DP

Chapter 5

Schedules of Reinforcement in Organizational Performance, 1971-1994: Application, Analysis, and Synthesis

Donald A. Hantula

The experimental analysis of behavior is concerned with identifying, to the extent it is possible, orderly relationships between the behavior of the intact individual and the environments within which the individual behaves, including organizational and social environments. Respondent behavior reflects effects of antecedent environmental stimuli on the individual's reflexive behavior. Operant behavior, on the other hand, is behavior of an individual that acts upon the environment, producing a consequent change in it or the individual's orientation to it (i.e., locomotion). Operant behavior typically exhibits sensitivity to its immediate, and sometimes delayed, consequences (Baum, 1973), and antecedent stimuli can evoke changes in operant behavior when they have reliably predicted its consequences. Relations among antecedents, operant behaviors, and their consequences (A-B-Cs) are described as three-term contingencies and contingencies of reinforcement by Poling and Braatz ( Chapter 2 , this book).

Operant behavior has captured the attention of applied researchers and behavior managers because changes this class of behavior produces on organizational environments can contribute to accomplishment of organizational objectives. And the effective management of behavior in organizations depends critically on its predictability and controllability. Thus, knowledge concerning the variables of which operant behavior is a function is essential to effective behavior management. One of the variables of which operant behavior is a function, and which has been thoroughly and systematically analyzed by basic behavior analytic researchers, is the relationship between a behavior and its consequence(s). These B-C contingecies or “if B then C” relationships are called schedules of reinforcement (Ferster and Skinner, 1957).

Schedules of reinforcement are environmental rules or E-rules (Baum, 1973). A full appreciation of their role in the orderly variation of behavior across environmental changes, e.g., changes in E-rules, is gained when their role in the description of operant behavior reinforcement, punishment, or extinction is understood. Zeiler (1977) succinctly presented this context for schedules of reinforcement as follows:

The word reinforcement refers to the effect of an operation; it does not describe an independent variable but is the interaction of an independent variable with behavior. By reinforcement is meant an increase in responding as a function of a stimulus event following a response. The stimuli having these effects are reinforcing stimuli or reinforcers. Schedules of reinforcement are the rules used to present reinforcing stimuli, (p. 202)

Unlike reinforcement per se, a schedule of reinforcement can be an independent variable. Manipulating schedules of reinforcement can, under certain conditions, produce predictable effects on dimensions of operant behavior such as changes in its rate, maintenance of its rate, establishment and maintenance in patterns of rate variation (high at some times and low at others), and change in operant behavior topography. Although they may be subtle and sometimes difficult to identify, work environments include E-rules that function as schedules of reinforcement (hereafter referred to as schedules) and as such exert powerful control of behavior within the work setting.

THE BASIC IMPORTANCE OF SCHEDULES

All events occur in time. All behaviors and consequences transpire across a temporal plane. However, despite the seemingly basic and selfevident importance of behavior's temporal dimension, it has been largely minimized and most often ignored in organizational behavior research. In the few instances in which time is considered seriously in organizational behavior research, it is conceptualized as a moderating variable (e.g. Gersick, 1988, 1989, 1994), but not as an independent variable in its own right. However, because organizations are formal structures or administrations of people that exist in space through time and are not momentary gatherings of individuals (Scott, 1992), this ignorance of behavior's temporal dimension poses a serious challenge to understanding and managing behavior effectively. Indeed, the behavior analytic perspective explicitly recognizes the significance of temporality and has historically addressed time as both a moderating variable and as an independent variable in its own right. Thus, the accumulated knowledge of schedules of reinforcement is among the most important features of the distinctive and distinguished contribution that behavior analysis can make to understanding and managing behavior in organizations.

Since publication of Schedules of Reinforcement (Ferster and Skinner, 1957), basic research in schedules has flourished to the point that schedule research may be the most highly developed and theoretically fertile area in the experimental analysis of behavior (Zeiler, 1977, 1984). This research is characterized by orderly data, sophisticated mathematical models, complex questions, and theoretical advances that illustrate the vast and profound influences schedules exert on individual behavior. However, despite some early interest in the academic and professional management literatures concerning how to apply reinforcement schedules to address organizational challenges (e.g., Aldis, 1961; Jablonsky and DeVries, 1972; Nord, 1969; Organizational Dynamics, 1973), schedule applications in organizations and in simulated organizational settings have lagged behind developments in the basic science of behavior or behavior analysis by decades.

Reviews of schedule research in organizations and in simulated organizational settings (Ayllon and Kolko, 1982; Latham and Huber, 1992; Mawhinney, 1986) find that the majority of past research has been limited to investigations comparing different simple schedules and their effects on job performance. In contrast, the basic literature in schedules has abandoned this comparative and parametric research in favor of such investigations as: preference and choice between different schedules, drug effects, dynamics of schedule performance, and synthesis of complex schedules (see Davison and McCarthy, 1988; Honig and Staddon, 1977; Zeiler, 1984 for further elaboration).

The focus of early research into effects of reinforcement schedules on task performance and the dominant focus of organizational behavior management has been at the front line or shop floor level of employees within work organizations (Balcazar et al., 1989; Hantula, 1992; O'Hara, Johnson, and Beehr, 1985). More recently, however, researchers have employed simulated organizational settings to address more complex questions about performance in organizations including financial decision making (e.g., Goltz, 1992, 1993; Hantula and Crowell, 1994a, b) leader effectiveness (e.g., Rao and Mawhinney, 1991), and perplexing theoretical issues surrounding these phenomena.

