behavior analysis is find- ing its heart. Journal of Applied Be- havior Analysis, 11, 203–214.

Wolf, M. M., Risley, T. R., & Mees, H. L. (1964). Application of operant condi- tioning procedures to improve the be- haviour problems of an autistic child. Behavior Research and Therapy, 1, 305–312.

Wolfe, L. H., Heron, T. E., & Goddard, Y. I. (2000). Effects of self-monitoring on the on-task behavior and written lan- guage performance of elementary stu- dents with learning disabilities. Journal of Behavioral Education, 10, 49–73.

Wolfensberger, W. (1972). The principle of normalization in human services. Toronto: National Institute on Mental Retardation.

Wolff, R. (1977). Systematic desensitization and negative practice to alter the after- effects of a rape attempt. Journal of Be- havior Therapy and Experimental Psychiatry, 8, 423–425.

Wolford, T., Alber, S. R., & Heward, W. L. (2001). Teaching middle school stu- dents with learning disabilities to recruit peer assistance during coopera- tive learning group activities. Learning Disabilities Research & Practice, 16, 161–173.

Wolpe, J. (1958). Psychotherapy by recip- rocal inhibition. Stanford, CA: Stan- ford University Press.

Wolpe, J. (1973). The practice of behavior therapy (2nd ed.). New York: Pergamon Press.

Wong, S. E., Seroka, P., L., & Ogisi, J. (2000). Effects of a checklist on self- assessment of blood glucose level by a memory-impaired woman with Dia- betes Mellitus. Journal of Applied Be- havior Analysis, 33, 251–254.

Wood, F. H., & Braaten, S. (1983). Develop- ing guidelines for the use of punishing interventions in the schools. Exceptional Education Quarterly, 3, 68–75.

Wood, S. J., Murdock, J. Y., Cronin, M. E., Dawson, N. M., & Kirby, P. C. (1998). Effects of self-monitoring on on-task behaviors of at-risk middle school stu- dents. Journal of Behavioral Educa- tion, 9, 263–279.

Woods, D. W., & Miltenberger, R. G. (1995). Habit reversal: A review of applications and variations. Journal of Behavior Therapy and Experimental Psychiatry, 26 (2), 123–131.

Woods, D. W., Twohig, M. P., Flessner, C. A., & Roloff, T. J. (2003). Treatment of vocal tics in children with Tourette syn- drome: Investigating the efficacy of habit reversal. Journal of Applied Be- havior Analysis, 36, 109–112.

Worsdell, A. S., Iwata, B. A., Dozier, C. L., Johnson, A. D., Neidert, P. L., & Thomason, J. L. (2005). Analysis of re- sponse repetition as an error-correction strategy during sight-word reading. Journal of Applied Behavior Analysis, 38, 511–527.

Wright, C. S., & Vollmer, T. R. (2002). Eval- uation of a treatment package to reduce rapid eating. Journal of Applied Be- havior Analysis, 35, 89–93.

Wyatt v. Stickney, 344 F. Supp. 387, 344 F. Supp. 373 (M.D. Ala 1972), 344 F. Supp. 1341, 325 F. Supp. 781 (M.D. Ala. 1971), aff’d sub nom, Wyatt v. Aderholt, 503 F. 2d. 1305 (5th Cir. 1974).

Yeaton, W. H., & Bailey, J. S. (1983). Uti- lization analysis of a pedestrian safety training program. Journal of Applied Behavior Analysis, 16, 203–216.

Yell, M. (1994). Timeout and students with behavior disorders: A legal analysis. Education and Treatment of Children, 17 (3), 293–301.

Yell, M. (1998). The law and special edu- cation. Upper Saddle River, NJ: Mer- rill/Prentice Hall.

Yell, M. L., & Drasgow, E. (2000). Litigat- ing a free appropriate public education: The Lovaas hearings and cases. Journal of Special Education, 33, 206–215.

Yoon, S., & Bennett, G. M. (2000). Effects of a stimulus–stimulus pairing proce- dure on conditioning vocal sounds as

reinforcers. The Analysis of Verbal Be- havior, 17, 75–88.

Young, J. M., Krantz, P. J., McClannahan, L. E., & Poulson, C. L. (1994). Generalized imitation and response-class formation in children with autism. Journal of Ap- plied Behavior Analysis, 27, 685–697.

Young, R. K., West, R. P., Smith, D. J., & Morgan, D. P. (1991). Teaching self- management strategies to adolescents. Longmont, CO: Sopris West.

Zane, T. (2005). Fads in special education. In J. W. Jacobson, R. M. Foxx, & J. A. Mulick (Eds.), Controversial therapies in developmental disabilities: Fads, fash- ion, and science in professional practice (pp. 175–191). Hillsdale, NJ: Erlbaum.

Zanolli, K., & Daggett, J. (1998). The ef- fects of reinforcement rate on the spon- taneous social initiations of socially withdrawn preschoolers. Journal of Ap- plied Behavior Analysis, 31, 117–125.

Zarcone, J. R., Rodgers, T. A., & Iwata, B. A., (1991). Reliability analysis of the Motivation Assessment Scale: A failure to replicate. Research in Develop- mental Disabilities, 12, 349–360.

Zhou, L., Goff, G. A., & Iwata, B. A. (2000). Effects of increased response effort on self-injury and object manipulation as competing responses. Journal of Ap- plied Behavior Analysis, 33, 29–40.

Zhou, L., Iwata, B. A., & Shore, B. A. (2002). Reinforcing efficacy of food on performance during pre- and postmeal sessions. Journal of Applied Behavior Analysis, 35, 411–414.

Zhou, L., Iwata, B. A., Goff, G. A., & Shore, B. A. (2001). Longitudinal analysis of leisure-item preferences. Journal of Ap- plied Behavior Analysis 34, 179–184.

Zimmerman, J., & Ferster, C. B. (1962). In- termittent punishment of S� respond- ing in matching-to-sample. Journal of the Experimental Analysis of Behavior, 6, 349–356.

Zuriff, G. E. (1985). Behaviorism: A con- ceptual reconstruction. New York: Co- lumbia University Press.

728

Index

Index Page references followed by "f" indicate illustrated figures or photographs; followed by "t" indicates a table.

2 20 questions, 258

A A-B design, 1-2, 15, 20, 178, 191, 196-200, 205-206,

215, 218, 616 A-B-A design, 1, 15, 196-197, 200 ABAB design, 197 Abbreviations, 74, 168 Abilities, 29, 69, 213, 287, 296, 453, 537, 550, 584,

590, 617, 619, 650 defined, 69

Abscissa, 149, 154-155 Absences, 505 Absolute value, 170 Abuse, 342, 379, 478, 492, 580, 678-679, 716, 722

alcohol, 342, 679 child, 379, 678, 716 drug, 679, 722 emotional, 478, 716 of children, 678-679, 722 reporting, 678-679 types of, 342

Academic achievement, 380, 591, 670, 691, 697-699, 709, 721

Academic development, 649 Academic performance, 87, 217, 385, 559, 569, 580,

687, 691, 706, 716-717 Academic skills, 69, 243, 315, 322, 632 Academics, 537 acceptance, 78, 140, 258, 318, 375-376, 378, 385,

389, 445, 472, 689 ACCESS, 4-5, 10, 12, 15, 18, 77-79, 82, 86, 90, 100,

116, 118, 125, 148, 151, 170, 186, 201, 213-214, 226, 240, 249, 253, 257, 259, 273, 291-293, 297, 306, 310, 313, 318, 337, 353, 356, 448, 484, 487, 489, 500, 502-504, 511, 517-518, 525-527, 529-531, 533, 539, 556, 559, 568, 571, 576, 581, 597, 603, 610-611, 640, 644, 678, 681, 693, 710, 718

Accessibility, 33 Accountability, 38, 73, 75, 85, 90, 681, 700

assessment for, 75 of teachers, 681

Accreditation, 29, 671, 686, 712 Accuracy, 1, 3, 19-20, 81, 93, 97, 101, 112, 115, 119,

122-125, 128-133, 140-144, 150-151, 156, 183, 192-193, 209-211, 230-231, 243, 254, 256-257, 259, 261, 267-268, 306, 436, 440, 442, 519, 533, 562, 570, 590, 601, 606-607, 620, 625, 643-644, 650-651, 662, 670, 672-673, 696, 703-704, 709-710, 712, 714, 724

Achieve, 5, 19, 23, 37-38, 63, 77, 81, 83, 93-95, 97, 102, 120, 140, 161, 179, 181, 191, 194, 238, 242, 246, 270, 272, 379, 437, 447, 460, 466, 477, 495, 497, 553, 567, 569, 577, 580, 586-587, 589, 598, 618-620, 627, 632, 634, 636, 649, 657-658, 661-663, 670, 681, 683

Achievement, 69, 73, 89-90, 104, 125, 209, 217, 380, 455, 463, 568-569, 580, 591, 670, 691, 697-699, 709-710, 714, 716, 721, 724, 727

academic, 69, 217, 380, 569, 580, 591, 670, 691, 697-699, 709-710, 716, 721, 724, 727

and homework, 463 in mathematics, 73 influences on, 714 tests, 69, 73, 89, 209

Acid, 37 Acquisition, 5, 10, 12, 32, 42, 63, 78, 93, 97-98, 104,

109, 123, 157, 183, 190, 211, 213, 215, 229, 231-232, 253, 286-287, 306, 317-318, 320,

322-323, 412, 417, 419, 425, 427, 429, 432, 498, 536-537, 540-541, 546, 550, 552-554, 626-627, 632, 636, 653, 659, 661-662, 680, 685, 688-692, 695, 698, 701-703, 705, 708, 713, 717, 720, 723-728

language, 32, 42, 78, 213, 215, 287, 498, 536-537, 540-541, 546, 550, 552-554, 626, 688-691, 703, 705, 708, 723-725, 728

vocabulary, 78, 726 ACT, 26-27, 32, 37, 61, 71, 93-94, 290, 312, 342, 345,

386, 456, 469, 559, 562, 573, 575, 585, 591, 605-606, 631, 650, 668, 679-683, 692, 697, 699-700, 703-705, 709, 718, 720

Acting out, 57, 70-71, 572, 726 actions, 25, 32, 46, 51, 59, 74, 115, 286, 315, 346,

356, 372, 429, 539-541, 552, 554-555, 572, 585, 664, 676, 678, 680

Activation, 49, 447 Active responding, 686 Activities, 1-2, 5, 18, 25, 33, 53, 73, 76, 78, 80, 96,

108, 113-114, 128, 140-141, 203, 227-228, 250, 259, 268, 291-293, 295-299, 313, 325-326, 338, 353, 360-361, 368, 376, 378, 380, 382, 431, 437-438, 443, 482, 491-492, 500, 507, 512, 514, 516-522, 524-529, 531, 537, 541, 555, 564-565, 568, 570-571, 574-575, 581, 597-598, 602, 604, 608, 611-612, 642, 649, 651, 654-657, 664, 674, 680, 690, 693-695, 697, 702, 705, 711, 728

developmental, 76, 78, 295, 353, 382, 431, 437-438, 491, 524, 529, 555, 602, 656-657, 694-695, 697, 702, 705, 711, 728

follow-up, 228, 325, 657, 697 instructional, 5, 78, 80, 108, 250, 378, 443, 492,

568, 642, 649, 654, 690, 693, 697, 728 learning, 18, 73, 78, 80, 250, 313, 325-326, 361,

431, 537, 541, 568, 570, 608, 649, 651, 654, 656, 693-694, 702, 705, 711, 728

ongoing, 5, 108, 250, 259, 376, 378, 380, 482, 524, 654, 664

planning, 228, 250, 259, 268 sequenced, 1, 73 space for, 680 varying, 18, 113, 642

Activity level, 727 Activity reinforcers, 291 Activity Schedules for Children with Autism, 710 Actors, 65, 640 Adaptation, 53, 83, 237, 254, 286, 299, 369, 409, 688 Adaptations, 52, 636 Adaptive behavior, 72, 76, 80, 90, 238, 388, 504, 511,

694, 706, 714 measurement of, 238

Adaptive behaviors, 79-80, 427, 482, 499, 700 Addiction, 64, 594, 649 Addition, 24, 31-34, 46, 51, 56, 59, 63, 65, 69-71, 78,

80, 94, 97, 103, 119, 132, 137, 141-142, 148, 150, 156-157, 160, 168, 170, 185-186, 188, 195, 197, 205-206, 213, 221, 231-232, 242, 249-250, 263, 281, 288, 304-305, 307, 309-310, 333, 357, 360, 364, 378, 380-381, 386, 392, 394-395, 397, 405, 412, 418, 422-424, 443, 448, 460, 462, 474, 479, 482, 484, 490, 492, 499, 504, 507, 513, 518-520, 523, 526, 528, 531, 533, 537-538, 540, 549, 551-552, 556, 559, 593, 603, 605, 607-608, 612, 616-617, 625, 627, 632, 639, 646, 652, 659, 668, 674, 681-682, 703

implied, 46, 56, 186 properties of, 46, 148, 309, 537, 540, 552, 556

Addition and subtraction, 422-424, 605, 627, 703 Adjectives, 32, 133, 537, 540, 552, 554, 701 Adjustment, 9, 74, 76, 90, 331, 604, 708, 726 Adjustments, 293, 333, 367, 381, 450, 572 Administration, 83, 90, 208, 217, 265, 575, 721

of assessment, 83

Administrative tasks, 598 Administrators, 65, 85, 90, 181, 206, 215, 217, 224,

233, 238, 261, 381, 670, 678 documentation, 678 educational, 206, 217, 670 school, 217, 233, 261, 670, 678

Adolescents, 110, 263, 356, 439, 443, 445, 671, 689, 702, 704, 710, 712, 715, 719, 722, 724, 726, 728

Adults, 34, 79, 87, 99, 108, 153, 233, 258-259, 266, 284, 287, 293, 300-302, 317, 336-337, 347, 351-354, 356, 360, 418, 437-438, 451-452, 501, 525, 527, 540, 555, 564, 566, 575, 590, 613, 625, 631-632, 636, 641, 654, 671, 680, 685, 688, 691-694, 714, 716-717, 719, 724

Adverbs, 537, 540, 552, 554, 701 Advertisements, 23-24, 95 Advertising, 669 advice, 27, 70, 160, 283, 293, 598, 617, 643 Advocacy, 371

effective, 371 Affect, 2, 36, 46, 54, 66, 72, 75, 118-119, 128, 183,

216, 262, 269, 278, 285, 299, 303, 341, 364, 368, 388-389, 397, 416, 425, 450, 475, 477, 480, 499, 520, 549, 585-586, 594, 607, 612, 620, 627, 681, 683

Affirmation, 1, 178, 190-192, 195 Age, 50, 53, 77, 79-80, 90, 108, 160, 253, 265, 271,

295, 307, 437, 463, 472, 485, 550, 568, 588, 593, 611, 632, 636, 654, 666, 687, 719, 721

mental, 79-80, 108, 636, 687, 719, 721 motivation and, 79

Agency, 345, 363, 370, 373, 379, 571, 665, 681 Agendas, 342-343, 550 Agents, 575, 619, 697, 722 Aggression, 14, 64, 158, 259, 272, 287, 303, 316, 320,

334, 340, 342, 354-355, 359-360, 365, 367, 370, 388, 395, 410, 429, 471-472, 475-480, 482-484, 486, 489, 492, 501, 504, 506, 509, 511, 513, 517-518, 524-526, 531-534, 537, 551, 599, 685, 697, 702, 707, 714, 716, 724-725

forms of, 287, 320, 355, 388, 478-479, 482, 533 aggressive behavior, 64, 254, 272, 320, 346, 355, 359,

395, 400, 403, 410, 489, 708, 715, 720 Agreement, 3, 10-11, 16, 19, 32, 41, 85, 117, 119, 121,

122, 129, 131-144, 256-257, 268, 367, 519-520, 602, 606, 657, 678, 691, 700, 715, 717

agreements, 119, 135, 138, 141, 559 Aides, 71, 261 AIDS, 5, 24, 42, 148, 341, 440-441, 574, 652, 679,

694, 696, 724 airplanes, 53, 63 Alcohol, 24, 342, 355, 679

function of, 24, 355 Alcoholism, 649 Alert, 131, 164, 303, 367, 491 Algebra, 64, 307, 417, 567, 659 Allergies, 88 Alternative communication, 151 Alternatives, 186, 337, 371, 515, 520, 666, 678, 680,

707, 714 Ambiguity, 363 American Association on Mental Retardation, 676,

687, 701 American Educational Research Association, 383 American Guidance Service, 694, 710 American Psychological Association, 28, 370, 471,

667-669, 671, 685-686, 689 American Sign Language, 214 Analysis, 1-5, 7-10, 13, 18, 20, 22-43, 44-49, 55,

57-59, 62-66, 68-69, 75, 80, 83, 85, 89-90, 92-93, 96-97, 100-101, 104, 110, 115-116, 118-121, 122-125, 127-128, 131, 133, 135, 140-143, 146, 148-159, 162-164, 168-170, 174-177, 178-188, 193-194, 196-202,

729

205-208, 212-213, 215-218, 220-221, 224-226, 228-229, 231-244, 245-274, 276-277, 280, 282-283, 285-289, 293-294, 301-302, 304-305, 308, 311, 317-322, 324, 327-328, 334-335, 337-338, 340-343, 344-346, 348, 352-354, 357-359, 363, 365, 369-371, 373, 374, 377, 382-384, 390-391, 395, 398, 400-405, 407, 408, 411, 415, 426, 434-435, 437-446, 449-450, 452-453, 454, 457, 459, 464-465, 467, 468-470, 472-475, 478, 481, 483-484, 486-490, 493, 495, 498-499, 501-502, 507, 510, 514-516, 518-519, 521-523, 525-526, 528-530, 532-534, 536-539, 543-544, 547-551, 554, 556-557, 558, 561, 563, 579, 581, 583-586, 589, 594, 599-600, 607, 612, 615-617, 622-626, 628, 630, 632-633, 636-638, 645, 653, 655, 657, 662, 664-675, 677-678, 680, 682-684, 685-728

story, 150, 152, 212, 525, 687, 689, 720 textual, 13, 18, 536, 539, 543-544, 548, 556-557,

719 Anecdotal reports, 320, 362, 457 Anger, 42, 395, 485, 697, 720

management, 42 Angles, 643 animals, 42, 50, 110, 279, 355, 455, 460, 713 Animation, 465 Anonymity, 673 Anorexia, 722 Anorexia nervosa, 722 antecedent variables, 48, 393, 405, 512, 533, 542, 553 Antecedent-based interventions, 498 Anthropology, 709 Antisocial behavior, 356, 537 anxiety, 17, 609, 614, 620, 697

management programs, 609, 620 APA Ethical Principles of Psychologists, 681

and Code of Conduct, 681 Applicants, 413 Application, 11, 16, 27-28, 31, 34, 37, 40-41, 43,

57-58, 63, 82, 85-86, 93, 188, 192-193, 200, 204, 211, 215, 217, 221, 233, 235, 237, 249-250, 255-256, 271, 273, 283, 321, 333, 335, 340, 345, 347, 351, 366, 368, 370-371, 373, 376, 382, 389, 421, 439, 466, 469, 476-477, 479, 491, 493, 502, 506, 524, 550, 559, 577-578, 580, 586, 607, 613, 619-620, 669, 682-683, 686-687, 695, 699-700, 702-704, 707-708, 710, 718, 720-723, 725-726, 728

Applications, 34, 36-37, 52, 133, 170, 224, 242, 258, 282, 305, 317, 322, 341-343, 354, 363, 371, 373, 379, 381, 389, 416, 455, 465, 467, 469-470, 476, 479-480, 482-483, 488, 503, 515, 538, 550, 557, 559, 561, 575, 587, 598, 603, 619, 636, 646, 688, 695-696, 698, 701, 704, 707, 709, 714, 716, 723-724, 727-728

Applied behavior analysis, 1-3, 5, 9-10, 22-43, 44-45, 47, 55, 59, 63, 65-66, 68-69, 85, 89-90, 92-93, 96, 101, 104, 110, 115, 118, 120, 122-123, 125, 127-128, 133, 135, 140-143, 146, 148-157, 159, 164, 169-170, 174-176, 178-184, 187-188, 193-194, 196-197, 199-200, 202, 205-208, 212-213, 215, 217, 220-221, 224-226, 228, 231, 234, 236, 239-241, 245-274, 276-277, 282, 285, 293-294, 301-302, 305, 311, 317-322, 324, 328, 334-335, 337-338, 340-343, 344, 352, 354, 358-359, 363, 365, 370-371, 373, 374, 377, 383-384, 390-391, 400-402, 408, 411, 426, 434, 440-441, 444, 446, 454, 464, 468-470, 472-473, 475, 481, 484, 486-490, 493, 495, 498-499, 501-502, 507, 510, 515-516, 536-537, 558, 563, 579, 583, 585-586, 589, 599-600, 615, 622-624, 626, 628, 637-638, 645, 653, 655, 657, 664-666, 668, 671, 674-675, 682-684, 685-728

data collection, 65, 110, 115, 155-157, 257, 719 employment, 691 preference assessments, 154, 301, 691, 693, 701,

708, 714, 719 reinforcement, 1-3, 5, 9-10, 28, 34, 40, 44, 59, 66,

69, 90, 155-157, 159, 183, 187, 200, 202, 205-206, 212-213, 221, 226, 239-240, 248, 250-251, 253-254, 257, 259-260, 262-263, 270, 276-277, 282, 285, 293-294, 301-302, 305, 311, 317-322, 324, 328, 334-335, 337-338, 340-343,

352, 354, 358-359, 363, 371, 373, 374, 377, 383, 391, 400-402, 411, 446, 454, 464, 469-470, 472-473, 475, 481, 484, 486-490, 493, 495, 498-499, 501-502, 507, 515-516, 536, 585-586, 589, 622, 624, 626, 628, 645, 653, 655, 666, 674, 685-728

stimulus control, 2, 44, 55, 66, 408, 411, 499, 586, 689, 694, 705, 708-709, 711, 717, 720, 724, 726-727

target behavior, 3, 5, 9-10, 24, 26, 68-69, 85, 89-90, 101, 110, 118, 120, 123, 125, 127-128, 133, 135, 140, 143, 149-151, 164, 176, 180, 182, 184, 187, 193-194, 197, 205-206, 208, 221, 234, 239-240, 246, 253-255, 258, 263, 267-268, 270-271, 273-274, 301, 305, 359, 470, 487, 493, 516, 585-586, 589, 599, 615, 623-624, 626, 628, 645, 653

Applied Behavior Analysis (ABA), 2, 22, 27, 34, 42-43 Applied behavioral analysis, 457, 696, 716-718 Appreciation, 65, 538, 652 Approaches, 32, 34, 37, 70, 79, 89, 91, 131, 140, 208,

217, 258, 295, 298-299, 331, 368-369, 373, 385, 389, 402, 435, 511-512, 539, 559, 566, 595, 607, 632, 634, 665, 691, 696, 698, 712

brief, 299, 402, 511, 698 Appropriate education, 83 Appropriateness, 93, 119, 205-206, 216, 218, 233-234,

237-238, 242, 244, 258, 270, 274 Area, 3, 10, 22, 25, 44-45, 68, 72, 75, 86, 90, 92,

113-114, 122, 146, 149, 164, 176, 178, 181, 191-193, 196, 217, 220, 224, 229, 245, 274, 276, 311, 324, 330, 337-338, 342-343, 344, 359, 374-375, 378-379, 390, 396, 400, 404, 408, 414, 426, 434, 449, 454, 458, 468, 481, 498-499, 510, 530, 536, 546, 550, 558, 583, 590, 601, 622, 650, 660, 664-665, 671, 673, 681-683

Arguments, 58, 183 Arizona, 34 Art, 40, 42, 227, 296, 462, 465, 716, 720

defining, 40 music, 296

Art activities, 227 Arthritis, 418 Articles, 27, 31, 36, 227, 256, 271, 499, 537 Articulation, 531, 533, 537, 553 Artifacts, 115, 121, 126, 128, 143, 240 Artist, 29 Arts, 414, 478, 561, 577, 608, 649-650, 656 Arts specialists, 478 Assertiveness, 589 Assessing, 18, 30, 72, 75, 79, 90-91, 102, 117,

122-144, 151, 180, 183-185, 213, 219, 238, 250, 255-259, 261, 263, 274, 278, 299, 301, 411, 429, 433, 435, 438, 441, 452-453, 587, 617, 625, 630, 635, 656, 659, 662, 672-673, 692-694, 706, 710, 714, 718, 723

Assessment, 3, 5-6, 8-11, 14, 16-17, 32-33, 40, 58, 68-75, 77, 79, 81-83, 89-90, 94, 97, 99, 101, 120, 129, 131-134, 137, 139, 142, 144, 149, 152, 155, 159, 200, 211, 213-215, 226, 229, 232, 245, 259, 263, 266-267, 271, 274, 276, 295-301, 307, 310, 320, 333, 336-337, 340, 342, 353, 363-364, 368-369, 382, 411, 418, 430, 437, 440-441, 447-448, 452-453, 470-472, 484, 488, 503, 507, 510-534, 550-551, 556, 562, 572-573, 577, 598-599, 601, 604, 608, 619, 635, 651, 658-660, 665-666, 669-672, 675, 677, 679, 681, 683, 685-686, 688-690, 692, 694-701, 704-709, 711, 713-720, 722-724, 727-728

Assessment:, 69, 132, 298, 688, 695, 701, 704, 708 alternate, 713 alternative, 5-6, 8, 11, 159, 310, 320, 340, 353,

368-369, 448, 453, 484, 507, 511-513, 516, 527, 530, 533-534, 677, 685, 696, 699, 704, 714, 716, 724

authentic, 263 authentic assessment, 263 behavioral, 3, 8-11, 14, 17, 32, 58, 68-71, 73-74,

77, 79, 81-82, 89-90, 94, 99, 101, 129, 131-134, 137, 139, 142, 144, 149, 152, 155, 159, 215, 274, 307, 320, 336-337, 364, 437, 441, 448, 452, 470-471, 510, 514-516, 519-521, 523-524, 529, 550, 604, 619, 658, 669, 671-672, 675, 681, 685-686, 688-690, 692, 694-701,

704-709, 711, 713-720, 722-724, 727-728 case example, 534, 708, 718 checklists, 3, 9, 69-70, 72-73, 89, 519, 533, 651 community, 3, 40, 70, 73, 99, 215, 507, 659,

665-666, 685, 689, 692, 694, 698-699, 707, 709, 711, 714, 716

components of, 159, 211, 437, 441, 452, 556, 577, 651

concerns about, 32, 681 continuous, 5, 73-74, 99, 226, 307, 336, 503, 507,

516-519, 522, 526, 533 contract, 3, 16, 562 criterion-referenced, 69, 73, 670 curriculum-based, 73, 670 cycle, 159, 635, 711 day-to-day, 75, 299, 665 decision making, 94, 562, 666, 677, 683 definitions, 89-90, 129, 132, 134, 142, 448, 470,

522, 700 demand for, 471 descriptive, 5, 58, 73, 75, 81, 90, 307, 472, 510,

514, 516, 518-519, 521, 533-534, 573, 686, 689, 695, 706-707, 711, 719, 724

direct, 5-6, 9-10, 14, 69-70, 73, 75, 81-82, 89-90, 94, 101, 131, 142, 155, 245, 263, 266, 274, 295, 300, 307, 310, 333, 342, 418, 516, 519-521, 529, 659, 683, 688, 695, 704, 706, 709, 722, 724

direct observation, 5, 70, 73, 90, 333, 516, 519-520, 704

early childhood, 686, 697, 723, 727 ecological, 6, 68, 70, 75, 90, 368, 701 essays, 133, 695 family, 69-71, 90, 133, 296, 598, 660, 666, 670,

677, 683, 714-715 formal, 5-6, 8-9, 17, 70, 430, 556, 666, 671 formative, 94, 120, 259 framework for, 301 functional, 5, 8-9, 14, 17, 32-33, 40, 58, 69-72, 74,

77, 81, 120, 152, 159, 200, 215, 226, 263, 266-267, 271, 274, 320, 353, 363, 382, 453, 470-472, 484, 503, 507, 510-534, 551, 556, 562, 573, 601, 604, 608, 669, 672, 681, 686, 688-690, 692, 694-696, 699, 704-709, 714-717, 719-720, 722-723, 727

functional behavioral, 716 group, 5, 8-10, 17, 73, 81, 263, 267, 299, 336-337,

488, 516, 531-533, 562, 572-573, 577, 608, 619, 635, 651, 670, 688-689, 698-699, 701, 704-708, 716, 718, 727-728

guidelines for, 310, 363, 484, 573, 671-672, 686, 688-689, 713, 728

health, 82, 470, 665, 681, 700, 713, 717, 727 HELPING, 81-82, 651, 666, 677, 692, 701 history, 6, 9, 14, 17, 58, 75, 90, 363, 411, 471, 520,

524, 550, 599, 635, 666, 695, 704, 715-716, 718

in grades, 296, 713 informal, 70 instrument, 33, 131, 259, 274, 707 integrated, 296, 608 intelligence tests, 73 interpretation of, 101, 129, 267, 696, 715 inventories, 73, 573 lifestyle, 68, 619 math, 73-74, 81, 97, 101, 131, 267, 297, 340, 382,

524-525, 562, 577, 604, 608, 635, 697, 706-707, 711

mathematics, 73, 97, 701 methods, 3, 6, 14, 32, 40, 58, 69-70, 75, 89-90, 94,

120, 129, 134, 139, 144, 152, 245, 259, 263, 266, 274, 295-296, 298-301, 310, 363, 437, 441, 452, 507, 510, 514, 516, 519-520, 533, 556, 562, 651, 660, 665-666, 670, 685, 692, 696-697, 704, 708, 714-715, 717, 723-724, 727-728

methods of, 40, 152, 298, 514, 516, 520, 651, 697 monitoring, 16, 69, 71, 75, 89, 94, 245, 524, 526,

528, 562, 598-599, 601, 604, 608, 619, 658, 683, 685, 689, 692, 699-700, 705, 707-708, 713, 719, 724, 727-728

monitoring and, 94, 601, 608, 689, 724 need for, 133, 271, 658-659, 677 needs assessment surveys, 71, 90 norm-referenced, 73, 670 objective, 6, 32-33, 70, 83, 89, 129, 133, 519-520,

659, 670

730

objectives and, 670 observation, 3, 5-6, 8, 10, 14, 16, 68, 70, 73-75, 90,

99, 101, 120, 129, 131, 133-134, 144, 259, 295, 297, 300, 333, 364, 488, 516, 519-520, 529, 532-533, 598-599, 601, 659, 695-696, 700, 704, 711, 719, 723-724

of motor skills, 704 of problem behavior, 5, 8, 152, 320, 363, 471,

513-515, 520-524, 526, 528-529, 532, 534, 700, 704, 711, 717, 724

of social skills, 694 of spelling, 722 of writing, 598 of young children, 688, 719 planning and, 89, 245, 259, 263, 266-267, 271, 274 postassessment, 430 preference, 17, 155, 213, 245, 276, 295-301, 310,

337, 364, 382, 484, 530, 562, 665, 681, 689-690, 694, 698-699, 701, 708, 714-715, 718-719, 727

principles, 3, 33, 40, 58, 266, 276, 368, 411, 524, 658, 666, 669, 672, 681, 683, 685-686, 688, 695, 704-705, 709, 711, 713, 717, 719, 722, 724, 727

problem, 5-6, 8-9, 69-71, 73, 75, 77, 81-82, 89-90, 97, 99, 131, 152, 259, 266, 320, 340, 342, 363-364, 369, 411, 470-472, 484, 488, 503, 507, 511-516, 518-534, 551, 562, 577, 608, 681, 686, 695-697, 700, 704, 706-707, 711, 714, 717-718, 722, 724

procedures, 3, 9, 14, 16-17, 33, 40, 58, 71, 73, 75, 79, 83, 89-90, 94, 99, 120, 129, 131-132, 134, 142, 211, 213-215, 259, 266-267, 271, 274, 276, 295, 297, 299, 307, 310, 320, 333, 363-364, 368-369, 382, 411, 418, 430, 437, 447-448, 452-453, 470-471, 488, 503, 507, 513, 516, 519, 521, 524, 530, 534, 551, 556, 572-573, 601, 635, 665-666, 669-670, 672, 675, 677, 683, 686, 688-689, 692, 696-699, 704, 706, 708, 711, 714-715, 717-718, 720, 723-724, 727-728

protecting, 677 purpose of, 69-70, 75, 213, 382, 437, 452, 521,

534, 666 purposes, 8, 69, 101, 131, 133-134, 142, 511, 513 rating scales, 9, 72, 519-521, 533, 688 reasons for, 132-133 reliability, 14, 17, 129, 131-133, 142, 144, 266, 271,

274, 520, 688-689, 692, 700, 704, 708, 714, 720, 722, 728

reliability of, 17, 129, 132-133, 142, 266, 274, 520, 708, 720

risk, 368-369, 516, 562, 573, 677, 689, 704, 728 science education, 724 scoring, 73, 129, 132-133 self-assessment, 16, 418, 601, 608, 651 service, 671, 683, 689, 694 social studies, 267, 297, 685 software, 723 special education, 73-75, 83, 267, 320, 526, 573,

604, 608, 677, 686, 697-698, 701, 708-709, 711, 715-716, 719, 722, 724, 727-728

standardized assessments, 551 standardized testing, 551 standardized tests, 69, 73 steps in, 8, 229, 437, 440-441 student achievement, 670 summative, 94, 120, 134, 670 supportive, 77, 82, 90, 706 task analysis, 8, 68, 229, 259, 437, 440-441,

452-453, 556, 665 technology, 32, 40, 75, 94, 142, 263, 271, 274, 659,

685, 689, 695-696, 698-699, 704, 706-707, 714-715, 719, 722-724

technology and, 695-696, 704, 723-724 threat, 82, 215, 484 time sampling, 10-11, 120, 137, 516, 599, 716, 719,

724 transition, 333, 665 validity of, 17, 142, 215, 259, 263, 266, 271, 274,

699 vocational, 320, 337, 452, 670, 688, 699, 719 work samples, 651

Assessment methods, 69-70, 75, 89-90, 295, 300-301, 363, 519-520, 533, 714

experimental, 301, 363, 533, 714 functional assessment, 69, 519-520, 533, 714 interviews, 69-70, 89, 519-520, 533

Assessments, 24, 69, 73, 75, 131-134, 139-142, 144, 154, 202, 211, 213, 254, 258, 262, 268, 274, 288, 295-296, 299-301, 306, 310, 316, 336-337, 353, 364, 372, 385, 469, 476, 479, 484, 499, 503, 507-508, 516, 518-522, 524-525, 527, 531, 533, 550-551, 562, 616, 666, 679, 691, 693, 701, 708, 714, 717, 719-720, 726

classroom, 73, 134, 254, 258, 268, 364, 385, 479, 484, 508, 516, 524-525, 527, 531, 691, 693, 701, 708, 714, 719, 726

comprehensive, 69, 551 of attention, 522, 525, 531, 708 of language, 550 of sleep, 520 of students, 693, 701 outcome, 262, 469, 516, 524 quality, 131-134, 139-142, 144, 274, 306, 310, 337,

364, 372, 476, 551, 720 Assets, 3, 69 Assignments, 110, 116, 134, 183-184, 278, 280, 291,

294, 385, 417, 561, 576, 599, 604, 606-607, 623, 625, 650, 694, 701, 724

extra, 184, 623 Assistance, 70-71, 102, 259, 316, 320-321, 418-420,

431, 437, 442, 452, 493, 505, 508, 514, 518, 526, 562, 589, 591, 604, 608, 625, 629, 636, 639, 651, 653, 656-657, 659-660, 691, 697, 728

cards, 639, 653, 657, 691, 697 resource room, 259, 653

Assistive devices, 507, 694 Assistive technology, 593 Association, 28-29, 32, 35, 48, 288, 368-370, 383,

