discussion question 6

JOURNAL OF APPLIED BEHAVIOR ANALYSIS

IMPROVING SOCIAL SKILLS AND DISRUPTIVE BEHAVIOR IN CHILDREN WITH AUTISM THROUGH SELF-MANAGEMENT

LYiNN KERN Koiwom, ROBERT L. KoEGia, CsmI HuRuLy, AND WIn"IAM D. F"A

UNIVERSITY OF CAORNIA AT SANTA BARBARA

The literature suggests that children with autism typically are unresponsive to verbal initiations from others in community settings, and that such unresponsiveness can lead to problematic social interactions and severely disruptive behavior. The present study assessed whether self-management could be used as a technique to produce extended improvements in responsiveness to verbal initiations from others in community, home, and school settings without the presence of a treatment provider. The results showed that children with autism who displayed severe deficits in social skills could learn to self-manage responsivity to others in multiple community settings, and that such improve- ments were associated with concomitant reductions in disruptive behavior without the need for special intervention. The results are discussed in terms of their significance for improved development of social skills in children with autism. DESCRIPTORS: autism, self-management, social skills, language, disruptive behavior

Although language researchers traditionally have been concerned with syntax (i.e., structure) and semantics (i.e., referential meaning), a shift in the field has gradually emerged that emphasizes a func- tional approach to understanding communicative competence and stresses a social perspective of lan- guage use (Kaiser & Warren, 1985). With the shift toward pragmatics, a greater emphasis is now being placed on the social interaction of commu- nicators within a given context, thus resulting in the study of larger units of conversation. Some questions of prime concern for pragmatic assess-

Funding for this research was provided in part by Cali- fornia State Department of Education Special Study Grant No. 42-03651-N27 1-00-33 that funded pilot work (USPHS MH28210 and MH39434 from the National Institutes of Mental Health) and by NIDRR Cooperative Agreement G0087CO234 from the U.S. Department of Education. The authors thank Alfred Bimbela, Yong Cho, Amanda Damron, Diane De la Riva, Sherri Goldstein, Trina Gravelle, Julie Jacobson, Kimberly Mullen, Teresa Paslawski, Deborah Ru- more Parks, Annette Smith, Jodi Smith, Kimi Stahler, Shel- ley Weichman, Laura Wilde, school personnel, and parents for their assistance throughout the study. In addition, we thank Rob Homer, Ted Canf, Wayne Sailor, and Glen Dun- lap for their feedback on earlier drafts of this study.

Requests for reprints should be addressed to Lynn Kern Koegel, Autism Research Center, Counseling/Clinical/ School/Psychology Program, Graduate School of Education, University of California, Santa Barbara, California 93106- 9490.

ment are: (a) Does the child relay a message ac- curately? (b) Can the child adhere to a topic without abrupt transitions to new topics? and (c) Does the child respond to questions and statements appro- priately (Cole & Cole, 1989)?

This shift in focus has been especially important in the treatment of children with autism, whose social and interpersonal relationships are pathogno- monic to the syndrome. Even in Kanner's follow- up study of his original sample approximately 30 years later (Kanner, 1971; Kanner, Rodriguez, & Ashenden, 1972), he observed that the original clients remained extremely aloof and continued to experience significant difficulties in interpersonal re- lationships. Relatedly, some researchers have sug- gested that social deficits may be the primary dif- ficulties in autism (Fein, Waterhouse, Lucci, & Snyder, 1985). Further adding to this problem is the fact that the smooth flow ofconnected discourse depends on the listener's ability to monitor the speaker's messages and to provide feedback to the speaker concerning their effectiveness (Fey & Leon- ard, 1983). This type ofdiscourse regulation during conversational turn-taking is necessary for fluent conversation. Although most typical children are taught to use appropriate communication during their preschool years, many children with disabil- ities who have difficulty with language skills at-

341

1992125,341-353 NumBER 2 (summER 1992)

LYNN KERN KOEGEL et al.

tempt to avoid interactions, thereby disrupting the communicative interaction. Because these disrup- tions occur frequently in children with autism, their ability to fit into integrated community settings is limited. Thus, the critical need for attention to social skills is obvious and of utmost importance.

Integration is now being considered as a primary goal in special education research, and it is now realized that mainstreaming and integrating indi- viduals with disabilities by simply placing them together in the same program or setting without support can be problematic. That is, such place- ment can result in poor social acceptance of the individuals with disabilities, negative rates of social interaction among their peers, poor self-concepts of individuals with disabilities, and a generally neg- ative attitude toward mainstreaming by regular ed- ucators (Gresham, 1986). It is dear that social skills need to be an integral component of the treatment program for children with autism if successfuil in- tegration is to be realized.

