Behavioral intervention for siblings of children with autism: A focus on skills to enhance play

Behavioral intervention for siblings of children with autism: A focus on skills to enhance play

BEHAVIOR THERAPY 24, 573-599, 1993 Behavioral Intervention for Siblings of Children with Autism: A Focus on Skills to Enhance Play DAVID A. CELIBERT...

2MB Sizes 0 Downloads 48 Views

BEHAVIOR THERAPY 24, 573-599, 1993

Behavioral Intervention for Siblings of Children with Autism: A Focus on Skills to Enhance Play DAVID

A. CELIBERTI

University of North Texas

SANDRA L. HARRIS Rutgers, The State University of New Jersey This study assessed the effectiveness o f a treatment program to teach siblings to use three sets of behavioral skills while playing with their brother or sister with autism. It was predicted that siblings could acquire behavioral skills, that these skills would maintain over time, and that the skills would generalize to untrained contexts. Three sibling pairs participated in a multiple-baseline design across three sets o f skills. Siblings were trained to elicit play and play-related speech, to praise play behaviors, and to prompt the child with autism when he or she failed to respond. Clear differences supporting the hypotheses were noted between baseline and posttraining sessions across all targeted behaviors for all three dyads. Generalization and follow-up probes indicated that the siblings were able to generalize the skills to a novel toy, and maintain the skills at 3-week, 6-week, and 16-week follow-ups. Ratings by naive undergraduates o f baseline and post-training segments documented the social validity o f the behavior changes and showed positive changes in the siblings' playtime experiences. These findings support the hypothesis that siblings can learn behavioral skills for use during cooperative play with their brother or sister with autism.

Behavioral parent training is an effective, integral aspect of the treatment of children with autism (e.g., Handleman & Harris, 1986; Koegel, Schreibman, This paper is based on a master's thesis submitted by the first author to the Department of Psychology at Rutgers University. Our special thanks to Jessica and Jimmy, Lauren and Annie, and Susan and Rick for their participation in this study. The present study was possible because of the dedication and diligence of many undergraduate research assistants. We would also like to thank Gwynne Moss, Jill Young, Rochelle Hausman, Melanie Marin, Diana Parra, Sandra Coronado, Christine Cowen, Deborah Rausch, Shawn Kelly, Peter Gager, and Jill Grigg for their assistance with the data collection, the interrater reliability tabulations, and the statistical analyses. Appreciation is also expressed to Jan S. Handleman, Jeanette Haviland, and Steve Gordon for their suggestions on an earlier version of this manuscript. Requests for reprints should be sent to David A. Celiberti, Department of Psychology, P.O. Box 13587, University o f North Texas, Denton, Texas 76203-3587. 573 0005-7894/93/0573-059951.00/0 Copyright 1993 by Association for Advancement of Behavior Therapy All rights of reproduction in any form reserved.

574

CELIBERTI AND HARRIS

Britten, Burke, & O'Neill, 1982). As primary caregivers in the natural environment, parents can play an important role in facilitating the acquisition and generalization of appropriate behaviors. In contrast to the many studies confirming the efficacy of parent training, there has been less systematic inquiry highlighting what the field of behavior therapy can offer the sibling of a child with autism (James & Egel, 1986; Lobato, 1983).