409, 447, 471, 667-671, 674, 676, 678, 685-687, 689-692, 695-699, 701-702, 708, 710, 712-713, 718, 721, 726

assumptions, 22-23, 25-26, 33, 42, 94, 148, 174, 178-195, 235, 243, 249, 569, 624

conclusions, 148, 174, 186, 624 error, 235

atmosphere, 597, 640 Attending, 34, 70-71, 74, 203, 206, 262, 346, 417-418,

429-430, 444, 474, 483, 525, 571, 673-674, 690

Attending skills, 418, 429-430 Attention, 8, 13, 24, 34, 55-56, 60, 69-71, 73-74, 82,

84, 90, 101, 103, 113, 117, 128, 150-154, 156, 168-169, 183-184, 187, 199, 202-207, 212, 215, 227, 254, 259, 262, 264, 266, 277-278, 280, 284-290, 293-294, 296, 301, 303, 307-308, 310, 316, 318, 320, 323, 356-357, 361-362, 367-368, 376, 382-383, 388-389, 396, 398-399, 417, 427, 431-433, 459-460, 464, 469-472, 477-479, 483, 489-490, 494, 501, 507, 511-512, 514-519, 521-529, 531-533, 538-541, 552, 562, 564, 575-576, 578, 580, 596, 600, 604, 606, 608-609, 613, 623, 636, 649-653, 655, 670, 674, 685, 687, 689, 692, 698-699, 704-705, 708-709, 712-714, 716-718, 724

and learning, 713, 716 from teachers, 187 negative, 13, 56, 69-71, 82, 84, 90, 117, 128, 262,

278, 316, 318, 320, 323, 356, 362, 367-368, 382-383, 389, 469-472, 477, 479, 483, 501, 511-512, 515-516, 533, 538, 552, 576, 608-609, 704-705, 709, 718

positive, 13, 56, 69, 82, 103, 117, 202, 254, 262, 266, 277-278, 280, 284-290, 293-294, 296, 301, 303, 307-308, 310, 316, 318, 320, 323, 356, 361-362, 368, 376, 382, 388-389, 459-460, 470, 472, 477, 479, 483, 490, 501, 511-512, 515, 523, 533, 564, 576, 578, 580, 596, 608-609, 649-652, 685, 687, 689, 698, 705, 708, 716, 718, 724

problems of, 202 selective, 522, 685 student, 34, 55, 73-74, 82, 84, 101, 103, 113, 117,

128, 156, 169, 184, 199, 203, 207, 212, 227, 254, 259, 262, 277, 280, 293, 296, 301, 318, 323, 356-357, 367, 376, 382, 388-389, 417, 427, 431-433, 460, 469, 477-478, 483, 489, 494, 507, 511, 516,

518-519, 521, 524-525, 562, 576, 580, 600, 604, 606, 608, 623, 636, 649-653, 655, 670, 685, 698, 705, 712-714

sustained, 82, 264, 303, 687, 689 theories, 712

Attention-deficit/hyperactivity disorder (ADHD), 266, 296, 562

Attenuation, 521, 707 Attitudes, 23-25, 27, 39-40, 42, 179, 727

teacher, 23-24, 40, 727 Attributions, 185 Audience, 2, 10, 35, 79, 180, 186, 249, 409, 536, 539,

543, 548-549, 557, 617 Audio, 110, 112, 116, 119, 130, 132, 654, 690, 711,

718 digital, 116, 119

Audio recordings, 116, 718 Audiotape, 104, 116, 119, 130, 132, 134, 366, 443 Audiotapes, 117-118, 385, 566 Augmentative communication, 552 Ault, 73, 442, 516, 689, 728 Austin, 42, 451-452, 456, 686-688, 695, 699, 701,

704, 706, 714, 717, 720, 723, 725, 727 Authentic assessment, 263 AUTHOR, 35, 148, 471, 635, 650, 669, 685-686, 688,

696, 709, 713, 720, 725 Authority, 23, 25, 70, 89, 234, 675-676

competent, 89 legal, 675-676

Authors, 20, 27, 34, 36, 45, 58, 60, 76, 110, 139, 148, 183, 189, 197, 203, 208, 224, 241, 256, 266, 268, 271-272, 278, 282-283, 289, 308, 319-321, 347-349, 358, 368, 413, 450, 476, 513, 561, 578, 596, 603, 613, 633-634, 648, 654, 678, 691, 720

Autism, 34, 40, 79, 94, 141, 151-152, 213-215, 225, 227-228, 288, 291, 307, 337, 340, 365, 418, 420-421, 427, 429, 445, 471, 477, 486-487, 489, 502, 529, 537, 546, 550-553, 555, 570, 590, 602, 623, 628, 632, 636, 644, 646, 650, 654, 667, 682, 685-686, 691, 694, 697-698, 701, 705-707, 709-713, 716-717, 719-720, 723-724, 727-728

Availability, 32, 119, 253, 281, 309, 318, 326, 332-333, 350, 357, 389, 393, 397, 401, 403-407, 410, 497, 499-500, 522, 559, 656

Average, 10-11, 19-20, 52, 82, 112, 156, 169-171, 174, 176-177, 181, 200, 203-204, 247, 258, 264, 273, 277, 293, 305-306, 327, 329-330, 332-333, 340, 345-346, 384, 394, 488-489, 493-495, 503, 520, 561, 578, 580, 591, 598-599, 601, 603, 606, 693

Averages, 578, 649 Aversion therapy, 710 Aversive conditioning, 713 aversive consequences, 345, 616 Aversive stimulus, 2, 8, 14, 44, 56-57, 59, 63, 66, 315,

338, 347, 388, 400, 471, 479, 540 Avoiding, 15, 240, 346, 388, 588, 672-673, 678 Awards, 328 Awareness, 2, 7, 9, 12, 56, 78, 82, 130, 143, 333, 614,

624, 674, 717 self, 9, 614, 674, 717

Awareness training, 9, 614

B B-A-B design, 1-2, 15, 20, 196-200, 205-206, 215, 218 Babbling, 52, 287, 546 Back, 34, 55, 74, 87, 124, 160, 197, 216, 219, 229,

277, 293, 345, 347-348, 360, 376, 379, 384, 406, 420, 428, 432-433, 437, 461-462, 472, 485, 527-529, 564, 585, 597, 599, 603, 611, 616, 629, 671, 681, 720

Background, 45, 64, 128, 180, 253, 271, 458, 598, 666 Background knowledge, 180 Backward chaining, 2-3, 59, 434-436, 442-446, 448,

450, 452-453, 699, 722 Balance, 14, 47, 49, 78, 217, 241, 296, 299, 368, 574,

666-667 Baltimore, 685, 695-697, 702, 704, 706, 711, 718,

723-724 Bar graphs, 148, 152-154, 176 Baseline data, 17, 19, 187-188, 190, 195, 200, 212,

221, 223, 229-231, 233-235, 294, 321, 376, 384, 387, 443, 456, 474, 491, 494, 497, 503, 530, 561, 568, 570, 615

AS, 17, 19, 187-188, 190, 195, 200, 212, 221, 223, 229, 231, 233-235, 376, 384, 387, 443, 456, 474, 491, 494, 497, 503, 530, 561,

731

568, 570, 615 Baseline data collection, 190, 195 BASIC, 1, 3-4, 7, 11-12, 16, 18, 23-27, 30-31, 34,

36-37, 40-43, 44-66, 73, 75, 85, 89, 93-95, 98-100, 102-103, 110, 120, 123, 137, 149, 160-161, 169, 174-175, 178-195, 215, 221, 224, 229, 243, 246, 258, 272, 278, 289, 291, 296, 298, 307, 312, 317-318, 322, 326, 329, 333-334, 336, 339-343, 345, 348, 350-351, 355, 364, 366-367, 370-373, 375, 378, 389, 391, 414-416, 425, 429, 465, 469, 473-474, 476, 482, 487-488, 496, 499, 508, 514, 537, 550-551, 553, 556, 570-571, 582, 587, 623, 630-631, 640, 644, 654, 661, 666, 669, 674, 683, 688, 692, 696-697, 702, 707-708, 715, 720, 723, 726

Basic research, 27, 36, 40-41, 43, 317, 322, 326, 329, 340, 342-343, 348, 351, 367, 371, 473-474, 476, 692

Basic skills, 73, 702 Beginning reading, 704 Beginning teachers, 294, 699 Behavior, 1-20, 22-43, 44-66, 68-91, 92-104, 108-121,

122-131, 133-144, 146-177, 178-195, 196-213, 215-219, 220-229, 231, 233-244, 245-274, 276-283, 285-298, 300-310, 311-323, 324-326, 328-343, 344-373, 374-389, 390-407, 408-414, 416-421, 424-425, 426-432, 434-453, 454-467, 468-480, 481-497, 498-509, 510-534, 536-557, 558-564, 566-570, 572-582, 583-620, 622-663, 664-684, 685-728

adaptive, 38, 52, 56, 72, 76-77, 79-80, 89-90, 238, 286, 308-309, 315, 362, 388, 427, 437, 482, 499, 504, 511, 524, 534, 681, 694, 700, 706, 713-714, 717

aggressive, 46, 64, 71, 83, 207, 254, 272, 320, 346, 355, 359, 362, 372, 387-388, 395, 400, 402-403, 410, 457, 478-479, 489, 500-501, 503, 513, 613, 688, 693, 697, 706, 708, 714-715, 720, 726

awareness of, 333, 624 challenging, 65, 83, 248, 259-262, 286, 368, 379,

499, 507, 511, 524, 623, 683, 686, 692, 722, 727

communication and, 76, 539 dangerous, 207, 238, 315, 322, 369, 373, 421, 459,

486, 494, 499, 588, 594, 641, 665 desired, 9, 12, 16, 69, 80-81, 83, 89-91, 102, 104,

127, 143, 161, 176, 179, 182, 184, 187-188, 204, 233, 242, 256, 322-323, 330, 361, 388-389, 396, 418, 437, 461, 466, 482, 485, 495, 497, 511, 513, 552, 554, 566-567, 573, 581, 586, 589, 595, 597-599, 601-602, 605-606, 608, 612, 615, 619-620, 625-627, 631-634, 636, 639, 642, 648-650, 654, 660-662, 670, 677

disruptive, 57, 70-71, 79, 93, 112-114, 187, 227, 260, 267, 277-278, 293-294, 315, 321-322, 345, 356-357, 359-361, 366, 376-380, 382-384, 387, 397-398, 420, 477, 479, 488-489, 492-493, 499, 505, 511, 519, 561-562, 568, 575-579, 606, 631, 651, 668, 686, 688-689, 691, 693-694, 697-698, 701, 704-706, 708, 713-714, 716, 719, 722, 727

environment and, 6, 33, 39, 45-48, 63, 75, 179, 184, 189, 297, 313, 587, 640, 657, 660

modification, 28, 32, 82, 94, 120, 203, 215, 242, 383, 439, 445-446, 453, 640, 685, 687-692, 694, 696-702, 704-710, 712-714, 716, 718, 720-727

observation of, 5-6, 25, 29, 42, 187, 195, 376, 397, 516, 519-520, 533, 552, 659

off-task, 99, 111, 127, 227, 267, 293, 333, 335, 485, 526-528, 561-562, 576, 606, 608, 616, 689, 692, 696, 699, 719

repetitive, 116, 151, 238, 250, 287, 320, 363, 472, 494, 641

replacement, 8, 63, 80, 259, 320, 323, 360, 391, 478, 482, 513, 524, 533-534, 552, 694

routines and, 40 self-esteem and, 589 simple, 2, 5, 13, 16, 26, 39, 42, 47, 52, 54, 63, 80,

86, 96, 140, 142, 147, 149-150, 152, 154, 170, 176, 186, 199-200, 211, 215, 233-235, 248-249, 254-256, 265, 271, 273, 278-279, 283, 293-295, 306, 343,

360, 393, 398, 403, 409, 411-412, 418, 429, 437, 465, 469, 531, 540, 546, 554-555, 560, 564, 587, 591, 601, 603, 613, 616-617, 643, 648-650, 652, 660, 681, 690, 710, 717, 719

social, 2-3, 9, 13, 16, 18, 23-24, 26, 29, 32-34, 36, 38-40, 42, 47-48, 52, 54, 59-60, 68-69, 72, 75-77, 79-80, 82-83, 89-91, 94-95, 117, 129, 141, 150, 156, 166, 176, 181, 184, 197, 202-206, 218, 226-227, 233, 237, 240, 246-247, 250, 253-254, 256-259, 261-264, 267, 270, 272, 274, 277, 279, 281, 285-288, 290, 293, 295-297, 301, 307, 309, 313, 315-316, 321-322, 325, 329, 333-334, 337, 350, 368-370, 376-378, 387, 397, 399, 402, 406-407, 416, 418, 421, 427, 431, 448, 455, 465, 470-472, 477, 482-483, 490, 499-501, 506-509, 511-512, 514-515, 517-518, 521, 527, 537, 539-540, 548, 550-553, 556, 561, 569, 574, 576-578, 580, 588, 591-593, 607, 609, 611, 614-617, 625-626, 628, 632, 646, 649, 651-652, 660, 665-668, 670-671, 680-681, 683, 685, 687-689, 693-694, 696-700, 704-706, 711-715, 717, 719-720, 722-728

target, 3-7, 9-16, 18-20, 24, 26, 38, 54, 62, 68-91, 93-94, 99-103, 108-113, 116, 118-121, 123-130, 133-140, 143-144, 147, 149-151, 158, 160-162, 164, 176, 180, 182, 184-185, 187-188, 190, 193-195, 197, 200, 203-206, 208-209, 216, 218, 221, 223, 227, 229, 233-234, 238-240, 242-244, 246, 253-255, 258-259, 263, 265, 267-268, 270-271, 273-274, 277, 279, 282, 287-288, 290, 292, 296-298, 300-301, 303-308, 310, 316, 322, 326, 330, 347, 349-350, 352, 355-356, 359, 366, 368, 372, 375, 379, 381, 386, 445, 448, 455, 461, 463, 466, 470, 476, 478, 487, 490-493, 496-497, 504, 509, 516, 518-519, 521-522, 553, 555, 561, 566-568, 570, 572-574, 577-582, 585-586, 588-589, 595-599, 601, 603, 605-606, 608-610, 612, 615-617, 619-620, 623-635, 639-646, 648-654, 656, 659-663, 665, 674, 681, 687, 689, 696, 700, 725, 728

triggers, 512 violent, 365, 494, 497, 722

Behavior change, 1, 3, 5, 8-11, 13, 15-16, 18, 23, 37-38, 40-41, 43, 44, 56, 58-59, 64-66, 68-69, 71, 73, 76, 78, 80-83, 89, 91, 94, 98, 102, 119-120, 124, 127-128, 140, 142, 144, 147-148, 150-151, 158-159, 164-166, 169-170, 173-177, 178-182, 186, 188, 191-195, 198, 203, 205-206, 208-209, 211, 216-217, 225, 229, 236, 238, 242, 244, 247, 250-251, 254-255, 258-259, 261-264, 268-274, 276, 282, 294, 304, 311, 316, 322-323, 324-325, 344-345, 347, 356, 361, 368-370, 374, 381, 384, 408, 410, 432, 434, 437, 448, 454, 459, 462, 468, 470, 477, 481, 492, 498-500, 507, 536-537, 558, 566, 568, 575, 580-581, 583, 586-587, 589-590, 592-593, 599, 606-607, 612, 617-619, 622-663, 664-665, 667-668, 672, 680-681, 690, 697, 699-702, 718

Behavior changes, 3, 20, 38, 43, 54, 63-64, 79, 94-95, 123, 159, 169, 174, 183, 185, 193-194, 200, 221, 223, 235-236, 238-239, 242, 251, 256-257, 259, 261, 269, 272, 274, 278, 280, 484, 589, 601, 615, 617, 620, 623-624, 630-633, 643, 648, 654, 658-663

Behavior contracts, 559 Behavior disorders, 259, 291, 505, 561, 590, 607,

687-688, 707, 714, 722, 726, 728 Behavior inhibition, 665 Behavior management, 28, 83, 200, 378, 580, 673,

686, 704 Behavior modification, 28, 32, 203, 439, 445, 640,

685, 687-688, 690-692, 694, 696-698, 700-702, 704, 706-710, 712-714, 716, 718, 720-725, 727

Behavior problems, 88, 334, 409, 479, 485, 491, 505, 531, 550, 659, 680-681, 690, 694-695, 698, 705, 707, 711, 726

Behavior, social, 72

Behavior support plan, 721 Behavior support plans, 256, 691, 722 Behavior therapy, 17, 370, 561, 614, 620, 686-687,

689, 693, 696, 700-701, 703-704, 706, 709-710, 713-714, 717-718, 722-723, 726-728

Behavioral approach, 39, 726 Behavioral assessment, 3, 68-70, 73, 82, 89, 510,

521, 669, 672, 688-689, 694-695, 700, 704, 706, 708, 716-718

Behavioral challenges, 39, 93, 247, 694 Behavioral consultants, 616 Behavioral contingencies, 44, 52 Behavioral contracts, 558, 722 Behavioral development, 29 Behavioral interview, 70-71, 82, 520 Behavioral observations, 521 Behavioral pharmacology, 550, 692 Behavioral principles, 52, 181, 308, 398, 466, 524,

559, 624 Behavioral problems, 80, 570 Behavioral rehearsal, 233 Behavioral sequence, 365, 437, 441, 463 Behavioral support, 718 Behaviorally disordered, 711-712 Behaviorism, 3, 11, 14, 22, 27-29, 31-35, 39-43, 57,

538, 543, 547, 584-585, 688, 691, 693, 698, 701-702, 707, 711-712, 716-717, 721, 723-724, 726, 728

Behaviors, 1-8, 10-16, 18-20, 23, 30, 34-38, 43, 46-57, 59-61, 63-66, 68-91, 93-95, 97-100, 102, 104, 108-110, 112-113, 115-119, 121, 123, 125, 127-131, 135-136, 138-140, 143-144, 150-152, 156, 158, 164-165, 167, 170, 176, 180, 183-186, 188, 190, 193-195, 198-201, 203-204, 206-208, 215-218, 220-225, 227, 229-230, 232-240, 242-244, 246, 250, 252-254, 256, 258, 260, 262-263, 265, 267-271, 277-279, 282, 284, 286-287, 289-293, 297-298, 300, 304-309, 315-318, 320-323, 324-325, 328, 332-337, 339-340, 342-343, 345, 347-348, 350, 353, 355-357, 359-362, 364-367, 369-370, 372, 376-384, 386, 388-389, 394, 404, 407, 409-410, 414, 416-418, 427-433, 436-438, 440, 442-446, 449, 452-453, 455-456, 458-463, 465, 467, 469-474, 476-480, 482-486, 488-494, 496-497, 499-502, 504-509, 511-513, 516, 518-522, 524-525, 527, 532-533, 537, 540, 543, 546-547, 551-553, 557, 561, 564, 566, 568-582, 584-587, 589-590, 592-593, 595-599, 602-606, 610-617, 619-620, 623-627, 629-634, 639, 641, 644-646, 648-653, 657, 659-663, 666-668, 674, 680-683, 685-689, 692-693, 697-698, 700, 703, 705-708, 710, 713-714, 716-718, 722, 725, 727-728

at-risk, 164, 728 bad, 125, 253, 315, 345, 369, 461, 587, 597, 603,

619, 710 communicative, 8, 484, 506-509 coping, 8 describing, 7, 13, 36-37, 69, 73, 93, 98, 236, 250,

282, 293, 307-308, 361, 443, 519, 648 desirable, 5, 10, 38, 57, 72, 79-80, 87, 89-90, 158,

188, 203, 206, 215, 233, 238, 243, 279, 286, 304, 320, 325, 337, 353, 355-356, 367, 378-379, 382, 389, 455, 478, 483, 496, 499, 504, 511-513, 520, 569-570, 573-574, 590, 595, 606, 612, 624, 629, 631-633, 639, 660, 667

enabling, 23, 94, 364, 589 nonverbal, 63, 112, 258, 279, 357, 427, 540, 543,

551-553, 557, 644, 687 on-task, 38, 80, 99, 110, 113, 127, 130, 320, 479,

485, 561, 590, 605-606, 613, 689, 693, 700, 705, 708, 728

self-management, 16, 75, 569, 575, 581, 584-587, 589-590, 592-593, 595-599, 602-606, 610-617, 619-620, 630, 634, 652, 657, 659, 663, 687, 692, 703, 705-706, 708, 710, 713-714, 716

SHARE, 2-3, 7-8, 46, 54, 60, 65, 88, 235, 243, 258, 263, 271, 336, 410, 455-456, 478, 547, 557, 590, 616, 627, 649, 660, 662

surface, 49, 81, 263, 473, 576, 579, 666, 703 verbal, 2, 5-6, 8, 10-11, 13, 15-16, 18, 20, 23, 36,

54, 63, 66, 70, 79-81, 90, 104, 183, 204, 208, 230, 252, 256, 258, 277, 279, 289,

732

293, 307-308, 316, 333, 350, 353, 356-357, 359, 361, 364, 366, 376, 388, 404, 407, 414, 418, 427, 429-433, 442-443, 445, 455, 460, 469, 477-479, 483-485, 492, 496, 507, 524-525, 537, 540, 543, 546-547, 551-553, 557, 561, 566, 572-574, 602, 612-613, 616, 620, 644, 652, 687-689, 692, 703, 707-708, 714, 717-718, 722, 728

Beliefs, 23, 26, 676 ability, 676 control, 23, 26

Belmont Report, 671, 713 Benefits, 24, 76, 78, 83, 94-95, 133-134, 148, 152,

158, 176, 179, 187, 194, 215, 248-249, 443, 540, 577, 584, 587, 589, 615, 619, 633, 662, 676-678

of teaching, 584, 662 Best practice, 369, 574, 715 Best practices, 615 Best treatment phase, 213, 218 Beverages, 291 Bias, 6, 11-12, 122, 124, 130, 141, 143, 164, 253-255,

704 system, 12, 124, 130, 143, 253-254 test, 254

Bibliography, 31, 685-728 Bills, 598, 625 Biography, 30 biological sciences, 24, 30 Bipolar disorder, 359, 531 Bishop, 376, 689 Biting, 9, 198, 287, 342, 346, 448, 479, 486, 531, 584,

598-599, 686 Blame, 661 Bleeding, 118, 263 Blending, 657 Blindness, 240 Blocks, 80, 86, 102, 204, 284, 308, 493, 519, 628-629,

655 Board of education, 292 Bolt, 104 Bonuses, 567 Bookmark, 596 Books:, 710

assignment, 647 picture, 461 recorded, 31, 385, 698

Borrowing, 102 Boundaries, 88, 104, 325, 683, 720 Boyce, 164, 689 Boys, 86, 128, 200, 259, 308-309, 328, 333, 337-338,

359, 386, 502, 612, 628, 644, 654, 689, 693, 715-717

Brain, 537, 550, 693-694, 697, 714, 723 research, 693-694, 697, 714, 723

Brain injury, 550, 693-694, 723 Breakfast, 421, 556 Bruises, 60, 118, 353, 524 Buffer, 387 Bullying, 82 Burns, 60 Busing, 601, 642

C Caffeine, 102, 342, 687 CAGE, 50, 286 Calculators, 108 Calendars, 330, 341 California, 439, 541, 716 Calorie, 608, 618 Campaigns, 23 Canada, 704, 718 Canon, 368 Capacity, 10, 217, 315, 393-394, 401, 427, 617,

675-677, 684 Capitalization, 561 Cards, 56, 98, 114, 183, 256, 267, 291, 307, 318, 330,

336, 421, 487, 499, 502, 506, 568, 570, 592, 606, 639, 649, 653, 657, 691, 693, 697, 706, 710, 713, 718

Career, 673 information, 673

Caregiver, 40, 86, 286-287, 368, 418, 432, 508, 521, 680, 696, 698

Caregivers, 9, 95, 287, 297, 366, 418, 428, 432, 491, 508, 513, 519, 522, 533, 546-547, 678, 680-681, 683

infants, 287

Caregiving, 413 Caring, 368, 670 Carrier, 86, 224 Cartoons, 296, 376, 464 Case examples, 524, 534 Case studies, 609, 666 Case study, 569, 609, 691, 697-698, 700-701,

704-705, 709, 718, 720 CAST, 95, 182, 465, 584 Categories, 249, 258, 265, 289, 293, 296, 336, 537,

630, 662, 711 Categorizing, 61, 662 Catheterization, 713 Causal relations, 24 Cause and effect, 32, 280, 309 Ceiling, 102, 128, 398, 404 Ceilings, 242 Centers, 380 Central tendency, 154, 171, 174 Cerebral palsy, 64, 225, 601, 681 Certification, 22, 29, 44-45, 58, 68, 92, 122, 146, 178,

196, 220, 245, 276, 311, 324, 344, 347, 374, 390, 408, 426, 434, 454, 468, 481, 498, 510, 536, 558, 583, 622, 664-665, 668, 671-674, 688, 712, 720, 722

competence and, 672, 674 Chains, 2-3, 8, 66, 227, 435-436, 442-444, 447, 450,

452-453, 597, 653, 703, 709, 727 Challenging behavior, 259, 261-262, 368, 499, 507,

524, 686, 727 Challenging behaviors, 260, 286, 511, 692, 722 Change, 1-20, 23-24, 26-27, 31, 33, 36-43, 44-48,

50-51, 53-54, 56-61, 64-66, 68-71, 73, 75-76, 78, 80-84, 88-91, 93-95, 98, 100-102, 115, 119-120, 123-124, 127-128, 130-131, 140, 142, 144, 147-151, 158-171, 173-177, 178-195, 197-198, 203, 205-206, 208-209, 211, 216-217, 221-222, 225, 227, 229, 234-244, 247, 250-251, 253-256, 258-264, 268-274, 276, 279, 281-285, 289, 294-296, 303-304, 308-310, 311-313, 316, 322-323, 324-327, 330, 333, 336, 344-347, 349-350, 355-356, 361, 364, 366-370, 372, 374-375, 381-382, 384, 388, 392-394, 397, 399, 403-404, 406, 408, 410, 420-422, 432, 434-437, 439, 448, 454, 459, 462, 468, 470, 476-478, 481, 491-492, 496, 498-500, 502, 504, 507-508, 512, 524, 527, 531-533, 536-537, 552, 558, 566-568, 573, 575, 580-582, 583, 585-590, 592-593, 597-599, 603, 606-607, 609, 612, 615-620, 622-663, 664-665, 667-668, 670, 672, 675, 677, 680-681, 683, 687-688, 690, 697-702, 715, 718, 724

acceptance of, 140, 258 attitude, 13, 26, 80, 649 constraints on, 180 continuing, 37, 89, 94, 190, 234, 403, 660, 683 decision making for, 677 essential, 3, 37-39, 45, 54, 61, 123, 128, 179, 182,

237, 279, 296, 436, 477, 649 facilitating, 333, 641-642 in schools, 568 motivation for, 448 planned, 10, 13, 18-19, 48, 75, 127, 182, 185, 188,

203, 229, 235, 243, 251, 253, 255, 273-274, 289, 295, 333, 374, 406, 524, 589, 612, 626-627, 631, 661, 667, 677

problems with, 124, 345, 603, 627, 635, 646 rate of, 3, 5-7, 10, 13, 16-17, 23, 43, 54, 60, 81, 84,

98, 100, 120, 150, 158, 176, 184, 237, 239, 242, 303-304, 316, 326-327, 333, 346, 355, 381, 436, 448, 470, 492, 496, 508, 606, 618, 636, 643, 646, 650, 652, 718

reciprocal, 653 resistance to, 3, 15, 186, 468, 476, 624, 646, 688,

724 science of, 3, 27, 31, 33, 40-43, 47, 181-182, 229,

264, 283 stages of, 131, 254, 308, 325, 507, 648, 700 sustainable, 667 theory of, 585, 636

Change agent, 586, 589-590 Change agents, 575, 619, 697 Changes, 1-6, 9-12, 15-16, 20, 23, 30, 36, 38, 41, 43,

46-51, 53-54, 56, 58, 60-61, 63-66, 79, 81, 91, 94-95, 97-98, 101, 116, 118, 123, 126, 128-129, 133, 140, 142-143, 147, 149-152,

154-155, 157-162, 166-169, 174-177, 179-181, 183-189, 191-195, 200, 202, 216, 219, 221-223, 227, 229, 233, 235-244, 246, 251-253, 256-257, 259, 261-262, 268-269, 271-272, 274, 278, 280-281, 294, 309, 312-313, 333, 343, 346, 348-349, 367, 391, 394, 399, 401, 409-410, 420, 435-436, 459-460, 464, 466, 484, 496, 511, 519, 531, 550, 586, 589, 601, 605, 615-617, 620, 623-624, 630-633, 643, 648, 654, 658-663, 676, 680, 685, 702, 704, 706, 713

economic, 23 Changing criterion designs, 184-185, 220-244, 249 Channels, 108, 706 Character, 587, 661 Characterization, 61 Charting, 17, 159-163, 176, 657, 698 Charts, 98, 132, 146, 148, 158-162, 169, 172-173,

176, 656, 687, 692, 720 comparison, 148, 159, 173, 176, 692, 720 data, 98, 132, 146, 148, 158-162, 169, 172-173,

176, 687, 720 equal-interval, 146, 158-159, 173 feedback, 148, 158, 176

Chats, 361 Cheating, 572 Checkers, 570 Checklist, 3, 68, 72-73, 90, 167, 377, 379, 418, 520,

559, 603, 685, 704, 709, 715, 728 Checklists, 3, 9, 69-70, 72-73, 89, 519, 533, 616, 641,

651 samples, 651

Chicago, 685, 688, 690, 699, 701-702, 710 Child Behavior Checklist (CBCL), 72 Child care, 39, 135, 710 Child development, 28, 293, 550, 685, 687-688, 704 Child needs, 555 Children, 24, 28, 34, 36-37, 39-40, 72, 75-76, 79, 83,

85-86, 94, 99, 102, 104, 108-109, 113-114, 116, 125-126, 131, 133, 141, 154, 158, 183, 203-205, 209, 211, 213-215, 227-229, 233-234, 236, 250, 252, 256, 258, 266, 268, 279, 287-290, 292-293, 296-297, 308, 317-318, 326-327, 350, 353, 356-357, 359, 364, 366, 376, 378, 380, 413-414, 418, 421, 427-429, 431-432, 438, 443, 456-458, 463, 485, 492, 505-506, 540-541, 546-547, 550-555, 560-566, 569-570, 576, 579, 581, 590-594, 596, 602, 604-608, 611, 613, 617, 620, 628-629, 631-632, 635-636, 641, 644-646, 649-650, 653, 655, 657, 659, 671, 673-674, 678-679, 682, 685-728

art of, 716 focus on, 214, 380, 593, 671, 687, 690, 692, 698,

700-701, 703-705, 709, 715-716, 724, 726-727

rights of, 688 self-evaluation, 602, 606, 608, 698, 704-705, 717,

719, 723 with ADHD, 688

Children with disabilities, 109, 719, 725 Chip, 302, 307, 530-531 Chips, 18, 290, 296, 302, 359, 530-531, 570-571, 574,

597, 635 Choice, 11, 26, 82, 93, 108, 198-199, 206, 234, 257,

259-260, 262, 295-298, 300, 303, 324, 336-339, 363, 368, 375, 381, 414, 419, 446, 463, 491, 494, 499, 525, 528, 530, 552, 602, 608, 611, 618, 633, 643, 647, 657, 680-681, 684, 694, 696, 700, 702, 705, 713-716, 718, 725

restricted, 525, 714 Choking, 200, 479 Chomsky, Noam, 538 Circles, 151, 154, 337, 444, 626 Citizenship, 604 Clarifying, 40, 79, 520 Clarity, 37, 98, 268, 499 Class discussions, 649 Classification, 289-290, 540, 543-545, 552, 595, 689,

707 Classroom, 9, 23, 34, 39-40, 48, 55, 57, 73, 78, 80,

87, 94, 108, 117-118, 134, 156, 181-185, 225-226, 228-229, 250, 254, 256, 258-261, 264, 266, 268, 278, 294, 297, 312, 330, 333, 338, 345-346, 356-357, 360, 362, 364, 375, 377-378, 382-383, 385, 388, 409, 436, 442, 444, 456-458, 462, 477-479, 483-485, 488, 493-494, 508, 511, 514, 516, 524-529, 531,

733

561, 564, 568, 570-571, 574-579, 581, 590-592, 594, 599-600, 602, 604-608, 610, 612-613, 617, 623, 625-628, 632, 635, 637, 641-642, 646, 649-653, 655-657, 685-693, 695-696, 698-699, 701-711, 713-714, 716, 718-719, 721-722, 724-727

arrangements, 48, 725 conference, 687, 691, 704, 710, 718, 727 displays, 156, 185, 568 environment in, 436, 594 organizing, 23, 592 visitors, 345

Classroom behavior, 87, 254, 259, 377, 477, 479, 485, 511, 599-600, 607, 687, 689-690, 711, 714, 722, 725

management system, 689 Classroom control, 294, 699 Classroom environment, 228

time, 228 Classroom management, 34, 704, 714, 718

positive behavioral support, 718 schedules, 704

Classrooms, 24, 118, 236-237, 259, 261, 266-268, 376, 443, 459, 561, 574-575, 580, 589, 591, 604, 636, 649-651, 686, 692-693, 697, 701, 704, 706, 708, 712, 717

behavior, 24, 118, 236-237, 259, 261, 266-268, 376, 443, 459, 561, 574-575, 580, 589, 591, 604, 636, 649-651, 686, 692-693, 697, 701, 704, 706, 708, 712, 717

behavior in, 24, 118, 259, 443, 686, 692-693, 697, 704, 708, 717

inclusive, 376, 697, 701 regular, 118, 259, 261, 574, 591, 650, 701, 717 special, 267, 376, 574, 591, 604, 649-650, 686,

693, 697, 701, 704, 708, 712 Clauses, 537 Cleaning, 87, 293, 307-308, 339, 382, 457, 651 Cleanup, 361 CLEAR, 19, 25-26, 30-31, 35, 58, 61, 85-86, 88, 91,

114-115, 133, 142, 144, 160, 162, 164, 171, 174, 181, 189-190, 199, 205, 209, 226, 231, 239, 249-250, 252, 255, 258, 260, 268, 272-273, 302-303, 361, 369, 371, 400, 402, 411, 432, 446, 466, 515, 520-522, 530, 533, 538-539, 550, 567, 573, 576, 601, 607, 623, 644-645, 667, 674, 677

Clicker, 4, 454-455, 460-461, 466, 716 Client records, 672 Clients, 39, 41, 60, 63, 70-71, 76, 78, 81, 85, 90,

94-95, 120, 238, 241, 257-258, 261, 270, 299, 367, 369-370, 381, 566, 568, 589, 599, 603, 605, 611, 613, 615-616, 620, 664, 667, 669, 672-673, 678, 680-681, 683, 694

behavior of, 60, 81, 90, 94, 120, 261, 568, 599, 605 context of, 63, 70, 76, 95 feedback for, 589, 694 responsibilities, 678

Climate, 52 Clinical assessment, 598, 619 Clinical judgment, 168, 333 Clinicians, 85, 227, 238, 244, 550, 589, 595 Clips, 108, 129, 261, 596 Closure, 314, 691 Clothing, 108, 312, 360, 568, 571

religious, 571 Clusters, 164, 442, 579 Coaches, 39, 224, 368, 640 Code of ethics, 667, 669, 696, 713 Codes, 7, 16, 83, 95, 110, 134, 259, 369-370, 516,

664, 667-668, 671, 684 Coercion, 315, 594, 677 Cognition, 695, 700, 709, 713, 726 Cognitions, 32 Cognitive abilities, 590 Cognitive approach, 537 Cognitive behavior therapy, 701, 709, 727 Cognitive Psychology, 33, 537 Cohesiveness, 577 COIN, 182, 209, 230, 419, 439, 565, 603, 639, 659 Collaborative learning, 635, 647 Collecting, 128-130, 143, 187, 190, 221, 231, 235,