Self-management (R. Koegel & Koegel, 1990) is one treatment technique that has been found effective with speech sounds of children (L. Koegel, Koegel, & Ingham, 1986), academic skills with learning-disabled children (Harris, 1986), and in the treatment of children with developmental dis- abilities. Teadcing self-management techniques al- lows the child to manage his or her own behavior in the absence of a treatment provider for poten- tially long periods (cf. Dunlap, Dunlap, Koegel, & Koegel, 1991; L. Koegel, Koegel, & Parks, 1991). The decreased need for staff assistance sug- gests that self-management techniques are ideal for persons with developmental disabilities in inte- grated academic and community settings (Lago- marcino, Hughes, & Rusch, 1989; Moore, Agran, & Foder-Davis, 1989; Ninness, Fuerst, Ruther- ford, & Glenn, 1991; Rhode, Morgan, & Young, 1983). The present study attempted to improve social

skills in children with autism in a number of in- tegrated community settings with as little clinician presence as possible. We were concerned primarily with these children's lack of response when other individuals attempted to interact with them. Be-

cause we wanted the children to be able to interact socially without the continual presence of a treat- ment provider, self-management was chosen as an intervention strategy. We also reasoned that if the children responded more consistently with a com- munication partner, the overall interaction would become more positive and escape- or avoidance- driven disruptive behaviors would decrease. Two primary questions were addressed. First, we asked if self-management could be used as a means of improving social communication skills (i.e., appro- priate responding to others' questions) in children with autism. Second, we asked if such skills were improved on an ongoing basis, would decreases in interfering disruptive behaviors occur when others (e.g., parents, peers, and adults) attempted to en- gage the children in social communicative inter- actions.

METHOD

Subjects Four children with autism participated. All were

diagnosed by outside agencies and referred to our clinic for treatment. Three were in classrooms for children with communicative handicaps, and 1 (Ian) was in a classroom for children with severe dis- abilities. During an initial intake interview, all 4 children demonstrated symptoms in accordance with the diagnostic criteria for autistic disorder described in the DSM III-R, including a lack of responsive- ness to other people, gross impairment in com- municative skills, bizarre responses to various as- pects of the environment, and ritualistic behavior. In addition, each exhibited lengthy histories of dis- ruptive behaviors (as described below).

Criteria for participation in this study were two- fold. First, the children had to be reported by par- ents and teachers as characteristically unresponsive to others' verbal initiations, despite the fact that language assessments indicated their receptive and expressive language scores to be at least at the 3-year-old level. That is, assessments indicated these children were able to combine at least three words, speak in a fairly intelligible manner, and understand most question types, such as yes/no, what, what

342

SELF-MANAGEMENT OF SOCIAL SKILLS

do, where, who, why, how, and when (it was noted during the sessions that Ian was unable to under- stand "where," although he did understand "what place"). Second, parent and teacher reports indi- cated the children's typical response to the verbal initiations of others to be disruptive behavior. In- dividual subject characteristics were as follows (see Table 2 for descriptions of individual children's disruptive behaviors).

Tony was 6 years 10 months old at the start of the study. Tony's data were collected over a 22- month period. His composite IQ score, as measured by the Stanford Binet, was 58. His nonverbal in- telligence score on the Leiter International Perfor- mance Scale was 100. On the Peabody Picture Vocabulary Test-R, he scored an age equivalent of 4 years 4 months, placing him at the 1st percentile. On the Assessment of Children's Language Com- prehension, he scored 94% on the vocabulary sec- tion and 100%, 80%, and 50%, respectively, on two, three, and four critical elements. His mother reported that he had numerous disruptive behav- iors, induding self-injury (picking at his skin to the point of drawing blood), inappropriate emotions such as laughing and crying too easily, frequent temper tantrums, physical aggression, and self- stimulatory behaviors. Adam was 11 years 1 month old at the start of

the study. His data were collected over a 16-month period. His composite IQ, as tested by the Stanford- Binet, was 85 and his IQ on the Leiter International Performance Scale was 70. On the Peabody Picture Vocabulary Test, he scored 3 years 7 months below his age level, placing him in the 1st percentile. On the Assessment of Children's Language Compre- hension he scored 98% on vocabulary and 90%, 80%, and 70%, respectively, on two, three, and four critical elements. Informal language probes indicated he was able to answer questions and use sentences of four or more words with correct syntax. His mother reported that he emitted numerous inappropriate behaviors, induding noncompliance, frequent temper tantrums, stubbornness, high lev- els of self-stimulatory behavior, and excessive pre- occupation with objects.

Howard was 6 years 10 months old and was

Table 1 Sample Questions for Each Child

Tony 1. What did you eat for lunch today at school? 2. Who are your best friends? 3. Did you have fun at school today?

Adam 1. Who drove you to school today? 2. How many tissues do you think are in this

pile? 3. What is your favorite video?

Howard 1. Why do you like Garfield? 2. Are you building any models now? 3. Did you take the bus to school today?

Ian 1. Did you eat lunch at Burger King today? 2. What place did you shop at today? 3. Who came and visited us last week?

Tony's identical twin. His data were collected over a 22-month period. His WISC-R yielded a fiul- scale IQ of69 and a verbal IQ of60. The Stanford- Binet yielded a fill-scale IQ score of 74. On the Leiter International Performance Scale, he received a nonverbal IQ of 111. On the Peabody Picture Vocabulary Test-R, he scored an age equivalent of 4 years 2 months, placing him in the 1st percentile, and on the Assessment of Children's Language Comprehension he scored 90% on vocabulary and 80%, 90%, and 80% on two, three, and four critical elements. His mother reported that he had occa- sional tantrums, frequently ran away, exhibited in- appropriate affect, and had poor concentration and attention.