Ayllon and Kolko (1982) supplied an earlier review of schedules of reinforcement in OBM research, and Latham and Huber (1992) provide a more recent review of application-focused schedule research. Rather than reiterate previous reviews, this chapter will focus on basic mechanisms of schedule performance, catalog past studies, and discuss recent advances and research in schedules of reinforcement in OBM. In this review, the research will be categorized as one of three levels of inquiry: application, analysis, and synthesis. Application studies are those in which schedules of reinforcement are used as tools to change important work performances. In analysis studies, ongoing work performances are examined to determine the schedules that may be exerting control. Synthesis studies are those in which known behavioral effects of schedules are used to create performances that resemble those occurring in organizations. This categorization scheme is a useful means to organize the research because schedules of reinforcement are not only applied tools that may be used judiciously by the OBM practitioner to manage important work performances, but are also important theoretical instruments that enable the OBM researcher to approach more complex and perplexing questions. Thus, other features of schedule research, such as subjects, settings, and performances addressed, should vary as a function of level of inquiry, as should the type of research questions addressed.

SCHEDULES OF REINFORCEMENT: THE BASICS

Schedules may be defined on the basis of responses, time, complexity, and periodicity (see Ferster and Skinner, 1957; Williams and Johnston, 1992; Zeiler, 1977, 1984). Although “schedules of reinforcement” implies a program or timetable for presenting reinforcement, in practice, the term has come to refer to not only the presentation of reinforcement, but also to the presentation of contingent punishing events and noncontingent events as well. Indeed, whether or not a consequent event will act as a reinforcer or punisher may change depending on the schedule under which it is delivered (see Morse and Kelleher, 1977). Schedules may have both direct-acting and indirect-acting effects on behavior. Direct-acting effects are generally those resulting from the response-reinforcer contingency of a particular schedule, such as the number of responses required to obtain reinforcement. Indirect-acting effects result from formal properties of a particular schedule, or often from the contextual impacts of a particular schedule or combination of schedules (Baum, 1973). For example, whether or not an individual will continue to invest resources in a venture that fails to return positive consequences is due both to the direct-acting effects of receiving repeated negative feedback during the venture's failure and the indirect effects of the pattern of positive consequences arranged by the schedule that was in effect prior to the venture's failure to continue delivering good returns (Goltz, 1992,1993; Hantula and Crowell, 1994a).

Schedules may operate singly. When they do they are called simple schedules. When schedules occur in concert with other schedules, they are called complex schedules. A complete account of complex schedules is beyond the scope of this chapter (see Catania, 1992 for a review). Examples of complex schedules often found in organizational research are concurrent schedules, in which two or more schedules are available simultaneously, and multiple schedules, in which two or more schedules operate sequentially. The following sections describe basic schedules of reinforcement and provide examples of their use in organizational contexts.

Response-Based Schedules

Response-based or count-based schedules are called ratio schedules. They refer to rules of administering a reinforcement contingent upon the completion of a certain number (count) of responses. These schedules are similar to piecerate pay systems. On ratio schedules, rate of response and rate of reinforcement are highly correlated; higher rates of response result in higher rates of reinforcement. Ratio schedules in which reinforcement is made available after every fixed number of responses (after the Nth response) are called fixed-ratio schedules (or FR-N, where N refers to ratio of responses to reinforcements received). Ratio schedules in which the ratio of response to reinforcements received (or N) varies about a mean of N responses per unit of reinforcement received are called variable (VRN) schedules of reinforcement. A continuous reinforcement schedule (CRF), in which reinforcement is made available after each response, is analogous to an FR-N schedule for which N = 1, i.e., an FR-1 schedule of reinforcement. VR schedules generally support steady responding, while performance on FR schedules is generally characterized by “break and run” patterns in which rate of response increases as the behavior progresses from far from contact with the reinforcement to very near the completion of the ratio. For example, work rate on an FR-50 schedule might be somewhat slow up to completion of twenty or thirty responses and then increase rapidly up to the moment the reinforcement is received. Ultimate receipt of a reinforcer is often followed by the work rate break or a pause. The break in rate of response following a reinforcement is so reliable it is referred to as a postreinforcementpause.

Latham and Dossett (1978) compared the effects of CRF and VR schedules on the performance of rodent trappers. An FR-1 (or CRF) schedule in which trappers were paid $1 for each rodent trapped was later changed to a VR-4. Each time a trapper caught a rodent and correctly specified the color of one of four marbles before drawing it from a bag, he was paid $4. It was found that the VR-4 resulted in more rodents trapped than did the FR-1 for experienced trappers, but the reverse was found for inexperienced trappers. Saari and Latham (1982) studied the same group of trappers four years later and found that the VR-4 resulted in more rodents trapped than did the FR-1 for all trappers. It should be noted that in these studies, the two schedules were equivalent in terms of their cumulative value. The FR-1 paid $1 with a probability of 1.0; thus the overall expected value on this schedule is $1. The VR-4 paid $4 with a probability of .25; thus the overall expected value on this schedule is also $1. Because the expected value for each schedule was equivalent, the results obtained were not due to the magnitude or amount of reinforcement, but resulted from the scheduling of the reinforcers.