Identification o f a Context for Sibling Involvement Like their parents, normally developing siblings may be uniquely suited to foster behaviors such as social responses, verbal interactions, and play by virtue of their comparable ages and the intensity of the relationship they share with their siblings with autism. The use of sibling resources is consistent with the belief that children with autism could benefit from intensive and varied exposure to appropriate models of language, play, and social behaviors beyond those present in the school setting. The development of positive sibling relations may also reduce the frustration inherent in diminished sibling interactions. Carr, Taylor, and Robinson (1991) demonstrated that the teaching behavior of adults can be reliably influenced by the behavior o f children. Carr and his colleagues found that participants were less likely to engage in interactions with a child with behavior problems such as tantrums, self-injury, and aggression than with a child who did not engage in these behaviors. This notion of "child effects" is particularly relevant to children with autism, who often pose challenging behaviors in situations when even seemingly benign demands are placed upon them. Given the pervasive deficits in social relatedness found in autism (e.g., Waiters, Barrett, & Feinstein, 1990), it is likely that the child with autism would not respond favorably to a sibling's interaction attempts. Historically, noncompliant behavior, lack of responding, and negative verbal statements can function as a "punisher" of any behavior that they follow, such as a sibling's attempts to elicit an interaction. Over time, James and Egel (1986) proposed that these patterns of interactions would likely culminate in a low rate of sibling interactions. It has been our clinical experience that many siblings may wish to interact with their brother or sister with autism, but are discouraged by a lack of responsiveness, diminished reciprocity, and the presence of maladaptive behaviors. Because it is likely that the typical sibling lacks the skills necessary to engage a noninteractive brother or sister, an intervention geared toward overcoming these obstacles is warranted. One might speculate that young siblings of children with autism are especially vulnerable to child effects because they lack experience and knowledge about autism and its management; and, given their developmental status, may feel threatened and intimidated by the puzzling behavior of their brother or sister. From a clinical standpoint, offering behavioral skills to siblings- much as we offer them to their p a r e n t s - i s indicated, so that diminished sibling interactions do not become a pervasive aspect o f the sibling relationship.

Previous Research Supporting Sibling Involvement The feasibility of using young siblings as intervention agents is suggested in the literature addressing the training o f nonhandicapped peers o f children

BEHAVIORAL INTERVENTION

575

with autism (e.g., Meyer et al., 1987; Odom & Strain, 1984). Exposure to normally developing peers has taken place in many environments, ranging from integrated preschool programs to mainstreamed educational placements (e.g., Egel & Gradel, 1988; Strain, Hoyson, & Jamieson, 1985). Odom and Strain (1984), in their evaluation of the treatment effects of peer-mediated interventions, noted that limited maintenance of the peers' behavior is typical when prosocial behavior is dependent upon external reinforcement. Fortunately, siblings probably have more intrinsic motivation to interact with the child with autism than does the classroom peer. The sibling relationship is typically less transient than peer relationships. In addition, although the classroom peer may have a multitude of playmates with whom to interact, the sibling has a more limited selection and therefore might be more invested in contributing to a more satisfying sibling relationship. Several studies have yielded promising data supporting the role of normally developing siblings. An early investigation by Colletti and Harris (1977) demonstrated that siblings could be trained in a circumscribed set of behaviormanagement skills to modify the behavior of children with autism and that parents could be reliable observers of their children's behaviors. Other sibling research has highlighted the role of siblings as "teachers." Schreibman, O'Neill, and Koegel (1983) successfully trained normally developing siblings to use behavioral techniques to teach their brother or sister with autism academically oriented tasks. Lobato and Tlaker (1985) taught a 21-year-old-sister to teach her 13-year-old brother with mental retardation basic self-care skills. Similarly, Swenson-Pierce, Kohl, and Egel (1987) found that siblings could be trained to teach domestic skills to their brother or sister with mental retardation. In these three studies, the contexts in which these siblings interacted with their handicapped brother or sister were task-specific and resembled the interactions that might resemble the educational placement of the child with autism.

Recognition of the Role of Siblings as Playmates Playtime may be a more appropriate focus for sibling involvement than academic work or self-help skills, because playtime interactions are age-appropriate for siblings and particularly relevant to a population of children whose play skills are poorly developed (Wulff, 1985). The feasibility of siblings as intervention agents for play behavior was supported by Powell, Salzberg, Rule, Levy, and Itzkowitz (1983), who demonstrated that parents could facilitate cooperative play between their mentally retarded and normally developing children. Although parents were taught to increase play interactions between the two siblings by using social praise and prompts, the parents did not teach specialized skills directly to the normally developing sibling. Similarly, James and Egel (1986) demonstrated that siblings could acquire modeling and prompting skills to increase reciprocal interactions. These researchers found that the increased rates of initiations and the child's responsiveness to these initiations endured at a 6-month follow-up and were generalized across settings. Although siblings were taught a variety of prompting and reinforcement techniques, James and Egel did not assess directly the siblings' skill acquisition, nor were children with autism included in this investigation. Coe, Matson, Craigie, and Gossen (1991) found that siblings of two children