Ian was 11 years 2 months old at the start of the study. His data were collected over a 27-month period. His IQ was measured by the Stanford-Binet at 58, with a verbal score of 53. His standard score on the communication domain of the Vineland Social Maturity Scale was 73. He was able to com- bine up to seven words with correct syntax, al- though he had difficulty attending to others, re- laying messages, and describing attributes or emotions of others. Further, he had difficulty with social norms, and had never been observed to apol- ogize for unintentional mistakes, follow rules of games, or respond verbally or favorably to the good fortune of others. Additionally, he exhibited nu- merous disruptive behaviors in social situations such

343

LYNN KERN KOEGEL et al.

as masturbation, inappropriate noises and singing, and tantrums and aggression toward his mother when she attempted to engage him in a social interaction.

Settings and Design Sessions were conducted in the context of a mul-

tiple baseline design across settings and subjects with a withdrawal for Adam and Ian. Settings were selected if reported as problematic by the children's parents. Three to four separate settings were used for 2 subjects, and two settings were used for the other 2 subjects. These settings induded a clinic setting and community settings for all subjects, a home setting for Tony, and a home and school setting for Adam. Treatment was not implemented at home for Howard and Ian because they typically were quite responsive in those settings during base- line measurements. Baseline measurements were collected in all settings, and initial self-management treatment sessions were conducted after school in a clinic on a university campus. The initial clinic treatment was conducted in a small room contain- ing a table, two chairs, and a one-way mirror, through which the sessions could be viewed. During community sessions, data were collected while each child was accompanied by adults (who were not involved in the treatment program) to a store to make a purchase (e.g., groceries, ice cream, cookies, etc.). The home sessions were conducted by the child's parent after school or in the evening, and communicative interactions with family members were measured. School sessions took place during lunch periods, when the child had ample oppor- tunity for communicative interactions, and data were collected on interactions with peers and the adult playground supervisor (who also were not involved in the treatment procedures).

Procedure Treatmentproviders and communicative part-

ners. Treatment was conducted by advanced un- dergraduate and graduate students, and was su- pervised by licensed speech and language specialists. All self-management training and reinforcers were delivered in the clinic. In the community settings,

the child interacted with undergraduate students who did not provide formal treatment in any of the settings. During the home sessions, the children most frequently interacted with their mothers, and on occasion with their fathers or siblings. At school, interactions occurred with a peer or the playground supervisor.

Baseline conditions. Prior to implementation of self-management treatment, baseline measures were recorded in each setting. During these sessions, pro- cedures were identical to all ofthe treatment sessions with one exception; the children were not required to self-manage their correct responses. Throughout the baseline and all experimental conditions, the communicative partner was instructed to ask similar types of questions. These were chosen from the children's regular activities, were age appropriate, and induded the same proportion of yes/no and information questions to control for task difficulty (see Table 1 for examples of the types of questions asked throughout all conditions). Data were re- corded on the percentage of appropriate responses emitted by the children with autism in response to the questions in each of the settings. In addition, because we were concerned about the children's disruptive behavior in the community, data were collected on disruptive behavior.

Withdrawal condition. To control for the pos- sibility that increased reinforcement alone during the self-management condition might be respon- sible for any improvement in the children's behav- ior, a withdrawal condition with reinforcement only (without self-management) was implemented for Adam and Ian. During this condition, reinforcers were provided on the same interval schedule pro- vided at that point of the self-management treat- ment. The children were told that they could still earn rewards for correct responding, but that the clinician would keep track of the correct responses so they did not need to wear the wrist counter. All other procedures were identical to the treatment procedures (i.e., type of questions, adult asking the questions, rewards, etc.).

Self-management training. Al training sessions were implemented in the clinic room. The exact procedures were as follows. Prior to the first treat-

344

SELF-MANAGEMENT OF SOCIAL SKILLS

ment sessions, several functional reinforcers were identified for each child. These typically consisted of small edibles that could be easily administered after correct self-monitoring response(s). In addi- tion, a wrist counter (i.e., an inexpensive golf coun- ter purchased from a sporting goods store) that could record up to 99 responses was used. To begin the initial steps, each child was taught to discrim- inate between a correct and an incorrect response. Specifically, the cinician demonstrated a question and an appropriate answer and showed the child how to record this response immediately on the wrist counter. Similarly, for incorrect responses, the linician demonstrated a variety of questions with- out a response or with an inappropriate response, and showed the child that a point could not be counted on the wrist counter for those behaviors.

Next, the wrist counter was placed on the child's wrist. To teach the child to record appropriate re- sponses, the cinician asked the child a question and then, following a correct answer, prompted the child to press the wrist counter. To do this, the clinician began by rewarding the child after one correct response was recorded appropriately on the wrist counter. Typically, within the first 30-min session, the child acquired this chain of responding (i.e., answering a question and recording the re- sponse on the wrist counter). Although some chil- dren required occasional verbal prompting to record all correct responses, prompting was completely faded after about 3 hr of treatment. As in baseline sessions, questions continued to vary systematically so that they included a combination of both yes/ no questions and information questions. To teach independence simultaneously with the

above procedures, the schedule of reinforcement was gradually thinned until the child recorded a large number of points (30 to 40) before earning a reinforcer. Also, during this time the child was taught to take or solicit his own reinforcer when he had earned enough points. This was accom- plished by systematically fading prompts to obtain a reinforcer. To do this, two goals had to be ac- complished. First, the child had to be able to rec- ognize on the wrist counter when the desired num- ber of points had been obtained. This was

accomplished by telling the child how many points he needed to earn for a reward, and by making sure he looked at the numbers on the wrist counter prior to obtaining the reward. Second, the adult's prompt to take a reinforcer was faded. This was accomplished by initially permitting the child to obtain his own reinforcer from the bag with verbal prompting. Next, the cinician simply made state- ments such as "What happens when you earn all of your points?" or "How many points did you earn?" Finally, the cinician provided minimal non- verbal prompts (such as glancing at the wrist coun- ter) until the child obtained the reinforcers without any prompts after the desired number of points had been earned.