The issue of ratio “stretching effects” is of some interest to organizational researchers (Mawhinney, 1986). Ratio schedules of reinforcement (for example, piece rate pay systems) may be thinned or “stretched” by increasing the ratio requirement for reinforcement over time, so that additional work or response is necessary to maintain the same rate of reinforcement (Ferster and Skinner, 1957). Clearly, schedule stretching in regard to people's wages is abusive because the organization reaps the increased rewards of additional work while the employee must labor harder just to maintain a similar level of income under the stretched (higher value of N on the FR-N or VR-N) schedule. However, ratio stretching may also be employed to facilitate the maintenance and generalization of behavior using social reinforcers such as praise and recognition, as well as by gradually exposing the individual to larger periods of time in which reinforcement is delayed, but ultimately delivered. Stretching of this sort has the advantage of forestalling the known effects of satiation with respect to social reinforcements.

McNally and Abernathy (1989) used titrating or stretching schedules to promote initial adoption and use of ATMs (automatic teller machines) by bank customers. In this study, ATMs were programmed to play a “game” in which the customer was to enter a two-digit number. If the customer “won,” a cash payoff of $5 to $100 was made. As individual customers used the ATM more frequently, the probability of “winning” decreased from a .50 probability for the first two uses, to .20 for the next eight uses, to .10 for the next ten uses, to .07 for the next twenty uses and finally to .05 for over forty uses. This titrating schedule arrangement resulted in increased use of ATMs, the goal of the project, as well as sustained use after these contingencies were withdrawn.

Time-Based Schedules

Using time-based schedules, reinforcement is delivered after a fixed period of time elapses (where FT refers to fixed time), or after a variable period of time elapses (where VT refers to variable time) whether or not any response has occurred. Consequences on FT and VT schedules are noncontingent with respect to specific operant responses. Although there is no correlation between rate of responding and reinforcement on FT or VT schedules, laboratory research indicates that these schedules do maintain low rates of response (Zeiler, 1977). While at first glance and VT schedules may not appear readily applicable to organizational settings, they may be useful in analyzing ongoing organizational activities such as pay delivery, performance reviews, inspections, and deadlines. Organizational events that occur regularly over time may evoke or facilitate behavior seemingly related to, but not maintained by these events. Called adjunctive behaviors (Falk, 1986; Thompson and Lubinski, 1986), these behaviors are not necessarily maintained by any reinforcing or punishing consequences delivered according to a schedule, but are evoked by the periodicity or pattern of the schedule itself.

Challenges in compensation administration may exemplify some of the differences between behavior maintained on time-based and response-based schedules. In most organizations employees are paid according to “standard” compensation plans in salaries or hourly wages and additional benefits, which may be fairly straightforward to administer, but may not be optimal in terms of motivating actual work performance. Such compensation plans provide money regularly but are largely noncontingent on job performance (Nord, 1969; Wallace and Fay, 1988), which is analogous to an FT or VT schedule. If any contingency is operating in this context, it is one of a compound schedule, in which fulfilling the requirements of one schedule (e.g., regular attendance) gives the employee access to the FT or VT salary schedule (see also, Organizational Dynamics, 1973).

George and Hopkins (1989) provide an interesting comparison of the effects of FT (hourly) pay vs. FR (performance-contingent) pay on the performance of waitpersons in three family-style restaurants. In this study, waitpersons’ pay was changed from an hourly wage (FT) to an FR-type schedule that provided the individual with 7 percent of individual receipts (a generally accepted industry standard). Policies regarding tips and all other terms of employment remained constant. After instituting the FR schedule, mean pay increased 20 to 30 percent, mean sales increased 18 to 36 percent, mean customers served per hour increased 18 to 26 percent, and labor costs were unchanged. A similar study (Wagner, Rubin, and Callahan, 1988) examined long-term (114 months) effects of FR pay in a unionized iron foundry in which employees were switched from an FT (hourly) wage to an FR wage based on gains over standard production hours. Production showed a positively accelerating increase (resembling a learning curve) over a period of years. The FR wage based on gains system eventually produced performance rates twice those maintained by the FT wages. This provides prima facie evident that the direct correlation between response and reinforcement under this ratio schedule continued to exert long-term effects on performance. At the same time labor costs and grievances declined significantly, suggesting ancillary effects of the FR wage schedule.

Although the vast majority of research in schedules of reinforcement has been directed toward the behavior of individual subjects, similar schedule effects may also be seen at more macro levels. Baum (1974) observed that schedule-based analyses described choice behavior of populations as precisely as it had described choice in individuals. Staw (1991) argues that it should not be surprising that organizational action can be explained by individuallevel principles and theories because, ultimately, organizations are collections of individuals. An example of schedule-induced organizational action is provided by Weisberg and Waldrop's (1972) analysis of some work habits of the United States Congress. In this study, cumulative numbers of bills passed were observed to increase slightly each month during the beginning of a legislative session, but increased rapidly as the session drew to a close, evidencing a scallop pattern of acceleration characteristic of FT or FI responding; a phenomenon whose existence in human affairs has been questioned by Hyten and Madden (1993). However, according to Sulzer-Azaroff and Mayer (1991), the schedule operating on the behavior of Congress is more properly characterized as a limited-hold schedule in which there is a restriction placed on an interval schedule requiring that, for reinforcement to occur, the response must occur within a particular time limit following the interval. Unless bills prepared during a given congressional session reach the floor prior to the regularly scheduled congressional recesses, reinforcement associated with their passage during the session is lost upon recess. The process recurs when the Congress reconvenes. Despite this terminological difference, the organization of Congress's work behavior around fixed intervals is striking and illustrates not only the powerful and pervasive effects schedules can exert over a period of years, but also the extent to which schedules may be determinants of cyclic or ritualistic behaviors in cultures and organizations (Falk, 1986).