321, 330, 381, 427, 447-448, 561, 568, 570, 576, 598, 619, 704, 707, 723

College students, 307, 331 Colleges, 671 color, 7, 17, 48, 61, 95, 98, 102-103, 169, 300, 355,

411-412, 415-417, 419, 421, 451, 545, 549, 642, 702

Colors, 98, 103, 412, 487, 642, 719 Com, 22, 45, 68, 92, 112, 122, 146, 169, 178, 196,

220, 245, 276, 311, 324, 331, 344, 374, 390, 408, 426, 434, 454, 468, 481, 498, 510, 528, 536, 558, 583, 605, 665, 671, 688, 716

Comics, 568 Commitment, 32, 36, 38, 413, 615-616, 620 Committees, 674 Communication, 1, 8, 37, 39, 65, 76, 79, 83, 94, 100,

151-152, 159, 179, 214, 227, 237, 250, 252, 257, 260, 262, 266, 290, 321, 342, 367, 427, 447, 483-484, 498-499, 501, 506-509, 520, 531, 533, 536, 539-540, 552, 562, 623, 678, 682, 689-690, 693-696, 698-699, 703, 706, 709, 711, 717, 719-720, 724-725, 727

behavior and, 8, 37, 76, 94, 100, 179, 237, 250, 262, 266, 483, 499, 507-508, 520, 539, 689, 706, 709

behavior as, 342, 483, 539, 696, 725 boards, 506, 719 disorders, 250, 520, 693, 699, 703, 717, 725 facilitated, 94, 682, 698, 727 good, 37, 39, 266, 520, 623, 678, 696, 719 impairments, 694, 724 language development, 689, 699 manual, 214, 447, 711, 720 parents, 39, 65, 79, 83, 257, 531, 533, 552, 562,

623, 682, 693, 699, 706, 717, 720 selection of, 76, 79, 83, 694 skill development, 83, 711 skills for, 706 skills training, 699, 703, 719 styles, 237 total, 8, 100, 250, 252, 257, 447, 562, 699, 711,

719 verbal and nonverbal, 427

Communication skills, 76, 427, 447, 499, 520, 623, 682, 698

instruction of, 698 Communications, 103, 717, 723 Community, 3, 26, 31, 40, 70, 73, 99, 116, 215, 230,

234, 237, 249-250, 258, 443, 507, 537, 602, 623, 627, 636-639, 641, 649-650, 659, 665-666, 674, 680, 685, 689, 692, 694, 698-699, 702, 707, 709, 711, 714, 716

agencies, 674 groups, 636, 694, 709 surveys, 714

Community settings, 70, 73, 234, 507, 650, 666, 680, 694

Community skills, 689 community-based instruction, 641, 685 Comparison, 7, 11, 14, 17, 26, 42, 52, 87, 102, 114,

131, 148, 151, 157, 159, 165, 167, 173-176, 182, 184, 190-192, 202, 208, 210, 213, 215, 217, 236-237, 246-248, 252-254, 263-266, 274, 318, 321, 326, 341, 414-417, 483, 501, 590, 685, 690-692, 696, 701-702, 704, 712, 714, 717-721, 724-725, 727

Comparisons, 7, 31, 41, 43, 93, 130, 151-153, 186-187, 191, 202, 237, 246, 249-250, 253, 259, 264, 416, 446, 465, 670, 722

Competence, 2, 8, 76, 90, 102, 133-134, 144, 256, 274, 648, 664, 669, 671-674, 684

clinical, 671, 674 credentialing, 671 maintaining, 672, 674 mathematical, 102

Competencies, 72, 83 Competency, 725 Competing, 3-4, 9, 26, 64, 69, 201, 217, 296, 305,

316, 322, 332, 417, 464, 502, 598, 614, 720, 725, 728

Competition, 65, 79, 279-280, 417, 579, 648 competitive employment, 691 Complaints, 382-383 Complexity, 45, 62-66, 81, 97, 128-129, 139, 185, 255,

270, 442, 450, 453, 455, 551, 635, 656, 676, 704

studies, 139, 185, 453 Compliance, 9, 86, 110, 129, 137, 236, 249, 255, 257,

266, 320-323, 378-379, 382, 385, 457, 472, 500, 502, 504-506, 509, 542, 561, 568, 606, 641-642, 680, 686, 690, 694-695, 699, 705, 709, 714, 721, 725

compliance skills, 378 Component analyses, 4, 272 Components, 4, 13, 20, 40, 64, 66, 81, 89, 159, 179,

182, 186, 194, 205, 209, 211, 229-232, 269,

280, 309, 339, 341, 355, 357, 360, 366, 370, 393, 401-402, 429, 433, 435, 437-438, 441, 447-449, 452, 455, 460, 463, 465, 482, 489, 496, 524, 526-527, 534, 541, 556, 559, 567-568, 577, 580-581, 586, 625, 629, 651, 656-657, 663, 716, 721

Composition, 20, 451, 550, 648, 726 Comprehension, 109, 132, 181, 542, 605 Comprehension skills, 181 Computation, 81, 237

instruction, 237 Computer programs, 593 computer software, 119, 168-169 Computer systems, 704 Computer technology, 467 Computers, 119, 121, 123, 329, 333, 335, 465-466,

695, 720, 723-724 operating systems, 119 software for, 119, 121, 723

Computing, 11, 135-136, 722 Concept, 4, 9, 16, 45, 54-57, 60-61, 79, 81, 93, 102,

124, 254, 257, 280, 283, 289, 291, 309, 345, 368, 391, 408, 412-414, 424, 455, 475, 480, 568, 586, 593, 607, 619, 629, 639, 676, 687, 691, 705, 711, 718, 726-727

Concept development, 705 Concept formation, 4, 408, 412-413, 424 Concept learning, 691 Concepts, 22, 29-30, 44-66, 102-103, 124, 178-179,

194, 276, 278, 283, 311, 344, 371, 374, 390, 408, 413, 426, 465, 468, 536, 622-623, 639, 661, 676, 689, 711, 713

guides, 179, 194 scientific, 29-30, 45-46, 63, 65, 124, 179, 194, 371,

713 spontaneous, 30, 468 teaching of, 689

Conceptual model, 39 Conceptual system, 33 Conceptual understanding, 58, 271-272, 499 Conceptualization, 465, 585 Conceptualization of self, 585 Conclusions, 12, 93, 126, 130, 136, 143, 147-148,

170, 174, 186, 202, 210, 247, 255, 272, 354, 381, 515, 522, 624

Conditional probabilities, 4, 24, 119, 518 Conditioned stimulus, 4, 9, 12, 15, 44, 47, 50-51, 66,

400, 409, 469 Conferences, 599-600, 674 Confidence, 24, 39, 125, 133, 142, 150, 170, 174, 177,

189-190, 193, 209, 222-223, 235, 254-255, 264, 271, 273, 371, 461, 666

Confidence level, 39 Confidentiality, 4, 75, 664-665, 669, 672, 677-679, 684

children and, 679 Conflict, 4, 41, 242, 664-665, 683-684 Conflict of interest, 4, 664-665, 683-684 conflicts, 83, 90, 227, 520, 537, 673, 683, 703 conflicts of interest, 683 Confounding variables, 8, 24, 42, 180, 185, 189, 194,

251-252, 254, 257, 269, 273 Confrontation, 572 Confusion, 57, 189, 280, 313, 381, 449, 499, 572, 587,

624 Conjectures, 235 Conjunctions, 227, 537 Connotation, 26, 62, 282, 345, 587 Connotations, 33, 35, 345 Consciousness, 27, 29, 33, 240 Consent, 10, 86, 370, 664-665, 672-673, 675-679,

684, 701, 703, 714 Consequence, 2-4, 14-16, 18, 44, 47-48, 53-59, 61-62,

66, 72, 203-204, 206, 209, 278, 280-284, 286-287, 297, 300, 304, 309, 312, 315-316, 330, 337, 340, 345-348, 351, 353, 356, 358-359, 361, 364, 366, 372, 377, 381-382, 388, 393, 396, 398-399, 401, 403, 405-407, 431, 470, 472, 474, 477, 480, 484, 492, 496, 499-500, 513, 516-518, 522-525, 527, 541-542, 544, 561, 575, 580, 582, 588, 594, 605, 608-609, 612, 616, 618, 645, 656, 692

Consequences, 1-4, 8-9, 12, 15-16, 24, 30-32, 40, 44, 48, 51-56, 58, 60-63, 65-66, 69, 73-74, 78, 88, 90, 95, 103, 178, 185, 210, 226, 253, 279-284, 286-287, 300, 306, 308-309, 316, 325, 335, 340, 345-346, 353, 356-357, 366, 369, 372, 386, 388, 393, 398, 406, 412, 416, 420, 425, 431, 469-470, 472-473, 476-477, 480, 484, 486, 492, 494, 499-500, 514, 516,

734

518, 520-522, 524, 539, 544, 546, 548-549, 552-556, 561, 569, 579, 586, 588, 590-591, 595, 597, 600, 604, 606-609, 611-618, 620, 628-629, 641, 644, 646, 655-657, 660, 663, 670, 675-676, 680, 688, 695-696, 701, 721, 724

reinforcement and, 4, 40, 54, 56, 58, 61-62, 66, 226, 282, 284, 300, 316, 325, 353, 406, 412, 477, 494, 539, 544, 556, 607-608, 611, 620, 644, 688

Conservation, 42, 256, 660 Consideration, 70, 83, 100, 109, 127, 186, 206, 217,

238, 272, 306, 316, 362, 369, 379, 381, 387, 453, 506, 514, 566, 571, 624, 633, 665, 667-668

Consistency, 14, 125, 129, 133-134, 136, 142, 189, 193, 255-256, 273, 343, 477, 499, 659-660, 680, 714

Constant time delay (CTD), 442 Constant time delay procedure, 420 Constructs, 30, 32, 43, 235, 585, 619, 709 Consultants, 370, 616 consultation, 29, 437, 561, 664, 672-673, 723, 727

organizational, 673 peer, 727 relationships, 29, 664, 672, 727

Contact, 6, 12-13, 15, 24, 33, 54, 64, 69-70, 79-80, 85-86, 94, 99, 115, 124, 130, 147-148, 176, 185, 187, 197, 203, 206, 229, 235-236, 247, 263, 266, 273, 289-290, 293, 298, 304, 307, 353, 355-360, 364, 376, 394, 420, 428-430, 432, 437, 446, 453, 455, 472, 478, 480, 491, 494, 503, 540, 546, 574, 597, 607, 611, 631, 634, 636, 643-644, 650, 652, 660, 662, 678, 683, 690, 699

Content, 22, 44-45, 68, 92, 122, 125, 146, 178, 180, 196, 217, 220, 245, 261, 264, 267, 276, 311, 324, 344, 374, 378, 390, 408, 426, 434, 454, 468, 481, 498, 510, 536, 556, 558, 583, 592, 594-595, 622, 632, 641-643, 664-665, 671, 685, 720

expectations, 267, 671 knowledge, 45, 125, 180, 264, 267, 595, 632, 641,

671 meaningful, 125, 180, 264, 267, 632

Context, 27, 30, 33, 35, 39-40, 47-48, 61, 63, 70, 73, 76, 88, 95-96, 100, 109, 123, 131, 143, 157, 166, 183, 188, 193, 211, 244, 248-251, 265-266, 304, 313-314, 316, 366, 369, 377, 381, 395, 399, 403, 409-410, 416-417, 419, 427-428, 432, 443, 483, 514, 518, 526, 541, 544, 564, 575, 635, 642, 665-666, 670, 676, 687, 690-691, 697-698, 700, 708-709, 714-715, 722-723, 725

vocabulary and, 35 Contextual variables, 266 Contingency, 1-6, 8-10, 12, 15-16, 18, 28, 30-31,

44-45, 53, 55-56, 61-64, 66, 111, 187, 203-204, 206, 211, 276, 278-282, 284-286, 291, 293, 296, 300-301, 303-310, 311-316, 318-319, 322, 325-327, 329, 333, 335-336, 339, 342, 345, 347-349, 351, 353, 355, 376-377, 385-387, 405, 407, 415-416, 431, 458, 465, 488, 491-492, 494, 497, 499-500, 508, 510, 522-523, 527, 541, 558-582, 586, 588-590, 594, 604-605, 608-609, 611-612, 615, 618, 622, 630-632, 643-650, 654-656, 662-663, 685, 687, 689-690, 693-694, 698, 702, 707-714, 718-719, 721, 725-726

Contingency contracting, 558-582, 618, 632, 694, 702, 711-712, 714

Contingency contracts, 296, 559-560, 563-565, 567 Continuing education, 674 Continuous measurement, 5, 7, 39, 43, 92, 113, 115,

122, 126-127, 181, 239 Continuous recording, 127, 157, 516-519 continuous reinforcement, 4-5, 307-308, 324-325, 336,

343, 475, 480, 497, 503, 648, 653 Contracts, 296, 382, 558-560, 562-567, 581, 592, 722

contingency, 296, 558-560, 562-567, 581 Control, 1-2, 4, 6-9, 11-13, 15-18, 23-24, 26-27, 29,

37, 39-43, 44, 47-48, 51, 55, 58-64, 66, 93, 104, 120, 128, 155, 170, 174, 177, 178-180, 182-187, 189-191, 193-194, 197, 201-207, 209-213, 215, 217-218, 221-223, 225, 229, 234-244, 245-246, 248-249, 251-257, 263-265, 268-270, 272-273, 278-282, 290, 293-294, 303-305, 307, 310, 336-337, 339, 341-342, 347-348, 350, 356, 371-372, 393,

398-399, 405, 408-425, 428, 440-441, 445, 449-450, 465, 482, 484, 491, 499-500, 504, 508, 514, 520, 522, 529, 531, 536, 538-540, 543, 546-557, 559, 561, 566, 568, 570, 574, 583-588, 590-592, 594, 596-598, 603, 607, 609, 611-614, 616-617, 619-620, 627, 629-630, 632, 639-642, 648, 652, 654, 660-663, 677, 680, 682, 685-690, 694-696, 698-700, 702-703, 705, 707-709, 711-714, 716-722, 724-727

impulse control, 16 self-control, 16, 336-337, 583, 585-587, 591-592,

594, 607, 612-613, 617, 619, 654, 688, 690, 694, 698, 713-714, 716-718, 720, 722, 724, 726

Control group, 246, 254 Control groups, 43, 246 Controlling stimulus, 411, 542 Controversial treatments, 94 Conventions, 135, 160-166, 537, 546 Convergence, 499, 508 Conversation skills, 636 Conversations, 119, 533, 539, 613, 632, 636

beginning, 613 Cooking, 30, 263, 601, 634, 636, 724 cooperation, 206, 381, 648, 660 Cooperative learning, 127, 656, 728

components, 656 Cooperative play, 99, 110, 131, 143, 203, 281, 301 Coordination, 648 Coping, 8, 726

behaviors, 8 Copyright, 1, 22, 31, 44, 48-49, 68, 92, 104, 122, 141,

146, 149, 151-164, 173, 175, 178, 196, 199, 201-203, 205, 207, 212-214, 220, 225-226, 228, 231, 234, 241, 245, 251-252, 261, 276-277, 285, 288, 292, 294, 301-302, 305, 311, 317-321, 324, 328, 334, 337-338, 344, 346, 352, 354, 358-359, 363, 365, 374, 377, 379, 383-384, 390, 408, 411, 426, 434, 438-441, 443-447, 451-452, 454, 456-457, 462, 464, 468, 471-473, 475, 481, 484-490, 493, 495, 498, 501-502, 507, 510, 536, 558, 560-561, 563-565, 569, 578-580, 583, 594, 596, 599-600, 602, 605, 610, 614, 622, 626, 628, 637-638, 640, 645, 647, 653, 655, 658, 664, 669-670, 673, 680, 685, 721

Correlation, 23, 42, 135, 179, 192, 194, 269, 350, 356, 400, 512, 516, 518-519, 606

Correlational studies, 23-24 Costs, 633, 665 Cough, 53 Council for Exceptional Children, 438, 443, 485, 605 Counseling, 293, 524

psychology, 293 Counting, 5, 8, 11, 53, 96-98, 100, 108-109, 115, 118,

126, 132, 135, 160-161, 170, 186, 286, 487, 604, 608, 706

Countoons, 604-605, 692 Courses, 666 Courtesy, 38 Courts, 83, 369, 381, 676 Creating, 30, 142, 169, 249, 373, 506, 587, 594-595,

598, 604, 606, 609, 612, 648-649, 690 Creativity, 39, 86, 695, 702, 709

fluency, 695 increasing, 695

Credentialing, 671, 722 Credentials, 675 Credibility, 237 Credit, 326, 573, 673 Criterion, 3-6, 8-10, 16, 19, 36, 38, 69, 73, 78, 92,

97-98, 101-104, 120, 128-129, 140, 142, 144, 184-185, 220-244, 249, 255, 258, 272, 291, 305-306, 310, 321, 334-337, 340, 343, 364, 380, 418, 420, 430-431, 436, 440-443, 452-453, 455-456, 458, 462-467, 475, 492-497, 501-502, 504-505, 527, 559, 562, 570, 573-574, 576-577, 579-581, 589-590, 602, 604, 611-612, 616, 618, 620, 631, 638-639, 646, 657, 670, 700

Criticism, 52, 204, 361, 538, 629, 652 Cubes, 461 Cue, 15, 256-257, 277, 333, 339, 418-419, 425, 429,

440-442, 444, 449, 453, 461, 596, 598, 604, 625, 653, 656-657

Cues, 112, 134, 339, 341, 417, 419, 429, 455, 508, 592-593, 595-596, 604, 616, 620, 651-652, 657

visual, 419, 595, 604 Cultural, 40, 52, 63, 253, 266, 286, 666, 683 Cultural practices, 52, 63, 286, 666 Culture, 48, 60, 594, 666, 668, 670, 682-683, 688, 698

high, 48, 688 influence of, 666 mainstream, 682 popular, 682

Cumulative records, 5, 148, 155-158, 176, 316-317 Cup drinking, 290, 705 Curiosity, 25, 27, 50 Curriculum, 34, 73, 94, 125, 180, 216, 267, 330, 368,

414, 417, 531, 556, 590-591, 605, 619, 635-636, 649, 655, 659, 667, 670-671, 690, 702

adaptations, 636 aligned, 667 basic skills, 73, 702 challenges to, 590 goals and objectives, 667 implementing, 94, 667 kinds of, 94 No Child Left Behind Act of 2001, 94 preschool, 34 social skills, 649 trends, 216, 267

Curriculum design, 34, 414, 636 Cursive writing, 76

D Daily instruction, 269 Daily living, 36, 72, 636, 716 Daily living skills, 72, 636, 716 Daily routines, 40 Daily schedule, 520 Dangerous behavior, 207, 315, 322 Data, 1-3, 5-7, 9-12, 15-20, 26, 29, 31-32, 37, 39, 41,

43, 65, 69-71, 73, 75, 79, 85, 90, 94, 96-104, 109-119, 121, 123-144, 146-177, 180-195, 197-202, 204, 208-210, 212-213, 215-218, 221-225, 227, 229-237, 239-243, 246-257, 259, 262-274, 277, 279-280, 287, 293-294, 296-298, 300, 305, 307-308, 316, 320-321, 333, 337, 340, 342, 358, 362, 364, 367, 370-371, 373, 376-377, 379, 381-382, 384, 387-388, 414, 430-433, 440-443, 445-448, 453, 456, 463, 471, 474, 476-477, 488-491, 494, 497, 503, 507, 510, 513, 515-520, 523, 528, 530, 532, 538, 561, 563, 567-570, 572, 574, 576, 578, 580-581, 584, 598, 601-602, 606-607, 615-617, 619-620, 627, 645, 650, 653, 657-659, 672-673, 681-683, 687, 689-690, 695, 700-701, 704, 706, 713, 715, 717, 719-720, 723

Data:, 135-136, 169, 615 validity, 7, 10, 15-17, 19, 79, 85, 123-127, 142-143,

180, 182, 184, 193-194, 215, 242, 246, 250-253, 256-257, 259, 262-267, 270-271, 273-274, 370, 387, 448, 520, 616-617, 687, 689, 700, 720

Data analysis, 119, 121, 269 Data collection, 65, 103, 109-110, 115-117, 119, 121,

129, 134, 155-157, 190, 195, 257, 367, 388, 442, 517, 672, 719

applied behavior analysis, 65, 110, 115, 155-157, 257, 719

during baseline, 156-157 Data paths, 113, 150-153, 158, 168-169, 171,

174-177, 194, 209-210, 213, 216, 218, 222, 227, 267, 269

Data points, 2, 5, 15-16, 19, 98, 149-150, 153-154, 157, 162, 164-165, 167-174, 176-177, 189-193, 195, 197-198, 204, 209-210, 221-223, 230, 233-234, 237, 243, 293, 337, 367, 377, 445, 489, 523, 602, 653, 658

Databases, 93, 593 Dating, 26, 368 Day care, 472-473, 479, 694 Daydreaming, 598, 643 Death, 31, 53, 588, 618 Debt, 345, 572 DEC, 160 deception, 673 Decision making, 94, 118, 127, 140, 146, 562,

666-668, 676-677, 683-684, 702, 710 ethical, 666-668, 676-677, 683-684 school-based, 710

Decision rules, 466, 496, 684 Decision-making, 162, 368, 379, 670, 677

735

curriculum and, 670 Decoding, 246 Deduction, 585 Definition, 2, 8, 15, 17-18, 22-43, 45-47, 51, 53, 58,

61, 68-69, 76, 85-89, 91, 93, 95, 98, 110, 112, 118, 120, 129, 132-133, 144, 161, 163, 180, 185, 188, 206, 246-247, 255, 267-268, 273, 278, 309, 312, 322, 327, 329, 331-332, 334-335, 345, 368, 372, 375, 381-382, 386, 389, 391, 405, 413-414, 425, 427-428, 432, 435, 452, 455, 462, 466, 469-470, 479, 490, 501, 537-539, 559, 568, 571, 575, 585-586, 599, 624, 660, 666, 672, 680

Definitions, 12, 60, 85-90, 118, 128-130, 132, 134, 142, 255, 306, 334, 393, 448, 470, 501, 522, 539-540, 543, 585-586, 623-624, 661, 680, 700

demonstrations, 31, 36, 234, 272, 293, 409, 594, 674 density, 156-157, 355, 713 Dependence, 471, 507

psychological, 471 Dependency, 62, 215, 290, 676 Depression, 277, 589 Depth, 95, 126, 312, 386 Description, 7, 15-16, 19, 23-25, 27, 31, 36-37, 42-43,

47, 52, 69, 73, 87-89, 93, 102, 120, 132, 135, 140, 160, 168, 175, 177, 179, 181, 194, 202, 239, 258-259, 272, 274, 280, 282, 307, 309, 312, 315, 381, 386, 401, 435, 482, 496, 503, 520, 537, 556, 559, 581, 588, 716

Descriptions, 3, 27, 33, 50, 72, 81, 88, 90, 119, 129, 148, 162, 187, 255-256, 272, 274, 283, 299, 410, 505, 520, 537, 539, 603, 681

Descriptive data, 75, 510, 516, 518 Descriptive research, 194 Descriptors, 134 Desensitization, 17, 583, 612, 614, 620, 727-728 Design, 1-3, 5, 7-8, 11-12, 15-16, 20, 34, 37, 39-41,

58, 64, 80, 88, 94, 101, 164, 168-169, 174, 178-180, 182-187, 190-191, 193-195, 196-219, 220-244, 245-246, 248-254, 257, 263-264, 267-274, 282, 292, 310, 316, 318, 320-321, 353, 362, 414, 455, 457, 489, 499, 523, 528, 530, 533, 562, 573, 581, 590, 601, 613, 615-616, 629-630, 632-633, 636, 644, 646, 688, 690, 692, 695, 698, 700-702, 704, 710-711, 714-715, 725-726

Designs, 2, 13, 15-16, 20, 37, 43, 178, 182-186, 191, 194-195, 196-219, 220-244, 246-250, 253, 263-264, 269, 273-274, 293, 304, 364, 574, 586, 591, 690, 696, 699-701, 704-705, 712, 719

Determinism, 5, 16, 22, 25, 27, 42, 179, 181, 194, 511, 594

Development, 9, 12, 19, 23-24, 27-29, 31, 34, 37, 39-40, 42, 54, 57, 78-80, 82-85, 90, 93, 97, 101, 187, 236, 240, 263, 267, 270-272, 286, 293, 309, 315, 322, 325, 334, 342-343, 354, 370-371, 380, 409, 411-418, 424-425, 427-428, 436, 439, 465, 482-483, 500, 505, 513, 516, 520, 538, 540, 546, 550, 553, 561-563, 567, 591, 594, 616, 633, 649, 664, 667, 672, 674, 681, 685, 687-689, 692, 696-697, 699, 703-705, 707-708, 711, 713-714, 716, 722-723

screening, 703 Developmental delay, 385, 524 Developmental disabilities, 34, 76, 79, 88, 94, 99, 102,

151, 156, 198, 214, 229, 263, 284, 287, 295, 300, 317, 353-354, 365-366, 382, 385, 429, 431, 437-438, 442, 456, 460, 487, 489, 491, 493-494, 499, 501, 505, 508, 550-551, 555, 580, 590, 601-602, 631, 653, 656-657, 665, 667, 671, 676-677, 686-689, 691-692, 694-706, 708-712, 714-715, 717-728

Devices, 24, 32, 73, 108-109, 112, 120, 128-129, 134, 148, 256, 507, 549, 603-604, 653-654, 694

Dewey, John, 591 Diabetes, 240, 695, 728 Diabetes mellitus, 728 Diagnosis, 280, 520, 531, 676, 717, 719 Diagrams, 62, 348, 710 Dialogue, 55, 717 Differences, 1, 41, 58, 63-64, 66, 99, 114, 131, 137,

151, 157, 167, 174, 208-209, 212, 215-217, 230, 246, 253, 300-303, 450, 500, 524, 624, 627-628, 645, 666, 682, 705, 717

Differential reinforcement, 2-6, 8, 10, 15-17, 20, 59, 100, 103, 250-251, 304, 310, 324, 334-336,

343, 353, 362, 366-367, 369, 373, 379-380, 385, 410, 412-413, 416, 424-425, 454-456, 458-459, 466, 477, 481-497, 500-501, 503, 506, 509, 524, 540, 549, 551-552, 569, 581, 655, 666, 674, 690, 693-694, 707-708, 715-718, 726

Differential reinforcement of alternative behavior (DRA), 5, 8, 304, 310, 366, 385, 481-483, 506, 509

Differential reinforcement of incompatible behavior (DRI), 5, 366, 481-483

Differentiation, 15-16, 87, 454-455, 466 Digestive system, 50 Digitized speech, 507 Dignity, 28, 665, 669, 676, 678, 680, 684, 721 Diligence, 27, 598 Dimensions, 10, 17, 22-23, 36, 38, 43, 46-48, 61, 65,

69, 79, 92-93, 95, 103-104, 119-121, 122, 149-152, 156, 176, 255, 261, 268, 316, 341, 411-412, 419, 436, 450, 456-457, 463, 465, 605, 616, 635, 641, 643, 670, 683, 687, 718-719

Diminishing returns, 699 Direct assistance, 419 Direct instruction, 75, 442, 593, 634, 659, 662, 667,

688 Direct observation, 5, 29, 70, 73, 90, 333, 516,

519-520, 641, 652, 662, 704 Direct observations, 69, 73, 82, 89, 521, 529, 682 Direct replication, 6, 15, 186, 245, 265-266, 274 Direct supervision, 643 Directions, 72, 75, 80, 84, 100, 262, 272, 333, 378,

398, 417, 485, 540, 593, 614, 641-643, 690, 695, 704-705, 708, 715

Directors, 29, 640, 668 Disabilities, 12, 34, 39-40, 76, 79, 82-83, 88, 94, 99,

102, 108-110, 114, 125, 151, 154, 156, 183, 198, 201, 211-212, 214, 229, 250, 258-263, 284, 287, 295-297, 300, 316-317, 353-354, 359, 365-366, 382, 385, 418, 429, 431, 436-439, 442-448, 455-457, 460, 472, 486-487, 489, 491, 493-494, 499, 501, 505, 507-508, 537, 550-551, 555, 568-569, 572, 576, 580, 590, 599, 601-603, 605, 613, 625-626, 631, 636-637, 640-641, 648, 650-651, 653-654, 656-660, 665, 667, 671, 676-677, 685-706, 708-712, 714-715, 717-728

ADHD, 296, 688, 696 developmental delays, 505, 650, 691 intellectual, 654, 711, 718, 721 visual impairment, 472

Disability, 12, 39, 83, 280, 437, 608, 676, 680, 689, 699, 701, 706, 724

Disabled students, 685, 701, 724 Discipline, 23, 34-37, 45, 85, 135, 179, 185, 272, 282,

361, 370, 501, 561, 623, 661, 665, 681, 706 Disclosure, 10, 675 Discourse, 541 discrimination, 4-6, 17, 57, 78, 102-103, 162, 208,

217, 345, 393, 408, 410-413, 416, 419, 421, 424, 450, 462, 499, 540, 544, 622, 636, 641, 694, 712, 718, 720, 722, 724

Discussion, 27, 34, 40, 54-55, 69, 96, 105, 115-116, 130, 164, 179, 232, 236, 246-248, 253, 264, 272, 274, 312-313, 356, 367, 376, 395, 442, 447, 512, 524, 538-539, 543, 551, 566, 584, 593-594, 607, 613, 624, 667, 675, 690

guided, 40, 594 Discussions, 34, 40, 57, 85, 253, 258, 280, 312, 369,

381, 649, 677 conceptual, 40, 280 issue, 253, 258, 369, 677

Displacement, 2, 8, 45-46, 65, 86, 97 Disruptions, 99, 104, 321, 356, 360, 387, 469, 471,

478, 483, 499, 516, 576, 697, 719 Distortions, 267, 549 Distractibility, 623 Distracting, 450, 461 Distraction, 152, 463-464 Distractions, 417, 432, 598 Distress, 447-448 Distribution, 15, 162, 164, 176, 291, 299, 571 Distributions, 115, 316-317, 519, 704 Diversity, 65, 79, 204-205, 308, 629, 698 Division, 28, 96-97, 160, 170, 190, 231, 562, 602, 630,

710 Division on Developmental Disabilities, 602 Divisors, 98, 101, 160

Doctrine, 368, 373 Documentation, 675, 678 Domain, 30, 40 Doubling, 98, 159-160, 173 Doubt, 13, 16, 22, 25-27, 37, 42, 140, 174, 179, 199,

237, 256, 283, 356, 371, 589, 683 Down syndrome, 531 Drawing, 5, 93, 98, 116, 132, 164, 167-168, 171-172,

174, 177, 227 Dressing, 451, 464, 504, 506, 596, 618, 656-657 Dressing skills, 451 Drill and practice, 282 Drowning, 23-24 Drug abuse, 722 Drugs, 24, 342, 355, 369

abuse, 342 Dube, 116, 413, 421, 685, 711 Durability, 15, 624 Duration, 5-7, 10-11, 13, 17-20, 37, 46, 54, 72, 74, 88,

92, 94-104, 106, 109-115, 117, 119-120, 122-123, 126, 128, 132, 136-137, 144, 152-153, 155-156, 178, 184, 186, 193, 199, 207, 243, 258-259, 268, 278, 297-298, 306-307, 316, 326, 330, 332-335, 337, 343, 350-351, 355, 360, 364, 369, 380-381, 389, 409, 419, 424, 429, 455-459, 463, 466, 471, 474-475, 488-489, 491, 493, 495, 503-504, 506, 529, 568, 574, 605, 643, 662, 691, 693, 725

Duration recording, 99-100, 113-115 Duty to warn, 679 Dyads, 728 Dynamics, 70, 465, 712

E Ear, 49, 409, 439-441, 521, 628, 723 Ear infections, 521 Early childhood, 686, 693, 697, 723, 727 Early childhood special education, 686, 693, 697, 727 Early intervention, 706, 710, 723, 726 Earth, 691 Eating skills, 660 Ecological assessment, 6, 68, 75, 90, 701 Economics, 437, 685 Economy, 10, 14, 18, 49, 53, 59, 130, 254, 290, 488,

558-582, 632, 686, 690-691, 697-698, 701, 716

Education, 1, 22, 28-29, 31, 34-35, 39, 42, 44, 57, 68, 73-75, 78, 80, 82-83, 92, 94-95, 101, 114, 116, 122, 146, 156, 178, 181, 183, 196, 220, 245-246, 259, 261, 264, 267, 276-277, 280, 292, 311, 320-321, 324, 344, 356, 374, 376-377, 390, 408-410, 413, 422-424, 426, 434, 436, 438-439, 443-444, 454, 457-458, 463, 468, 481-482, 492, 498, 505, 510, 526, 536, 550, 558, 569-571, 573-574, 576, 583-584, 589-591, 594, 602, 604, 608, 619, 622-623, 626-627, 632, 636-637, 642, 647, 649-651, 653, 656, 658, 664-665, 668-671, 674, 677, 682, 684, 685-694, 696-713, 715-728

at home, 116, 492, 574, 591, 627, 632, 670 evidence-based practices, 267 exemplary, 267 for teachers, 75, 95, 594, 651, 685, 701, 709 funding for, 34 global, 591 perspectives on, 34, 623, 686, 708, 712, 717,

724-726 prevention and, 39, 727 records, 1, 73, 156, 602, 684 right to, 95, 668-670, 677, 684, 686, 688, 725 supports, 576, 707, 709

Education programs, 594 Education Week, 706 Educational administration, 721 Educational goals, 217 Educational research, 383, 705, 725

applied, 383, 705, 725 Educational technology, 346, 699

effective use of, 346, 699 Educators, 75, 79, 83, 85, 95, 210, 282, 381, 591, 695 Effective instruction, 69, 318

generalization, 69 maintenance and, 69

Effective teaching, 69, 695 Effectiveness, 1, 3, 5, 7, 9, 11, 14-15, 17-19, 26-27,

37-38, 59, 85, 88-89, 94-95, 120, 179, 183, 200, 204, 212-213, 217-218, 224, 227, 236,

736

238, 244, 246-247, 249-250, 253, 266, 270, 281-285, 289-291, 300-302, 304, 306, 309-310, 316, 333, 345-346, 349-352, 354-356, 361, 363-364, 366-369, 371-372, 377, 381, 387, 391-401, 403, 405-407, 410, 416, 435, 463, 469, 476-477, 479, 484-485, 494, 496, 500, 502-504, 506-508, 511, 513, 524, 551, 568, 577, 581, 586, 588, 590, 595, 601-602, 606-607, 611, 613, 615, 642, 650, 657, 668-669, 674, 682, 694, 698-699, 701, 708-709, 712-713, 716-718, 720, 722, 725

Efficiency, 2, 41, 78-79, 94, 102, 119-120, 154, 215, 417, 449, 455, 460, 465-466, 478, 480, 506, 590

Effort, 25, 33-34, 38, 60, 64, 75, 80, 101, 119, 135, 148, 164, 169, 174, 185, 187, 194-195, 197, 206, 236, 255-256, 258, 265, 267, 270-271, 273-274, 293, 301, 306, 342, 349, 356, 359, 361, 387, 396, 398, 401, 406, 473, 476, 480, 484, 496, 500, 516, 561-562, 572, 575, 591, 597, 615, 623-624, 627, 633, 642-643, 648-649, 652, 660, 663, 697, 703, 716, 728

Ego, 9, 32 Eighth grade, 718 eKidTools, 592 Elaborating, 549 Electric shock treatment, 708 Electrical shock, 665 Elementary school, 110, 183, 296, 330, 577, 581, 604,

611, 687, 695, 697, 704, 707, 719-720 Elementary students, 113, 156, 211, 277, 377, 603,