All of the self-management training and almost all of the fading of prompts and reinforcers were completed within the first few sessions (the sessions in which prompting began and in which the fading of the reinforcement schedule was completed to either fixed-ratio 30 or 40 are indicated in Figures 1 and 2).

Self-management in community, home, and school settings. At the designated point in the multiple baseline design, self-management was in- troduced into the community, home, and school settings by telling the child that points could now be earned in these settings. That is, the child was permitted to wear the wrist counter in the new setting, and the treatment provider instructed (i.e., prompted) the child to continue monitoring re- sponses. All subsequent trials then took place with individuals who were not involved in either the self-management training or in providing prompts or reinforcers.

Self-Management 2 condition. For Howard, a slight variation in the program was made after he continued to show variable responding following the implementation of self-management. The typ- ical order of conditions was always a community session followed by a clinic session, in which the child completed earning the total number of points necessary to obtain a reinforcer for that day. How- ever, it appeared that Howard was not discrimi- nating the contingency between points earned in the community and the subsequent reinforcer de-

345

LYNN KERN KOEGEL et al.

Table 2 Disruptive Behaviors Measured in Community Settings for Each Subject

Tony Self-injurious behavior (head-banging with dosed fist, intense picking at skin in arm, neck, and facial areas often to the point of drawing blood)

Tantrums (crying and/or throwing self on ground) Running away from communicative partner Stereotypic twirling of hair lasting for more than 3 s, usually accompanied by leaving communicative part-

ner

Adam Spinning the entire body in cirdes while standing or walking Lying on the ground; hands inappropriately placed on head or over ears while rapidly and repetitively

tapping head with hands Yelling, screaming, or other out-of-context vocalizations at an extremely inappropriately high-intensity level

Howard Running away from the communicative partner Tantrums (screaming, crying, and/or throwing self on ground) Shouting words out of context at an inappropriately high-intensity level Self-injurious behavior (head-banging with a dosed fist)

Ian Pretending to play an imaginary instrument or singing at an inappropriately high-intensity level Delayed echolalic phrases, typically from a television commercial or program (repeated inappropriately loud-

ly with inappropriate voice quality) Hitting objects (e.g., hitting cars with a dosed fist) Walking or running away to avoid the communicative partner

livered later in the clinic session. Therefore, in the Self-Management 2 condition, to emphasize the contingency between points earned in the com- munity and the reinforcer Howard obtained, the order of sessions was reversed for one session. That is, Howard began that day in the clinic session, and then continued earning points in the subsequent community session so that he reached criterion on the number of points needed while in the com- munity, and then obtained the reinforcer.

Dependent Variable Definitions Data were recorded for all responses or the ab-

sence of a response to every question asked by another individual in each setting. Data were re- corded continuously throughout all sessions (i.e., 30 min for the clinic sessions, 15 to 20 min for the community sessions, and 10 to 20 min for the school or home sessions). Each data category is defined below.

1. An event recording system was used to record appropriate and inappropriate responses to ques- tions. Appropriate responses were defined as any verbal response or appropriate attempt at a response that was related to the stimulus (question) and occurred within 3 s of the stimulus. Each time one

of these events occurred, the observer marked a precoded data sheet.

2. Inappropriate responses were defined as the child either not responding at all (within 3 s of the question; typically if the child did not respond within this time period, no response occurred), making a response that was unrelated to the ques- tion, or making an echolalic response.

3. Because disruptive behavior was less discrete than the above responses (i.e., it did not always have a discrete onset or offset) and its duration was as important as its frequency, a continuous time- interval recording system was used to record dis- ruptive behavior. That is, disruptive behavior was recorded in each community setting in continuous 1-min intervals any time a child exhibited one of the disruptive behaviors (defined individually for each child) listed in Table 2. Then the percentage of intervals with disruptive behavior was calculated by dividing the number of 1-min intervals in which disruptive behavior occurred by the number of in- tervals with disruptive behavior plus the number of intervals without disruptive behavior and mul- tiplying by 100.

4. The accuracy ofself-recording (i.e., ifcorrect responses were tallied on the wrist counter and

346

SELF-MANAGEMENT OF SOCIAL SKILLS

incorrect responses were not tallied) was calculated in all settings for all subjects with the exception of Tony, whose accuracy was calculated in only the clinic and community settings. Accuracy was cal- culated separately for occurrences and nonoccur- rences by dividing the number of unprompted ac- curate tallies by the sum of the accurate and inaccurate tallies.