Time and Response-Based Schedules

For combined time and response-based schedules, referred to as interval schedules, reinforcement is made available after a period of time elapses, but procurement of the reinforcer depends on occurrence of a response. Interval schedules may be either fixed (FI or fixed interval) or variable (VI or variable interval). The interval in fixed and variable interval schedules is a specified interval of elapsed time, t. Thus, FI and VI are typically designated Fl-t and Vl-t, where t refers to fixed elapsed time and average elapsed time respectively. Interval schedules maintain steady rates of response, with FI schedules sometimes evoking a positively accelerated “scallop” curve of response rate. Although only a single response is required for reinforcement on interval schedules, once the required time has elapsed, these schedules can engender many more responses than are necessary to obtain reinforcement. Because interval schedules maintain regular and steady rates of response, they have become popular as a means of establishing stable baseline response rates. Stable baseline response rates can serve as tools in basic behavior research for investigating the operation of many other variables and processes. Introducing these other variables in the context of the otherwise steady-state baseline behavior rate can reveal their effects as observed departures from the baseline behavior rates coincident with their introduction.

Alavosius and Sulzer-Azaroff (1990) compared the effectiveness of performance feedback delivered under CRF and FI schedules in the acquisition and maintenance of safely performed health care routines. Subjects were given feedback on either CRF or weekly FI schedules regarding the proportion of each patient health care routine performed safely. Both schedules led to better levels of performance than did written instructions. Although the performance reinforced under the CRF schedule reached mastery level in fewer work days than did the performance reinforced under the FI schedule, an equivalent number of feedback messages were needed to achieve mastery under each schedule.

The VI schedule is especially useful for studying the dynamics of behavior in probabilistic or equivocal environments. A properly programmed VI schedule (see Fleshier and Hoffman, 1962; Hantula, 1991) holds the probability of reinforcement constant over any given time horizon while keeping the occurrence of reinforcement at one time uncorrelated with the occurrence of reinforcement at any other time. Thus, a VI schedule allows a researcher to examine the effects of overall probability or density of reinforcement in a given situation or context, without the confounding of correlations between reinforcer deliveries.

For example, Hantula and Crowell (1994b) used multiple VI-VI schedules to study the effects of an alternative investment opportunity on financial decision making under conditions of equivocality and failure. Returns on investments were first provided on equal value multiple VI-VI schedules, and later one of the investments ceased to yield returns, while the other remained unchanged (a multiple VI-EXT schedule). (Recall that extinction [EXT] is termination or cessation of a previously operating schedule of reinforcement.) Previous research indicated that investment allocations escalated on an EXT schedule when EXT was preceded by experience with an intermittent schedule (Goltz 1992, 1993; Hantula and Crowell, 1994a). However, these studies did not include an alternative investment. McCain (1986) showed that presence of an alternative investment can attenuate escalation in a failing course of action (EXT context for one alternative) but did not document how funds might be reallocated to the alternative (if at all). In Hantula and Crowell (1994b), investing in the unchanged schedule alternative (i.e., the one that continued to deliver rewards) increased 1.5 times the original level while investing in the failing (changed to EXT schedule) alternative nearly ceased. This finding is consistent with basic research in behavioral contrast (Reynolds, 1961). But this finding was entirely unexpected and would not have been predicted based on the existing literature regarding decision-making behavior.

SCHEDULES OF REINFORCEMENT: THE RESEARCH

Table 5.1  summarizes past schedule research from 1971 to 1994 in organizations or under simulated organizational conditions. Included in this table are all empirical schedule-based experiments performed in actual organizations, schedule-based analyses of behavior in organizations, and laboratory studies that employed a schedule-based analysis of performance in a simulated organizational setting, or which addressed directly theoretical concerns in organizational research. In general, it may be concluded from these studies that (1) reinforcement schedules are an effective means for managing work performance; (2) reinforcement schedule effects on work performance in the field are generally similar to those found in organizational laboratory simulation research; (3) differences in schedule parameters have mixed effects, although the presence or absence of a schedule of contingent reinforcement accounts for the largest effects; (4) there is a good deal of promise in further application and investigation of reinforcement schedules in organizations and in simulated organizational settings; and (5) this promise of schedule research for organizational studies has gone largely unfulfilled.