605, 650, 653, 689, 692, 720, 728 E-mail, 62, 545, 681 Embarrassment, 240 Emotion, 471, 548, 697 Emotional and behavioral disorders, 613, 704, 723 Emotions, 395, 547, 698

anger, 395 Empirical data, 520 Empiricism, 6, 16, 22, 25, 27, 42, 93, 120, 179, 710 Employers, 36, 95, 120 Employment, 540, 691 Empowerment, 680

goal, 680 Encouragement, 23, 268, 325, 327, 580, 594, 612,

616 Endowment, 49, 584 Energy, 15, 17, 46-47, 65, 180, 256, 295, 409-410,

574, 595 light, 65, 409 sound, 409

Energy conservation, 256 Engagement, 15, 72, 93, 99, 108, 113, 240, 267, 291,

297-298, 310, 333-334, 336-337, 364, 502, 511, 610, 654-655, 680, 699, 705, 710, 724

English, 131, 284, 409, 462, 539-540, 542-543, 578, 688-689, 693

Old, 462 Standard, 131, 540

English language, 543 Enrichment, 73, 718 Enthusiasm, 204, 308, 341, 461, 629 Environment, 1-2, 4, 6-12, 14-15, 30-33, 35, 38-39,

44-48, 51-54, 56, 58-61, 63, 65-66, 69-70, 73, 75-77, 79, 83-87, 90, 93, 115-116, 118-119, 121, 129-130, 134, 140, 150, 153, 179-181, 184-185, 187, 189-191, 194, 202-203, 206, 216, 221-222, 228-229, 233, 235, 237, 251, 259, 263-264, 270-271, 273-274, 289, 297, 309, 313, 315, 326, 337, 347, 360-361, 366, 368-369, 372, 375-378, 380-381, 389, 394, 396, 398, 404-405, 407, 409, 417, 427, 432, 436, 439, 449-450, 453, 462, 469, 477-478, 484, 491, 501, 504, 508-509, 511, 514, 516, 519, 522, 533, 539-540, 546-547, 551-552, 556, 562, 566, 575, 584, 586-587, 589-590, 594-598, 617, 619, 624-627, 631, 636, 639-641, 648, 650, 652, 656-657, 659-661, 663, 668-669, 675, 680, 684, 695-696, 716, 723

arranging, 405, 562, 680 home, 75, 116, 185, 297, 313, 315, 381, 508, 533,

546, 594, 596, 627, 631, 659-660 least restrictive, 368-369, 375, 680 outdoor, 376

Environmental control, 409 Environmental factors, 52, 194, 722 Equal intervals, 10, 149, 159, 165, 176, 335, 493, 497 Equal opportunity, 56, 127, 244

Equating, 57 Equations, 63, 444, 465, 567 Equipment, 83, 119, 129, 168, 230, 326, 335, 353,

403-404, 594, 670, 681 Error, 3, 11-12, 19, 57, 80, 95, 124-125, 128-132, 137,

143, 168, 212-213, 235, 245, 269-270, 274, 318-320, 323, 360-361, 415-416, 418-420, 425, 438, 442, 477, 513, 589, 610, 613, 650, 674, 688, 700, 716, 718, 721, 728

Error correction, 319, 323, 416, 589, 650, 688 Errors, 64, 101, 128, 131-132, 143, 212-213, 230,

269-270, 283, 305, 318, 360-361, 419, 443, 562, 603, 637, 639-641, 651, 662, 670, 696, 721, 724

correcting, 128, 131 grammar, 361

ESCAPE, 2, 6-8, 14, 24, 34, 40, 56, 64, 69, 80, 186, 198-199, 257, 259, 261, 311-312, 314-318, 320-323, 340, 354-355, 363-364, 367, 370, 372, 378, 382, 394, 401, 406, 410, 468, 471-472, 474-475, 477-479, 482-484, 487, 500-502, 504-506, 509, 511-516, 518, 521-529, 532-533, 553, 584, 598, 603, 609, 616, 674, 686, 693, 695, 698, 705-706, 714-716, 718, 722

Estimation, 170, 633, 727 Ethical and legal issues, 571 Ethical codes, 7, 95, 369, 664, 667, 671 Ethical considerations, 65, 68, 70, 178, 196, 220, 241,

245, 322-323, 345, 367, 373, 389, 664-684 Ethical guidelines, 671 Ethical issues, 215, 312, 367, 581, 665-666, 669, 675,

684, 700, 723, 725 privacy, 669, 684

Ethical principles, 668-669, 681, 684, 685-686, 713 Ethical responsibility, 373, 673-674 Ethics, 7, 27, 234, 368, 370, 479, 664-669, 671,

682-683, 685-686, 692, 696, 700, 706, 710, 713, 717, 721

APA, 669, 685-686 codes, 7, 370, 664, 667-668, 671 codes of, 7, 370, 664, 667-668 ethical dilemmas, 665 integrity, 669 principle, 368, 370, 669 professional boundaries, 683 value, 7, 27, 700

Ethnic, 636 Ethnic groups, 636 Ethnicity, 253, 636 Ethologists, 110 Etiology, 723 Evaluation, 16, 69, 82, 85, 88, 94, 96, 113, 120, 133,

153, 178, 196, 202, 220, 231, 238, 244, 245, 250, 259, 271-273, 333, 343, 352, 363-364, 370, 373, 381, 389, 437, 447, 453, 474, 478, 521, 534, 561-562, 567, 574, 583, 601-602, 606, 608, 612, 616, 618-619, 669-670, 672, 681, 686, 689-690, 693, 695-705, 707-708, 712-719, 722-724, 726, 728

Evaluations, 264, 364, 606-607, 679 Event recording, 7, 18-19, 92, 99, 108-109, 113, 115,

120, 135-137, 140, 144 Events, 1-2, 5, 9-11, 14, 18, 22-27, 30-33, 36-37,

40-43, 45-48, 51, 54, 59-62, 65, 70-73, 75, 78, 86, 90, 93-96, 100, 108, 110, 119-120, 125, 128, 133-135, 143-144, 150, 167, 169-170, 179-181, 185, 187, 189, 193-195, 202, 209, 215, 240, 253, 280-281, 283, 285, 291, 293, 295, 312, 315-316, 320, 322-323, 325, 337, 347, 349-350, 353, 367, 372, 377, 391, 393, 395, 397-401, 406, 410, 416-417, 424, 499-500, 508, 511-514, 516, 518-522, 532-533, 536, 539, 542, 544, 547-549, 552, 554, 557, 571, 584, 586, 589, 594-595, 597, 603, 611, 614-615, 619-620, 623, 627, 630, 641, 669, 674, 683, 691, 709, 712

intersection, 10, 86 stimulus, 1-2, 5, 9-11, 14, 18, 27, 30, 40, 42, 45-48,

51, 54, 59-62, 65, 78, 90, 96, 100, 110, 120, 280-281, 283, 285, 295, 312, 315-316, 322, 337, 347, 349-350, 353, 367, 372, 377, 391, 393, 395, 397-401, 406, 410, 416-417, 424, 499-500, 508, 512, 516, 520, 542, 544, 547-549, 552, 554, 557, 586, 595, 597, 614, 619, 627, 630, 641, 691, 709

subsequent, 5, 23, 45, 60, 70, 100, 120, 170, 189, 195, 291, 512, 516, 518, 520, 586, 595,

615, 619 Evidence, 2, 17, 24, 27, 29-30, 40, 64, 68, 94-95, 103,

120, 126, 142-143, 181-182, 188, 193, 207, 209-210, 213, 221, 233, 252, 255, 258, 262, 267, 270-271, 288, 307, 355, 367, 382, 397, 400, 428, 442, 446, 511, 515, 561, 568-569, 574, 591, 609, 612, 637, 646, 665, 671, 680, 682, 690, 702, 713

Evidence based, 94, 120, 682 Evidence-based interventions, 94 Evidence-based practice, 94, 702 Evidence-based practices, 267 Evolution, 13, 16, 19, 49, 53-54, 59, 65, 289, 400, 584,

689, 725 Exceptional, 173, 305, 438, 443, 485, 546, 584, 594,

605, 685, 690, 692-694, 696-697, 701-702, 705, 707-710, 712, 714, 716-717, 719-722, 724, 726-728

Exceptions, 50, 164, 168, 170, 342, 479, 525, 559, 566

Exclusion, 7, 13, 374-375, 377-378, 380-381, 389, 719 Exemplars, 103, 412-413, 428, 635, 702, 713 Exercise, 42, 69, 81, 86, 128, 180, 326-328, 356,

359-360, 397, 418, 543, 575, 580, 585-586, 589, 595, 608, 616-618, 692-694, 705, 708

Exercises, 305, 457, 545, 726 Exhibits, 17, 187, 666, 681 Expectancy, 179, 676 Expectations, 6, 12, 25, 79, 128, 130, 143, 164,

254-255, 267, 273, 567, 569, 576, 593, 601, 611, 671

classroom behavior, 254 Experience, 16, 36, 39, 60, 64-65, 76, 139, 190, 202,

206-207, 232, 256, 260, 274, 297, 345, 355, 362, 364, 369, 395, 409, 428, 431, 437, 455, 460, 465, 474-475, 562, 567, 591, 612, 618, 640-641, 661, 666, 671, 673, 680-681, 684, 694

experiences, 9, 48, 60, 64, 83, 128, 305, 345, 350, 400, 448, 463, 587, 619, 634, 641, 649, 666, 674-675, 683

Experimental control, 4, 7, 11-12, 26, 37, 41, 43, 174, 178-180, 182, 184-185, 189, 191, 194, 197, 202-207, 209-210, 213, 217-218, 221-223, 234-244, 248-249, 251-252, 270, 272-273, 278, 482

demonstration of, 179-180, 182, 197, 205-207, 209, 218, 223, 235-238, 240-244, 248-249, 273

Experimental design, 1-3, 7-8, 11-12, 15, 94, 174, 178-180, 182, 186, 191, 193-195, 202, 208, 212, 218, 221, 224, 237, 246, 249-253, 257, 264, 267, 269-270, 272-273, 353, 533, 590

Experimental group, 246-247, 254 Experimental research, 155, 257 Experimentation, 16, 23, 25, 27, 40, 42-43, 45, 93,

148, 179, 182, 186, 197, 216-217, 248, 264-266, 270, 537, 710

Experiments, 8, 14-15, 20, 25-26, 31, 39-43, 50, 59, 63, 96, 165, 170, 174, 179-180, 182-183, 185-187, 191, 193-194, 197, 201-202, 209, 215, 221, 224, 235, 237, 239-240, 246, 249-250, 253-255, 263-268, 270-271, 273-274, 277, 293, 342, 356, 409, 492, 590, 601, 616, 682, 720

Expert, 262, 437, 449, 676, 695 Experts, 259, 261, 274, 437, 452, 463 Explanation, 26, 29, 31-33, 52, 135, 160, 162, 221,

280, 286, 304, 351, 361, 366, 470, 584, 645, 712

Explicit teaching, 256 Exploratory play, 680 Expressive language, 414, 539, 550 Expressive Vocabulary Test, 551 Expulsion, 7, 667-668 Expulsions, 445 Extensions, 200, 207, 381, 464-465, 544, 546-547,

713 Extinction, 4-8, 14-17, 44, 50-51, 54, 57, 59, 66, 186,

232, 266, 289, 303-305, 310, 325, 331, 336, 340, 343, 346, 348, 353, 357-358, 366-367, 369, 373, 379, 381-382, 392-393, 396, 401-407, 411-412, 455-456, 459, 463, 466, 468-480, 482-483, 485-486, 490, 497, 499-502, 504, 507-509, 512, 522, 524, 538, 553, 611, 624, 628, 644-646, 675, 688, 693, 696, 698-699, 702-704, 706-707, 710, 715, 717-719, 722, 726-727

Extra credit, 673

737

Extraneous variables, 7, 26, 185, 195, 236-237, 253 Eye contact, 266, 289-290, 356, 376, 429-430, 455,

636, 652, 699 eyes, 49, 201, 314, 472

F Fables, 698 FACES, 111, 605, 666, 704, 708 Facets, 623

conditions, 623 facial expression, 402, 641-642 Facilitated communication, 94, 682, 698, 727 Facilitating, 116, 333, 503, 555, 641-642, 722, 726 Facilities, 598 Factoring, 63 Factors, 1, 3, 8, 23-26, 32-33, 39-40, 42, 52, 63, 65,

68-69, 71, 75, 83, 97, 123, 128, 140, 143, 170, 177, 182, 191, 193-195, 208, 239, 241-242, 247, 251-253, 265, 269-271, 273, 316, 324, 345, 350, 362, 366, 380, 409, 416, 425, 435, 445, 450, 453, 460, 487, 494, 499, 520, 560, 587, 613, 619, 633, 635, 642-643, 656, 659, 665, 676, 680, 708, 722

Facts, 23, 29, 42, 63, 81, 97, 147-148, 150, 179, 194, 327, 330, 333, 339, 632, 635, 655, 689, 709, 711

Fading, 7, 40, 141, 153, 227-229, 408, 419, 421-425, 445-446, 459, 466, 479, 486-487, 492, 551-554, 674, 686, 689, 699, 704, 706, 712, 715-716, 718-720

Failure, 38, 94, 168, 235-237, 240, 255, 265, 315, 317, 371, 373, 387, 403, 499, 537, 551, 555, 567, 588, 593, 616, 623, 628, 697, 704, 720, 727-728

repeated, 240, 265, 387 Fairness, 267 Falls, 168, 177, 492, 594 Families, 76, 95, 563, 679, 693, 706, 727

needs, 76 step, 563

Family, 63, 69-71, 90, 133, 201, 236, 258, 296, 325, 339, 380, 384, 386, 499, 561, 563-565, 597-598, 623, 631, 660, 666, 670, 676-677, 683-684, 710, 714-715, 726

Family dynamics, 70 Family members, 71, 90, 258, 296, 325, 597, 660,

676-677, 683 Family Process, 715 Fast-food restaurants, 259, 625, 641 Fathers, 36 Fax, 160, 162, 681 fear, 17, 395, 459, 547, 614, 620, 665 Federal statutes, 370 Feedback, 104, 128-130, 133, 143, 148, 158, 176,

183, 205, 211, 256, 258, 268, 282-283, 307-308, 330, 361, 370, 381, 383, 420, 443, 496-497, 570, 579-580, 589, 591, 601, 608, 616, 626, 646, 648-649, 651-652, 672, 690-691, 694, 696, 702, 707, 709-711, 715, 722, 724-725

and practice, 129, 143, 256, 282, 649, 702, 710 evaluative, 601 facilitative, 710 general, 183, 256, 307-308, 361, 443, 570,

648-649, 651, 691, 702, 709, 711, 722 immediate, 148, 176, 308, 420, 496, 591, 616, 646,

690-691 mediated, 256, 709 seeking, 589

Feeding, 50, 250, 256, 287, 317-318, 323, 362, 421, 618, 668, 692, 713, 715

tube, 317 feelings, 14, 27, 33, 65, 87, 400, 546, 584, 589, 595,

603, 612, 706 Fees, 669, 672, 678 Females, 104, 164, 414, 632 Fiber, 15, 46 Fiction, 7, 22, 32, 587, 607, 617 Fidelity, 14, 19, 37, 83, 245, 255-257, 273-274, 486,

660, 669, 702, 727 Field testing, 582 Fifth grade, 330, 681 Fighting, 293, 355, 386, 561, 572, 686, 725 File, 678 Files, 598, 678 Findings, 7, 15, 17, 23-27, 34, 36, 40, 42, 45, 63, 69,

148, 170, 180, 184, 191, 193, 200, 246, 255, 263-267, 270-271, 274, 277, 285, 291, 316, 342, 350-351, 354, 385, 562, 617, 673-674,

682, 697 presentation of, 15, 24, 63, 316

fire, 47, 247, 545, 552 Fish, 40, 565, 584 Fixed schedules, 475, 480, 648 Flash cards, 98, 307, 318 Flexibility, 229, 241, 246, 249, 270, 273, 375, 559, 580 flight, 115, 444, 711, 716 Floors, 116, 242, 601 Flow, 606, 620, 641 Fluency, 63, 88, 97, 126, 128, 643, 656, 689, 693, 695,

703, 706, 711, 720 calculating, 97 oral reading, 126, 128, 695

FOCUS, 6, 15, 23, 38, 40-41, 69-70, 80, 89, 100, 123-125, 150, 167, 169, 184, 197, 204, 214, 229, 246, 264, 273, 282, 315-316, 380-381, 436, 460, 494, 513, 522, 554, 567, 573, 589, 593, 613, 624, 659, 671, 687, 690, 692, 698-701, 703-705, 709, 715-716, 724, 726-727

Folders, 571, 625 work, 625

Food, 1, 3, 7, 14, 19, 30-32, 49-50, 53, 57-60, 66, 81-82, 102, 183, 226, 250, 259, 263, 277, 281, 283-284, 286-287, 289-291, 302, 312, 315, 317-318, 347, 349-350, 360, 385, 391-399, 401, 403-407, 409, 412, 421, 429, 435, 442, 444-445, 447-449, 459, 471-472, 492, 494, 501, 540, 546, 548, 552, 568, 570-571, 596-597, 605, 614, 618, 625-626, 637, 639-642, 652-653, 657, 685, 693, 695, 698, 704, 709, 714-715, 718-719, 728

cooling, 394 Food preparation, 442, 698 Forgetting, 470, 633 Formative assessment, 94, 120, 259 Forms, 3-4, 7, 9, 11-13, 16, 29, 32, 43, 52-53, 56, 58,

61, 63, 66, 72, 85-87, 101, 110, 129, 133, 148, 153, 204-205, 221, 224, 227, 243, 283, 287-288, 291, 293, 307, 315, 317-318, 320, 323, 350, 355-356, 361, 377, 380, 382, 388, 396, 402, 405-406, 409, 413, 418, 427, 431, 455, 469-470, 478-479, 482, 506, 512, 520, 523, 533, 540-541, 548, 551-552, 557, 571, 585, 595, 603-604, 608, 623, 628-629, 632-633, 636, 648, 653-654, 661-662, 678, 698

Formula, 135-137, 140, 163, 171, 305, 701 Formulas, 249, 305 Forward, 3, 8, 18, 39, 286, 308, 359, 434-436, 438,

442-443, 446, 448, 452-453, 631, 711 Forward chaining, 3, 8, 18, 434-436, 438, 442-443,

446, 452-453, 711 Foundations, 34, 419, 538, 687, 700 Fowler, 75, 97, 590, 599, 646, 650, 654, 687, 696,

705, 723 Fractions, 625 Frames, 604, 710 Free play, 281, 515, 523, 526, 528-529, 532 Free time, 26, 83, 282, 290, 295-298, 312-313,

384-385, 387-388, 488, 492, 559, 561, 568, 573, 576, 580-581, 590, 608, 685, 704

Free writing, 100 Freedom, 28, 413, 680, 696, 721 Frequency, 1-3, 5-8, 10-15, 17, 19, 31, 46, 48-49, 51,

53-59, 61-63, 65-66, 70, 72-73, 81, 90, 92-93, 96, 101, 104-105, 109, 112-113, 115, 117, 120, 127, 130, 132, 138-141, 149, 151, 156, 159-163, 173, 176-177, 184, 187, 198, 202, 204, 206, 213, 226, 229, 236, 243, 250, 253, 258, 278-281, 283-285, 287-289, 291, 294, 298, 303-304, 306, 309-310, 312-313, 316-317, 338, 342, 345-352, 356-361, 366-367, 369, 372, 375-376, 381-382, 388-389, 391-401, 405-406, 409-410, 424, 427, 455-457, 459, 462-463, 466, 469-470, 473-475, 477, 480, 484, 486, 491, 494, 496, 499, 516, 522, 532, 541, 546, 551, 553-554, 575, 578, 585-586, 588-589, 595, 597, 600, 606, 608-611, 613-614, 620, 624, 629, 643, 650, 652, 657, 670, 686, 692, 701

of sound, 104 Frequency distributions, 316-317 Frequency polygon, 149 Frequency recording, 516 Frequently asked questions, 571 Freudian, 549 Fromm, Erich, 286

Frustration, 594 Fun, 229, 461, 555, 612, 649 Functional analysis, 8, 29, 46, 48, 153, 194, 198, 202,

207, 226, 287, 320, 357, 382-383, 472-474, 478, 483-484, 486, 489, 499, 501-502, 510, 514-516, 518, 521-523, 525-526, 528-529, 532-534, 674, 693-694, 703, 705-707, 710, 714-715, 717, 723, 726-727

Functional assessment, 9, 69, 72, 74, 510-511, 514, 516, 519-521, 524, 526, 529, 533, 604, 608, 672, 686, 692, 694-695, 705, 708-709, 714-715, 723

Functional Assessment Checklist for Teachers and Staff (FACTS), 709

Functional assessments, 507 Functional behavior analysis, 215, 353 Functional behavior assessment, 5, 8, 69, 226, 266,

363, 470-471, 503, 507, 510-534, 688 Functional communication training, 1, 8, 151-152, 159,

257, 260, 262, 266, 367, 484, 498, 501, 506-509, 562, 690, 693-696, 699, 706, 720, 725

Functional communication training (FCT), 8, 266, 367, 484, 498, 506, 509

Functional relationships, 62 functional skills, 669, 680 Functioning, 34, 69, 72, 82-84, 90, 204, 296, 348, 362,

366, 394-395, 402, 407, 410, 414, 420, 543, 568, 571-573, 650, 659

Functions, 2-4, 8-9, 12-13, 19, 46, 48, 52, 56, 59-61, 65-66, 69, 89, 93, 120, 152-153, 179, 194, 221-222, 249, 287, 289, 296, 303, 306-307, 310, 313, 349-350, 356-359, 372, 394-395, 397-400, 403, 409-410, 424, 452, 499-500, 508, 511-513, 516, 520-522, 524, 526, 529, 531-533, 537, 539, 543, 549, 557, 591, 596, 674, 676, 700, 703, 705, 711, 727

Funding, 34, 224, 594 Furniture, 577, 639, 643 Fusion, 630

G Galvanic skin response (GSR), 49 Gambler, 80

compulsive, 80 Games, 78, 81, 98, 224-225, 296-298, 313, 459,

571-572, 598, 688, 726 Gastrostomy tube, 471 Gaze, 298 Gender, 253, 636 General education, 78, 156, 183, 259, 261, 267, 443,

526, 570, 574, 576, 627, 632, 636-637, 649-651, 653, 686, 692-693, 706, 708, 712, 720, 727

general education classroom, 78, 156, 183, 570, 574, 576, 632, 650-651, 653, 686, 692

Generalization, 4-5, 8-10, 12, 14-18, 34, 59, 69, 93, 117, 123, 127, 153-154, 156, 216, 218, 230-231, 233, 238, 244, 261, 263, 270-271, 370, 396, 408, 410-412, 414, 424, 442-443, 445-446, 448, 484, 507-508, 524, 544, 546, 548, 557, 573, 589-590, 594, 613-614, 619, 622-646, 648-663, 682, 685-692, 694-699, 702, 704-706, 708-709, 714, 717, 722-725

enhancing, 691, 694 math instruction, 691 of behavior, 4-5, 8, 12, 14-18, 34, 59, 93, 117, 123,

127, 153-154, 156, 216, 218, 231, 233, 238, 244, 261, 263, 270-271, 370, 396, 410-411, 442, 446, 484, 507-508, 524, 589, 594, 619, 622-646, 648-663, 685-688, 690-692, 695, 698-699, 702, 704-706, 708-709, 714, 717, 722, 724-725

Generalizations, 656 Generativity, 695 Genes, 52 Genres, 306 Geography, 649 Gifted students, 581 Girls, 308, 335, 494-495, 574, 629, 646, 650, 655, 720 Gloves, 470 Glucose, 240-241, 418, 685, 695, 728 goal setting, 101, 562-563, 601, 608, 619, 690,

710-711, 719 Goals, 23, 39-41, 68, 75, 81, 83, 89, 130, 143, 217,

224-225, 257-258, 262, 264, 274, 336, 381, 417, 432, 562, 564, 567, 579, 581, 584, 587, 589-591, 594, 601, 616, 619, 664, 667, 670,

738

680, 684, 693, 709, 726 chart, 224, 601, 616 clarifying, 40 lesson, 417, 590-591

Goals and objectives, 667 Goal-setting, 562, 707 Gold standard, 126 Golden Rule, 668 Good Behavior Game, 210, 266, 386, 577-580, 688,

712, 719 Government, 23, 31, 42, 264

relations, 23, 31, 42 Grades, 23, 25-26, 81, 281, 283, 296, 328, 567, 590,

652, 702, 713 Grading, 73, 591, 646, 648, 670

level, 73, 670 multiple, 646

Graduated guidance, 230, 419-420, 425, 442 Graffiti, 116, 227, 249, 713, 726 Grammar, 20, 361, 537, 546, 610, 655, 688, 690, 710,

720 Graph, 2, 5, 9-10, 15-16, 31, 93, 141-142, 146-158,

164-171, 174, 176-177, 188-189, 191, 200, 242, 247, 267, 277, 331-332, 367, 373, 514-515, 530, 591, 601-602, 609, 616, 618, 657, 721

Graph paper, 168 Grapheme, 630 Graphic displays, 98, 142, 146-177, 185, 221, 249,

674 Graphics, 152, 169, 724 Graphs, 5, 146-154, 157-159, 164-172, 174-177, 188,

197, 210, 221, 236, 247, 259, 302, 367, 616, 690, 704, 715

Grasping, 50, 53 Greek, 585 Grid, The, 329-330 Grief, 25 Grooming, 574 Group cohesiveness, 577 Group instruction, 236, 267, 339, 477-478, 516, 670,

701 Group size, 217 Group therapy, 376 Grouping, 613 Groups, 11, 43, 164, 181-182, 184, 221, 227, 246-248,

253-254, 256, 263-266, 274, 336-337, 385-386, 411, 564, 566, 574-575, 579-580, 636, 642, 647, 652, 656, 694, 709, 721, 727

Groups:, 246 Growth, 36, 187, 253 Guessing, 235 Guidance, 183, 187, 195, 228-230, 307, 318, 361,

418-420, 425, 429, 431-433, 442-444, 448, 460, 466, 615, 675, 694, 710

guided notes, 180, 261, 718, 727 guided practice, 594, 626, 635 Guidelines, 7, 73, 76, 94, 96, 139, 165-166, 190, 235,

241, 243, 246, 278, 299, 305, 310, 345, 361, 363, 366, 369-370, 372, 375, 387, 431, 433, 435, 455, 460, 463, 467, 476, 482, 484, 490, 494, 501, 559, 563, 566-567, 571, 573-574, 580, 582, 584, 593, 603, 611, 615, 620, 648-649, 667-668, 671-673, 675, 683-684, 686, 688-689, 713, 728

Guidelines for Responsible Conduct, 668, 671-673, 684, 688

Guides, 36, 179, 194, 235, 418-419, 441, 683 Guilford Press, 687, 690, 700 Guilt, 64, 561, 595, 598, 603, 720

H Habituation, 9, 44, 50, 364, 372 Hand washing, 512, 522-524 Handicapped children, 158, 596, 701, 718, 727 Handling, 72, 80 Handwriting, 103-104, 254, 283, 421, 457, 706, 723

cursive, 103-104, 457 manuscript, 104, 457 problems, 457

Happiness, 125, 547, 698 Hardware, 119, 121, 543 Harlow, 429, 700 Harvard University, 690, 716-717 Head Start, 703 Health, 39, 42, 82, 95, 258, 470, 486, 588, 618,

630-631, 665, 673, 676, 678, 680-681, 684, 700, 703, 713, 717, 727

exercise, 42, 618

Health and well-being, 673 Health care, 676, 703 Health insurance, 681, 700 Health Insurance Portability and Accountability Act

(HIPAA), 681, 700 Health problems, 618 Hearing aids, 440-441, 724 Hearing disorders, 703 Hearing loss, 64 heart, 33, 49, 95, 195, 240, 257, 399, 588, 603, 617,

728 Height, 95, 165, 167, 176, 457, 464 Helping, 41, 80-82, 238, 293, 402, 593, 628, 634, 649,

651-652, 657, 661, 666, 677, 680, 692, 701 helping others, 666 Hemingway, Ernest, 603 Hierarchy, 420, 568, 581, 614, 620, 667 Higher education, 709 Highly qualified, 94 High-probability request sequences, 505, 705 Hippocrates, 368 Hippocratic Oath, 368 Histogram, 2, 152 History, 2, 4, 6, 9, 12-14, 17-19, 23, 27, 42, 44, 51-52,

54, 57-58, 60-61, 64, 66, 75, 90, 104, 140, 183-184, 201, 225, 279, 281-282, 284, 287, 289, 309, 315, 327, 335, 348-349, 363, 371-372, 380, 389, 393-394, 396-397, 399-404, 406-407, 410-412, 417, 424, 428-429, 432, 471, 476, 482, 500, 504, 508-509, 520, 524, 537-538, 540, 546-547, 550, 557, 584, 587, 599-600, 623, 635, 641, 644, 649-650, 666, 674, 695, 704, 715-716, 718, 721, 726

Holes, 330, 570-571, 700 Home, 40-41, 75, 78, 116, 166, 185, 240, 287, 293,

296-297, 313, 315, 320, 350, 364, 381-383, 403, 418, 437, 442-443, 472-473, 479, 489, 492, 502, 508, 531-533, 545-546, 559, 564, 571, 574, 577, 581, 585, 591-592, 594, 596, 614, 627, 629, 631-632, 635, 654, 659-660, 665, 670, 673, 679, 681, 683, 686, 694, 698, 708, 711, 714, 718

Homework, 48, 116, 183, 217, 291-293, 314-315, 330, 337, 463, 499, 562-563, 565-566, 572, 590, 592, 599, 604, 627, 685, 701, 712, 714, 719, 724

Homework assignments, 116, 183, 599, 701, 724 homework completion, 314-315, 562, 714 Homework performance, 562-563, 712 Honesty, 26-27, 413 Hope, 42-43, 94, 182, 189, 633, 643, 654 Horse, 264, 455, 460, 541, 565 Human development, 713 Human relations, 669 Human rights, 385 Human services, 71, 75, 80, 85, 368, 373, 697, 728 Humanistic education, 591 Humiliation, 676 Humor, 549 Hyperactivity, 101, 207, 254, 266, 296, 482, 524, 526,

562, 689, 705, 713, 719 Hypoglycemia, 240 Hypotheses, 23, 70, 235, 337, 470, 511, 514, 516,

519-525, 527-529, 532-534, 707 Hypothesis statements, 522, 525 Hypothesis testing, 235

I Id, 9, 32 IDEAL, 126, 180, 191, 237, 304, 320, 357, 369, 591,

651, 722 Ideas, 38, 95, 461, 682 identity, 14, 414, 458 ignorance, 189, 248 Illinois, 36 Illiteracy, 64, 537 Illness, 470, 676 Illustration, 104, 172, 198, 353, 386, 422-424, 435,

446, 449-451, 463, 474 Illustrations, 224, 250, 421 Imitation, 8-9, 13, 79, 108-109, 213-215, 418,

426-433, 444, 455, 479, 536, 541, 544, 548, 555, 557, 644, 687, 697, 702, 712, 725, 728

Immediacy, 54, 101, 279, 309, 337, 350, 387, 389, 427-428, 432, 608

Impairment, 64, 472, 492 Implementation, 69, 118, 130, 238-239, 241, 244, 256,

274, 282, 370, 379, 387-388, 448, 466,

485-486, 516, 562, 564, 570, 573, 579, 582, 589, 592-593, 601, 615, 672, 675, 691, 703, 719

changing criterion designs, 238-239, 241, 244 reversal design, 238, 244

Importance, 26, 33-34, 37-38, 40, 43, 45-46, 54, 60, 76, 78-79, 82-85, 88-89, 103, 124, 126, 130, 197, 210, 213, 215-216, 235, 238, 246, 249, 258-259, 268, 270-274, 279, 286, 293, 309, 313, 316, 369, 373, 375, 396, 403, 435, 464, 490, 507, 608, 615, 625, 631, 633, 636, 640, 646, 649, 659-660, 683, 702

Improvisation, 709 Impulse control, 16 Inappropriate verbalizations, 570, 711 Incentives, 703 Incidental teaching, 448, 552, 680, 691, 700, 710 Inclusion, 33, 217, 242, 269, 274, 559, 575 Inclusive classrooms, 701 Income, 81, 571, 665 Independent functioning, 72, 82-84, 90 Independent living, 436 independent practice, 81, 594 independent variable, 1-2, 5, 7-11, 13, 15-16, 18-20,

24, 26, 42, 146, 149-151, 153, 163, 167, 169, 174, 176-177, 178, 180-195, 197-198, 200-201, 206, 211, 218-219, 221-225, 227, 229, 232-238, 240, 242-244, 246-257, 264-265, 267, 269-274, 277, 280, 553, 587, 624, 727

Independent work, 526-527, 643, 719 Indexes, 117 Indiana, 30, 36 Indications, 142, 459 Indiscriminable contingencies, 10, 634, 644-648, 662,

697 Individual differences, 63-64, 66 Individualized Education Program, 570, 623, 642, 665 Individualized education program (IEP), 570, 665

target behaviors, 570 team members, 665

Induction, 624 Inductive reasoning, 195 Infants, 29, 287, 289, 429, 439, 687, 715

crying, 289 environment, 289, 439 hearing, 287 stimulation, 287, 289, 715

Inferential statistics, 235 Inflection, 462 Inflections, 104 Influence, 4, 10, 12, 23, 32, 38, 40, 42-43, 45-46, 48,

52, 63, 75, 93, 114, 119, 133-134, 144, 153, 180, 182, 186, 189, 193, 199, 215, 236-237, 252-253, 257, 273, 279-280, 283, 286, 315-316, 320, 327, 331-332, 341-342, 345, 350, 391, 417, 476, 480, 494, 500, 502, 575, 585, 588-589, 600, 613, 619, 624, 626, 651, 661, 664, 666, 677, 680, 683, 703, 705, 707-708, 711-712, 717-718

Information, 3-4, 8-9, 32, 45, 70-73, 75, 82-83, 85, 89-90, 96-97, 100, 102, 112, 125, 127, 130, 133, 137, 147, 150, 160, 163-165, 173, 177, 179, 181, 195, 200, 205-206, 212, 238, 240, 246, 250, 259, 265-266, 268, 282-283, 296, 329, 369, 381, 401, 407, 441, 511-512, 514, 516, 518-522, 524-527, 529-534, 537, 540, 543, 551-552, 555, 571, 592-593, 604-606, 615, 627, 633, 659, 671, 673, 675-679, 682, 684, 723-724

policies, 369, 381, 571, 677-678 Information processing, 32 Informed consent, 10, 370, 664-665, 673, 675-678,

684, 703 Inner speech, 593 Innovation, 33 Inquiry, 3, 23-24, 39-40, 89, 264, 371, 466, 665, 673

naturalistic, 40 In-service training, 206 Instruction, 2, 10-12, 16, 18, 34, 63, 69, 73, 75, 78, 80,

87, 102, 108-109, 117, 127, 148, 187, 190, 195, 211, 217-218, 228, 230-232, 236-237, 243, 267, 269, 283, 307-308, 318, 321, 323, 327, 336, 339-340, 342, 345, 378, 396, 407, 414-415, 417-420, 422, 425, 429, 441-442, 447-448, 469, 477-478, 487, 492-493, 502, 504, 506, 513, 516-518, 531, 541, 568, 577, 583, 587, 591, 593, 596, 599-600, 608, 612-613, 620, 623, 625-627, 629, 631-632,

739

634-642, 644, 648, 651-653, 657-663, 667, 670, 685-686, 688-691, 695, 697-698, 701-703, 705-706, 708, 711, 713-716, 718-722, 725

accountable, 689 adequate, 10, 318, 378, 541, 667 data-based, 447, 701 delivery of, 308, 327, 487, 504 differentiation of, 87 in preschool classrooms, 686 indirect, 78, 80, 283, 307, 642, 706 individualized, 623, 642, 648, 670, 708 learning strategies, 593 sequencing, 429, 627, 636 sheltered, 531 strategy, 2, 16, 18, 69, 187, 190, 195, 232, 323,