Reliability Reliability measurements were taken across all

settings and conditions for all subjects. For given pairs of observers, one observer was rotated unsys- tematically (from a pool of 14 observers) through- out the course of the study to prevent observer drift. During 69 sessions (i.e., 22% of the total number of sessions), two independent observers recorded data on the child's responses to questions. Percentage ofagreement between the observers (cal- culated separately for occurrences and nonoccur- rences) was calculated by dividing the number of agreements by the sum of the number of agree- ments and disagreements and multiplying by 100 for all sessions that had at least three occurrences or nonoccurrences. Additionally, reliability was cal- culated for disruptive behavior during 21 separate sessions (i.e., during 1 5% ofall the sessions in which disruptive behavior was recorded), with an agree- ment defined as both observers recording the be- havior as either not disruptive or disruptive in any given 1-min interval.

The percentage of agreement for recording oc- currences of appropriate responses to questions av- eraged 96% (range, 56% to 100%, with 68 of the 69 sessions above 80%). The percentage of agree- ment for recording nonoccurrences averaged 87% (range, 56% to 100%, with 33 of the 43 sessions above 80%). The percentage of agreement for re- cording disruptive behavior averaged 89% (range, 40% to 100%, with all but five of the sessions at 100%). The reliability for recording the accuracy of the children's self-management responses was calculated for 47 sessions (i.e., during 21% of all sessions in which accuracy was recorded). Percent- age of agreement averaged 96% (range, 75% to 100%).

RESULTS

Responses to Questions from Others Figures 1 and 2 show the percentage of appro-

priate responding to others' questions for all 4 chil- dren in all settings. Data were collected in one or more (but not necessarily all) settings in each ses- sion. The data showed that during baseline in each setting, the children responded appropriately only infrequently and inconsistently when others at- tempted to interact with them. For example, during baseline in the clinic, the 4 children responded appropriately only 59%, 61%, 49%, and 35% of the time. Further, in the community, home, and school settings, the children typically responded at even lower levels. Note that because home sessions occurred relatively infrequently, home and school settings were plotted together on the graphs, with home settings marked by an open square. With the initiation of self-management procedures, how- ever, all 4 children showed relatively rapid im- provement (typically within a few sessions) in the clinic, and remained at improved levels in that setting for the rest of the self-management condi- tions. It is also noteworthy that the children learned the self-management procedures relatively quickly. The arrows in Figures 1 and 2 indicate the sessions in which prompts were provided to teach the child the chain of responding to a question, recording the response on the wrist counter, and obtaining a reinforcer. The final arrow indicates the approxi- mate point at which the reinforcement schedule was thinned to its peak level of fixed-ratio (FR) 30 or 40. Typically, prompting was faded and the reinforcement schedule was thinned to FR 30 or 40 within a few sessions. In contrast to baseline, responding typically reached 90% to 100% after completion of self-management training. In addi- tion, the withdrawal of self-management proce- dures for Adam and Ian resulted in decreases in appropriate responding, followed by subsequent rapid improvements when self-management pro- cedures were reintroduced.

In accordance with the multiple baseline design, self-management procedures also were introduced in all of the targeted settings. The results in the home and community settings paralleled the im-

347

LYNN KERN KOEGEL et al.

Baseline

l 5 10 15

Self-Management

f Prompt Tony CLINIC

I I I I I I I I I I I I

20 25 30 35 40 45 50 55 60 65 70 75 M_-b

I I I I I I 5 10 15 20 25 30 31

COMMUNITY I I I I I I I I

4045 50 55 60 65 70 75 Prompt

HOME 1l I I I I l1

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75

Prompt

FRII LL30

1 1 ll1

Self-Management

Adam CLINIC I I I I I I I I I I I

5 0 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Prompt

4~~~~~ 5 10 15 2 0 25

IUMMUNI I T I I I I I I I I I I I

30 35 40 45 50 55 60 65 70 75 80 85 90

Prompt

SCHOOL/HOME I I I , ,I I I I I I I I I

5 10 15 20 25 W 40 45 50 55 60 65 70 758 85 90 SESSIONS

Figure 1. Appropriate verbal responses for Tony and Adam, who each received self-management training in three settings (Adam also had several data recorded in his home, indicated by open squares) in a multiple baseline design (with a withdrawal in one setting for Adam).

provements in the clinic setting (Figures 1 and 2). That is, responses during baseline sessions in the home and community settings tended to be rela- tively low, with the children typically responding

less than 50% of the time when someone asked them a question. For Tony, Howard, and Ian dur- ing baseline in the community setting, there may have been some slight initial cross-setting gener-

100 80 60 40 20

100 80 60 40 20 C

100 80 60 40 20 0

100 80 60 40 20 Al

CIf w Cn z 0 C,)On wa:

w CE 0 a: al 0L

LL 0 w

z wL cca:I wL

I I I I I I I I I

Baseline

K 100 80 60 40 20 0

100 80 60 40 20

348

I 4

I I I I

SELF-MANAGEMENT OF SOCIAL SKILLS

10O0 80 60 40 20

Baseline

I

100 - 80 -

60 _ 40 F 201_ 0

5 1a 15 20 25 30 35 40 45 Prompt

5 10 15 20 25 30 35 40 45 !f

w

25

Prompt Baseline | Self-Manage 100 80-

60 40 _ FR40 20 -

10 15 20 :