TABLE 5.1. Summary of Schedule-Based Field, Simulation, and Analog Studies

Authors and Year	Outcome or Performance	Subjects and Method	Results	Level of Inquiry
		FIELD STUDIES
Weisberg and Waldrop (1972)	Passing bills	Members of U.S. Congress; cumulative number of bills passed in each legislative session over fourteen-year period examined.	Negatively accelerating work habits (scallop curve) resembling FT performance.	Analysis
Ayllon and Carlson (1973)	Absenteeism	Employees of distributing company; FR schedule, gave tickets for weekly cash lottery to employees not tardy or absent for the entire week.	On-time attendance increased to over 80 percent; employee survey indicated employees liked lottery and attributed positive attitudinal changes in work environment to it.	Application
Everett, Hayward, and Meyers (1974)	Bus riding	Campus bus patrons; tokens given to bus riders under CRF and VR3 schedules.	27 percent increase in riders under CRF and 30 percent increase under VR3 over baseline.	Application
Pedalino and Gamboa (1974)	Absenteeism	Unionized manufacturing and distribution employees; FR schedule, weekly, then biweekly poker games.	18 percent decrease in absenteeism under baseline, no difference in weekly or biweekly games.	Application
Pierce and Risley (1974)	Task completion	Adolescent, minority recreation aides; hourly vs. CRF wages based on percentage of tasks completed.	Percentage tasks completed doubled under CRF wages; providing written job descriptions and termination threats ineffective under hourly wage.	Application
Yukl and Latham (1975)	Tree planting	Tree planters—”marginal workers” paid under hourly wage plus CRF or VR incentives. Reported schedule parameters innacurate.	Higher rate of planting under CRF and VR2 than no incentive; rate under CRF higher than VR2 (actually VR2000) and VR4 (actually VR4000).	Application
Yukl, Latham, and Pursell (1976)	Tree planting	Tree planters seasonal workers paid under hourly wage plus CRF or VR incentives. Reported schedule parameters inaccurate.	Higher rate of planting under CRF and VR2 than no incentive; rate under CRF higher than VR2 (actually VR2000) and VR4 (actually VR4000). Employees preferred CRF.	Application
Pritchard etal. (1976)	Performance on programmed training in basic electronics	Males in late teens; pay for tests passed on hourly, FR3, VR3,orVR3-variable amount schedules.	FR and VR schedules resulted in higher performance and earnings than hourly pay, subjects more satisfied with FR and VR pay than hourly with highest levels under FR.	Application
Latham and Dossett (1978)	Rodent trapping	Unionized beaver trappers; CRF vs. VR4 schedule.	Trapping rate higher under CRF than VR4 for inexperienced trappers, reverse for experienced trappers. Employees preferred VR4.	Application
Dickerson (1979)	Placing bets	Gamblers in U.K. betting offices, Fl schedules.	Betting rate increases as a function of time to race; "scallop" curve for high-frequency gamblers, linear curve for low-frequency gamblers.	Analysis
Pritchard, Hollenbeck, and DeLeo(1980)	Performance on programmed training in basic electronics	Males and females in late teens; pay for tests passed on hourly, CRF, or VR-variable amount schedules.	Higher performace under CRF or VR than hourly, more effort exerted under CRF.	Application
Saari and Latham (1982)	Rodent trapping	Unionized mountain beaver trappers; CRFvs.VR4 schedules; replication of Latham and Dossett (1978).	CRF increased number of rodents trapped 50 percent over baseline, VR4 increased 108 percent over baseline, VR4 schedules perceived as including many motivating job enrichment variables.	Application
Gaetani, Hoxeng, and Austin (1985)	Machining metal parts	Machinists; hourly wage compared to CRF wage (standard wage plus 5 percent of dollar value exceeding standard).	CRF increased productivity 174 to 210 percent over hourly wage.	Application
Evans, Kienast, and Mitchell (1988)	Auto repair	Automobile service mechanics; hourly pay baseline compared with hourly pay plus 1 of 2 lotteries: (1)VR10 lottery for state lottery tickets plus weekly drawings for cash, (2) CRF lottery for state lottery tickets, both lotteries paid off in variable amounts.	Both lotteries increased output > 100 percent over baseline; lottery 2 more cost effective than lottery 1; employee acceptance of lotteries did not appear to be related to performance.	Application
Wagner, Rubin, and Callahan (1988)	Manufacturing iron castings	Unionized foundry employees; compared hourly (FT) wage to CRF based on standard production hours, divided among work groups according to the worth of individual jobs.	Production increased in positively accelerated curve over a period of years, mean increase 103.7 percent over FT wage. Labor costs decreased 33 percent, grievance rate decreased 64 percent 3 years into the CRF pay condition.	Application
George and Hopkins (1989)	Food sales	Waitpersons; compared hourly pay baseline to CRF pay (7 percent of sales), policies regarding tips unchanged.	Under CRF pay per hour increased; sales increased 37 to 61 percent over previous year, customers served/hour increased 18 to 26 percent over baseline; labor costs unchanged.	Application
McNally and Abernathy (1989)	Automatic teller machine (ATM) use	Bank customers; no incentive baseline, individually titrating cash payoff VR schedules for ATM use.	Number of cards used, number of uses per card and ATM transactions as percentage of teller transactions doubled during VR; no reversal when VR removed.	Application
Alavosius and Sulzer-Azaroff(1990)	Patient care	Directcare medical service employees; concurrent CRF and Fl feedback delivery.	Either CRF or Fl more effective than written instructions in learning care routines; more rapid acquisition under CRF, maintenance equal under CRF or Fl.	Application
SIMULATION AND ANALOG STUDIES
Cherrington, Reitz, and Scott (1971)	Clerical work	Junior business students; hourly pay plus contingent (Fl) or noncontingent (FT) pay; "appropriate" (rewards for high performance and none for low performance) and "in appropriate" (reverse).	Output increase for all rewarded subjects, higher for "approximately" rewarded; positive correlations between satisfaction and performance for "appropriately" rewarded subjects.	Application
Yukl, Wexley, and Seymour (1972)	Clerical work	Female student employees; cash payment and schedules: baseline hourly pay, incentive .5 VR2, .25 VR2, or .25 CRF, plus hourly.	Higher output under any incentive schedule than baseline; .5 VR2>.25VR2or CRF, no loss in quality when quantity increased.	Application
Berger, Cummings, and Heneman (1975)	Clerical work	Female student employees; cash payment and schedules: baseline hourly pay,.5VR2,.25 VR2, or .25 CRF, plus hourly pay.	Higher output under any schedule (CRF or VR) than baseline; .5 VR2>.25VR2or CRF.	Application
Farr (1976)	Clerical work	College students; compared hourly pay, individual and/ or group (n = 3) CRF pay.	Output lowest with hourly pay; highest output with combined group and individual CRF pay; hourly subjects set goals below what was achieved by subjects on CRF pay.	Application
Mawhinney, Dickinson, and Taylor (1989)	Playing computer games or trigger pull task as indicator of “intrinsic motivation”	College students: VI schedule for trigger pulls throughout, no pay baseline for computer games, CRF pay for number of preferred games played for one session, reversal to no pay.	CRF pay increased number of preferred games played (quantity) and number of points earned per game (quality) for skill and chancebased games, decreased duration of chancebased games, and decreased number of other games played; baseline levels of game playing and points earned recovered when pay withdrawn, no “undermining” of “intrinsic motivation.”	Analysis
Stoneman and Dickinson (1989)	Assembly	College students; base pay plus CRF individual or group (n = 2,4,5, or 9) incentive.	No difference in output between individual or group pay; greater variability in output as a function of group size.	Application
Frisch and Dickinson (1990)	Assembly	Freshman and sophomore college students; base pay or base pay plus incentives (10,30,60, or 100 percent) of base pay; total compensation held roughly equal for all subjects.	Earned incentive pay half of available; greater output under all incentive pay conditions than base pay alone, no difference between incentive pay conditions.	Application
Hantula (1990)	Investment decisions	Senior business and engineering students; investing in stock under mult VI-VlandVI-EXT schedules.	Equivalent investing under VI-VI, contrast effects (escalation in VI, decrease in EXT) under VI-EXT, effects reversed with schedule changes.	Synthesis
Rao and Mawhinney (1991)	Leadership	Male college students in “superior” and “subordinate” dyads; FT and reciprocal FR schedules for superior and subordinate.	Low rates of leader and subordinate response under FT, high rates under reciprocal FR.	Synthesis
Goltz (1992)	Investment decisions	College students; acquisition of stock investing on CRF, FR2, and VR2 schedules; schedule effects examined in extinction.	Acquistion typical of schedule in effect, persistence, and escalation of investing during extinction for VR subjects; 2nd experiment showed subjects with no experience with task behaved as if under VR schedule.	Synthesis
Oah and Dickinson (1992)	Check proofing	Freshman and sophomore college students; pay under either exponential or linear CRF schedule.	More money earned under exponential schedule, no difference in output between either schedule.	Application
Skaggs, Dickinson, and O'Connor (1992)	Playing video games as indicator of “intrinsic motivation”	College students; no pay baseline, CRF pay for number of preferred games played for 5 sessions, reversal to no pay; replication of Mawhinney, Dickinson and Taylor (1989).	CRF pay increased number of preferred games played (quantity), decreased number of points earned per game (quality) and decreased number of other games played; baseline levels of game playing and points earned recovered when pay withdrawn, no “undermining” of “intrinsic motivation.”	Analysis
Goltz(1993)	Allocation of organizational resources	Freshman and sophomore business students; replication and extension of Goltz (1992); acquisition of resource allocation on CRF, FR2, and VR2 schedules; schedule effects examined in extinction.	Acquisition typical of schedule in effect, persistence, and escalation of allocations during extinction for VR subjects, CRF and FR subjects did not persist or escalate allocations; schedule effects accounted for most of the variance, “responsibility” effects accounted for none.	Synthesis
Hantula and Crowell (1994a)	Investment decisions	Freshmen and sophomore college students; replication and extension of Goltz (1992); acquisition of resource allocation on CRF and VR2 schedules; schedule effects examined in extinction.	Acquistion typical of schedule in effect, persistence, and escalation of investments during extinction for VR subjects only, 2-step escalation/deescalation function found.	Synthesis
Hantula and Crowell (1994b)	Investment decisions	Senior business and engineering students; stock investing on equal VI-VI schedules, then VI-EXT.	Equal investing under VI-VI; behavioral contrast effects evident, investing in EXT declined precipitously, investing in VI stock escalated.	Synthesis