447, 613, 635-636, 640, 648, 652, 662-663, 688, 690, 697-698, 706, 718

targeted, 18, 78, 117, 127, 187, 195, 236, 307, 442, 516, 568, 577, 596, 620, 632, 660, 662-663

unit, 18, 102, 415, 442, 698 Instructional activities, 642 Instructional design, 636 Instructional environment, 625, 640 Instructional examples, 630, 635, 648 Instructional games, 81 Instructional objectives, 670 Instructional procedures, 211, 320, 413, 670 Instructional programs, 78, 642 Instructional strategies, 659, 682 Instructive feedback, 690, 702 Instrumentation, 24 Insulin, 240, 695 Insurance, 629, 678, 681, 700 Integration, 79, 272, 723 Integrity, 14, 18, 117, 121, 206, 245, 255-257, 267,

272-274, 455, 669, 672, 698, 715, 719 Intellectual disabilities, 654, 711, 718

moderate, 711 severe, 654

intelligence, 7, 73, 466, 537, 550, 716 Intelligence tests, 73 Intensity, 10, 12, 48, 56-58, 72, 83, 103-104, 107, 120,

306, 338, 347, 349-351, 355, 364, 372, 419-420, 456, 471, 520, 532, 699, 727

Intensity/magnitude, 350 Interaction, 2, 13, 40, 45-47, 53, 61, 65, 82, 110, 141,

179, 181, 203, 206, 227, 284, 313, 322, 376-377, 402, 407, 511, 517-518, 537-540, 550, 553, 556, 562, 642, 650, 666, 683-684, 692, 697, 706, 716, 723

Interactions, 4, 12, 36, 60, 70, 75-76, 78, 90, 100, 125, 132, 141, 166, 227, 254, 289, 295, 334, 338, 356, 380, 401-402, 407, 427, 455, 466, 483, 511-513, 539-540, 551, 561, 568, 576-577, 580, 584, 586, 625, 636-637, 673, 681-682, 684, 689-690, 697-698, 703, 705-706, 714-715, 719

Interference, 12, 196, 216, 218, 673, 711 intermittent schedules of reinforcement, 4, 10,

325-326, 334, 336, 341, 343, 366, 475, 480, 644-645, 662

Internal validity, 7, 10, 15, 178, 180, 194, 215, 250-252, 257, 267, 273-274

Internet, 95, 598 conduct, 95

Interobserver agreement (IOA), 3, 10, 122, 133, 136, 143, 268

Interobserver reliability, 700 Interpersonal relations, 40 Interpretation, 3, 12, 26, 29, 70, 96, 101-102, 129,

131, 134, 140, 150, 165, 174, 177, 247, 267, 269, 272, 274, 393, 395, 406, 519-520, 696, 705, 715

Interpreting, 20, 75, 90, 98, 135, 146-177, 202, 246, 248, 269, 501, 514, 522, 525, 527, 529, 532-533

Interrater reliability, 714, 720 Interspersed requests, 9, 505, 509, 702 Interval recording, 10-11, 13, 18, 20, 92, 98, 110-115,

120, 130, 135, 600, 700 Intervention, 1, 3-5, 8-9, 11-12, 15-16, 18, 29, 38-39,

65, 68-71, 74-77, 79-81, 83-84, 86, 88-90, 94-95, 99, 104, 117, 120, 127-128, 147-148, 150, 158, 161, 169-170, 179, 182, 185-187, 190, 195, 197-198, 200, 205-208, 210-213, 215-218, 224, 229, 232-234, 236-238, 240-241, 245-246, 248, 254-263, 269, 274,

278, 287, 291, 296, 303, 308, 316, 318, 334-336, 340, 346-347, 355-356, 358-359, 365-370, 372-373, 406, 412, 427, 434, 443, 453, 469-471, 474, 477-479, 483-486, 488-492, 494, 496-497, 498-509, 511-513, 518, 521-524, 526-529, 531, 533-534, 550-552, 554, 560-562, 566, 570, 572, 574, 577-578, 580-581, 584, 587, 592, 595, 598, 601-602, 604-605, 609, 611-617, 619, 623-626, 629-633, 636, 641, 648, 650, 652, 655-661, 663, 664-668, 674, 677, 681-684, 685-689, 691, 693, 695, 697, 701, 703-704, 706-711, 713, 720, 722-723, 726

Intervention:, 206, 366, 503, 522, 524 amount of, 503 discrete trial training, 562 facilitated communication, 94, 682

Intervention programs, 39, 69, 550, 554 Interventions, 22-24, 28-29, 37-40, 43, 58, 69-71, 73,

94, 100, 120, 123, 148, 151, 164, 169, 178, 196, 202, 206, 208, 210, 216-217, 219, 220, 233, 245-246, 250, 256, 258, 262, 267, 273-274, 282, 288, 295, 299, 310, 316, 318, 329, 333, 337, 342, 345, 347, 353-354, 356, 361, 366-373, 398, 417, 428, 448, 473, 482-484, 486, 498-509, 511-513, 523-524, 526, 533, 550, 559, 586, 592, 609-610, 615-617, 623-624, 629-630, 641-642, 644, 656-657, 661, 663, 666, 669, 671-674, 680-682, 685-686, 691, 693, 701, 706, 708, 710-711, 713-714, 717, 719, 721-728

antecedent-based, 498 behavior reduction, 367 consequence-based, 372

Interviewing, 71, 641, 708 Interviews, 9, 69-70, 72-73, 89, 380, 519-522,

524-525, 527, 531, 533 initial, 69, 73, 89, 380, 520, 522

Intonation, 537 Intraverbals, 16, 288, 536, 541-542, 548-549, 555,

557, 723 intrinsic motivation, 325 Introduction, 150, 169, 188, 190, 192-194, 198, 213,

218, 221, 226, 236-237, 242, 265, 271, 305, 326, 336, 338, 353, 362, 419, 428, 493, 594, 701, 709-711, 714-715, 717, 727

introspection, 27, 29 Investigator, 180, 182, 184-186, 188-190, 192-193,

204-205, 209, 215, 217, 224, 235, 242, 252-255, 264, 272, 505

Invitations, 683 Iowa Tests of Basic Skills, 73, 702 IPods, 654 Iron, 612 IRT, 6, 10, 16, 18, 92, 100-101, 103, 106, 110,

118-120, 136-137, 144, 334-335, 491, 493-497

caution, 101 uses, 6, 16, 136

Issues, 28-29, 35, 40-41, 43, 69, 119, 185, 207, 215, 260, 267, 270, 312, 345, 362, 367, 370, 381, 471, 571, 581, 607, 615, 665-666, 668-669, 675, 678, 684, 691-692, 698, 700, 708-710, 713, 721, 723, 725-726

controversial, 345, 362 Italics, 46, 57, 60 Items, 5, 18, 72-73, 101-102, 104, 136, 154, 156,

213-214, 226, 257, 259, 290-291, 296-297, 299-300, 302-303, 336-337, 353, 418, 449, 487, 500, 502, 520-521, 525-526, 552, 562, 564, 566, 568-574, 581, 585, 590, 595, 597, 602, 605-606, 611, 637, 641, 651, 657

specifications, 590

J Job satisfaction, 81 Joint, 52, 64, 647 Journal articles, 499 Journal of Applied Behavior Analysis (JABA), 36, 43,

674 Journal writing, 561 Journals, 27-29, 94, 168-169, 272 Judges, 86, 262 Judging, 36, 76, 82, 108, 142, 255, 369 Judgment, 37, 75, 88, 94, 120, 128, 131, 168, 271,

333, 463, 479 judgments, 76, 140, 148, 170, 176, 258, 709-710 Judicial decisions, 690 Junior high school, 184

Justice, 413, 669, 712 juvenile delinquency, 649

K Kansas, 36, 160, 162, 699, 715, 719 Kelley, 100, 155, 186, 201, 207, 250, 290, 359-360,

476, 483-484, 503, 562-563, 685, 693, 696, 705, 707, 711, 713, 725

Key terms, 22, 44, 68, 92, 122, 146, 178, 196, 220, 245, 276, 311, 324, 344, 374, 390, 408, 426, 434, 454, 468, 481, 498, 510, 536, 558, 583, 622, 664

Keyboard, 98, 682 Keyboarding, 88 Keystrokes, 128 Kicking, 460, 474, 479, 485-486, 489, 531 Kindergarten, 99, 291, 462, 578, 604 Kindergarten children, 604 Kindergarten students, 291, 578 Knots, 78 Knowledge, 10, 13, 23, 25-26, 29, 32-33, 38, 40, 42,

45, 47, 58-59, 65, 70, 83, 85, 93, 110, 120, 124-125, 127, 130, 143, 179-180, 190, 194, 256, 264-265, 267, 272, 274, 283, 340, 342, 369, 371, 373, 398, 401, 403, 466, 537, 548, 561, 575, 585, 587-588, 595, 632, 639, 641, 649, 658-659, 661, 666, 670-672, 674-678, 684, 689, 702, 707, 712

domains, 40, 712 of subject, 127, 264 prior, 58-59, 65, 124, 180, 190, 403, 632, 659, 666,

675 prior knowledge, 180 topic, 585, 632 vertical, 10

Knowledge base, 23, 26, 33, 340, 342, 371, 661 Kodak, 207, 265, 502, 705

L Labels, 81, 93, 120, 149-150, 167-169, 176, 505, 509,

586, 595, 670, 680 situation, 168

Laboratory research, 30, 185, 277, 290, 314, 371, 465, 492, 723

Language, 8, 13, 16, 20, 32, 34, 36, 38, 40, 42, 47, 56, 69, 78-79, 83, 135, 213-215, 235, 248, 250, 278-280, 282-283, 287, 296, 325, 362, 381, 402, 404-405, 407, 409, 414, 430, 455, 471, 498, 521, 536-543, 545-546, 550-557, 561, 566, 577, 589, 608, 617, 623, 626, 642, 649-650, 656, 674, 677, 682, 688-691, 693-694, 699-700, 703, 705, 708-709, 714-716, 719, 722-725, 728

acquisition, 32, 42, 78, 213, 215, 287, 498, 536-537, 540-541, 546, 550, 552-554, 626, 688-691, 703, 705, 708, 723-725, 728

animal, 32, 402, 404, 555, 725 body, 47, 250, 287, 430, 555, 557, 677 clear, 250, 402, 521, 538-539, 550, 623, 674, 677 delay, 16, 278-280, 283, 296, 608, 691, 709, 725 difference, 280, 542, 617, 656 empowering, 38 informative, 38 play with, 79, 278, 405, 649 playing with, 649 production, 79, 279, 455, 542, 642 receptive, 405, 414, 471, 539, 542, 545, 550-551,

555-556 written, 8, 16, 83, 414, 536-537, 539, 541-542, 545,

551, 556, 561, 714, 728 Language acquisition, 32, 42, 287, 498, 536-537, 546,

550, 723 Language arts, 414, 561, 577, 608, 649-650, 656 Language development, 78, 689, 699, 708, 714, 722

procedure, 689, 699, 708 use, 78, 689, 699, 708, 714, 722

Language instruction, 541 Language learning, 79 Language skills, 38, 40, 47, 78, 83, 296, 521, 551, 556 Language systems, 409 Languages, 409, 691 Large group, 183, 589 Latency, 5, 10, 13, 15, 17-18, 54, 92, 96, 100-103,

109-110, 119-120, 136-137, 144, 156, 243, 278, 282, 392, 405, 409, 424, 440-441, 455-457, 466, 552, 605, 643-644, 662, 727

Laundry skills, 229, 231, 724

740

Law, 11, 26, 29, 31, 83, 94, 181, 291, 324, 338-339, 667, 679, 690, 701, 725, 728

case, 701, 728 Law of effect, 701 Leaders, 42 Leads, 52, 63, 71, 79, 86, 133, 148, 240, 280, 283,

312-314, 322, 369, 588, 606, 609, 666, 680 Learned behavior, 30, 396, 427, 431, 538, 550, 589,

625, 639, 644, 662 Learners, 74, 215, 258, 267, 295-296, 306-307, 362,

402, 407, 417-419, 427, 429-432, 443, 455, 462, 492, 494, 496, 550-554, 570-575, 591, 604, 617, 631, 634, 639, 645, 648, 654, 661-662, 680, 691, 708, 714

active, 362, 431, 591 auditory, 604

Learning, 4, 9, 12, 18-19, 28, 30-32, 34, 39, 48-50, 58, 63, 66, 73, 78-80, 82-83, 88, 90, 94, 103, 110, 127, 160, 162, 180-181, 183, 211-212, 243-244, 249-250, 261, 263-264, 267, 277, 279-281, 289, 307-309, 312-315, 317-318, 320, 322-323, 325-326, 334, 343, 345, 361, 370-371, 373, 394, 397, 399-400, 403, 406-407, 414, 419, 429-431, 441-442, 444, 446, 450, 460, 493-494, 504, 537, 539, 541, 547, 556, 568-570, 584, 589-590, 592-594, 603, 605, 608, 619-620, 630-631, 633, 635-636, 641, 647, 649-654, 656, 661, 667, 670, 678, 685, 688-689, 691, 693-694, 698-702, 705-706, 708-709, 711-713, 715-716, 720, 722-724, 726-728

Learning:, 685, 702, 727 areas of, 702 connected, 110 discovery, 31, 181, 361 distance, 691 enjoyable, 589 events, 9, 18, 30-32, 48, 73, 78, 90, 94, 110,

180-181, 280-281, 312, 315, 320, 322-323, 325, 397, 399-400, 406, 539, 547, 584, 589, 594, 603, 619-620, 630, 641, 691, 709, 712

incidental, 32, 691, 700 mastery, 103, 243, 441, 670 observable, 31, 547, 589 readiness, 32 real-world, 263, 371 scenarios, 584 to learn, 34, 49, 73, 82-83, 103, 394, 429, 446, 450,

541, 593, 633, 635, 649, 667, 711 Learning activities, 649 Learning disabilities, 110, 183, 211-212, 261, 263, 460,

493, 537, 568-569, 590, 603, 605, 650, 653, 656, 667, 685, 688, 691, 693-694, 701, 708, 711, 720, 723-724, 727-728

Learning disability, 280, 608, 689, 699, 701, 706 Learning environment, 504 Learning environments, 82 Learning experiences, 9 Learning goals, 83 Learning groups, 647, 652, 656, 709 Learning opportunities, 79 Learning problems, 631, 641, 702 Learning process, 446 Learning rates, 243-244 Learning Strategies, 592-593 Least restrictive alternative, 368, 373, 387, 680 Lecture, 180, 261, 475, 599 Legal context, 676

and, 676 Legal issues, 571, 666, 675 Legibility, 165 Legitimacy, 94, 120, 123 Leisure, 36, 81, 83, 99, 227, 250-251, 291, 298, 353,

357, 385, 418, 487, 492, 499-500, 502, 521, 524, 526-527, 571, 575, 633, 654, 693, 719, 728

Lesson plans, 591 free, 591

Lessons, 114, 118, 529, 641, 697, 706 Let Me Hear Your Voice, 710 Letter formation, 457, 724 Letter writing, 598 Letters, 47, 76, 82, 104, 108, 197, 201, 305, 413-414,

421, 502, 542, 598, 623, 649, 687 Level, 1-2, 4-7, 10, 12, 15, 17, 19-20, 23-26, 34,

37-39, 42, 48, 57, 59, 64, 66, 70, 72-73, 75, 78, 81-83, 93-94, 97, 101-104, 111, 120, 125, 128, 130, 133-134, 140, 142-144,

146-150, 152-153, 166-171, 174-175, 177, 179-181, 185, 187-195, 198, 200, 205-206, 208-210, 212-213, 215-217, 221-225, 227, 229-231, 233-235, 237, 239-244, 246-249, 251, 253, 255-259, 261-263, 270-274, 278, 281, 291, 293-295, 304-305, 309-310, 334, 337, 346, 348, 364, 366, 368-369, 376, 381-382, 385, 387, 392-393, 397, 399, 405-406, 414, 416-418, 420, 430, 432, 437-438, 440, 443, 452, 455-460, 462-467, 471, 473-476, 491, 496-497, 514, 529, 531, 549, 558, 562, 568-570, 572-575, 580-581, 592-593, 599-600, 602, 604, 611, 633, 636, 650, 656-657, 659, 663, 670-671, 676-677, 689, 719-720, 722, 727-728

Lexicon, 537 Library, 296-297, 330-331, 507, 571, 584, 587 Licensing, 73, 668 Licensure, 668, 671

certification and, 671 Lighting, 7, 48, 75, 185 Likert scale, 3, 72, 520 Limitations, 41, 127, 136, 142, 168, 202, 208, 234,

238, 243-244, 264, 272, 289, 299, 304, 345, 361, 371, 418, 459, 465-466, 514, 516, 518-521, 534, 672

Limits, 53, 56, 89, 102, 236, 256, 335, 350, 439, 497, 588, 643, 664, 678, 683, 696, 700

Lincoln, Abraham, 126 Line graphs, 148-149, 164, 176 Line of best fit, 163 Lines, 10, 98, 149-150, 161-164, 167-168, 171-174,

176-177, 193, 231, 241, 292, 296, 377, 419, 421, 505, 564, 602, 613, 641, 658, 681, 687, 727

Linguistics, 537-539, 556 Liquids, 290, 431, 445 Listening, 109, 291, 382, 417, 459, 525, 565, 618, 715

discriminative, 417 Literacy, 550, 703, 712 Literature, 24, 39, 58-59, 62, 82, 96, 100, 110, 114,

135, 156, 164-165, 170, 179, 183, 190, 197, 199, 207-208, 221, 224, 239-240, 266, 271-272, 287, 295, 341-342, 346-348, 366, 369, 413, 427-429, 447-448, 463, 465, 470, 474, 480, 559, 568, 586, 615, 619, 624, 630, 666, 674, 681-682, 695, 709, 713

graphic, 156, 164-165, 170, 208, 221, 239, 674 Literature review, 272 Litigation, 381 Living environment, 369 Local Education Agency, 571 Locker room, 224 Locus of control, 584 logical reasoning, 190 Logical thinking, 591 logos, 653, 657 Longitudinal study, 562 Long-term goals, 83, 667 Loss, 15, 18, 64, 81, 95, 128, 170, 240, 270-271, 279,

333, 343, 345, 347, 349-350, 375-377, 382-383, 386-389, 513, 572, 584, 589, 608, 610, 617-618, 665, 676, 683

feelings of, 589 Lotteries, 330 Lottery, The, 330 Love, 65, 395, 725 Lunchtime, 284, 513 Lynch, D., 708

M machines, 128, 327, 636-639, 653, 657 Magazines, 571 Magnitude, 5, 10, 49, 54, 63, 92, 94, 103-104, 107,

118, 120, 132, 151, 171, 178-179, 193, 241-242, 244, 278, 293, 306, 310, 316, 322, 337-338, 348-351, 364, 369, 372, 386-387, 389, 392, 398, 405, 455-458, 463, 466, 473, 476, 480, 484, 643-644, 662, 695, 702, 707

maintenance, 3, 5, 8-11, 15, 20, 34, 58, 69, 78, 80, 82, 93, 123, 127, 129, 154, 156, 183, 197, 205, 261, 263, 270-271, 286-287, 314, 317, 320, 325, 362, 370, 448, 465, 484-485, 505, 507-508, 524, 546, 568-569, 573, 589-590, 613-614, 619, 622-663, 670, 680, 682, 685, 687-688, 695, 697, 702, 705-706, 708-709, 711, 717, 719-720, 724-725

of learning, 34, 317, 325, 619, 636, 708, 720, 724 Malapropism, 16

Malpractice, 95 Management, 16, 28, 34, 42, 75, 83, 200, 246, 293,

314, 378, 569, 573-575, 580-581, 583-620, 630, 634, 652, 654, 656-657, 659, 663, 673, 678, 686-687, 689-693, 696, 702-710, 712-714, 716, 718, 726

Mand, 9, 11, 213-215, 307, 405, 407, 448, 536, 539-545, 548-557, 689-690, 694, 711, 720, 723

Mandates, 575 Manual dexterity, 648 Manual prompts, 228 Marketing, 258 Marking, 108, 112, 134, 330, 560, 570, 586, 596 marriage, 69, 179 Mastery, 103, 125, 128, 243, 435, 437-438, 440-441,

452, 572, 670, 682 Matching, 11, 14, 17, 152, 154, 259, 324, 334,

338-339, 408, 414-417, 425, 431-432, 487, 555-556, 607, 620, 699, 702, 710, 713, 720, 723, 728

Materials, 80, 83, 87, 108, 110, 131, 199, 203, 216, 227, 257-259, 262, 277, 291-292, 297, 353, 355, 357, 416-417, 420-421, 439-441, 483, 485-487, 502, 511, 514, 528-529, 533, 571, 591, 593-595, 603, 619-620, 625, 630, 640, 642, 654, 670, 697, 710, 726

adaptation of, 83 Math calculation, 97 Math instruction, 691 Math skills, 76, 125 Mathematics, 24, 73, 97, 414, 463, 646, 701, 710 Mathematics performance, 646 matter, 7, 20, 24-27, 29, 31, 33, 35, 37, 45-46, 52, 56,

62, 83, 94, 104, 123-124, 126-127, 179-180, 185, 202, 221, 224, 246, 263-264, 277, 283, 286, 332, 539, 547, 572, 631, 646, 649, 661, 682, 697, 715, 721-722

Maturation, 202, 216, 242, 252-253 Mealtime skills, 82 Mean, 2, 7, 11, 19-20, 47, 55, 80-81, 83, 85, 97,

99-101, 104, 115, 122, 124, 135-137, 140-141, 144, 154-156, 170-171, 174-175, 177, 187, 195, 201, 210, 213, 224, 226, 229, 235-236, 242, 246-248, 259, 261, 265, 278, 286, 288, 293-294, 327-328, 332, 346, 350-351, 362, 377, 382, 386-387, 391, 418, 447, 464, 487, 491, 494-495, 530, 537, 539, 555, 561, 567, 577, 580, 601-602, 639, 650, 660, 666, 683, 687

Meaning, 7, 19, 25, 34, 37, 52, 62, 69, 88, 93, 95, 142, 148, 169, 176, 185, 203, 221, 231, 282, 286, 309, 326, 328, 345, 412, 429, 470-471, 479, 502, 504, 537, 539-540, 542-543, 547, 549, 624, 650, 677, 708

of words, 537, 539 shades of, 412

Meaningfulness, 76, 85, 264, 267 Meanings, 7, 10, 16, 62, 96, 135, 288, 539, 624 Measurement, 1, 3, 5-7, 9-14, 18-19, 24-25, 31, 36-39,

41, 43, 45-46, 68-69, 75, 86, 92-97, 99-102, 104, 108, 110, 112-121, 122-144, 147, 149-150, 158, 165-166, 168-170, 176-177, 179, 181, 185, 187, 189-191, 193-195, 209, 216, 218, 221, 223-224, 229, 231-240, 243-244, 246-247, 249, 252-254, 257, 262, 267-268, 271-273, 432, 475, 488, 491, 567, 599, 605, 662, 670, 691-692, 716, 727-728

attribute, 36, 45 in algebra, 567 interobserver agreement (IOA), 3, 10, 122, 133,

136, 143, 268 of time, 3, 5-7, 9-10, 13-14, 18-19, 75, 92, 94-97,

99-101, 104, 110, 112, 115-120, 128, 135, 144, 149-150, 165-166, 168, 170, 176-177, 187, 190, 216, 237, 244, 247, 267, 475, 488, 491, 605, 716

parameters of, 104, 127 proficiency, 19, 97, 101 social validity, 38, 68, 257, 262, 267, 727-728 terms, 6, 14, 18-19, 37-38, 45-46, 68, 75, 92, 94,

96, 101-102, 104, 110, 115, 120, 122, 135, 150, 166, 168, 170, 177, 185, 271, 567

variables, 3, 7, 9-10, 14, 19, 24, 36, 38, 41, 43, 45, 69, 75, 93, 104, 120, 128, 133, 140, 147, 150, 158, 165, 169-170, 176, 179, 181, 185, 187, 189, 191, 193-195, 216, 218, 223, 234-238, 240, 244, 246-247, 249,

741

252-254, 257, 267, 271-273, 432, 475, 692

Measurement error, 11-12, 124-125, 128-132, 143, 716

random, 11-12, 124 source, 11, 131-132

Measurement scales, 126, 128, 143 Measurements, 9, 15, 18, 93-94, 99, 110, 119,

123-124, 128, 132-134, 142, 147, 150, 155, 165-166, 169-170, 176, 185, 190, 243, 253, 491, 682, 722

cost, 15, 119, 142 Measures, 2-3, 5, 10, 12-13, 15, 17, 19, 25, 36, 70, 73,

81, 93-104, 109-111, 113-115, 117-118, 120, 123-135, 137, 139, 141-144, 147, 150-151, 153-154, 162, 167-168, 170-171, 174-177, 181, 183-185, 187-195, 197-198, 200-201, 209, 212-213, 216, 221-223, 226-227, 229, 231, 236, 240, 243, 246-247, 254, 257, 261, 264, 267, 298, 359, 364, 458, 462, 482, 516, 553, 573, 605, 617, 627, 659, 670, 693, 704, 716, 726

Measures of central tendency, 174 Media, 101, 571, 673, 715

examination of, 715 magazines, 571

Median, 100-101, 154, 163, 170-171, 174, 177, 590, 646

Mediating variables, 30 Mediation, 2, 287-288, 309, 366, 373, 419, 501, 509,

538, 556, 632, 677, 702 Mediators, 549, 612, 620 Medical history, 471 Medications, 254, 520 Meetings, 616, 677

telephone, 616 Memory, 32, 102, 418, 550, 687, 715, 728 Mental health, 676 Mental retardation, 34, 99, 108, 110, 149, 152-153,

156-157, 200, 212, 225, 231, 266, 301-302, 308, 320, 336, 351-352, 357, 359-360, 362, 411, 414, 418, 429, 438, 442-443, 445, 449, 474, 478, 483, 486, 492, 494-495, 502, 531, 574, 590, 601-602, 613, 635-636, 639, 642, 644, 650, 657, 676, 687, 689-693, 696-699, 701-708, 710-714, 718-720, 722, 727-728

Mentors, 674 Messages, 62, 94, 629, 682 Metaphors, 462, 537 Michigan, 36, 676, 693, 697, 699-700, 705, 709, 711,

718, 720 Microsoft Excel, 169, 690, 721 Middle school, 104, 183, 561, 656, 685, 714, 721, 728 Milieu teaching procedures, 726 Milk, 2, 24, 34, 277-278, 297, 330, 361, 421 Minnesota, 690 Misbehavior, 15, 18, 57, 346, 353, 356, 360-361, 372,

377-378, 381-382, 385, 388, 493, 496, 499, 579, 693

Misconceptions, 725 Missouri, 594 Mobility, 667 Mode, 8, 127, 427, 432, 541-542, 556 Modeling, 206, 218, 230, 233, 256, 284, 307, 355-356,

372, 418, 420, 425, 442-444, 448, 462, 572-573, 587, 639, 650, 652, 713, 727

of self-control, 587, 713 Model-lead-test, 579 Models, 32, 164, 168, 224, 235, 264, 388, 418,

427-433, 441, 443, 469, 628, 639, 650, 685 Modification, 28, 32, 82, 94, 120, 203, 215, 242, 383,

439, 445-446, 453, 640, 685, 687-692, 694, 696-702, 704-710, 712-714, 716, 718, 720-727

Momentary time sampling, 11, 13, 18, 92, 110, 113-115, 120-121, 516, 599, 603, 719, 724

Momentum, 3, 9, 498, 505, 509, 709, 713 Money, 60, 72, 270, 290, 307-309, 315, 325, 330, 339,

342, 376, 382, 386, 397, 439, 568, 570, 585, 605, 618, 623, 626, 628

Monitoring, 16, 69, 71, 75, 89, 94, 130, 238, 241, 244, 245, 351, 361, 389, 463, 493, 504, 524, 526, 528, 562, 575-576, 579, 581, 583, 590, 592-593, 598-610, 612, 615, 618-620, 650, 656-658, 680, 683, 685, 689, 692-693, 699-700, 705, 707-708, 710, 713, 719, 724, 727-728

progress, 16, 69, 89, 463, 562, 581, 603, 605, 680, 705

Monitors, 580 Morningside model, 703 Morpheme, 697 Morphemes, 537

grammatical, 537 Most-to-least prompts, 420, 425 motion, 48, 51, 54, 88, 286, 449, 457, 463, 529-530 MotivAider, 112, 528, 605, 696 Motivation, 9, 48, 79, 129, 278, 281, 309, 325, 357,

391, 400, 448, 476, 501, 504, 506, 509, 520-521, 523, 568, 574, 595, 614, 682, 692, 695, 712, 714, 720, 722-723, 728

achievement, 568, 714 intrinsic, 325 states, 48, 309, 400, 520, 568

Motivation Assessment Scale, 520, 692, 695, 714, 720, 722, 728

Motivation to learn, 682 Motor skills, 704, 721 Mouse, 14, 648 Movement, 9, 45-47, 50-51, 54, 56, 86-87, 118,

286-287, 298, 353, 359, 364, 418-420, 427, 430-431, 458, 461, 463, 465, 472, 540, 546, 548, 552-554, 557, 634, 680

Movement cues, 419 Movies, 507, 561, 565, 603

showing, 603 Multiple baseline design, 5, 11-12, 16, 215, 217,

220-224, 227, 229, 232-240, 242-244, 249, 601, 701-702, 726

Multiple disabilities, 125, 295, 625, 697-698, 705, 708, 717

Multiple probe design, 12, 190, 220, 229-232, 235, 243

Multiple-baseline designs, 185 Multiplication, 63, 78, 97, 170, 231, 463, 632, 635, 702 Multiplication and division, 97

sequences, 97 muscle relaxation, 17, 614, 620 muscles, 46, 49, 53, 472, 613 Music, 56, 80, 104, 116, 201-202, 284, 291, 295-296,

325, 353, 376, 382-383, 396, 428, 431, 478, 589, 640, 642, 654, 692

listening to, 291, 382 rock, 428

Music teachers, 478, 640 Myths, 686, 725

N Nail biting, 287, 342, 448, 686 National Association of School Psychologists, 671,

701, 713 National Association of Social Workers, 671, 713 National Institute of Child Health and Human

Development, 713 National Reading Panel, 78, 713 Native language, 287 Natural consequences, 670, 680 Natural environment, 1, 9, 14, 70, 73, 76-77, 79, 84,

86-87, 90, 129, 153, 259, 263, 274, 337, 366, 484, 491, 514, 516, 519, 522, 533, 547, 552, 575, 636, 639-640, 657, 680, 716, 723

Natural environments, 308, 340, 342, 366, 514, 666, 680

Naturalistic observations, 297 Nature, 1, 19, 23, 25, 33, 38-39, 47, 56, 60-61, 64, 93,

95, 102, 120, 123-124, 128, 133-134, 142, 144, 164, 169-170, 179, 181-182, 185-187, 206, 216-218, 221, 234-235, 238, 243, 248, 269-270, 278, 286, 289, 295, 309, 313, 320, 345-346, 371-372, 399, 402-403, 410, 456, 460, 467, 519, 548, 550-551, 566, 625, 645, 649, 661, 669, 676, 678, 688, 695, 698, 703, 720

Nausea, 554 needs, 38, 71, 74, 76, 78, 82, 89-90, 97, 99, 113, 174,

183, 188, 325, 351, 368, 386, 436-437, 446, 453, 478, 494, 497, 524, 540, 551, 555, 571, 588, 594, 609, 625, 633, 635, 639, 643, 646, 649, 666, 670, 685, 714

Needs assessment, 71, 82, 90, 685 Negative reinforcement, 6, 13-14, 44, 56-58, 62, 66,

69, 257, 260, 262, 278, 282, 311-323, 347-348, 355-356, 368, 371-372, 383, 410, 448, 469-472, 477, 479, 483-484, 501-502, 504, 509, 511-512, 515-516, 520, 561, 581, 603, 608-609, 620, 667, 688, 691, 702-703, 705, 707, 709, 711, 720, 722, 725

Nervous system, 708

Net, 186, 267, 385, 459 Neuroscience, 727 New Math, 81 New York, 439, 680, 686-702, 704-708, 710, 712,

714-716, 718-721, 723-727 News, 24 NICHD, 713 Nicotine, 588 Nintendo, 525 No Child Left Behind, 94, 725 No Child Left Behind Act, 94 No Child Left Behind Act of 2001, 94 Noise, 19, 75, 104, 152-153, 294, 312, 314-316, 364,

376, 404, 461, 479, 492, 500, 530, 572, 581, 640, 686, 702, 707, 710

Nominal scales, 691 Nonmaleficence, 669 Nonverbal behavior, 258, 539, 541, 543 Nonverbal behaviors, 63, 427, 543 Norm, 69, 73, 368, 670 Norming, 73

population, 73 Norms, 79, 368, 463 Notes, 74, 77, 101, 141, 180, 221, 261, 297, 428, 574,

577, 587, 599, 623, 681, 709, 718, 720, 727 anecdotal, 74

Notetaking, 261, 727 Notification, 388 Nouns, 32, 537, 540 Numbered Heads Together, 647 Numbers, 34, 72, 93, 130, 147, 156, 168, 184, 241,

248, 309, 329-330, 333, 377, 383-384, 464, 603, 627, 634

Nurse, 34, 43, 95, 568, 686 Nutrition, 52

O Obesity, 412, 588, 711 Object, 1, 7, 11, 14, 18-19, 33, 46, 50, 86, 93, 108,

149, 206, 226, 277, 287, 291, 299, 355, 391-392, 397, 405-407, 412, 427, 430, 461, 472, 502, 542, 553-554, 568, 570, 585, 595-596, 665, 674, 686, 696, 718, 720, 728

Object relations, 108 Objective, 6, 23, 25, 29, 32-33, 42, 70, 80, 83, 85,

87-89, 91, 93, 129, 133, 187, 195, 209, 255, 264, 412, 429, 432, 458, 464, 519-520, 567, 595, 609, 615, 623, 630-631, 633, 656, 659, 670, 682, 712

Objective observation, 6, 25, 42, 633 Objectives, 148, 206, 591, 608, 667, 670, 672, 675,

711 measurable, 670 observations of, 591

Objectivity, 25 Observation, 1, 3-6, 8, 10, 12-14, 16, 20, 23, 25, 29,

42, 46-47, 61, 68, 70, 73-75, 90, 96, 99-101, 110-117, 119-120, 122, 126-131, 133-135, 138, 140, 144, 150, 156-158, 169-170, 179-180, 184-185, 187, 195, 203, 234, 237, 257, 259, 268, 293, 295, 297, 300, 333, 364, 374, 376-377, 389, 392, 397, 442, 488, 516, 519-520, 529, 532-533, 552, 569, 576, 589, 598-599, 601, 626, 633, 641, 650, 652, 659, 661-662, 695-696, 700, 704, 711, 719, 723-724

focused, 150, 442, 519 of students with disabilities, 114, 576 quantitative, 10, 120, 724 self-monitoring, 16, 75, 598-599, 601, 650, 700,

719, 724 Observational learning, 444, 698, 727 Observational recording, 120, 140 Observations, 6, 8, 14, 23, 61, 69, 73, 75, 82, 89-90,

97-98, 126-129, 132-134, 143, 147, 157, 166, 170, 185, 187, 190, 197, 206, 229, 231, 239, 254, 258, 268, 271, 278, 297, 310, 502, 516, 521, 529, 532, 561, 570, 591, 641, 682-683

Observer training, 128-129, 143 Occupational therapist, 261 Office for Civil Rights, 700 Office of Special Education Programs, 594 Ohio, 35, 74, 87, 109, 111, 117, 210, 257, 260,

262-263, 422-424, 549, 604, 618, 640, 689, 701, 704, 708, 712, 721, 725, 727

Ohio State University, 35, 74, 87, 109, 111, 117, 210, 257, 260, 262, 604, 689, 701, 708, 712, 721, 725, 727