100 80 60 40 20

S-M 11 Follow-up

CLINIC I I I I I I 0 55 60 65 70 75

0

COMMUNITY I I I I I I 50 55 60 65 70 75

Prompt Self-Management

CLINIC I I I I

30 35 40 45 50 55 60 65 70 75 80 85 90 95

I I I I I I I I I I I

PromptIe COMMUNITY

I I I I ILJ

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95

SESSIONS

Figure 2. Appropriate verbal responses for Howard and Ian, who each received self-management training in two settings in a multiple baseline design (with a withdrawal in one setting for Ian). Howard received self-management sessions in two different orders: Order 1 was the typical order, and Order 2 (diamond symbol) reversed the clinic and community settings for one session in order to emphasize the response-reinforcer contingency (see text).

alization after the self-management procedures were introduced in the clinic setting. However, this gen- eralization was relatively brief, and responding re- turned to approximately the initial baseline level before the self-management intervention was in- troduced in the community setting. Then, after self- management procedures were introduced, 3 of the 4 children showed increases in responding in every setting. Howard also showed immediate improve- ment during Self-Management 2, after the order of self-management sessions was reversed for one session in the community setting. In short, only after the introduction of self-management did rel- atively prolonged high levels of responding occur. This was especially dear in the home setting, where

no generalization occurred following implementa- tion of treatment in the clinic.

Accuracy of Self-Recording The accuracy of the children's tallying correct

and incorrect responses was calculated separately for each child. In general, the children averaged relatively high levels of accuracy in self-recording their responses, with increasing percentages of cor- rect recording as their training progressed. Tony's accuracy for tallying correct responses averaged 84% (range, 50% to 100% during taining). Adam's accuracy for correct responses averaged 72% (range, 0% to 100% during training); Howard's accuracy averaged 72% (range, 0% to 100% during train-

Self-Managemnent I

FR30 Howard Pr t I_

I I I I I I

349

cn w Cu) z 0a- Cl) w

a:

w C- a.0~ a: 0 a. a.

LI. 0 w 0

w

w a.

I'* . .ILetwi- .UP.,

is I I I 0.4 AP El 1.L N 9He VIn 17

.. ..

ni- I I I m I0%1

-

PC

6

LYNN KERN KOEGEL et al.

ing); and Ian's averaged 89% (range, 19% to 100% during training). Accuracy for recording nonoc- currences (i.e., the child did not make a tally when an incorrect response occurred) averaged 76% (range, 0% to 100% during training) for Tony, 69% (range, 0% to 100% during training) for Adam, 87% (range, 50% to 100% during training) for Howard, and 95% (range, 50% to 100% during training) for Ian.

Disruptive Behavior To assess whether increased and sustained re-

sponding to questions from others could result in a general reduction in disruptive behavior, results for each of the 4 children's disruptive behaviors in the community settings are plotted in Figure 3. The data show that, following the implementation of the self-management procedures, disruptive be- havior was much lower in the previously problem- atic community settings than it was before the in- tervention. This was true for Tony, Adam, and Ian with the regular order of sessions. Howard's dis- ruptive behavior followed a pattern similar to his appropriate responding described above. That is, his baseline level of disruptive behavior was very high, and it did not decrease during the initial implementation of the self-management treatment (Self-Management 1). However, following the im- plementation of the self-management sessions in the reversed order, Howard's disruptive behavior was completely eliminated (i.e., it decreased dra- matically from 42% in Self-Management 1 to 0% in Self-Management 2).

DISCUSSION

Overall, the results of this study showed that the lack ofsocial responsivity that is so characteristic in autism can be successfully treated with self- management procedures, requiring minimal pres- ence of a treatment provider in the children's nat- ural environments. Other studies corroborate these findings, suggesting that self-management may be an especially promising intervention for children with autism when minimal therapist presence is

desired (e.g., Johnson, 1991; R. Koegel & Koegel, 1990; Stahmer & Schreibman, 1992).

The second finding, that collateral reductions in disruptive behavior occurred when the children's responsivity improved, relates to several bodies of literature (Hunt, Alwell, Goetz, & Sailor, 1990). For example, Carr and Durand (1985) suggested that as more effective communication skills are learned, disruptive behavior commonly decreases. In the present study, it appeared that when the children responded consistently to their commu- nicative partner, the interaction not only was more normal in appearance but it also provided a more coherent and fluid conversation (in contrast to the disconnected interactions that occurred when the children were responding only inconsistently during the baseline and withdrawal conditions). The more fluid and consistent responding permitted many opportunities for the nondisabled partners to adjust to the child's competency level and to assist the child's communication. Thus, the conversational interactions during the treatment conditions were very likely much less aversive, and less likely to be associated with escape- or avoidance-driven disrup- tive behavior (e.g., Carr & Durand, 1985).