After experiencing a fallow period, research in organizational applications of schedules of reinforcement was the focus of renewed attention in the 1990s that promises to continue with vigor into the new millenium. The reasons for this resurgence are unclear. Latham and Huber (1992) attribute the revitalization of the research to an increased interest in performance-based compensation systems, which are de facto schedules of reinforcement; possibly this renewed interest is a reflection of an increased activity in schedule research and application in applied behavior analysis as a whole (e.g., Mace et al., 1994; Neef et al., 1992; Pierce and Epling, 1995), or perhaps interest in schedules may be schedule dependent, and is what Zeiler (1984) has alluded to as a sleeping giant that is waking again. Nevertheless, the rejuvenation of schedule-based application and research in organizational behavior management bodes well because it strengthens the ties between the basic science (the experimental analysis of behavior) and one of its applied branches. This trend should stimulate both theoretical progress and the development of more effective technologies (Mace, 1994).

APPLICATION, ANALYSIS, AND SYNTHESIS

Application studies are those in which schedule-based interventions have been designed to change organizationally important performance. This work is often experimental or preexperimental in nature. For example, Pedalino and Gamboa (1974) used a lottery system on an FR schedule to decrease absenteeism as did Evans, Kienast, and Mitchell (1988) on CRF and VR schedules to increase mechanics’ productivity; and Gaetani, Hoxeng, and Austin (1985) employed a CRF (piece rate) pay system to increase productivity in a metal machine shop.