576

CELIBERTI AND HARRIS

with autism could acquire prompting and reinforcement skills and use them in a variety o f play-oriented contexts without direct parental monitoring. The present study, like that of Schreibman et al. (1983), entailed the training of a generalized set of behavioral skills. However, our intervention was geared toward improving the sibling's experiences as a playmate o f a child with autism. Our primary objective was not necessarily to improve the functioning o f the children with autism, but rather to alter siblings' behavior so as to make their interactions with their brother or sister with autism more reinforcing. Specific targeted behaviors were assessed individually throughout the baseline, training phases, and follow-up sessions. If one wants to track specific skills and there is limited spontaneous acquisition, a multiple-baseline design across skills is more descriptive than amalgamating skills and using a multiplebaseline-across-subjects design. A multiple-baseline-across-skills design allows for better assessment o f the functional relationship between specific training phases and corresponding increases in siblings' specific skill acquisition. This design identifies the spontaneous generalization of treatment effects, and assesses the maintenance o f treatment effects as the sibling progresses through training. Behavior therapists have become increasingly sensitized to the social validity of treatment modalities and treatment effects (Bernstein, 1989; Hawkins, 1991; Irvin & Lundervold, 1988; Schwartz & Baer, 1991). Thus, the present study assessed social validation o f the treatment effects. Aside from depicting behavior change, we felt it was important to demonstrate that the siblings' skill acquisition and subsequent changes in play topography were noticeable and appreciated by naive observers. To this end, undergraduate students viewed segments o f videotaped sibling play in both baseline and posttraining conditions and rated them on several dimensions.

Method Subject Recruitment The children with autism were recruited from three classrooms serving preschool-age children at the Douglass Developmental Disabilities Center (DDDC). The teachers were asked to identify children who had a nonhandicapped sibling at least five years old and not more than one year younger or six years older than the child with autism. Nine potential families were identified and informed by letter of the study. To be selected for participation, the child with autism had to engage in little, if any, cooperative play and the siblings had no known behavioral or academic problems. Of the families who met eligibility criteria, four parents were interested in the study. One family was dropped after the intervention was initiated because the child with autism was very aggressive toward his sister whenever she attempted to interact, and baseline probes could not be collected safely. A modified intervention was offered to this family but no systematic data were collected.

Participants Children. The three sibling dyads who participated in the study were from intact families. Each dyad consisted o f one older, normally developing sister