Implications for Social Integration The marked improvement in the children's be-

haviors (both increases in social communicative re- sponses and decreases in disruptive behavior) has major implications for social integration of these children in community and educational settings. There is little question that quality of social inter- action is a predictor of social acceptance (Asher & Renshaw, 1981; Dodge, 1983; Hartup & Sancilio, 1986; Quay &Jarrett, 1984). Without appropriate social skills in their repertoire, children with autism or behavior problems have significantly greater dif- ficulty being socially integrated into their schools and neighborhood communities (cf. Haring, 1990). In contrast, increases in social interactions due to social skills competence is likely to result in greater opportunities to be exposed to appropriate behavior models, spontaneous tutoring, and/or social rein- forcement for appropriate behavior (Cowen, Ped-

350

SELF-MANAGEMENT OF SOCIAL SKILLS

Baseline

5 10 15 20

Baseline

I I I

Self-Management

| |W~~IAdam]

25 30 35 40 45 50 55 60 65 70 75 80 85 90

Self-Manage I

I I I

5 10 15 20 25 30 35 40 45

Self-Management 11

I Howard | I-Tq~ I I 1

50 55 60 65 70 75

Baseline

5 10 15 20

Self-Management

l I I I I I I I 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95

SESSIONS Figure 3. Percentage of disruptive behavior (which was untreated) for each child before and after the self-management

procedures were implemented for only verbal responses in the community settings. Consistent with the data for appropriate verbal responses, Howard received self-management sessions first in the typical order (Self-Management 1) and then with the order of sessions reversed for one session to emphasize the response-reinforcer contingency (Self-Management 2; see text).

erson, Babigian, Izzo, & Trost, 1973; Parker & ment of other social language skills. Children who Asher, 1987; Strayhom & Strain, 1986). Studies have difficulties with effective communication, es- in our own and other laboratories are examining pecially children with autism, usually are deficient these issues. in more than one social aspect of their speech (Gar-

For example, the success the present study dem- fin & Lord, 1985). One particularly intriguing oc- onstrated in treating a specific social aspect of lan- currence in the present study was the anecdotal guage encourages future study in the self-manage- observation of an increase in verbal initiations by

cc 100 0 >: 80 I 60 wm 40 W 20 =0a3-

cc cn

o 100 W 80 < 60 Z 40w C) 20 w o- 0

100 80 60 40 20 o

351

352 LYNN KERN KOEGEL et al.

2 of the children following increases in their ap- propriate responses to questions from others, and 3 of the 4 children began to initiate conversations outside of the treatment environment without the self-monitoring wrist counter. This observation is consistent with the hypothesis discussed in other studies (e.g., R. Koegel, Frea, & Surratt, in press) that social skills directly involved in conversation may be part of a relatively large response dass. Also, although the fading of the self-monitoring device was not addressed in this artide, we believe it might be particularly easy with children who achieve generalization with little programming. Support for this speculation is provided by Stahmer and Schreibman (1992), who achieved fading rel- atively quickly under such conditions. The tech- niques and types ofchildren for which this stimulus fading can be most successful will be an interesting and important area for future research.

The role of the wrist counter in the naturalness or quality of the children's interactions is another issue that deserves discussion. During the initial treatment sessions, the conversational flow was somewhat disrupted as the children looked fre- quently at their wrist counters. However, this ap- peared to be directly related to the amount of prompting and the dense schedule ofreinforcement. As the fading of prompts and reinforcers was com- pleted, the children looked at their wrist counters infrequently, and the counters became a relatively unobtrusive part of the environment, with a cor- responding improvement in the naturalness of the conversational exchanges.

The independence that self-management permits for children with autism, and the stimulus gener- alization it promotes, make this technique partic- ularly promising. The fact that it is teachable in a relatively short time and provides rather quick re- sults makes it especially useful in previously difficult treatments environments (e.g., the classroom). The fact that self-management permits much less direct contact with the treatment provider again makes this ideal for the dassroom environment as well as for the community and home, where the child is provided with greater opportunity to interact nat- urally with family members and peers.

An important question for fuither investigation is how to fade self-management materials opti- mally, as well as fading the reinforcement role of the adult. Natural reinforcement from social inter- action will occur for different children with different amounts of mediation or programming. This re- quires further investigation in the context of a self- management treatment package, but appears to be an especially important area for future research.

REFERENCES Asher, S. R., & Renshaw, P. D. (1981). Children without

friends: Social knowledge and social skills training. In S. R. Asher & J. M. Gottman (Eds.), The development of children'sfriendships (pp. 273-298). New York: Cam- bridge University Press.

Carr, E. G., & Durand, V. M. (1985). Reducing behavior problems through functional communication training. Journal of Applied Behavior Analysis, 18, 111-126.

Cole, M. L., & Cole, J. T. (1989). Effective intervention with the language impaired child. Rockville, MD: As- pen Publishers.

Cowen, E. L., Pederson, A., Babigian, H., Izzo, L. D., & Trost, M. A. (1973). Long-term follow-up of early detected vulnerable children. Journal of Consulting and Clinical Psychology, 41, 438-446.

Dodge, K. A. (1983). Behavioral antecedents ofpeer social status. Child Development, 54, 1386-1399.

Dunlap, L. K., Dunlap, G., Koegel, L. K., & Koegel, R. L. (1991). Using self-monitoring to increase independence. Teaching Exceptional Children, 23, 17-22.

Fein, D., Waterhouse, L., Lucci, D., & Snyder, D. (1985). Cognitive subtypes in developmentally disabled children: A pilot study. Journal of Autism and Developmental Disorders, 15, 77-95.