Application studies make up 72 percent of extant schedule research. That the majority of the research is application studies should be expected because the bulk of studies in the organizational behavior management tradition have been focused on application research, often with line-level or shop floor employees (Hantula, 1992; O'Hara, Johnson, and Beehr, 1985). Similarly, 70 percent of the application studies reviewed were conducted in field settings, and 30 percent in laboratory simulations or analog settings. Application studies have appeared regularly throughout the twenty-year period reviewed.

Analysis studies are those in which ongoing organizational activities, group and individual performances, and environmental events are examined for temporal regularities to determine the types of schedules that are controlling the performance of interest, such as the effects of regularly occurring deadlines on work rate. These studies are often correlational, but may also involve experimental research. For example, Weisberg and Waldrop (1972) and Dickerson (1979) observed and recorded the behavior of the U.S. Congress and British horse racing bettors respectively and determined that in both cases the subjects’ behavior exhibited the temporal regularities expected on FT or FI schedules of reinforcement. In an example of an experimental laboratory study, Mawhinney, Dickinson, and Taylor (1989) arranged CRF and VI schedules for video game playing and trigger pulling to examine the alleged “undermining of intrinsic motivation” by “extrinsic” reinforcers and found typical schedule performance, but no “undermining of intrinsic motivation.”

Analysis studies make up 12 percent of the research reviewed and were equally divided between field and laboratory settings. Half of the analysis studies appeared after 1989. Although not included in  Table 5.1  as empirical studies, three important conceptual papers appeared that broached a schedule-based analysis of ongoing organizational activities. Two (Gowen, 1990; Mawhinney and Gowen, 1990) explored a schedule analysis of gainsharing programs in organizations, a theme reiterated by Latham and Huber's (1992) review of schedule research, which interpreted schedule research in terms of compensation and incentive management. Redmon and Lockwood (1986) outlined a qualitative approach to analyzing both intra- and extra-organizational activities based on the matching law (Herrnstein, 1970; Baum, 1973).

Synthesis studies are those in which schedules are manipulated to model or build performances that are hypothesized to be controlled by those particular schedules in organizations so that the performances may be studied in the laboratory. This research is experimental and often highly theoretical. Examples of such research include Rao and Mawhinney's (1991) study of leadership as dyadic exchange, which found that interdependent FR-like exchanges of reinforcers in addition to FT pay increases both supervisor and subordinate performance, and noncontingent (FT) pay fails to support any appreciable level of supervisor/subordinate interaction and performance. Another line of research has investigated how investors will escalate resources expended in failing courses of action (i.e., “throwing good money after bad”). This phenomenon is the result of an interaction between an intermittent (VR) reinforcement history (Goltz, 1992, 1993; Hantula and Crowell, 1994a) and the onset of extinction (EXT). It can also be the result of behavioral contrast effects that occur when investment returns previously programmed as VI-VI schedules change to a VI-EXT arrangement (Hantula, 1990; Hantula and Crowell, 1994b).

Synthesis studies make up the remaining 16 percent of the research and are 100 percent laboratory based. All of the synthesis studies reviewed have appeared recently (since 1989), perhaps reflecting a growth and maturation of the OBM field as a whole as researchers address more complex and theoretical issues, while at the same time not abandoning the effective applications that are its stock-in-trade. Both application and synthesis studies share in common a truth criterion of pragmatism. In application studies, the focus is on superimposing a particular schedule or arrangement of schedules onto the present environment to achieve organizationally important outcomes. Issues of generalizability in application studies concern whether or not a particular schedule arrangement will have the same effects for a different performance or in a different organizational context.

Although synthesis also involves superimposing a schedule or arrangement of schedules onto a present environment, there are subtle but important differences between the two levels of inquiry. Unlike application studies, synthesis studies are concerned with constructing schedule arrangements that closely model or produce behavior found in organizations by first creating “reinforcement” histories for subjects, and then studying the effects of schedules or other variables while controlling for confounding variables that cannot be controlled in field settings. Issues of generality are concerned with whether a particular schedule arrangement effectively produces the behavior under examination, and then whether these effects can transfer into the organizational environment.

THEORETICAL ISSUES AND FUTURE DIRECTIONS

Schedules of reinforcement provide a theoretically rich basis from which behavior in organizations may be managed and analyzed. More successful applications to meet ongoing organizational challenges may be designed and built with deftly scheduled consequences. However, the real promise of schedule research in organizational performance lies in the more theoretical realm. Because any behavior or event of any real import in organizations occurs over time, schedule effects are paramount in making sense of activities in some context. Indeed, hypothetically constructed mechanisms can often become trivial when schedule and temporal matters are considered (Hantula, 1992).