742

Older adults, 418, 685, 692 Ongoing process, 147 Onset, 6, 8, 13, 15, 50, 59, 96, 100, 110, 116, 120,

199, 278, 307, 314, 349-350, 372, 376, 387, 393, 397-399, 401, 404-406, 435-437, 445, 457, 474, 476, 478, 503, 568, 609, 690

Onsets, 312, 398 On-task behavior, 99, 110-111, 113-114, 127, 130, 320,

485, 590, 605-606, 609, 613, 689, 691, 693, 699-700, 705, 708, 728

Ontogeny, 4, 9, 12-14, 44, 52-53, 60, 63, 66, 689 Open-ended questions, 296 Openers, 636 Operant behavior, 2, 4, 8-9, 12, 15, 18, 28, 30-31, 34,

43, 44-46, 51-53, 55, 62-63, 66, 88, 278, 409, 536, 597, 607, 686, 692, 702-703, 712, 717, 721-723, 727

Operating system, 119 Operational definition, 255, 273 Opinions, 6, 25, 93, 120, 258-262, 274, 362, 667 Oral reading, 33, 94, 98, 100, 126, 128, 132, 212-213,

360, 695, 721 Ordinate, 149, 155 Oregon, 36, 637, 709, 727 Org, 162, 669-670, 685-686, 692, 706, 713 Organization, 63, 160, 162, 537, 580, 585, 587, 591,

667-668, 684, 694 Organizational behavior, 673 Organizations, 27-29, 368, 370, 667-668, 671, 673,

684 Organizing, 23, 45, 85, 592 Orientation, 33, 537, 584, 623 Orthography, 132 Osteoporosis, 418 Outcomes, 8-9, 19, 36-39, 52, 68, 76, 79-80, 83, 87,

94-95, 103-104, 115-116, 122, 127, 130, 140-141, 155, 170, 177, 179, 193, 250, 253, 257-259, 261-263, 270, 274, 282, 345, 387, 389, 417, 434, 448, 499, 511, 520, 562, 576, 586, 588-591, 594, 616, 623-624, 627, 629-634, 658, 660-663, 665, 667, 670, 674, 680-681, 684, 687, 698, 700, 704, 717, 726

predicting, 19 Outliers, 171 Outlines, 27, 104, 677 overcorrection, 13, 15, 216, 344, 356, 360-362,

364-365, 367, 372, 375, 610, 666, 676, 685-686, 691, 697, 707, 712, 714, 721-722

positive practice, 13, 15, 344, 360-361, 372, 610, 686, 721

restitutional, 13, 15, 344, 360, 372 Overlap, 69, 174, 206, 209-210, 213, 218, 435, 629 Overlearning, 637 Overweight, 588, 596 Oxymoron, 705

P Pacing, 46 Painting, 65, 103, 227 Palm, 50, 88, 108, 294-295, 357, 479 PANs, 601 Pantomime, 429 Paradigm, 29-30, 45, 52, 142, 270, 276, 689, 720 Parents, 2, 9, 36, 39, 42, 65, 70-72, 78-79, 83, 85, 90,

95, 133, 158, 187, 206, 210, 217, 224, 233, 238, 240-241, 256-259, 261, 270, 287, 293, 297, 325, 337, 360-362, 364, 378, 381, 383-385, 387, 409, 428, 432, 470-471, 473-474, 477-478, 482, 491-492, 519, 531, 533, 546-547, 552-553, 559-560, 562-565, 567, 576, 581, 593-594, 596, 609, 617, 623, 657, 660, 665-666, 670, 677, 682, 691, 693, 699, 701, 706, 713, 717-718, 720

expectations of, 79 involvement, 660 involvement of, 660 participation by, 83, 90, 706 response of, 259

Parsimony, 13, 16, 22, 25-27, 42, 179 Partial interval recording, 600 Partial-interval recording, 13, 18, 92, 110, 112-115,

120 Participants, 10, 18, 34, 36, 39, 41, 64-65, 83, 85, 94,

116, 129-130, 140, 150, 152-154, 156, 176, 183-184, 206, 213, 217, 221, 224-225, 227, 229, 232, 236, 243, 248, 259, 263-264, 271, 274, 284-285, 290, 300, 303, 306-307, 314, 316, 319, 326-329, 331, 333, 336-338, 341, 351, 353, 362, 364-365, 367-368, 389, 414,

418, 429, 483, 486, 490, 496, 505-508, 564, 567-571, 573-575, 581, 589, 601, 603, 606, 608, 617, 627, 636, 644, 646, 656, 673

debriefing, 673 Participation, 83, 90, 113, 267, 277, 369, 398, 431,

433, 446, 531, 562, 570, 677, 691-692, 697, 706, 710, 713

Past experiences, 128 Path, 2, 5, 19, 146, 149-152, 156-157, 166-171,

176-177, 179, 188, 190, 198, 200, 213, 217, 223, 267, 460, 712

Pathology, 95, 539 Patience, 648 Patterns, 63, 73, 98, 131, 142-143, 146, 158, 162,

169, 172, 175, 184, 187-190, 195, 201, 218, 266, 338, 515, 519-520, 522, 629, 641, 656, 688

number, 73, 98, 158, 184, 187, 195, 519, 629, 641, 656

patterns of behavior, 63, 98, 146, 522 Pavlov, Ivan, 30, 698 Paying attention, 117, 156, 356, 576, 604, 606 Peabody Individual Achievement Test, 73, 710 Peabody Picture Vocabulary Test, 694 Peak, 617 Peck, 97, 199, 207, 281, 382, 384, 387, 411, 415-416,

435, 686, 690, 692 Peer assistance, 656, 728 Peer interaction, 556, 716 Peer pressure, 577 Peer review, 674, 680, 683 Peer support, 692 Peer tutoring, 166, 462, 699, 701, 724

classwide, 701, 724 Pencils, 228, 359, 570 People with disabilities, 12, 40, 79 Perception, 183, 378, 537, 682 Perceptions, 169, 378, 717 Performance, 5, 9-10, 12-13, 16, 18, 29, 32, 34, 39,

48, 70, 73, 81, 87, 89, 91, 96-98, 101-103, 106, 115, 120, 125, 128-129, 132, 134, 140, 147-148, 151, 153-154, 156-158, 160-162, 165, 167, 169-171, 174-177, 181-183, 187-188, 190, 194-195, 200, 204, 209-210, 213, 216-218, 221, 223-225, 229, 231-234, 242-244, 246-248, 253-254, 256, 258-259, 267-268, 270, 273-274, 279, 282-283, 292, 305-308, 318-320, 322-323, 332-333, 335-337, 339-341, 343, 375, 381, 385, 389, 418-420, 425, 428, 430, 432-433, 435-436, 440-445, 448, 450, 452-453, 456-460, 463-465, 491, 504, 559, 561-563, 567-570, 573, 575-576, 578-582, 588-591, 599, 601-604, 606-609, 611-613, 615-616, 619, 625, 627, 630-631, 634-640, 642-648, 650, 652, 656, 659-663, 670, 678, 683, 687, 691, 693-695, 697-698, 703, 705-706, 710-712, 714-717, 719-721, 724, 728

level of, 10, 12, 48, 70, 97, 101-103, 120, 125, 128, 134, 140, 147, 167, 169-171, 174, 177, 187, 190, 194-195, 200, 210, 213, 216, 221, 223, 229, 231, 233-234, 242, 244, 246, 248, 253, 256, 258-259, 270, 273-274, 305, 381, 420, 432, 440, 459, 463, 491, 562, 580, 599, 602, 611, 650, 656, 659, 663

measurement systems, 101, 120, 128, 246 Performance criteria, 10, 91, 102, 568, 580, 582,

590-591, 604, 611-612, 615, 619, 638, 648, 670

performance feedback, 129, 256, 691, 694 Performance goals, 590 Performance standards, 590 Period, 3, 5-13, 17-18, 20, 34, 56, 74-75, 80, 88, 92,

96, 98-101, 105, 109-113, 115-117, 120, 126-127, 130, 134-135, 144, 148, 150, 156-158, 160-161, 163, 167-168, 170, 176-177, 186-187, 190, 195, 203-204, 209, 215-217, 219, 237, 240, 244, 247, 256, 258, 266, 268, 278, 281, 284, 293-294, 297, 299, 301, 326-327, 329, 345, 347, 349, 366, 375-378, 380, 382, 384-385, 402, 407, 410, 413, 436, 457, 460, 478, 488, 492, 506, 513, 516, 519, 522, 526, 529, 533, 562, 568, 570-573, 591, 599, 603-604, 606-611, 613, 644-647, 650-651, 657-658, 674

Periodicals, 578 Perpendicular lines, 10, 149, 176 Perseverance, 27, 128, 477

Persistence, 15, 448, 475-476, 624, 685, 690, 698 Personal experience, 190, 666 Personal relationships, 683 Personal space, 378 Personality, 73, 712 Personality inventories, 73 Personnel, 34, 83, 129, 568, 670, 698 Peterson, 32, 73, 95, 204, 250, 255-257, 264, 429,

510, 516, 549-550, 605, 680-682, 687-690, 692, 696, 701, 709, 715-716

Philosophy, 3, 12, 23, 27-29, 31-33, 39, 41, 43, 369, 538, 547, 585, 666, 682, 691, 701, 705, 710, 712

contemporary, 29, 43 Phobias, 17, 614, 620 Phoneme, 78, 307

segmentation, 78 Phonemes, 537, 541-542 Phonemic Awareness Skills, 78 Phonics, 246 Photographs, 63, 116, 263, 566, 654 Physical contact, 13, 293, 357-358, 376, 420, 540 Physical disabilities, 691, 708 Physical education, 376-377, 457-458, 463, 571, 719,

727 Physical environment, 116, 326, 417, 511, 540 Physical fitness, 463, 716 Physical proximity, 411 Physical restraint, 349, 364, 366, 507, 585, 721 Physical safety, 370, 680 Physical science, 35 Physicians, 34 Physiological factors, 520 Picture, 14, 17-18, 24, 69, 75, 114, 123-124, 126, 130,

136, 139, 143, 154, 174, 190, 206, 209, 213-215, 227, 231, 240, 248, 299, 414-416, 437, 461, 506, 508, 521, 533, 541, 547-548, 550, 553-555, 612, 632-633, 689, 694, 699, 709, 724, 726

recipes, 724 Pictures, 16, 154, 296, 299, 378, 414, 418, 502, 529,

547, 551, 566, 571, 596, 641, 653, 657, 667 PILOT, 460, 545 P.L, 704 Placement, 86, 168-169, 299, 450-451, 483, 574, 670 Planned ignoring, 13, 289, 374, 376, 389, 524 Planning, 69, 71, 83, 88-89, 148, 174, 180, 182, 187,

193, 195, 228, 245-274, 450, 476, 478, 592, 595, 631-632, 644, 662, 667, 720

inadequate, 265 of questions, 260, 264, 267, 274, 632

plants, 65, 225, 418, 618 Platform, 459 Plato, 95, 701 Play, 32, 48, 56, 64, 74-75, 79-80, 85-86, 98-99, 101,

108, 110, 131, 143, 153, 199, 203-204, 224, 257-258, 265, 267, 278, 281, 288, 291-292, 296-298, 301, 308, 315, 317, 322, 337-338, 383, 405, 427-429, 459, 479, 489, 493, 514-515, 522-523, 525-530, 532, 540, 543, 549, 554, 559, 563, 571, 573, 580, 584, 589, 629, 642, 646, 649, 652, 667, 674, 680, 686, 694, 702, 710, 712, 718, 726

absence of, 110, 288, 308, 315, 322, 427-429, 489, 589, 642, 674

blocks, 80, 86, 204, 308, 493, 629 pretend, 652

Play area, 338 Plays, 25, 56, 170, 209, 218, 271, 319, 327, 348, 394,

401, 409, 413, 428, 448, 537, 541, 543, 546, 553, 614, 644

Plot, 166, 168, 209, 463, 704, 724 Plots, 208 Poetry, 549 Point of view, 186 Pointing, 20, 52, 82, 98, 296, 346, 419, 427, 538, 553 Policies, 363, 369-370, 381, 571, 677-678 Policies and procedures, 369 Policy, 369-370, 373, 379-380, 666, 669 Political activity, 283, 286 Pooling, 249 Population, 52, 73, 181-182, 184, 246, 264, 274, 636 Populations, 83, 381, 674, 712 Positioning, 416 Positive behavior, 29, 278, 388, 523, 564, 578, 596,

610, 686, 707, 721, 723-724 Positive behavioral support, 718 Positive correlation, 350 Positive learning environment, 504

743

Positive Practice, 13, 15, 344, 360-361, 372, 610, 686, 721

positive practice overcorrection, 13, 15, 344, 360, 372, 610, 721

Positive reinforcement, 13-14, 18, 40, 44, 56-58, 62, 66, 69, 202, 213, 226, 276-310, 312-313, 316-320, 322-323, 347, 360, 368-369, 371, 373, 374-376, 381-382, 385, 388-389, 401, 407, 437, 460, 466, 470, 477, 479, 483-484, 490, 501, 509, 511-512, 515, 560-561, 584-585, 607-608, 676, 687, 716

schedules of, 13, 40, 57, 276, 296, 300-301, 308, 310, 368, 374

secondary, 66, 289, 437 Positive reinforcers, 18, 61, 315-316, 375, 378, 385,

387-389, 402, 407, 482 Postassessment, 430-431 Post-test, 444 Posttest, 102, 246-247, 253 Pot, 126, 601 Potential, 3, 10, 24, 39, 42, 48, 60, 69-70, 72-73,

75-77, 79, 82-84, 90, 100, 119, 128-129, 179-180, 184, 187, 194, 204-205, 209, 218, 229, 234, 237, 244, 250, 252-253, 256-257, 265, 267, 273, 278, 289, 294-297, 300, 303-304, 310, 337, 357, 361, 369, 371, 380, 385, 387, 389, 414, 446, 449, 466, 484, 490, 494, 508, 512, 514, 519-523, 549, 551, 564-565, 588-589, 591, 615-616, 640, 653, 660, 665-668, 671, 675-679, 683, 687, 696, 711, 723

Poverty, 64 power, 38, 190, 234, 248, 290, 293, 403, 414, 580,

607, 696, 722, 724 destructive, 696

Practice, 11, 13, 15, 19, 25-27, 31-32, 36, 40-41, 43, 59-60, 78, 81, 85, 94-96, 102, 108, 115-116, 118, 124-125, 128-129, 139, 143, 149, 164, 166, 175, 178, 190, 195, 202, 209, 211, 216, 224-225, 231, 234, 236-238, 242, 244, 249, 254-257, 264, 267, 272, 279-280, 282, 290, 299, 313, 318-319, 323, 325, 335, 337, 339, 344-345, 351, 360-361, 369, 372, 417, 428, 458-459, 461, 465, 478, 499, 516, 519, 524, 533, 561, 564, 574, 579, 583, 589, 591, 593-594, 606, 609-610, 612, 614, 617, 619-620, 626, 635-636, 639-641, 648-649, 652, 659-660, 662, 664-671, 674, 678, 680, 682, 684, 686, 689-691, 700, 702-703, 706, 710, 715, 720-721, 726, 728

acts, 118, 282 termination of, 94, 313, 524, 678

Practice effects, 13, 178, 190, 195, 216, 242, 254, 267 Praise, 23, 34, 56, 59-60, 86, 109, 132, 140-141, 156,

167, 183, 186, 203-206, 227, 230, 268, 282, 287, 290, 293, 295, 307-310, 315, 318, 325, 333, 350, 356-357, 361, 385, 402-403, 407, 430-433, 436, 442, 460, 483, 488, 491-492, 494, 505-506, 512, 526-527, 541-542, 553-555, 561, 569, 573-574, 580, 586, 589, 596, 600, 603, 608-610, 615-616, 618, 628-629, 635, 645, 649-653, 656-657, 660, 663, 691-692, 705, 712, 720, 723-724

Precision teaching, 28, 160-163, 688, 692, 706-708, 716, 727

Predicates, 537 Predicting, 15, 19, 26, 172, 190, 209, 213, 291, 310 Prediction, 2, 13, 15-16, 20, 23-24, 27, 29, 39, 42,

62-63, 93, 120, 178-179, 187, 189-195, 197-201, 209, 218, 221-223, 233, 235-236, 238-239, 250, 269, 273, 369, 547, 615-616, 699, 705, 722

Predictive validity, 699 Preference assessments, 154, 295-296, 299-301, 306,

310, 691, 693, 701, 708, 714, 719 prejudice, 683 Premack Principle, 13, 15, 276, 291, 297, 309, 561,

589 Preplanning, 580 Preposition, 710 Prepositional phrases, 537 Prepositions, 227, 537, 540, 552, 554 Preschool, 24, 34, 102, 108, 126, 203-204, 233, 289,

293, 308, 350, 412, 436, 444, 456, 477, 479, 482, 505, 612, 629, 650, 653, 673, 686-687, 689, 692-694, 697, 700, 711, 715-719, 722-723

Preschool children, 24, 34, 102, 108, 204, 233, 289, 308, 350, 505, 650, 653, 686-687, 692, 694,

715, 718-719, 723 Preschool classrooms, 24, 686, 697 Preschoolers, 250, 439, 505, 566, 590, 650, 687, 689,

691, 705-706, 714, 728 aggression, 714

Prescribing, 668 Presence, 6-7, 11-14, 17-19, 23, 26, 55, 61, 66, 75,

79, 90, 110, 113, 118-119, 127, 129, 149-150, 174, 179, 183-184, 186, 189, 191-193, 197, 209, 216, 223, 227, 230, 236, 242, 248, 253, 256, 267, 273, 281, 284, 295, 301, 303-304, 307, 310, 312, 314-316, 322-323, 347-349, 353, 355, 372, 393, 395, 397-398, 400-406, 409-412, 416-417, 419, 424-425, 435-437, 442, 444, 449-451, 453, 456, 458, 466, 500, 518-519, 525, 537, 540-541, 543, 546-549, 553, 557, 566, 589, 597, 616, 624-625, 627, 630, 639-640, 642, 644, 653, 662, 674, 680, 690, 692, 727

Presentation, 2, 6, 8-9, 12-13, 15, 19, 24, 47, 50, 56-57, 59-61, 63, 66, 78, 98, 102, 132, 154, 156, 169, 185, 200, 203, 217, 256-257, 278-280, 284, 287, 298-299, 301, 303-305, 309-310, 312-316, 320, 322-323, 339, 344-373, 401, 420, 427, 429-430, 432, 442, 451, 482, 503, 524, 570, 574, 597, 685, 693, 704, 715, 727

Presuppositions, 130 Pretest, 102, 129, 209, 246-247, 253, 430 Prevalence, 475, 707 prevention, 39, 42, 362, 511-513, 533, 694, 703, 711,

714, 727 Preventive maintenance, 129 Primary sources, 682 Priming, 460, 466 Principle of reinforcement, 59, 278 Principles of Behavior Modification, 688 Print, 76, 594 Printing, 168-169, 592 Privacy, 669, 680, 684 Private practice, 671 Probability, 1, 3-4, 9, 13, 16, 20, 23, 42, 56, 79, 163,

179, 181, 192-195, 208, 217, 242, 246, 255, 264, 270-271, 278, 291, 297-298, 309, 343, 348, 356, 378, 382, 389, 451, 456, 469, 479, 498, 501, 504-506, 509, 510, 518-519, 527, 585, 587-588, 615, 619, 640, 642, 686, 693, 705, 721, 727

Probe sessions, 229-230, 263, 626, 638-639 Problem behavior, 4-6, 8-9, 12-13, 15, 20, 24, 29, 63,

82, 84, 99, 152-153, 170, 198-199, 227, 250-251, 257, 259-260, 262, 316-318, 320, 335, 340, 346, 351-352, 354-355, 357-367, 370, 372, 375, 401, 403, 406-407, 469-472, 474, 476-478, 480, 482-494, 496-497, 500-504, 506-509, 511-516, 518-534, 608-610, 613, 620, 686, 695, 697, 700, 704, 706-707, 711, 714, 717, 722, 724-725

self-injurious, 153, 170, 262, 351-352, 354, 359, 362-365, 367, 471-472, 478, 485-486, 489, 492, 494, 497, 501-502, 506, 511, 513, 700, 706-707, 722

stereotypic, 250, 320, 359-360, 489, 494, 707 Problem solving, 15, 102, 409, 537, 592, 594, 616,

710 Problem-solving, 592, 702 Procedural Safeguards, 369-370, 373 Procedures, 3, 9, 12, 14, 16-17, 28, 31, 33-34, 37-40,

42-43, 58-59, 62, 66, 71, 73, 75, 79, 83, 85, 89-90, 92, 94, 99, 103, 108-110, 114-115, 118-120, 122-124, 126-132, 134, 142-143, 150, 156, 167, 179, 184-185, 195, 211-216, 228, 246, 250, 255-256, 258-262, 264-268, 270-274, 276, 284, 289, 293-295, 297, 299, 303-304, 307, 310, 311, 318-320, 324-325, 327, 329, 333, 335, 343, 344, 361-370, 373, 374-375, 379-382, 384-385, 388-389, 402-403, 407, 408, 410-414, 416-422, 425, 429-431, 434-437, 442, 446-448, 452-453, 454, 459-460, 465-467, 468-471, 475-477, 479-480, 481-483, 485-489, 491-497, 498-501, 503, 507-509, 513, 516, 519, 521, 524, 530, 534, 536, 547-548, 551-556, 558-560, 564, 568, 570, 572-573, 576, 581-582, 583, 586-587, 589, 592, 594, 601, 603, 606-607, 609, 611-615, 618, 622, 624, 626, 634-635, 638, 641-642, 646-647, 656, 661, 663, 665-667, 669-670, 672, 674-678, 680, 683, 686-689, 691-692, 696-699,

702-704, 706, 708, 711-712, 714-715, 717-718, 720-721, 723-724, 726-728

process of teaching, 541 Processing, 32, 537 Product, 2, 4, 8, 11-13, 18-19, 24-25, 40-41, 49, 51-52,

59, 64, 66, 79, 81, 92, 97, 101, 115-116, 118-119, 121, 133-135, 142-143, 147, 176, 254, 266, 280, 284, 286-287, 289, 309, 325, 348-349, 415, 424, 437, 452, 455, 466, 488, 541-542, 550, 556-557, 567, 587, 594, 606

Productivity, 39, 111, 118, 156, 183, 254, 350, 388, 590-591, 598, 601-602, 605-606, 613, 653, 689, 691-692, 698, 700, 707-709

Products, 19, 24, 61, 63, 81, 86-87, 115-119, 121, 129, 133-134, 139-140, 142, 254, 287-288, 360, 544, 566, 587, 606, 624, 689, 703, 712

Professional associations, 7 Professional behavior, 684 Professional boundaries, 683 Professional competence, 664, 671, 674, 684 Professional development, 664, 672, 674 Professional ethics, 713 Professional issues, 35 Professional organizations, 27, 368, 370, 667-668,

671, 684 Professional Psychology: Research and Practice, 710 Professional relationships, 677 Professional standards, 68, 664, 667-668, 674, 684,

706 Professionals, 80, 83, 85, 368, 381, 666, 671, 678,

682-683, 701, 720 Profiles, 674 Program effectiveness, 88 Program planning, 83 Programmed instruction, 34, 689 Programming, 14, 154, 156, 215, 308, 335, 366, 466,

493, 501, 569, 611, 622, 627, 631, 636, 640-642, 644, 647-648, 652-654, 657, 660, 662-663, 667, 677, 685-686, 690-691, 693-695, 697, 704, 711, 717, 720, 723

Programs, 20, 29, 34, 36, 39-41, 56, 69, 71, 76, 78, 80-81, 83, 85, 89-90, 94-95, 116, 168-169, 243, 256-258, 272, 292, 294, 312, 322, 325-326, 345, 369, 379, 386, 402, 405, 407, 414, 432, 536, 550, 554, 556, 559, 569, 574-576, 586, 589, 591-594, 609, 611-612, 615-616, 620, 639, 642, 656-657, 660, 663, 666, 669-672, 680-681, 686, 688, 702, 712

regular classrooms, 574 remedial, 69, 686, 712

Progressive time delay, 420, 701 Project, 162, 308, 422-424, 483, 595, 608, 616-617,

704 Projection, 162-163, 594 Projects, 64, 161, 296, 577, 591-592, 611, 616, 708 Promotions, 328 Prompts, 5, 16, 46, 102, 133, 227-230, 287, 291, 307,

310, 359, 361, 417-420, 422, 425, 429, 431-433, 443, 445, 448, 450, 452, 455, 471, 493-494, 500, 502, 507, 514, 527, 543, 552-556, 573, 586, 595-597, 604-605, 612-613, 618-620, 628-629, 642, 650, 652-654, 656-657, 660, 663, 688-689, 692, 694-695, 697, 701, 713-714, 716, 723, 726

topics, 697, 723 Property, 16, 45, 56, 60-62, 149-150, 263, 281, 286,

306, 320, 412, 428, 479, 492, 513, 524-526, 547, 550, 557

Propositions, 93 Prosocial behavior, 576, 721 Prosody, 537 Protecting, 677-678 Protocol, 6, 118, 240, 297, 427, 429, 433, 679, 686,

716 Psychiatric disorders, 336, 492 Psycholinguistics, 537-538 Psychological measurement, 691 Psychologists, 30, 668-669, 671, 681, 684, 685-686,

699, 701, 713 Psychology, 27-29, 33, 42, 57, 101, 181, 184, 203,

246, 261, 264, 293, 391, 471, 516, 537, 539, 578, 671, 685, 687, 689-691, 693-700, 704-722, 724, 726-727

social psychology, 471, 698, 712, 727 Psychotherapy, 31, 707, 728 Psychotic disorders, 34 Psychotropic medication, 676 Publication Manual of the American Psychological

Association, 685

744

Publications, 379, 439, 445, 640, 674, 705, 713, 727 Publishing, 34, 36, 281, 462, 673, 688, 705 Punctuation, 561 Punctuation marks, 561 Punishment, 2, 4-5, 8-9, 12-15, 18-19, 34, 44, 53-54,

56-59, 61-62, 64, 66, 76, 82, 200-201, 206, 244, 250, 266, 270, 272, 281-283, 293, 313-315, 322, 338, 344-373, 374-389, 390-391, 393, 396-400, 406, 459, 462, 470, 482, 487, 491-492, 499-500, 513, 516, 533, 536, 544, 557, 594, 603, 607-610, 616, 620, 631-632, 662, 666, 672, 686-687, 689-691, 694, 696-699, 702, 707-708, 710, 712-714, 718-719, 722, 724-726, 728

Punishment procedure, 8, 14, 359, 361, 364, 368-370, 373, 375, 385, 389, 390, 393, 396-398, 406, 491

Purchasing, 591, 595 Puzzles, 716

Q Quadriplegia, 667 Qualifications, 54, 278, 678, 720 Quality, 16, 65, 93-94, 101, 122-144, 255, 265, 267,

274, 284, 304, 306, 310, 327, 337-338, 364, 372, 412, 455, 466, 476, 480, 502, 506, 551, 587, 595, 601, 619, 643, 648, 650-651, 659, 680, 695-696, 702, 712-713, 720

Quality control, 128, 712 Quantities, 13-14, 17-18, 95-99, 101, 115, 120, 187,

337 Quantity, 13, 19, 93-95, 101, 103, 120, 170, 184, 194 Quasi-experimental designs, 690 Questioning, 42, 190 Questionnaires, 9, 71, 90, 258, 274, 519-520, 533 Question(s), 249, 273 Questions, 7, 22-23, 31, 33, 35, 37, 41, 45, 68, 70-71,

73-76, 79-80, 82, 88-90, 92-93, 95, 100, 108-109, 116, 118, 120, 122, 124, 126, 139-140, 142, 146, 169, 177, 178-180, 193, 195, 196, 216, 220, 227-229, 235, 238, 245-246, 250, 257-260, 263-264, 267-270, 272, 274, 276, 296, 311, 324, 344, 371, 374, 386, 390, 401-402, 407, 408, 417, 426, 434, 454, 466, 468-469, 477-478, 481, 485, 494, 498, 510, 520-521, 526, 528, 536, 539-540, 542-543, 551-553, 555-556, 558, 571-572, 583, 602, 605, 612, 615-616, 625-626, 632-633, 642, 650, 665-668, 673, 675, 677, 679-683, 715, 725, 727

closed, 169, 626 easy, 75, 108, 238, 528, 551-552, 555, 571, 632,

725 formulating, 70 generating, 572 harder, 417, 555 inferential, 235 investigating, 180, 238 leading, 89 leads, 71, 79, 666, 680 moving on, 551 on tests, 269 poor, 124, 296, 417, 668 scaling, 169, 267 subject matter of, 539 what if, 126, 193

Quizzes, 180, 332 Quotation marks, 654

R Race, 29, 339, 588 Radiation, 409 Radical behaviorists, 33 Radio, 201, 362, 618 Random numbers, 329, 333 Random selection, 211 Range, 3-4, 8, 13, 17, 19-20, 42, 50, 52, 54, 63, 66,

69, 72-73, 75-76, 79, 89, 97-98, 100-101, 103, 106, 116, 124, 128, 135, 137, 140-142, 148-149, 154, 165-171, 174, 176-177, 186, 188-189, 195, 209-210, 229, 241-242, 246, 249, 259, 261, 267, 273, 278-279, 287, 290, 305, 315, 322, 328, 350, 362, 372, 379, 446-448, 456, 458, 460, 482, 484, 486, 489, 515, 521, 531, 561, 590, 595, 601-602, 616, 629, 634-639, 641, 646, 650, 654, 662, 666, 675, 713, 727

Rape, 728

Rash, 512, 546, 555 Rates, 3-6, 8, 10-11, 16-17, 19, 23-24, 66, 82, 96-100,

109-110, 119, 126-127, 138-139, 144, 149, 156-157, 160, 173, 206-207, 211, 226, 240, 243-244, 268, 282, 284, 289, 291-293, 302-303, 310, 316, 318, 320, 324, 327-329, 331-332, 334-336, 340, 343, 355, 357-358, 362, 366, 379, 383, 471, 481-482, 485, 490-494, 496-497, 503, 508, 514, 522, 526, 529, 581, 590-591, 601, 606, 613, 636, 708, 726-727

Rates of learning, 211 Rating scales, 9, 72, 519-521, 533, 688 Ratio scale, 159 ratio schedules of reinforcement, 326, 335 ratios, 14, 96, 101, 327-330, 571, 590 Reaching, 33, 79, 307, 331, 337, 348, 404, 418, 540,

551, 606, 638-639, 674 Reactivity, 12, 14, 68, 75, 90, 119, 122, 130, 143, 164,

237, 254, 268, 599, 689, 700, 704-705, 708, 713

reactivity to measurement, 254 Readiness, 29, 32 Reading, 33, 45, 53, 55, 69, 73, 78-80, 94, 96, 98,

100, 102, 109, 116, 124, 126, 128, 132, 160, 166, 179, 181, 211-213, 216, 239, 243, 280, 282, 291-294, 297-298, 306, 318-319, 330, 360, 382-383, 414, 463, 538-539, 542, 556, 566, 570, 574, 578-579, 590, 598, 605, 608, 632, 634, 643, 646-650, 655, 667, 674, 686, 689, 692, 694-695, 704-705, 713, 719-721, 728

acceptable, 128, 463, 598, 667, 695 assisted, 705 cumulative, 166, 211, 319 difficulties, 280, 719 effective instruction, 69, 318 pointed, 69, 538 to students, 73, 294, 632, 694, 705, 721 wide, 73, 79, 98, 116, 578, 590, 632, 634, 646, 650

Reading comprehension, 109, 132, 181, 542, 605 Reading fluency, 128, 695 Reading instruction, 713 Reading performance, 216, 319 Reading skills, 73, 78, 566 Readings, 704, 709, 719 Reasoning, 1-2, 13, 16, 20, 27, 183, 187, 190, 192,

194-195, 216, 232, 236-237, 247, 249, 264, 269, 280, 309, 357, 585, 619

Recall, 221, 315, 401, 520, 596, 647, 667, 689 Receiving, 4, 18, 46, 273, 281, 289, 300, 302, 325,

330-332, 338, 345, 376, 402-403, 405, 407, 429, 441, 459, 572, 608, 613, 631, 639, 641-642, 645, 648, 651, 657, 659, 669, 677, 684

Receptive language, 405, 471, 539, 542, 545, 556 Recess, 376, 382, 385, 387-388, 436, 523, 577-578,

649 Reciprocal inhibition, 728 Reciting, 625 Recognition, 421, 653 Recommendations, 68, 128, 139, 164-165, 371, 450,

476, 502, 611, 665, 685, 717, 726 Reconstruction, 28, 728 Record keeping, 669 Recorders, 129, 131, 183, 598 Recording, 1, 7, 10-11, 13, 18-20, 26, 68, 73-74, 90,

92, 98-100, 104, 108-117, 119-121, 122, 127-130, 132, 135-137, 140-142, 144, 150-151, 157, 161, 164, 168, 184-186, 199, 257, 268, 293, 297-298, 440-441, 458, 516-519, 526, 533, 537, 560, 562, 570, 578, 580-581, 590, 598-601, 603-604, 606, 608, 618-619, 650, 657, 689-690, 692, 695, 700, 705, 707-710, 712, 717, 724

Recordkeeping, 592 Records, 1, 5, 13, 16, 20, 73, 97, 110, 112-113, 117,

121, 134-135, 138, 148, 155-158, 176, 297-298, 316-317, 381, 388-389, 516, 565, 580, 586, 598, 602, 615, 672, 679-681, 684

strength of, 316, 580 Recruitment, 508, 608, 650-652 Recycling, 86, 116, 585, 690 Redirection, 359, 362, 442, 507, 699 Reference, 3, 45-46, 51-52, 56, 61-63, 65, 87, 96, 98,

100-101, 189, 278, 286, 399, 586, 619 Referral, 71, 240, 673, 681 Referrals, 664, 672 Reflecting, 669

Register, 35 Regression, 172, 177 Regrouping, 78, 97, 528, 603, 635, 694 Regularity, 72, 506 Regulations, 673, 677, 681 Rehabilitation, 376, 686 Reinforcement, 1-20, 28, 34, 40, 44, 52, 54, 56-62, 64,

66, 69, 75-80, 82, 84, 90, 100, 103, 108, 111, 130, 155-157, 159, 167, 183, 186-187, 195, 200-206, 212-213, 218, 221, 226, 229, 233, 239-240, 242, 244, 248, 250-251, 253-254, 257, 259-260, 262-263, 270, 276-310, 311-323, 324-343, 345-348, 350-364, 366-369, 371-373, 374-383, 385-389, 391-407, 409-417, 424-425, 427-429, 432, 435-437, 442-443, 445-446, 448, 450, 452-453, 454-456, 458-460, 462-467, 469-480, 481-497, 498-504, 506-509, 511-516, 518, 520-524, 527, 530, 533-534, 536, 538-544, 546, 548-549, 551-557, 560-561, 568-569, 571-576, 580-581, 584-590, 594, 601-603, 606-613, 616, 619-620, 622, 624, 626, 628-635, 641-650, 652-656, 658, 660, 662-663, 666-667, 672, 674, 676, 680, 685-728

consequences and, 469 differential, 2-6, 8, 10, 13, 15-17, 20, 59, 66, 100,

103, 186, 195, 218, 248, 250-251, 304, 307, 310, 318, 321, 324, 334-336, 340, 343, 353, 362, 366-367, 369, 373, 379-381, 385, 393, 406, 410, 412-413, 416, 424-425, 454-456, 458-459, 465-466, 477, 481-497, 499-501, 503, 506, 509, 524, 538, 540, 549, 551-552, 569, 580-581, 586, 601, 606, 616, 655, 666, 674, 690, 692-694, 705, 707-709, 715-718, 726

extinction, 4-8, 14-17, 44, 54, 57, 59, 66, 186, 289, 303-305, 310, 325, 331, 336, 340, 343, 346, 348, 353, 357-358, 366-367, 369, 373, 379, 381-382, 392-393, 396, 401-407, 411-412, 455-456, 459, 463, 466, 469-480, 482-483, 485-486, 490, 497, 499-502, 504, 507-509, 512, 522, 524, 538, 553, 611, 624, 628, 644-646, 688, 693, 696, 698-699, 702-704, 706-707, 710, 715, 717-719, 722, 726-727

of alternative behaviors, 367, 477, 500 of incompatible behaviors, 64, 693 thinning of, 351