Fey, M. E., & Leonard, L. B. (1983). Pragmatic skills of children with specific language impairment. In T. M. Gallagher & C. A. Prutting (Eds.), Pragmatic assessment and intervention issues in language (pp. 65-82). San Diego, CA: College-Hill.

Garfin, D. G., & Lord, C. (1985). Communication as a social problem in autism. In E. Schopler & G. B. Mesibov (Eds.), Social behavior in autism (pp. 133-149). New York: Plenum Press.

Gresham, F. M. (1986). Strategies for enhancing the social outcomes of mainstreaming: A necessary ingredient for success. InJ. Meisel (Ed.), Mainstreaming handicapped children: Outcomes, controversies, and new directions (pp. 193-218). Hillsdale, NJ: Erlbaum.

Haring, T. G. (1990). Social relationships. In L. Meyer, C. A. Peck, & L. Brown (Eds.), Critical issues in the lives ofpeople with severe disabilities (pp. 195-217). Baltimore, MD: Paul H. Brookes.

Harris, K. R. (1986). Self-monitoring of attentional be- havior versus self-monitoring of productivity: Effects on

SELF-MANAGEMENT OF SOCIAL SKILLS 353

on-task behavior and academic response rate among learning disabled children.Journal ofApplied Behavior Analysis, 19, 417-423.

Hartup, W. W., & Sancilio, M. F. (1986). Children's friendships. In E. Schopler & G. B. Mesibov (Eds.), Social behavior in autism (pp. 61-77). New York: Plenum Press.

Hunt, P., Alwell, M., Goetz, L., & Sailor, W. (1990). Generalized effects of conversation skill training. Journal of the Association for Persons with Severe Handicaps, 15, 250-260.

Johnson, J. (1991). The use of self-management proce- dures for maintaining responsiveness of autistic indi- viduals in natural environments. Unpublished doctoral dissertation, University of California, Santa Barbara.

Kaiser, A. P., & Warren, S. F. (1985). Pragmatics and generalization. In S. F. Warren & A. K. Rogers-Warren (Eds.), Teaching functional language: Generalization and maintenance of language skills (pp. 392-442). Baltimore, MD: University Park Press.

Kanner, L. (1971). Follow-up study of eleven autistic chil- dren originally reported in 1943. Journal ofAutism and Childhood Schizophrenia, 1, 119-145.

Kanner, L., Rodriguez, A., & Ashenden, B. (1972). How far can autistic children go in matters ofsocial adaptation? Journal ofAutism and Childhood Schizophrenia, 2, 9- 33.

Koegel, L. K., Koegel, R. L., & Ingham, J. C. (1986). Programming rapid generalization of correct articulation through self-monitoring procedures. Journal of Speech and Hearing Disorders, 51, 24-32.

Koegel, L. K., Koegel, R. L., & Parks, D. R. (1991). How to teach self-management to people with severe dis- abilities. Santa Barbara: University of California.

Koegel, R. L., Frea, W. D., & Surratt, A. V. (in press). Self-management as a strategy for modifying problem behavior. In E. Schopler & G. B. Mesibov (Eds.), As- sessment and management of behavior problems in au- tism. New York: Plenum Press.

Koegel, R. L., & Koegel, L. K. (1990). Extended reductions in stereotypic behavior of students with autism through

a self-management package.Journal ofApplied Behav- ior Analysis, 23, 119-127.

Lagomarcino, T., Hughes, C., & Rusch, F. (1989). Using self-management to teach independence on the job. Ed- ucation and Training in Mental Retardation, 24, 139-148.

Moore, L., Agran, M., & Foder-Davis, J. (1989). Using self-management strategies to increase the production rates of workers with severe handicaps. Education and Training in Mental Retardation, 24, 324-332.

Ninness, H. A. C., Fuerst, J., Rutherford, R. D., & Glenn, S. S. (1991). Effects of self-management training and reinforcement on the transfer of improved conduct in the absence of supervision. Journal of Applied Behavior Analysis, 24, 499-508.

Parker, J. G., & Asher, S. R. (1987). Peer relations and later personal-adjustment: Are low-accepted children at risk? Psychological Bulletin, 102, 357-389.

Quay, L. C., & Jarrett, 0. S. (1984). Predictors of social acceptance in preschool children. Developmental Psy- chology, 20, 793-796.

Rhode, G., Morgan, D., & Young, K. (1983). General- ization and maintenance of treatment gains of behavior- ally handicapped students from resource rooms to regular classrooms using self-evaluation procedures. Journal of Applied Behavior Analysis, 16, 171-188.

Stahmer, A. C., & Schreibman, L. (1992). Teaching chil- dren with autism appropriate play in unsupervised en- vironments using a self-management treatment package. Journal of Applied Behavior Analysis, 25, 447-459.

Strayhorn, J. M., & Strain, P. S. (1986). Social and lan- guage skills for preventive mental health: What, how, who, and when. In P. S. Strain, M. J. Guranick, & H. M. Walker (Eds.), Children's social behavior (pp. 287- 330). London: Academic Press.

Received August 11, 1991 Initial editorial decision November 5, 1991 Revision receivedJanuary 30, 1992 Final acceptance February 20, 1992 Action Editor, Samuel Odom