Two recent examples are the research in “undermining intrinsic motivation” and escalation of commitment. Both of these phenomena were said to be contrary to a behavioral analysis of complex organizational behavior, and further, although both were said to influence behavior over time, they were hypothesized to result from hypothetical mechanisms measured at one point in time. In the case of “undermining intrinsic motivation,” previous research employing oneshot measurement strategies found evidence that “extrinsic” rewards dilute one's “intrinsic motivation” to perform an activity (Deci and Ryan, 1985). Research using schedules of reinforcement showed that undermining effects, if they occur at all, are fleeting and unlikely if rewards are presented repetitively, rather than once (Mawhinney, Dickinson, and Taylor, 1989; Skaggs, Dickinson, and O'Connor, 1992). Similarly, escalating commitment to a failing course of action is purported to result from either self-justification/ego-defense mechanisms (Brockner, 1992), or cognitive framing effects (Whyte, 1993). Schedule research, on the other hand, indicates that escalation is yet another manifestation of the behavioral phenomenon called “extinction burst.” The extinction burst is often observed when contingencies change from an intermittent reinforcement schedule to an extinction condition (Goltz, 1992; Hantula and Crowell, 1994a). Schedule research also suggests that if self-justification processes are operating within an experimental context, they do not operate past the first decision in a series of decisions (Goltz, 1993). Thus, in both of these examples, an imputed hypothetical causal mechanism may actually be an artifact of ill-considered temporal factors. That the operation of these hypothetical causal mechanisms may be artifacts is not evident unless and until a more contextually rich schedule-based analysis is brought to bear on the behavior in question.

In sum, schedules of reinforcement provide a ready source of method and theory to examine contextual determinants of behavior, and to design more effective interventions. This review has only scratched the surface of the promise and possibilities of schedules of reinforcement and their use in application, analysis, and synthesis of organizational performance. In the case of application, schedules are readily applicable to designing performance-based and incentive pay systems, which are becoming the focus of increasing concern (Wilson, 1995). However, beyond the readily apparent use of schedules in designing incentive systems, schedules may be applied creatively to many organizational challenges from managing consumer behavior to employee drug use.

In terms of managing consumer behavior, while Foxall (1990) broaches the subject of schedules in marketing, and McNally and Abernathy (1989) provide a compelling example of their use, the utility of schedules in this domain remains to be discovered. For example, sophisticated computerized point-of-sale technologies can be used to reinforce purchases with coupons or other discounts delivered according to appropriate schedules. Further, with the growth of “cybershopping” via the Internet and online services (Cronin, 1994), issues such as delay and probability of reinforcement, whether product delivery or system response, come to the forefront. These basic issues, which are essentially schedule properties, may well provide the impetus for further activity in schedule research in this context. Interestingly, just like the reawakened interest in schedules, interest in behavioral applications in marketing is on the rise (Foxall, 1992, 1994). Whether or not these events are causal is open to question, although their mutual resurgence is certainly cause for optimism.

Concerning the issue of employee drug use, schedules may provide an alternative to current drug testing practices, which are extremely controversial (Crant and Bateman, 1990). Presently, drug tests involve analyses of bodily fluids for metabolites correlated with specific drugs; however, such tests are viewed as invasive, and cannot determine whether an individual is impaired by drug use. In addition, the problem of false positives, even if extremely improbable, presents a serious ethical dilemma. Schedules have been widely utilized as baselines for studying drug effects (Dews, 1963). Schulz (1991) describes a computer game that is used to identify fatigued truck drivers; perhaps similar schedule-based games could be used to determine whether an employee may be under the influence of drugs. Because schedule performances are stable over time, significant deviations from an individual's “behavioral fingerprint” that resemble changes in schedule performance correlated with use of certain drugs could signal the need for a “for cause” drug test. Such a behavioral test would be both less invasive and objectionable than current drug tests, and it would also lower the probability of false positives dramatically because only individuals who show evidence of behavioral impairment would be tested. Perhaps a new specialty of organizational behavioral pharmacology may emerge from these types of applications.

In terms of analysis and synthesis, schedules may further bridge gaps between the fields of OBM and finance. Ferguson (1989) has presented a behavioral analysis of stock market crashes based on stimulus generalization and escape behavior that complements recent schedule research concerning determinants of continued investment under conditions of no return (Goltz, 1992; Hantula, 1990; Hantula and Crowell, 1994a, b; O'Flaherty and Komaki, 1992). Currently, the stock market is conceptualized as a white noise or random walk model in which today's stock price does not allow prediction of tomorrow's stock price (Malkiel, 1973). Interestingly, a properly programmed VI schedule provides a similar state of affairs in which the probability of each reinforcement delivery is uncorrelated with other deliveries (Fleshier and Hoffman, 1962; Hantula, 1991). Previous research has employed VI schedules to synthesize investment behaviors (Hantula, 1990; Hantula and Crowell, 1994b) and perhaps further schedule-based analyses of stock performance and investor behavior may lead to intriguing interdisciplinary investigations.

Advances in schedule research that have not been presented in detail, such as probability differential or “response deprivation” models (Podsakoff, 1982), and the matching law (Herrnstein, 1970), also hold a good deal of promise for future research and application. The latter provides a readily applicable quantitative tool for analyzing and managing behavior (Pierce and Epling, 1983, 1995; Redmon and Lockwood, 1986) and a solid link to economics (Hursh, 1980, 1984; Kagel, Battalio, and Green, 1995). Extensions of schedule research to other “organizational” disciplines including cultural analysis (Falk, 1986; Harris, 1980) and training (Pritchard et al., 1976) are promising, as are explorations of job satisfaction (Cherrington, Reitz, and Scott, 1971; Latham and Dossett, 1978; Mawhinney, 1989), and rational choice theory (Herrnstein, 1990a, b) await further research and application. The question to be addressed is not whether schedules are operating in a given context, but which schedules are operating. Indeed, Dews (1963) has suggested that just as osmosis is ubiquitous in its operation in physiology, schedules operate similarly in regard to behavior—wherever and whenever they can, they will.

Reference:

Redmon, W. K. (2001). Handbook of Organizational Performance, 1st Edition.