BEHAVIORAL INTERVENTION

577

and one younger child diagnosed with autistic disorder according to DSMIll-R criteria (American Psychiatric Association, 1987). The children were originally diagnosed by an outside agency not associated with the DDDC and confirmed by an experienced clinical psychologist at the Center. The three siblings were enrolled in regular public school classes and had never been referred for any behavioral or academic difficulties. The experimenter met with each sibling to assess the youngster's interest and enthusiasm as well as an absence of overt resentment about participation. Informed consent was obtained from the parents and the siblings. All three children with autism displayed gross deficits in their play skills. The topography o f their independent play was characterized by poorly developed skills ranging from unfocused and preoccupied activity with a stereotypic repertoire o f behavior (e.g., lining up, stacking, or holding objects) to ritualistic and isolated play with a limited set o f toys. According to parental and sibling verbal reports, the children with autism engaged in little, if any, cooperative play with their siblings. During sibling interactions, the three children with autism engaged in maladaptive behaviors such as pushing, hitting, and grabbing and in verbal noncompliance such as tantrums, yelling, and overt, verbal rejection of the siblings' attempts to interact. The siblings involved in this intervention described frustration and disappointment regarding their inability to play with their brother or sister with autism. Despite the behavior problems, all three siblings were tolerant o f the child with autism and receptive to learning new ways to interact. The first dyad consisted o f Jimmy, a 4-year, ll-month-old boy with autism, and Jessica, his 7-year, 9-month-old sister. Connie, their younger sister, a 2-year, ll-month-old girl enrolled in a specialized preschool program for children with communication handicaps, also participated in baseline and generalization probes with Jessica. Jimmy's score on the Stanford-Binet 1V (Thorndike, Hagen, & Sattler, 1986), administered 7 months prior to training, was 78. The second dyad consisted of Annie, a 4-year, 3-month-old girl with autism, and Lauren, her 10-year, 3-month-old sister. Annie's score on the Stanford-Binet 1V, administered 9 months prior to training, was 68. The third dyad consisted of Rick, a 4-year, 10-month-old boy with autism, and Susan, his 8-year, 2-monthold sister. Rick's score on the Stanford-Binet IV, administered 8 months prior to training, was 51. Experimenter. The first author, a graduate student in clinical psychology with over two years of intensive experience with children with autism, conducted all o f the sibling training sessions and videotaped the final 10 minutes of each session, during which no training took place. Observers. Two research assistants who had at least two semesters o f experience working with children with autism scored the videotapes. The coders were naive to the experimental hypotheses and remained uninformed as to the condition and stage of training in which the sessions were recorded. Social Validation Raters. 17 undergraduate students, enrolled in a field work course at the DDDC, voluntarily participated in the social validation procedures. This group included 13 women and 4 men who were assigned to classrooms with older children and adolescents with autism and had no previous clinical contact with any o f the children in the study. These raters were naive

578

CELIBERTI AND HARRIS

to the purpose of the study, and their consent was obtained prior to their participation. A second group of 45 undergraduate students (39 women and 6 men) served as a replication sample.

Setting All baseline, training, and follow-up sessions were conducted in the family room or playroom of the participants' home. This setting remained constant across all phases of the study for two of the three sibling dyads. For Rick and Susan, the location was changed to the living room midway through training.

Materials The toy used during the data collection probes at the end of each session was the Li'l Playmates Farm Play Set. The novel toy was the Fisher-Price Little People Airport. The set of training toys included the Mattel Disney Mickeytown Circus Playset, the Kenner Play Doh Modeling Set, the Lucky Star Enterprise & Co., Ltd., wood blocks, the Fisher Price medical kit, assorted nonconnective puzzles, and one Nerf ball.

Design A multiple baseline across three sets o f skills with replications across three siblings was used to assess the acquisition of the targeted skills. Each set of targeted skills corresponded to a treatment condition. These skills included Set A, eliciting appropriate play and play-related language; Set B, praising appropriate responses; and Set C, prompting the child with autism when he or she did not respond and correcting incorrect responding. Following a three-session baseline, training for the first set was introduced, while the second and third sets of skills remained in baseline. Once criterion for the first set was met, training for the second set commenced and the third set remained in baseline. Continued, intermittent reinforcement for the first set of skills was provided; however, training no longer emphasized the first set of skills. Once criterion was met for the second set, training for the third set commenced. Mastery of a particular skill was determined to have occurred if the sibling demonstrated the target skill at a level equal to or higher than 80% for two consecutive training sessions. For Set A, eliciting responses, mastery was related to the quality rather than the quantity of elicitations. An attempt to elicit a response was considered "adequate" only if all o f the following conditions were met: The child's eye contact with the sibling or the stimulus materials was present during the onset of the attempt; the sibling's attempt was not unnecessarily repeated to the child with autism; and the child was given ample time to respond (5 seconds). Mastery o f Set B was achieved when the sibling praised at least 80% o f the child's appropriate responses over two consecutive sessions. Mastery of Set C was achieved when the sibling prompted the child in at least 80% of the instances when he or she failed to respond appropriately. For Jessica, the first sibling to participate in the intervention, the order of training was Sets B, A, and C. Lauren and Susan were trained in the order of Sets A, B, and then