Reinforcement history, 2, 75, 90, 389, 500, 540 Reinforcement ratio, 324, 326 Reinforcement schedules, 14, 108, 159, 300, 310,

334, 342-343, 465, 501, 576, 692, 702, 704, 707, 713, 725

Reinforcers, 2-4, 9-10, 13-15, 17-19, 47, 53, 56, 58-61, 66, 69, 76, 78-80, 83, 90, 101, 253, 276, 278, 280-282, 287, 289-291, 293-297, 299-303, 306-310, 311, 315-316, 322, 328, 336-337, 340, 345, 347, 349-350, 353, 364-367, 375-378, 380, 382, 385-389, 391-398, 402-403, 405-407, 416, 425, 431, 433, 435, 462, 468-470, 476, 479-480, 482-485, 490, 496, 500-503, 507-508, 511, 520, 523, 530, 533-534, 539-540, 546, 550-553, 561, 568-574, 577, 580-582, 588, 591, 594-595, 602, 610, 631, 633, 635, 643, 647-650, 657, 660, 663, 672, 681, 685, 691, 694-696, 698, 708, 717-718, 722, 725, 728

reinforcing successive approximations, 455, 458 Rejection, 298, 715 Relapse, 707 RELATE, 340, 377, 409, 473, 515, 526, 533, 538, 579,

671 Relationship, 2, 4, 17, 23, 72, 77, 84, 95, 125, 143,

149, 254, 264, 270, 279, 301, 340, 361, 377, 400, 409, 412-413, 425, 436, 453, 461, 559, 566-567, 680, 683, 709, 712, 719, 721

benefit, 680 Relationships, 7, 29, 41, 62, 101, 147-148, 164, 176,

264, 303, 339, 552, 664, 672, 677, 683-684, 706, 727

healthy, 29 relaxation, 17, 287, 597, 614, 620 Releasers, 32 Reliability, 14-15, 17, 26, 36, 42, 119, 122-125,

128-133, 142-144, 172, 193, 195, 256, 265-266, 268, 271, 274, 520, 688-689, 692,

745

700, 704, 708, 714, 720, 722, 728 illustration, 172 interrater, 520, 714, 720 meaning, 142, 708 measurement error, 124-125, 128-132, 143 of assessments, 142

Religion, 31, 537, 668 Remembering, 57 Reminders, 595, 604 Repeated reading, 282 Replacement behavior, 8, 80, 259, 320, 323, 482, 524,

534 Replay, 116, 133 Replication, 2, 6, 13, 15-17, 20, 22, 25-27, 42, 85, 178,

186-187, 191, 193-195, 197-199, 201, 209, 218, 221, 223, 229, 233, 235-236, 239-240, 245-246, 248, 250, 265-266, 269, 271-274, 358, 384, 682, 686, 689, 697-698, 700, 714, 717, 722, 724

Reporting, 5, 96-97, 99, 126, 132-135, 137, 139-142, 144, 200, 267, 616, 670, 678-679, 689

Reports, 12, 37, 82, 102, 125, 130-131, 142-143, 148, 182, 209, 296, 320, 338, 362, 457, 607, 615, 670, 672, 682, 689, 692, 698, 708, 713-714, 721

library, 296 Representation, 51, 95-96, 100-101, 126, 130, 142,

155, 157, 166, 169, 171, 184, 208, 267, 450, 687

Representations, 50, 57, 664 Representativeness, 267 reproduction, 14, 49-50, 53, 65, 394 Research, 7, 10, 13, 16-17, 20, 23-24, 26-28, 30-31,

34, 36-37, 40-43, 50, 54, 59, 63, 75, 78-79, 82, 89, 94, 101, 113, 117, 124-126, 130-131, 133, 139-140, 142, 147-148, 155, 160, 162, 174-175, 179-187, 189, 193-195, 197, 207, 213, 224, 233-235, 238, 245-274, 277-279, 290, 293, 305-306, 312-314, 317, 320, 322, 326, 329, 340-343, 345, 348, 350-351, 356-357, 363, 367, 369-373, 383, 400-401, 414, 419, 421, 439, 446, 448, 465, 469, 473-474, 476, 492, 499, 504, 508, 511, 514-516, 519-520, 538, 551, 554, 575, 586, 590, 598, 605, 612, 614-616, 628, 636, 645, 650-651, 654, 661, 666, 669, 671, 673-675, 677, 681-682, 686-711, 713-728

findings, 7, 17, 23-24, 26-27, 34, 36, 40, 42, 63, 148, 180, 184, 193, 246, 255, 263-267, 270-271, 274, 277, 342, 350-351, 673-674, 682, 697

instrumentation, 24 on writing, 718 sampling, 10, 13, 20, 113, 126, 140, 264-266, 516,

716, 719, 721, 724 scholarly, 30, 598, 616 theory and, 13, 41-42, 688, 695, 703, 710

Research design, 233, 715 Research literature, 207, 448, 666, 674, 681, 713 Research methods, 139, 197, 246, 263-264, 692, 697,

715, 723-724 Research on teaching, 266 Research reports, 37, 82, 131, 142, 708 Research-based methods, 682 Resistance, 3, 15, 186, 359, 378, 419, 468-469,

475-476, 480, 624, 646, 688, 724 end of, 15

Resolution, 616, 686 Resource rooms, 261, 717 Resources, 3, 42, 65, 69-70, 83, 89, 94, 127, 142,

182, 189, 202, 224, 233, 238, 243-244, 253, 265, 295, 365, 442, 460, 467, 566, 574, 591, 593, 633, 641, 660, 667

Respect, 5, 9, 15, 18, 33, 47-48, 52, 95, 100, 125-126, 162, 176-177, 186, 236, 246, 248, 265, 278, 281, 296, 299, 309, 312, 316, 356, 375, 377, 391, 393-394, 396-400, 402-403, 406, 437, 447, 449, 455, 465, 512, 514, 533, 541, 555, 568, 582, 612, 667, 669, 678, 680, 682-684

Respondent conditioning, 4, 14-15, 44-45, 47, 49-51, 60, 66, 409, 543

Responding, 1-4, 6-7, 10-17, 20, 54, 60, 64, 66, 81, 89, 93, 96-98, 100, 103-104, 112, 120, 126-127, 147-148, 151, 157-159, 171, 174, 178, 181, 186-193, 195, 197-198, 200-202, 205-210, 212-213, 215-216, 218, 221-225, 227, 231, 236-237, 239-240, 242-244, 248, 252-254, 273, 278, 281, 283, 297, 300, 302-310, 312, 314-315, 322, 326-327, 329,

331-332, 334-341, 343, 348, 351-352, 355-359, 361, 365-366, 372, 392, 396, 401, 405, 409, 411-412, 414, 416, 421, 424-425, 445, 447, 452, 463, 470, 474-475, 477, 481-482, 484, 487, 492-497, 505-506, 508, 515, 534, 539, 555-556, 570, 587, 619, 624-625, 628, 632, 635-636, 639, 641, 644-646, 654, 662, 670, 683, 686, 690, 692, 694, 696, 702, 705, 707, 711, 713, 721, 727-728

context for, 100 Response, 2-20, 27, 29-31, 37, 39, 42-43, 44, 46-66,

70, 74, 79, 81, 85-88, 90, 92-93, 95-98, 100-104, 106, 108-110, 112-115, 117-118, 120-121, 126-128, 131, 133-134, 136-137, 142, 144, 146, 148, 151, 153-154, 156-157, 159-160, 162, 165, 170, 173, 176, 183-184, 189-190, 192, 194, 197, 202-203, 208-209, 211-212, 216, 221, 227, 229-230, 233, 236, 238, 240, 242-243, 247, 251, 259, 262, 264, 267, 271, 276, 278-284, 286-288, 291-292, 295, 297, 300-310, 312-316, 319-322, 326-335, 337-341, 343, 344-362, 365-366, 368, 372-373, 374-375, 381-389, 391-393, 396-398, 401-407, 409-422, 424-425, 427-433, 435-438, 440-445, 447-453, 454-467, 469-470, 473-476, 478-480, 482-486, 491-497, 498-501, 504-509, 512, 516, 519, 521, 523, 537-557, 561, 566-567, 570, 572-573, 585-586, 588-591, 595-597, 606-614, 618-620, 622-625, 627-630, 632, 634-636, 638-640, 643-646, 648-650, 652-656, 659-663, 668, 670, 676-677, 683, 685-688, 691-695, 697, 699-703, 705-711, 713, 716-720, 722-728

Response cards, 114, 183, 267, 691, 693, 697, 706, 710, 713, 718

Response cost, 3, 15, 58-59, 142, 347, 374-375, 381-389, 397, 406, 482, 567, 570, 572-573, 610, 676, 686, 695, 703, 716, 726-727

Response cost procedure, 383, 386-388, 397, 570, 572, 610

Response generalization, 8-9, 12, 15, 271, 622-624, 628-630, 632, 634, 636, 654, 661-663

Response maintenance, 8-9, 11, 15, 622-625, 627-630, 634, 644, 646, 656, 659, 661-662, 719

Response patterns, 338 Response prevention, 362 Response prompts, 16, 133, 307, 310, 361, 418-420,

422, 425, 431-433, 445, 448, 452, 493-494, 573, 586, 595-597, 612, 618-619, 629, 650, 653-654, 660

Responses, 3-8, 10-17, 19-20, 29, 31, 33-37, 40, 46-47, 49, 51-54, 56, 60, 63, 65-66, 78-80, 85, 87, 89-90, 93, 96-98, 101-104, 106-109, 112, 115, 118-120, 125-126, 131, 135-137, 140-141, 147, 149-152, 155-159, 161-162, 176, 183, 190, 200, 211, 221, 226-234, 236, 238, 240, 243, 249, 258-259, 264, 273, 278-279, 281, 284-285, 288-289, 296, 298, 300-303, 305, 307, 309, 312-319, 322, 326-343, 345, 348-349, 351, 353-358, 360-362, 365-367, 372, 389, 392-393, 398-399, 409-414, 416, 419-420, 424-425, 429-432, 435-438, 440, 442, 444, 446, 449-453, 455-457, 459-460, 463, 465-466, 470, 473-477, 483, 486-487, 490-497, 501, 503-506, 509, 520, 538-539, 541-543, 545-551, 553, 555-557, 567-568, 570, 574, 584-586, 588-591, 596-597, 602-603, 605-606, 609, 612, 620, 623, 626, 628-630, 635-636, 644-645, 650, 652-655, 660-663, 689, 697-698, 707, 712, 720, 728

selection, 16, 52, 63, 65-66, 78-79, 85, 211, 249, 278, 303, 333, 337, 416, 494, 497, 588, 636, 655, 698

Responsiveness, 9, 571, 724 Restatement, 572 Restating, 257 restitutional overcorrection, 13, 15, 344, 360, 372 Retention, 157, 267, 282, 670, 689-690, 713 Retrieval, 32 Reversal design, 1, 12, 15-16, 20, 183, 191, 196-208,

215, 218, 221-222, 229, 233, 238, 243-244, 310, 321, 362, 530, 646

Reversal designs, 196, 218, 221 Reversals, 197, 200, 207, 218, 242, 250 Reversibility, 17, 197, 292, 414

Revision, 63, 449, 573, 694 Rewards, 10, 59, 209, 254, 265, 290, 296, 489, 524,

559, 561-565, 567, 579, 581, 587, 589-590, 608-609, 611, 615-616, 644-646, 648, 662, 687, 695

Rewriting, 156 Right to Effective Behavioral Treatment, 668-669, 684,

725 Ripple effect, 630 Risk management, 678, 691 risks, 95, 270, 369, 513, 676-678 Robotics, 465-467, 719 Role playing, 129, 626 Role-play, 652 Role-playing, 625, 641, 650, 652 Roles, 197, 209, 218, 540, 543, 584, 642, 680 Roots, 516 Rounding, 348 Routines, 8, 40, 47, 60, 455, 514, 526, 531, 651 Rubber band, 362, 596, 609 Rubric, 110 Rules, 16, 25, 93, 96, 115, 118, 120-121, 164, 169,

182-183, 246, 249, 252, 259, 280, 307, 333, 368, 380, 382, 385-386, 388, 437, 466, 492, 496, 566-567, 570-572, 576-577, 582, 588, 606, 616, 618, 630, 666, 668, 677, 679, 683-684, 691, 709

Rush, 298, 331, 521, 699, 715

S Sadness, 547 Safety, 38-39, 42, 82, 86, 167, 186, 233, 249, 266,

370, 458-459, 461, 486, 614, 630, 659, 665, 667, 678, 680, 684, 688, 692, 697, 701, 712, 728

of self, 233, 659, 697, 728 plans, 370 precautions, 370

Safety issues, 678 Salaries, 83 Saliva, 353 Samples, 110-111, 114, 117, 127, 154, 259, 651, 699 Sanchez, S., 727 SAT, 104, 204, 212, 287, 357, 539, 562, 584, 598, 629 Satiation, 3, 5, 15, 44, 59, 66, 155, 284-285, 303, 306,

309, 394, 398-399, 406, 432, 552, 649, 704 Satisfaction, 81, 258, 325, 576 Scale, 3, 46, 72, 75, 128, 149, 151, 156, 159, 165-170,

172, 176, 209, 218, 232-234, 243, 258, 262, 264, 268, 447, 456, 513, 520-521, 570, 586, 606, 618, 692, 695, 703, 706, 714, 720, 722, 728

Scales, 9, 72, 126, 128, 143, 159, 166, 169, 176, 258, 274, 519-521, 533, 589, 688, 691

ordered, 533 Scaling, 151-152, 157, 160, 164, 166-167, 169, 267

dimensional, 151 Scanners, 119, 129 Scanning, 128 Scapegoats, 576 Scatterplot, 15, 146, 162, 164, 176, 519, 521 Scatterplots, 148, 162, 164, 516, 519, 533 Schedules, 1, 3-4, 6, 10-14, 28, 40, 57, 108, 110, 127,

155-156, 158-159, 212, 227, 266, 276, 295-296, 300-301, 308, 310, 311, 324-343, 344, 351-352, 355, 357-358, 366, 368, 374, 418, 421, 431, 465, 467, 475, 480, 488-490, 493-494, 496-497, 498, 501, 503-504, 507, 524, 568, 575-576, 633, 644-645, 648, 653-654, 662, 680, 686, 688, 690-693, 695-697, 702, 704, 707-710, 713, 716, 724-726

of reinforcement, 1, 3-4, 6, 10-14, 28, 40, 155-156, 212, 276, 296, 300-301, 308, 310, 324-343, 355, 357-358, 366, 368, 465, 475, 480, 489-490, 493-494, 496-497, 498, 501, 503-504, 507, 524, 568, 575, 633, 644-645, 648, 654, 662, 686, 688, 693, 696, 702, 704, 708, 724-725

Scheduling, 31, 126, 139, 185, 623 Schemes, 633 Schneider, M., 718 School, 36, 72, 78, 81, 83, 88, 94, 99, 104, 110, 113,

125, 131, 147, 165-166, 168, 180, 183-184, 200, 202-203, 209-210, 217, 227, 233-234, 237, 240, 246, 248, 254, 261, 266, 277, 281, 286, 289, 291, 293, 296-298, 313-315, 330, 335, 337, 350, 353, 359, 376, 380, 382-383, 385, 388, 427, 457, 462, 488-489, 507, 516,

746

520, 522, 525, 550, 559-561, 563-567, 570-571, 574, 576-578, 580-581, 584, 590-592, 594, 599-600, 602, 604, 606, 611, 613, 617-618, 623, 626-627, 631-632, 635-637, 639, 641, 645, 649, 652, 656, 659-660, 665, 667, 670-671, 676-679, 682-683, 685-689, 692, 695, 697-707, 709-710, 712-714, 716, 718-723, 726-728

School activities, 113 School board, 380 School counselor, 599 School day, 209, 227, 293, 522, 559, 599-600, 606 School district, 246, 385, 665, 667, 677 School districts, 94 School lunch, 565 School psychologist, 261, 701 School psychologists, 671, 701, 713 Schooling, 60, 595 Schools:, 727

descriptions of, 33 Science, 2-3, 7, 11, 16, 22-31, 33-36, 39-43, 47,

62-63, 123, 125, 143, 179-183, 194, 229, 246, 248-249, 264, 267, 283, 342, 345, 371, 455, 465-466, 537-538, 543, 577, 585, 591, 682, 686, 688, 691, 695, 697-698, 700, 703-706, 716, 719, 721-722, 724, 727-728

in the curriculum, 125 new, 3, 7, 26-27, 29-31, 34, 36, 63, 183, 229, 264,

267, 455, 465-466, 585, 686, 688, 691, 695, 697-698, 700, 704-706, 716, 719, 721, 724, 727

Science instruction, 697 Sciences, 24-25, 30, 52, 181, 184, 246, 264, 647, 658 Scientific knowledge, 23, 25-26, 29, 85, 93, 120,

124-125, 672 Scientific method, 25, 27, 591, 682-683, 721 Scientific research, 28, 94, 179, 181, 248, 713, 720 Scientists, 23-27, 33, 42, 47, 50, 85, 93, 110, 120, 148,

179, 181, 183, 189, 269-270, 371 Scope, 34, 36, 41, 69, 89, 247, 257, 371, 437, 556,

559, 633, 659, 662 Scores, 48, 73, 84, 86, 102, 132, 140, 144, 169, 171,

180, 183, 209-211, 231, 246-248, 262-263, 520, 561, 604, 607, 617, 700, 718

Scoring, 73, 87, 116-117, 129, 132-133, 233, 607 Screening, 69, 89, 148, 233, 521, 703, 706 screws, 104, 404-405 Script, 141, 227-229, 256, 274, 421, 706, 719 Sculpting, 103 Search, 33, 38, 60, 179, 193, 236, 247, 593, 607 Seating, 48, 75, 294, 417, 499, 512 Seating arrangements, 48, 75, 512 Seatwork, 109, 111, 329, 457, 477-478, 492, 579,

604-605, 643, 646-647, 650-651, 653, 690, 708

Second Step, 573 Secondary school, 710 Secondary students, 83, 250, 260, 262-263, 592, 650,

685, 691, 723 Section, 27, 45, 78, 82, 164, 189, 246, 256, 258, 274,

278, 289, 295, 299, 317, 345, 356, 391, 397, 412, 428, 438, 441-442, 493, 554, 568, 571, 587, 605, 607

Security, 26, 404, 615 Segmentation, 78 Seizures, 240 Self, 9, 11, 16, 26, 36, 38, 64, 71-72, 75, 79-81, 83, 87,

90, 94, 96, 117-119, 125, 149, 151, 153, 164-165, 167, 170, 179, 182, 184, 186, 189, 200, 207, 209, 212, 225, 227, 233-234, 237-239, 254, 256, 259, 262, 287, 289, 296, 316-317, 320, 326, 336-337, 341-342, 351-354, 357, 359-360, 362-365, 367-369, 373, 378, 380, 403, 411, 418, 437-438, 446-448, 452, 469, 471-472, 474-475, 478-480, 485-486, 489, 492, 494, 497, 501-502, 504, 506, 508-509, 511, 513, 526-528, 534, 543, 548, 558, 562, 564-565, 569-570, 575-576, 579-581, 583-620, 625, 630, 633-634, 636-637, 642, 650-654, 656-659, 663, 665-668, 670, 674, 680, 683-684, 685-700, 702-728

constructing, 149, 151, 153, 164-165, 167, 170, 437, 452, 614, 625, 702

discovering, 182 Self-assessment, 16, 418, 601, 608, 651 Self-awareness, 9, 614 Self-concept, 81, 254 Self-control, 16, 336-337, 583, 585-587, 591-592, 594,

607, 612-613, 617, 619, 654, 688, 690, 694, 698, 713-714, 716-718, 720, 722, 724, 726

Self-control training, 336-337, 694 Self-determination, 586, 680, 685, 696, 702, 717 Self-determination skills, 685, 702 Self-directed learning, 685, 702 self-esteem, 589 self-evaluation, 16, 259, 583, 601-602, 606, 608, 612,

616, 618-619, 698, 704-705, 717, 719, 723 Self-injurious behavior, 96, 118-119, 149, 153, 170,

189, 200, 209, 262, 287, 342, 351-354, 362-363, 367-368, 411, 471, 478, 486, 489, 492, 501-502, 511, 513, 665-667, 692-694, 696, 703, 706-708, 710, 713, 718-719, 721-723, 727

Self-injurious behaviors, 359, 364, 472, 506 Self-instruction, 16, 583, 596, 612-613, 620, 702, 706,

718-719 Self-instructional training, 636, 697, 703, 715 Self-knowledge, 585 Self-management, 16, 75, 569, 575, 581, 583-620,

630, 634, 652, 654, 656-657, 659, 663, 687, 690-692, 702-703, 705-706, 708-710, 712-714, 716

generalization and, 589-590, 614, 619, 630, 634, 652, 654, 656-657, 659, 663, 692, 702, 705-706

self-monitoring, 16, 71, 75, 526, 528, 579, 581, 583, 590, 592-593, 598-610, 612, 615, 618-620, 650, 656-658, 683, 689, 692-693, 699-700, 705, 707-708, 710, 713, 719, 724, 727-728

Self-monitoring skills, 605, 692 Self-monitoring techniques, 593, 603 self-observation, 16, 598 Self-recording, 184, 186, 526, 570, 598-601, 603-604,

606, 608, 650, 657, 689-690, 692, 705, 707-708, 712, 724

Self-reinforcement, 607-608, 612-613, 616, 620, 688, 691, 698, 700

Self-stimulatory behavior, 357, 380, 508, 697, 717 self-talk, 548, 593 Semantics, 537 Sensation, 609 senses, 47 Sensitivity, 193, 248, 341-342, 365, 456, 667, 709, 713 Sensory reinforcement, 521, 717, 725 Sensory systems, 298 Sentences, 58, 250, 524, 537, 539, 561, 608, 623, 701 Separation, 209, 216, 378, 494, 692 Sequence, 1-3, 7, 9, 11-12, 16-17, 20, 25, 48, 54,

69-70, 73-74, 97, 111-112, 127, 132-133, 167, 196, 200, 202, 208-209, 212, 215-216, 218, 226, 229-232, 236-237, 243, 248-250, 256, 267, 273-274, 286, 298-299, 329-330, 339, 361, 365, 367, 372, 401, 403-404, 407, 418-419, 430, 432-433, 435-438, 440-450, 452-453, 458-459, 463, 488, 498, 501, 503-506, 509, 514, 555, 559, 574, 579, 591, 604, 612-614, 625, 641, 652, 659, 693

Server, 403 Setting, 5, 7-16, 18, 37, 48, 51, 61, 65, 74, 80, 83, 89,

91, 101, 117, 126, 131, 134-135, 140, 182, 184-187, 189, 193, 195, 197, 218, 221, 225-227, 229, 233, 236-238, 240, 242-243, 249, 252-253, 256, 263, 265, 267, 271, 273-274, 282, 305, 337-340, 350, 361, 375-376, 378, 381, 383, 388, 391, 416, 442, 445, 448-450, 453, 457, 469, 484, 491-492, 497, 503, 507, 512-514, 521, 562-563, 568-569, 571, 574, 576-577, 587, 589, 591, 593-594, 596-598, 601, 605, 608, 611, 613, 616, 619, 622-632, 634-637, 639-646, 648-650, 652-654, 659-663, 689-691, 694-695, 697, 700, 707-708, 710-711, 714, 717, 719

Severe disabilities, 83, 250, 260, 262, 296-297, 316, 359, 418, 439, 445-448, 455, 507, 660, 665, 690-692, 698, 721-722

Sex, 395-396, 399 Sexual behavior, 394 Sexual relations, 666 Shapes, 455, 487 Shaping procedure, 456-458, 462, 464, 467, 696 Shared control, 592 Sharing, 47, 132, 412-413, 450, 455, 466, 615, 629,

660, 678, 726 Shock, 19, 60, 315, 349-350, 355, 401, 410, 609, 665,

690, 708, 719, 725, 727 Siblings, 71, 297

Sidman, Murray, 182, 294, 308 Sight words, 211, 307, 326, 688, 701, 720 Sign language, 8, 13, 16, 213-215, 430, 541, 545, 551,

557, 690, 724 Signals, 16, 36, 133, 143, 188, 301, 350, 376, 382,

404, 598, 604 Significance, 23, 29, 38, 69, 75-77, 90, 148, 150, 166,

169, 176, 237, 247, 258, 262, 267, 269-272, 274, 277, 588, 615, 668

statistical, 38, 148, 247, 269-270, 274 Significant others, 3, 36, 69, 71-72, 76, 82-84, 89-90,

187, 270-271, 295-297, 300, 310, 370, 476, 478, 508, 574, 618, 650, 658, 660, 663

Signifying, 62, 197 Signs, 86, 95, 152, 227, 249, 263, 333, 395, 418, 422,

489, 506, 523, 531, 540, 545, 584, 641, 667, 691, 724-725

Silence, 281, 461 Silent reading, 647, 689 Simplification, 256 Simulations, 465 Simultaneous prompting, 720 Singing, 287, 541, 546, 618 Single-subject studies, 265 Size, 7, 17, 39-40, 48, 61, 125, 160, 217, 242, 265,

326-327, 329, 411-412, 419, 421, 450, 463, 540, 548, 557, 584, 588, 639, 642

Skill acquisition, 97, 231, 680, 703 Skill development, 82-84, 90, 97, 465, 482, 711 Skill mastery, 670, 682 Skill training, 685 Skill(s), 224 Skills, 3, 24, 38, 40-41, 47, 53, 63, 69-70, 72-73,

75-76, 78-80, 82-83, 86, 89-90, 101, 104, 109, 125, 127, 129, 141, 164, 170, 181, 190, 213-215, 224, 227, 229, 231-232, 238, 243, 246, 253, 256, 258-259, 261, 263, 266, 296, 307, 315-316, 322, 337, 360-362, 367, 369, 378, 416-418, 425, 427-430, 432-433, 436-438, 441-443, 447, 451-452, 455, 460, 463, 465, 484, 499, 507, 516, 520-521, 524, 531, 541, 551, 556, 559, 562, 566, 571-572, 577-578, 580, 584, 589-594, 604-605, 612, 619, 623, 625-627, 632-636, 639-642, 646, 649, 653-656, 658-661, 663, 666-667, 669-671, 674-675, 680-682, 685, 689-692, 694, 697-699, 701-706, 711-716, 718-719, 721, 724-725

attending, 70, 417-418, 429-430, 571, 674, 690 conceptualization, 465 fine motor, 83, 429-430 practicing, 129, 591, 627, 636, 666, 671, 674, 682 prosocial, 721 receiving, 429, 441, 572, 639, 641-642, 659, 669 retention of, 689 self-help, 83, 633, 642, 654 sending, 315 speaking, 296

Sleep, 48, 117, 320, 384, 387, 394-396, 399, 406, 470-471, 500, 520, 555, 598, 686, 714, 716

Slides, 625, 641 Slips, 291, 330, 382, 549, 574, 600-601 Slope, 5, 17, 98, 157, 160, 175-176, 332, 411 Small group, 18, 46, 109, 339, 576, 581, 594, 641,

662, 725 Small groups, 337, 574, 647, 694 Small-group instruction, 339, 477-478 SMART, 26, 87, 117 smiles, 48, 136, 289, 308, 362, 402, 660 Smith, S., 722 Snacks, 46, 227, 290, 489, 512, 521, 571, 646 Social environment, 2, 52, 551-552 Social history, 402 Social interaction, 82, 141, 206, 284, 313, 322, 377,

402, 407, 517-518, 537, 539-540, 550, 556, 697, 706, 723

peer interaction, 556 Social issues, 29, 713 Social learning, 32, 712 Social needs, 368 Social problems, 698 Social psychology, 471, 698, 712, 727 social reinforcement, 202-204, 206, 233, 277, 287,

293, 316, 333, 521, 548, 626, 685, 700, 722, 726

Social reinforcers, 13, 293, 309, 376, 511, 588 Social relationships, 552 Social sciences, 184, 246, 264 Social skill instruction, 577

747

Social skills, 3, 38, 76, 79, 129, 258, 337, 416, 577-578, 625, 649, 680-681, 685, 694, 699, 705, 711

Social skills instruction, 577 Social skills training, 258, 699 Social studies, 261, 267, 297, 580, 685 Social validity, 16, 38, 68, 79, 89, 91, 256-259,

261-263, 267, 270, 274, 370, 387, 448, 508, 609, 616-617, 665, 667, 687, 696, 700, 720, 727-728

Social values, 680 Social workers, 671, 713 Socialization, 499 Sociology, 181 Software, 110, 116, 119, 121, 168-169, 465, 591-594,

723 ability to evaluate, 591

Software programs, 116, 168-169, 591-592 Solid foods, 317 Solitary play, 99 Solutions, 34, 39, 42, 202, 667, 683, 719 Songs, 541, 555 Sony, 201 Sorting, 108, 229, 336-337, 649 Sound, 1, 4, 27, 38, 45, 48-51, 54-55, 59, 78, 95, 99,

102, 104, 113, 185, 189, 195, 201, 208, 212, 217, 237, 269, 281, 284-285, 287-288, 349, 371, 398, 403-404, 409, 427, 432, 444, 447, 453, 455, 458, 460, 466, 479, 483, 530, 543, 546, 553, 613, 635, 652, 667

Sounds, 33, 37, 51, 54, 63, 78, 102, 112, 281, 284, 287, 368, 397, 409-410, 458, 471-472, 537, 546, 554-555, 635, 640, 728

speech, 287, 458, 471, 537, 546 Space, 2, 8, 45-46, 65, 97, 112, 115, 152, 160-161,

167-168, 185, 272, 353, 378, 380, 496, 585, 680, 682

Spanish language, 248 Spanking, 356 Speaking, 32, 59-60, 102, 204, 242, 281, 283, 286,

296, 347, 409, 456, 458, 462, 548, 611, 645 Special education, 35, 73-75, 78, 83, 183, 267, 292,

320-321, 463, 526, 573-574, 590-591, 594, 604, 608, 627, 649-650, 656, 677, 686, 693, 697-698, 701-702, 708-712, 715-716, 719, 722, 724, 727-728

Special education programs, 594 Special education services, 83, 608 Special education students, 78, 267, 292, 320-321,

591, 650, 711 Special education teacher, 463, 573, 627, 649-650 Special education teachers, 75 Special educators, 79 Special needs, 714 Specific hypotheses, 522 Speech, 79, 104, 213, 287, 360-361, 447, 455-456,

458, 463, 471, 477-478, 507, 511, 524, 537, 539, 546, 551, 553, 593, 610, 627-628, 691, 703, 707, 712, 720

speed, 5, 62-63, 123, 129, 157, 164, 193, 212, 293, 345, 348, 362, 436, 453, 460, 656, 670, 725

Spelling, 96, 102, 116, 131-132, 141, 154, 156-157, 169, 183, 209-211, 267, 278, 360, 427, 432-433, 539, 542, 561, 580, 605-606, 645, 650, 691, 711-714, 719, 722, 724

transitional, 724 Spelling tests, 169, 209 Spillover, 630, 723 Spontaneous recovery, 7, 17, 468, 474-475, 480 Sports, 42, 80, 286, 306, 571 Spreads, 38, 589, 623 Stability, 94, 170, 172, 174, 177, 189-190, 193, 233,

237, 242-243, 267, 303, 520, 692 Staff, 71-72, 83, 90, 200, 233, 248, 259, 261, 308,

351, 370, 457, 462, 492, 524, 531-533, 635, 649-650, 656, 660, 675, 677, 679-680, 709, 719-720

credentials, 675 Staffing, 520 Stages, 18, 131, 254, 307-308, 325, 343, 431, 456,

460, 464, 467, 507, 516, 547, 648, 690, 700 Stages of learning, 307-308, 325, 343 Standard setting, 700 Standard units, 8, 14, 99, 166

important, 99 Standardized assessment, 259, 274 Standardized assessments, 551 Standardized tests, 69, 73, 262 Standards, 7, 68, 89, 95, 103, 120, 131, 255, 373,

519, 590, 664-665, 667-669, 671, 674-675, 683-684, 685, 694, 696, 706, 725

Stanford University, 728 Starches, 131 State and local, 571 States, 4, 13, 27, 29, 36-37, 45-46, 48, 87, 125, 131,

190, 291, 296, 309, 322, 334-335, 400, 459, 466, 519-520, 540, 548, 554, 568, 588, 616, 640, 665, 671, 674-678

court system, 677 courts, 676

Statistical methods, 269-270, 696 Statistics, 135, 148, 235, 248, 699 Statutes, 370, 677 Stereotypic behavior, 342, 360, 494, 685, 692, 707 Stimulus, 1-19, 27, 29-30, 40, 42, 44-51, 53-66, 78,

90, 96, 98, 100, 110, 118, 120, 153, 155, 203, 208, 213, 250, 276, 278-281, 283-285, 287-289, 295-307, 309-310, 312-316, 322, 337-339, 344-373, 374-389, 390-393, 395-407, 408-425, 427, 432, 435-437, 444-446, 449-453, 457, 459-460, 466, 469, 471-472, 479, 484, 486, 492, 499-500, 508, 512, 516, 520, 530, 538, 540-544, 546-557, 566, 585-586, 595, 597-598, 609, 614, 619, 622, 624-625, 627, 629-630, 632, 634-636, 638-642, 648, 652, 654, 659, 661-663, 680, 686, 689, 691-694, 699, 701, 703-706, 708-709, 711, 714-715, 717-720, 724, 726-728

Stimulus control, 2, 4, 6-7, 11-13, 17, 44, 55, 61, 66, 281, 307, 348, 408-425, 445, 499-500, 508, 520, 538, 543, 546-547, 549-557, 566, 586, 597-598, 627, 629-630, 632, 639-641, 654, 661-662, 689, 694, 705, 708-709, 711, 717, 720, 724, 726-727

Stimulus events, 47-48, 60-61, 65, 96, 100, 350 Stimulus fading, 40, 153, 421-425, 459, 466, 486, 492,

686, 704 Stimulus generalization, 4, 8, 17, 408, 410-412, 414,

424, 546, 548, 624, 640, 692, 706 Stimulus shaping, 422-424, 703 Stomach, 46-47, 49, 547 Stop, 55, 74, 86, 109, 111, 249, 277, 283, 315,

355-357, 360-362, 386, 388, 432-433, 469, 473, 489, 521, 525, 545, 597, 599, 609, 644, 649, 692, 697, 699-700, 705, 709, 718, 720, 725

Storage, 32, 537 Stories, 166, 541, 565, 721 Story writing, 687 Strategies, 28, 39, 68, 81, 83, 94, 125, 174, 178-195,

246, 248, 260-261, 318, 320, 361, 380, 434, 442, 499-500, 512, 520, 533, 552, 562, 573, 580, 583, 592-593, 610, 612, 623, 633-634, 650, 654, 659, 661-662, 665, 682, 685, 701, 704-705, 707, 711-712, 715, 728

deliberate, 361 intended, 81

StrategyTools, 592 Stress, 350, 412 Strips, 240-241, 570 Strokes, 8, 85, 98, 104, 108, 248, 291 Structural linguistics, 537 Structuralism, 32-33 Structure, 33, 53, 59-60, 63, 109, 202, 378, 537-538,

556, 576, 592, 678, 689, 695, 702, 709 Structured interviews, 9, 519, 533 Student achievement, 670 Student behavior, 87, 130, 277, 340, 504, 600, 710,

725 Student input, 594 Student performance, 34, 73, 323, 570, 698, 710 Student progress, 419 Student success, 670 Students, 3, 23, 25, 31, 35, 39-40, 43, 48, 55, 57, 61,

65, 70-71, 73-74, 78, 80-81, 83, 87, 93-97, 101-104, 108, 111-114, 116, 127, 131, 133, 141, 152, 154, 156, 180-181, 183-185, 190, 198-199, 202, 206, 209-214, 217, 221, 229-232, 237-238, 242-243, 247-250, 254, 259-263, 266-269, 277, 282-283, 285, 291-294, 296, 306-307, 312-313, 316, 319-321, 329-333, 335, 356-357, 367, 376-378, 380, 382, 384-388, 413, 417-418, 420, 436, 438, 440-445, 447, 450, 455-458, 462, 465, 469, 477-479, 483-485, 488, 499-500, 507, 520, 525, 527-528, 562, 569-573, 575-581, 589-594, 598-600,

603-609, 611, 613, 615-616, 619, 625-627, 631-632, 636-643, 646-654, 656-657, 659-660, 670, 672, 675, 680, 685-686, 688-695, 697-699, 701-703, 705-708, 710-714, 716-717, 719-724, 726-728

antisocial, 356, 722 conferences with, 599 differences between, 209, 230, 627, 705 distinctions, 96, 579 exceptional, 438, 443, 485, 594, 605, 685, 690,

692-694, 697, 701-702, 705, 707-708, 710, 712, 714, 716-717, 719-722, 724, 726-728

in regular classrooms, 650 reluctant, 649 self-evaluations, 606

Students with disabilities, 83, 114, 443-444, 576, 626, 631, 637, 640, 651, 727

Studies, 23-24, 28, 34, 36, 39, 42, 75, 85, 90, 94-95, 98, 104, 113, 127, 133, 139, 170-171, 182-185, 191, 194, 197, 200, 207-208, 224, 227, 236-237, 249, 253, 261, 265-267, 271-272, 277, 289, 291-293, 295, 297, 306, 318, 325, 351, 353, 355-356, 359, 361-362, 365-367, 385, 419, 429, 447, 449, 453, 463, 483, 515, 519-520, 580, 590, 605-606, 608-609, 613, 631, 636, 642, 646, 654, 666, 682, 685, 689, 705, 724

D, 39, 94, 98, 170-171, 183, 237, 261, 277, 365, 429, 449, 453, 580, 609, 685, 689, 705, 724

G, 23-24, 34, 42, 75, 90, 94, 98, 104, 113, 133, 170-171, 182-184, 191, 197, 200, 207-208, 224, 227, 236, 249, 253, 261, 265-267, 271, 277, 289, 291-293, 295, 297, 351, 353, 355-356, 359, 365-366, 385, 419, 429, 447, 463, 483, 519-520, 580, 590, 605-606, 608, 613, 631, 636, 642, 646, 654, 666, 682, 685, 689, 705, 724

Style, 103, 107, 264, 450, 499, 543, 549, 639-640, 654, 685, 702, 706

Subtraction, 56, 63, 102, 170, 231, 422-424, 528, 603, 605, 627, 694, 703

Successive approximation, 16, 229, 243, 454, 457, 464, 553

Suggestions, 62, 73, 123, 387, 603, 615, 617, 620, 630, 650

delivering, 62 Superego, 9, 32 Supervision, 24, 207, 370, 389, 437, 579, 591, 643,

654, 664-665, 672, 680, 694-695, 714 Support, 9, 29, 58, 94-95, 180-181, 206, 224, 245,

256-257, 259, 261, 266, 287, 349, 371, 431, 477, 499, 520, 593, 595, 608, 614, 616, 643, 665, 667-668, 670, 675, 677, 680-681, 683-684, 691-692, 717-718, 721-722

Supporting, 68, 75, 191, 203, 559, 665, 716 Surveys, 71, 90, 296, 391, 714 Susceptible, 19, 49, 51-52, 202, 218, 254, 289, 309,

347, 349 Suspension, 522, 524 Sustained silent reading, 647, 689 Sustained Silent Reading (SSR), 647, 689 Swallowing, 445-446 Swanson, 282, 723 Switch, 110, 158, 199, 215, 284, 300, 302, 314, 345,

376, 388, 447, 450-451, 472, 499, 530-531, 655, 702

Symbols, 150, 159, 161, 167-169, 177, 234, 519, 566, 637, 720

Symmetry, 14, 17-18, 408, 414, 425 Synonyms, 282, 289, 349-350 Syntax, 35, 537, 546, 691 Synthesis, 695, 705 System, 3, 10, 12, 18, 32-33, 50, 64, 83, 110, 116-117,

119, 124-125, 128-131, 133-134, 143-144, 150, 162, 170, 182, 185, 191, 193, 197, 200, 214, 249, 253-254, 267-268, 290, 345, 362-363, 386, 445, 460, 465-466, 494, 558, 566, 568-575, 580-582, 591, 599, 608-609, 612, 659, 665, 667, 677, 686-687, 689-690, 698, 701, 708-709, 711-712, 717-719, 722-724

Systematic instruction, 591, 629, 644 Systematic replication, 15, 17, 245, 265-266, 271-272,

274, 358, 384, 689, 697, 714 Systems, 9-10, 33-34, 37, 47, 68, 101, 109, 119-121,

128, 139, 143, 245-246, 298, 409, 447, 506,

748

537, 558, 568-569, 574-575, 614, 664, 678, 682, 688, 691, 696, 700, 704, 709, 714, 719

breakthrough, 682 dynamic, 47, 128, 246 human, 33-34, 119, 128, 143, 537, 688, 704, 709,

714

T Tables, 95, 353, 449, 529, 601, 639-640, 642, 652 Tactics, 2, 16, 28, 31, 37, 39-40, 43, 58-60, 63, 164,

170, 174, 179, 185, 187, 189, 191, 195, 197, 199, 206, 215, 221, 233, 235-236, 238, 242, 246, 248-252, 269, 272-273, 304, 308, 345, 347, 356, 368, 500, 507-508, 524, 537, 579, 584-586, 588, 590, 592, 595, 607, 612, 615, 617-620, 623-624, 631, 633-634, 640, 643, 652, 654, 656, 658, 660-663, 704, 720

Talking, 1-2, 53, 55, 79, 81, 93-94, 109, 117, 204, 206, 259, 282, 293, 335, 342, 383-384, 409, 458, 485, 494, 521, 525-526, 548, 566, 568, 576, 579, 585, 601, 604, 611, 613, 620, 627, 634, 651, 699, 717

Tangible reinforcers, 291, 431, 433, 552 Tangles, 683 Tantrums, 37, 99, 316, 470-471, 473-474, 476,

478-479, 492, 511-513, 516, 518-519, 522-526, 551-552, 561

tape recorders, 129 Tardiness, 485 Target behaviors, 16, 38, 54, 68-91, 93, 99, 108-109,

112-113, 127, 129, 188, 200, 206, 218, 238, 242, 246, 254, 265, 270-271, 290, 304, 379, 386, 568, 570-573, 579-582, 589, 596, 605-606, 615-616, 627, 631-633, 641, 645-646, 651, 662, 674, 681, 687, 689, 700, 725

TASH, 686 Task analysis, 2, 8, 18, 64, 68, 229, 256, 259,

434-435, 437-446, 449-450, 452-453, 459, 556, 625, 665

Task performance, 307, 320, 322, 562 Tasks, 9, 16, 24, 38, 47, 65, 73, 78, 81, 93, 108, 117,

121, 129-130, 198-199, 259, 262, 284, 296, 306, 316, 320, 326, 329-330, 380, 437, 442, 446-447, 452, 459, 465, 471, 486-487, 500, 504-506, 508-509, 511-512, 514, 525, 527-529, 531, 533, 562-565, 573, 587-590, 592-593, 597-598, 601, 603-605, 612-613, 616, 619, 646, 650, 654, 671, 711, 725-728

Taxes, 257 Taxonomy, 727 Teacher, 6, 23-24, 34, 40, 48, 62, 70-73, 75, 78, 80-81,

83, 86-87, 89, 94-95, 97-100, 102-103, 108-110, 113, 115-118, 130, 132, 140-141, 154, 156, 181-184, 187, 198, 201, 203-206, 210, 215, 227-229, 233, 236-237, 254-256, 259, 261-262, 277-278, 280, 282, 291, 293-297, 301, 303, 305-308, 316, 325, 327-331, 333, 339-340, 345-347, 356-357, 362, 375-376, 378-383, 385, 387-388, 410, 412-413, 417-420, 427, 429-430, 436-437, 440-445, 456, 458, 460, 462-463, 469, 477-478, 483, 485, 488-489, 492-494, 499-500, 502, 504-506, 509, 511, 513, 518, 520-521, 524-529, 541, 554-555, 561, 566-567, 570-571, 573-580, 586, 589-591, 593-594, 599-605, 607-609, 612-613, 620, 623, 625, 627, 629-630, 632, 634-635, 639, 641-643, 646-653, 655, 658-660, 665, 668, 672, 678, 683, 685, 690-695, 698-699, 701, 705-708, 710-712, 716, 720, 723, 725, 727

Teacher aides, 71 Teacher education, 692, 698, 716 Teacher training, 236 Teachers, 9, 23-24, 36, 39, 42, 46, 58, 70-73, 75, 80,

87, 94-95, 98, 108, 117, 126, 162, 187, 203-204, 206-208, 215, 217, 224, 227, 233, 236-238, 244, 258-259, 261, 264, 268, 293-294, 296, 306, 318, 325, 329-330, 333, 342, 356, 364, 366, 378-379, 381, 383, 388, 409-410, 417-418, 430, 442, 455, 459, 463, 473-475, 477-480, 482, 491, 499, 506, 519, 522, 524, 526-527, 529, 533, 553, 561-564, 572, 576, 580, 591, 593-594, 599, 604, 617, 625, 631, 634, 636, 640-643, 647, 649, 651-652, 656, 660, 668, 670, 681-682, 685-686, 699, 701, 704, 708-709, 712, 721

caring, 670 educators, 75, 95, 381, 591

experience of, 36 general education, 259, 261, 526, 576, 636, 649,

651, 686, 708, 712 head, 87, 98, 224, 418, 474, 478 highly qualified, 94 individualized instruction, 670 participation of, 562 Spanish-speaking, 409 substitute, 649

Teaching, 2-3, 8, 12, 18, 28, 34, 36, 39, 52, 69, 73, 76, 78-80, 82-83, 94-95, 102-103, 141, 148, 156, 160-163, 173, 214, 216, 228, 234, 237, 250, 256, 266, 337, 345, 362, 367, 370, 402, 407, 412-414, 421-422, 429, 432-433, 436-443, 445, 448, 455, 458, 462, 465, 473, 485, 494, 507-508, 512, 522, 524, 533-534, 541, 547, 551-556, 584, 587, 589, 591, 593, 605-607, 616-617, 619, 622-623, 626, 630-636, 638-643, 650-653, 657, 659-660, 662-663, 667, 670, 672, 680-681, 685, 687-703, 705-716, 718-720, 722-728

Teaching:, 643, 708, 727 Teaching Children to Read, 713 Teaching skills, 507, 670 Teaching styles, 237 Teaching team, 660 Teaching Tolerance, 720 Teams, 209, 267, 578-580, 590 Teasing, 520, 608 Techniques, 9, 26, 34, 37, 59, 63, 83, 92, 133, 137,

144, 152, 165, 197, 204, 206, 208, 218, 221, 243, 255-256, 278, 295, 304, 366, 369, 371, 379, 419, 435, 448, 482, 487, 504, 549, 553, 566, 569, 575, 585, 587, 589, 591-595, 603, 614, 617, 624, 631, 634, 644, 674, 686, 688, 692, 694-695, 702, 710, 715, 719, 726

Technology, 23-24, 27, 29, 32, 37, 39-43, 64, 75, 85, 94, 119, 142, 148, 164, 179, 181-182, 193-194, 236, 257, 263-264, 269-272, 274, 345-346, 371, 467, 593, 617, 623, 659, 661-662, 685, 689, 695-696, 698-699, 703-704, 706-707, 714-715, 719, 722-724, 726

assistive, 593 computers, 119, 695, 723-724

Telephones, 332 Television, 297-298, 312, 376, 459, 464, 511, 559,

571, 597-598, 649 temperature, 7, 48-49, 59, 95, 126, 185, 312, 315,

349, 394-396, 400, 406, 643 Terminal behaviors, 80 Termination, 6, 12, 56-57, 59, 66, 94, 101, 110, 120,

192, 278, 294, 305, 312-314, 316, 322, 347, 370, 376, 380, 397, 401-402, 404, 512, 522-524, 568, 657, 672, 678-679

Terminology, 58, 283, 315, 404, 586, 623-624, 709 Test, 8, 17, 24, 27, 37-38, 46, 48, 69, 73, 96, 101, 113,

116, 125, 132, 148, 209-210, 224, 231, 233, 240-241, 246, 254, 263, 269-270, 294-295, 300, 302, 304, 307, 411, 413, 415, 425, 437-439, 443-446, 514, 521-522, 524-526, 528, 534, 551, 570, 572-573, 579-580, 590, 599, 604, 632, 659, 667, 682, 689, 694, 699, 710, 724, 727

Test items, 101 Test scores, 73 Testimonials, 23, 95, 673 testing, 3, 7, 43, 50, 88-89, 235, 258-259, 267, 274,

411, 414, 425, 522, 525, 528-529, 532, 550-551, 582, 676, 708, 720, 727

field, 274, 582, 720 Tests, 34, 42, 69-70, 73, 81, 89, 148, 169, 183, 209,

235, 240-241, 246, 261-263, 269-270, 274, 296, 307, 411, 414, 425, 438, 539, 551, 567, 592, 675, 684, 702

essay, 592 Expressive Vocabulary, 551 Problem with, 270 select, 81, 235, 414, 675 studying for, 567

Text, 5, 8, 13, 24, 31, 33-34, 45, 47, 53, 58, 88, 96, 164, 166, 168-169, 181, 183, 189, 221, 228, 246-247, 257, 313, 482, 536-537, 539, 541, 544, 548, 557, 575, 596, 705

Textbooks, 57-58, 477, 499, 537, 692, 724 The Parent, 70, 315, 356, 360, 369, 427-428, 432,

456, 458, 541, 678 Theater, 640 Theatre, 47

Theme, 546 theories, 27, 32, 95, 537, 585, 695, 712 Theories and philosophies, 95 Theory, 7, 13, 26-27, 32, 41-43, 235, 272, 334, 537,

550, 585, 587, 636, 682, 688, 690-691, 695, 699-700, 702-703, 710, 712-715, 720

Therapist, 15, 40, 69, 75, 81, 86, 88, 149, 153, 183, 226, 254-255, 261, 291, 295, 306, 320, 351, 353-354, 357, 364, 372, 381, 411, 417, 471-472, 483, 490, 492, 494, 500, 502-504, 508-509, 514, 562, 586, 589, 614, 627, 683, 712, 714, 727

Therapy, 17, 29, 80, 94-95, 342, 370, 376, 397, 561-562, 568, 614, 620, 626, 628, 644-645, 669, 676, 686-687, 689-690, 693-694, 696-697, 700-701, 703-706, 709-710, 713-714, 717-718, 722-723, 726-728

depth, 95 Think, 27, 35, 45, 57, 65, 88, 183, 371, 402, 462, 542,

550, 585, 591-593, 598-599, 609, 627, 629, 649, 661, 677, 700, 717

Thinking, 20, 33, 43, 45, 53, 183, 375, 537, 542-543, 547-549, 591, 652, 712

thinning, 15, 159, 266, 324, 333, 336, 343, 351-352, 366, 486, 491-492, 495-496, 503-504, 507-508, 699, 707

Thomas, 34, 357, 458, 590, 604, 606, 609, 664, 688, 698, 703, 709, 725

Thompson, 24, 42, 85, 127, 133, 149, 158-159, 197, 200, 203-204, 229-231, 287-288, 303-306, 353-354, 356, 364, 366, 369, 418, 472, 482, 501, 503, 506, 508, 521, 644, 691, 695-696, 699, 702-704, 711, 717, 719, 723-724, 727

Threats, 123-125, 128, 143, 180, 252, 255, 257, 524 Threshold, 94, 99, 104, 362, 376, 456, 458 Thrust, 286 Time, 1-3, 5-20, 26, 30-32, 34, 38, 40, 42-43, 45-46,

48-50, 53-60, 64-65, 71-75, 81, 83, 89, 92, 94-101, 104, 108-121, 123-124, 126-137, 139-140, 142, 144, 148-153, 155-161, 163-166, 168-170, 172, 174-177, 181-182, 185-195, 198-204, 206-209, 214-217, 219, 221, 223, 228, 232, 236-240, 242-244, 247-248, 250-251, 253, 255-259, 263, 266-271, 277-279, 282-284, 286, 289-291, 295-301, 303-304, 306, 308-310, 312-315, 322, 325-327, 329-335, 337, 340, 342-343, 345, 347, 349-351, 353, 359-361, 364-366, 368-369, 372, 374-382, 384-389, 391-392, 394, 397-399, 401-404, 406, 409-410, 412, 417, 419-420, 422, 425, 427, 429-430, 432, 436, 439-442, 446, 448-450, 452, 455-457, 459-465, 470-471, 473-476, 478-479, 482-483, 485-497, 498, 501-509, 511, 513-514, 516-519, 521-530, 533-534, 538, 541-542, 546, 548, 551-552, 559-562, 564, 567-568, 570-578, 580-581, 584, 586-591, 593, 596-601, 603, 605-613, 615-617, 620, 623, 629-630, 633, 635, 640-641, 643-646, 648-649, 652, 654-657, 661, 663, 666-668, 675, 677, 679, 683, 685, 689-691, 695, 697, 699-702, 704, 709-714, 716-717, 719-720, 724-726

elapsed, 5, 10, 15, 96, 100-101, 120, 130, 150, 237, 244, 291, 326, 330, 334-335, 340, 351, 366, 457, 489, 645

engaged, 13, 40, 99, 108, 110-111, 113-114, 121, 209, 250, 259, 263, 277, 297-298, 333, 353, 364, 376, 457, 483, 486, 488, 517-518, 524-529, 533, 570, 589, 598, 609, 615-616, 655-656

for responses, 11, 334-335, 455, 465, 644 on task, 109-111, 457, 478, 528-529, 599, 606,

612, 717 to think, 462, 598, 649, 677 units, 8, 14, 18, 81, 97, 99, 126, 151, 158, 166,

442, 452, 548, 551, 655-656 Time out, 186, 482, 689, 691, 717, 726 Time sampling, 10-11, 13, 18, 20, 92, 108, 110,

112-115, 120-121, 126-128, 135, 137, 140, 516, 599, 603, 716, 719, 724

Title, 81, 704 Toileting, 362 Token, 10, 18, 34, 46, 59, 130, 186, 229, 254, 259,

282, 284-285, 290, 488, 491, 558-582, 607, 630, 632, 645, 647, 657, 686, 689-691, 693, 695, 697-698, 701-702, 711, 713, 716, 718-719

Token system, 254, 571-572, 574, 718

749

Token systems, 574 Tone, 47-48, 289-290, 314, 348, 350, 361-362, 439,

538, 604, 606, 642-643 Tone of voice, 361, 642-643 Tools, 33, 39, 93, 124, 264, 274, 403-404, 407, 465,

591-594, 676, 695, 710 for teaching, 407

Toothbrushing, 707 Topics, 20, 537, 544, 559, 617, 677, 686, 693, 697,

703, 705, 723, 727 Topography, 5, 10, 15, 18, 46-48, 50, 52-54, 64-66, 68,

86-88, 90, 92, 103-104, 106-107, 120, 151, 187, 233, 278, 427-428, 432, 437, 455-460, 465-466, 473, 480, 512, 520, 522, 533-534, 537, 543, 556, 639, 643, 662, 682

Total Communication method, 250 touch, 47, 49, 53, 287, 289, 337, 419, 447, 461, 472,

513, 597, 614, 628, 644, 659-660, 698, 715 Touching, 86, 104, 117, 254, 257, 259, 289, 346,

419-420, 427, 430, 456, 483, 485, 526, 531, 533, 597, 644

Tourette syndrome, 183, 728 Toys, 79, 186, 213, 225, 287, 291, 297, 318, 337, 384,

396, 484, 489, 522, 525-527, 540, 552, 568, 572, 654

Traffic patterns, 641 Training, 1, 4-6, 8-9, 11, 14, 17-18, 29, 124, 128-130,

133-134, 143, 150-152, 154, 156-157, 159, 167, 169, 183, 206, 211-215, 227, 229-231, 233-234, 236-237, 250, 252, 255-258, 260, 262, 266, 268, 271, 293, 300-301, 336-337, 356, 360-361, 367, 369-370, 393, 402-403, 405-406, 408, 411-416, 418-420, 424, 427-433, 435, 437-438, 440, 442-443, 445-453, 454-456, 460-463, 465-466, 483-484, 487, 498, 501, 505-509, 516, 519, 536, 540, 547, 550-557, 562, 572-573, 592, 594, 601-602, 608, 613-614, 622-629, 631-632, 635-642, 644, 650-654, 656, 659, 662-663, 664, 666, 669-674, 680, 684, 685-690, 692-699, 702-707, 710-716, 718-720, 722-725, 727-728

peer tutor, 256 script, 227, 229, 256, 706, 719 self-instructional, 636, 689-690, 697, 703, 715

Traits, 27, 64 Transcription, 13, 18, 536, 539, 542-544, 548, 556-557 Transfer, 80, 400, 419-421, 425, 547, 551-556, 624,

650, 689, 707-708, 714, 720, 722, 724, 727 Transfers, 110, 547 Transition planning, 592 Transitions, 98, 650-651, 686 Transitivity, 14, 17-18, 408, 414-415, 425 Translation, 400, 691 Transparencies, 256 Travel, 168, 436 Treatment, 1-4, 6, 9-14, 17-20, 28, 38, 40, 63, 71, 73,

75-76, 79, 83, 89, 94-95, 102, 114, 117-118, 120-121, 124, 127, 130-131, 140, 143, 148-150, 152-153, 159, 165, 167-168, 170, 175-176, 178, 180, 182-195, 196-198, 200-202, 204-219, 220-223, 225-227, 232, 235-242, 244, 245-247, 249-251, 253-264, 270-274, 277, 290, 292, 313, 318, 320-321, 325-326, 334, 336, 340, 346, 351, 353, 356-357, 359, 362-370, 373, 376, 378, 380-384, 391, 409-411, 413, 421, 445, 457, 471-474, 477-480, 482, 484-486, 488-495, 497, 500-504, 507-509, 512-513, 515, 520, 563, 569, 581, 589, 598, 614-615, 620, 624, 626, 630, 646, 659, 665-670, 672, 674-684, 685-687, 690-699, 701-702, 704-720, 722-728

lifestyle changes, 680 Treatment acceptability, 258-260, 370, 714 Treatment failures, 369 Treatment methods, 89 Treatment plan, 83, 381 Treatment settings, 492, 589 Treatments, 1, 11, 16, 27, 39, 94-95, 102, 120, 148,

164, 183, 186, 196-219, 249-250, 252, 255, 257, 270, 273-274, 289, 304, 318, 321, 353, 362-363, 368-370, 391, 469, 477, 499, 501-502, 516, 524, 608, 636, 668, 676-677, 682, 685, 688, 701, 704, 711, 714, 718-719, 721

Trend, 1-3, 5, 16-17, 19-20, 92, 98, 146-150, 168, 170-175, 177, 185, 188-189, 195, 197, 209, 212, 235, 294, 316, 346, 474, 727

Trend lines, 98, 171-173, 177, 727 Triangles, 151, 444, 626 Truth, 19, 23, 32, 257, 269, 631, 661 Tube feeding, 317 Tuning, 355 Turns, 203, 205, 314, 371, 398, 403, 428, 435, 461,

511, 612, 632, 651 Tutors, 653, 688-689 Type II errors, 269

U Unconditioned stimulus, 4, 12, 15, 19, 44, 49-51, 66,

393, 395, 399, 409, 469, 540 Understanding, 7, 16, 20, 23-24, 27, 31-33, 35, 38-40,

42-43, 45, 47, 55, 58, 60-61, 63, 70, 93-94, 109, 113, 134, 154, 164, 169, 177, 179-180, 182, 194, 246-247, 249, 260, 264, 271-272, 278, 280-281, 286, 313, 342-343, 395, 403, 405-407, 409-411, 414, 427, 465-466, 473, 499-500, 513, 516, 524, 537, 542-543, 548, 551, 573, 581, 585-587, 595, 607, 616, 619, 648, 662, 677-678, 681, 688, 695, 701, 715, 719-720

Undoing, 14 United States, 36, 131, 548, 588, 671 Units, 8, 14, 18, 63, 81, 93, 97, 99, 126, 151, 158, 166,

437, 442, 452, 537, 539, 548, 550-551, 655-656, 674

Universities, 36, 671, 674 U.S. Department of Education, 94, 594 Utensils, 82, 403, 407 Utilization, 710, 728 Utterances, 87, 458, 483-484, 539

V Vacation, 82, 168, 371 Validity, 7, 10, 13, 15-17, 19, 38, 42, 68, 79, 85, 89, 91,

122-127, 142-143, 178, 180, 182, 184, 193-194, 215, 242, 246, 250-253, 256-259, 261-267, 270-271, 273-274, 345, 370, 387, 448, 508, 520, 609, 616-617, 665, 667, 687, 689, 696, 699-700, 720, 727-728

criterion, 10, 16, 19, 38, 142, 184, 242, 258, 616, 700

face, 182, 262 threats to, 123-125, 143, 180, 252, 257

Values, 1, 7, 10, 12-14, 17, 19, 24-25, 61, 98-99, 124, 126, 128, 130-133, 142-144, 149-151, 159, 162-163, 166-168, 171, 174, 176-177, 178, 185-186, 188-189, 195, 209, 211, 295, 310, 336, 519, 642, 668, 680, 721

philosophy, 12 Variables, 2-4, 7-10, 14-16, 19, 23-24, 26, 30, 32-34,

36, 38, 40-43, 45, 47-48, 50-51, 56, 58-59, 62-66, 69-71, 75, 83, 93, 104, 120, 128, 133, 140, 147-148, 150, 152, 158, 162, 165, 169-170, 174-176, 178-189, 191, 193-195, 196-197, 199, 201-202, 205-206, 208, 215-216, 218, 220, 223, 227, 234-238, 240, 244, 245-258, 263-267, 269-274, 278, 281, 283, 341-342, 350, 356, 363-364, 366, 368-369, 371, 389, 391, 393, 395, 400, 403, 405-407, 409, 427-428, 432, 470, 472, 475, 480, 498, 512, 514-516, 520, 531, 533, 537-539, 542-544, 547-551, 553-557, 561, 568, 584-587, 603, 607-608, 619-620, 641, 643, 656, 663, 681-682, 692, 707-709, 711, 715, 725

antecedent, 2-4, 7-9, 14-15, 19, 30, 42, 48, 51, 62, 65-66, 71, 185, 187, 191, 195, 281, 342, 366, 369, 391, 393, 405, 409, 427-428, 432, 498, 512, 514, 516, 533, 539, 542, 548, 553-557, 584-585, 619, 641, 656

isolation of, 180 measurement, 3, 7, 9-10, 14, 19, 24, 36, 38, 41, 43,

45, 69, 75, 93, 104, 120, 128, 133, 140, 147, 150, 158, 165, 169-170, 176, 179, 181, 185, 187, 189, 191, 193-195, 216, 218, 223, 234-238, 240, 244, 246-247, 249, 252-254, 257, 267, 271-273, 432, 475, 692

ranked, 83 scaled, 16, 128, 152, 169, 176, 267

Variance, 127, 698, 700 Vegetables, 290 Ventilation, 378, 571 Veracity, 291 Verbal ability, 316

Verbal behavior, 2, 9-11, 16, 20, 28, 63, 66, 70, 80-81, 90, 213, 280, 288, 405, 407, 427, 477, 483-484, 536-557, 561, 620, 690-692, 694-695, 699, 703, 707-709, 711-712, 715, 717-723, 726, 728

content and, 556 Verbal cues, 652 Verbal praise, 23, 204, 333, 431, 433, 573-574, 616,

656 Verbal reprimand, 293, 353, 357, 364 Verbal self, 296, 692 Verbal skills, 79, 307, 531, 541, 566 Verbs, 32, 133, 147, 537, 540, 552 Vermont, 685 vicarious reinforcement, 630, 704 Victim, 345, 360, 479, 679 Video, 116-119, 129-130, 201-202, 224, 257, 259,

261-262, 297-298, 427, 594 Video games, 297-298 Videos, 552 Videotape, 104, 116-117, 132, 254, 427, 463, 695, 723 Videotapes, 117, 133, 257, 645 Videotaping, 116, 130 Vignettes, 129 Virtue, 54, 279, 281, 682 Vision, 47, 240, 417 Visual aids, 696 Visual cues, 604 Visual discrimination, 712 Visual impairment, 64, 472, 492

causes, 64 Visual impairments, 102, 263, 715, 724 Visual prompts, 419 Visualizing, 62 Vitamins, 94 Vocabulary, 35, 78, 281-282, 548, 551, 694, 726

sight, 694 Vocal tics, 728 Vocational training, 623, 650 Voice, 55, 104, 117, 158, 254, 287, 325, 357, 361,

429, 456-459, 530, 546, 593, 629, 641-643, 696, 710, 719

Voice volume, 104, 117, 254, 456-458, 696 Voicing, 287 Volume, 31, 104, 117, 254, 286, 293, 312, 445,

456-459, 492, 516, 538, 616-617, 645, 670, 696, 709, 727

volunteering, 259 Volunteers, 128, 478 Vomiting, 49, 445 VORT, 439 Vort Corporation, 439 Vowel sounds, 102 Vowels, 413, 707

W Wages, 331 Walden Two, 28, 31, 721 Walls, 37, 249 wants, 57, 78, 80, 82, 95, 130, 180, 182, 284, 297,

444, 449, 459, 524, 540-541, 551, 567, 574, 581, 584-585, 596-598, 603, 606, 612, 629, 651, 665

Warmth, 287, 289, 394, 546 Warning signs, 584 Washington, 36, 293, 379, 676, 685-687, 689,

700-701, 703, 713-714, 721, 725 Watson, 27, 29-30, 42, 227, 236, 249, 585-586,

597-598, 615, 724, 726 Wealth, 381 Weed, 64, 629 Weight, 45, 48, 81, 95, 451, 559, 584, 589, 605,

617-618, 711, 722 Welfare, 78, 631, 666, 668-669, 684, 713 Well-being, 666, 673, 683 West Virginia, 36, 700 White, R., 694 Whole, 15, 18, 20, 52, 92, 97, 110-112, 114-115, 120,

130, 137, 183, 201, 248, 386, 442, 461-462, 489, 547, 566, 576-577, 587, 593, 646-647, 652, 683, 691

Whole-class instruction, 652 Windows, 119, 353, 594 Wisdom, 661 Withdrawal, 1, 18, 20, 49, 56, 58, 61, 66, 150,

192-193, 196-198, 200, 205-206, 218, 244, 278, 310, 318, 375-376, 381-383, 387-389, 476, 551, 553, 574, 588, 656-658, 663, 673, 676, 686-687, 710, 719

750

Withdrawal designs, 196, 719 Women, 150-151, 336, 502, 613, 636, 667, 705 Word analysis, 212-213 Word identification, 421 word problems, 562, 627, 630, 635 Word reading, 319, 694, 705, 728 Words, 8, 16, 25-26, 31, 35, 37-38, 45, 56, 58, 60, 71,

87, 94-96, 98, 102-104, 109, 116, 123-124, 126, 131-132, 140, 150, 156-158, 161, 166, 181-183, 203, 209-212, 255, 264, 267, 283, 286, 307-308, 318-319, 326, 329, 356, 358, 367, 401, 405, 414, 417-418, 421, 431, 449, 455-458, 483, 502, 524, 531, 533, 537, 539-543, 546-547, 549, 551-555, 587, 603, 606-608, 611, 616, 625, 627-628, 630, 632, 634-635, 642-643, 656, 661, 667, 677, 688, 701, 720

base, 26, 102, 661 classes of, 60, 109, 421 exception, 45, 156, 367, 642 generative, 587 manipulation of, 26, 37, 405, 455, 502

Work, 2, 23, 25, 35-37, 39-40, 43, 50, 55-56, 59-60, 64-65, 74, 76, 80-81, 83, 87, 93, 96, 108, 118, 123, 129, 140-141, 147, 181, 197-199, 238, 246, 248, 257, 259-260, 262-263, 267-269, 274, 277-278, 281, 283, 289-290, 296-297, 302, 307-308, 313-314, 316-318, 320, 323, 327, 329, 331, 336, 340-341, 350, 355-356, 366-367, 371, 388, 400, 404, 410, 413, 419, 427, 455, 462, 465, 469, 483, 486-487, 499, 521, 524-529, 531, 533, 537-538, 543-544, 552, 556, 560-562, 564, 584-585, 588, 590-591, 596, 598, 601-611, 613, 616, 623-625, 634, 639, 641-644, 646-647, 650-652, 661, 666, 668, 671-672, 674, 678, 680, 683, 692-693, 695, 698, 703-704, 706-707, 713, 715-717, 719-720, 724-725

Work habits, 623 Workbooks, 87 Workplace, 39, 185, 371, 589, 673 Worksheets, 74, 116, 129, 134, 141, 208, 268-269,

291, 561, 605-606, 646, 650 Wrist counter, 108, 361, 610, 708 Writers, 282, 477, 598, 607 Writing, 31, 55, 59, 63, 74-76, 87, 95, 97, 100, 103,

117, 128, 156, 168, 272, 277, 281-283, 292-294, 296, 485, 522, 524, 539, 542, 545, 556, 561, 564, 566, 572, 582, 587-588, 598, 603, 608, 616, 623, 646, 648, 651, 682, 687, 689, 694, 696, 701, 709-710, 712, 718

as punishment, 59 explanatory, 587 form of, 59, 95, 282-283, 293, 522, 556, 587, 616 kid, 55, 603 right, 55, 95, 103, 564, 598, 651 talking and, 282 to solve, 603

Writing skills, 566, 701 Written language, 83, 537, 561, 714, 728

X X axis, 149, 158-160, 165, 167, 176

Y Y axis, 16, 149, 152, 156, 158-160, 167, 171, 176,

487, 489 Yale University, 712 Yoga, 589, 699 Young children, 39-40, 109, 234, 356, 376, 418, 432,

540, 546, 604, 606-607, 646, 671, 687-688, 694, 703, 716-720, 725-726

Z Zero, 6, 19, 99, 102, 104, 116, 149, 160-161, 171-172,

177, 190, 226, 231, 240, 314, 348, 357, 359, 363, 376, 412, 456, 469-472, 479, 487-488, 490, 492, 494, 497, 500, 608, 657, 721

Zoloft, 531

751

752

Cover
Table of Contents
Glossary
1. Definition and Characteristics of Applied Behavior Analysis
2. Basic Concepts
3. Selecting and Defining Target Behaviors
4. Measuring Behavior
5. Improving and Assessing the Quality of Behavioral Measurement
6. Constructing and Interpreting Graphic Displays of Behavioral Data
7. Analyzing Behavior Change: Basic Assumptions and Strategies
8. Reversal and Alternating Treatments Designs
9. Multiple Baseline and Changing Criterion Designs
10. Planning and Evaluating Applied Behavior Analysis Research
11. Positive Reinforcement
12. Negative Reinforcement
13. Schedules of Reinforcement
14. Punishment by Stimulus Presentation
15. Punishment by Removal of a Stimulus
16. Motivating Operations
17. Stimulus Control
18. Imitation
19. Chaining
20. Shaping
21. Extinction
22. Differential Reinforcement
23. Antecedent Interventions
24. Functional Behavior Assessment
25. Verbal Behavior
26. Contingency Contracting, Token Economy, and Group Contingencies
27. Self-Management
28. Generalization and Maintenance of Behavior Change
29. Ethical Considerations for Applied Behavior Analysts
Bibliography
Index