BEHAVIORAL I N T E R V E N T I O N

579

C, because it was determined that administering Set A as the first set of skills was clinically important and laid the groundwork for the acquisition of skills in Sets B and C. Dependent Measures Data were collected from all baseline, training, follow-up, and generalization sessions. These data reflected the sibling's attempts to elicit play and playrelated language and ability to elicit adequately (Set A), to praise appropriate responding (Set B), and to prompt nonresponding and correct inappropriate responding (Set C). Data reflecting the child's play responses to the sibling's elicitations were also collected so that all the opportunities for the sibling to use the skills associated with Sets B and C could be identified. These data for the child with autism are presented as the frequency of desired responses during each session. Data for four measures of sibling behavior are represented by percentages. "Elicitation attempts" was computed by dividing the number o f intervals in which the sibling initiated an attempt, regardless o f the overall quality o f the attempt, during each session by the total possible number o f intervals. The "elicitation attempts" data reflect a quantitative index of interaction. By contrast, a qualitative index was obtained reflecting the percentage o f attempts that met the criteria addressed in training. "Elicitations attempts which met criteria" was computed by dividing the number of attempts that met the criteria over the total number o f attempts. "Praising" was computed by dividing the number of instances that the sibling praised the child with autism for a play or play-related language response by the number of times that the sibling had the opportunity to praise (equal to the number of correct responses by the child with autism). "Prompting" was computed by dividing the number of instances that the sibling prompted the child with autism for failure to respond by the number of times that the sibling had the opportunity to prompt (equal to the number o f incorrect responses and failures to respond by the child with autism). This method of presenting the dependent variables was chosen because it was sensitive to fluctuating rates of elicitation across sessions and provided an accurate representation of how the sibling was responding to the child's behavior. Curriculum o f Training 1 Skill Set A. This set of skills involved how to deliver play-related commands effectively. As discriminative stimuli, the commands took the form o f directives and models that required the child with autism to perform an appropriate toy manipulation (e.g., "Put farmer on horse.") or required play-related speech from the child (e.g., "Make the sound o f the horse."). These skills included: (1) Obtaining eye contact and appropriate attending behavior prior to delivering commands, requests, and directions; (2) delivering verbal commands in an audible, articulate, simple, and discrete format; (3) allowing the 1 Details about the curriculum are available from the first author.

580

CELIBERTI AND HARRIS

child ample time to respond (5 seconds rule); and (4) generating a wide range o f commands and requests during play. Skill Set B. These skills involved how to use social praise as a consequence for appropriate responses. The sibling learned how to deliver praise contingent upon appropriate play-related speech and appropriate play behavior. These skills included: (1) Delivering social praise for every instance of an appropriate response to the sibling's request; (2) delivering social praise in a direct and clear fashion; (3) delivering social praise contingent upon appropriate responding (within 3 seconds of the child's response); (4) using a wide range o f praise statements; (5) incorporating praise statements that were specific to the nature o f the child's response; (6) modifying the length and pitch of praise statements according to both the appropriateness and "novel" nature of the child's response; and (7) accompanying social praise with physical contact. Skill Set C. These skills involved how to respond when the child with autism did not comply with a request. This involved the following skills: (1) Withholding prompts until the child had 5 seconds to respond, gave the incorrect response, or emitted incompatible behavior; (2) providing feedback to the child every time he or she failed to perform the desired response; (3) providing feedback contingent upon inappropriate or incorrect responding or after passage of 5 seconds; (4) providing a variety of prompts to the child; (5) maximizing the child's attention while prompting the desired response; and (6) providing positive feedback for the child's partial attempt. The skills targeted for acquisition by the siblings are among a larger group o f behavioral skills addressed in the parent training offered at the Douglass Developmental Disabilities Center. Skill Set A was included so that the sibling could learn how to elicit play and play-related language from the child with autism. We expected that the effective delivery of commands would increase the child's level of compliance and appropriate responses and therefore make the play interactions more enjoyable for the sibling. Skill Set B was taught so that the sibling would be prepared to praise cooperative play responses as they begin to emerge. We expected that as the child with autism engaged in more appropriate play, this would serve as a "reinforcer" for the sibling. Because siblings typically abandon their efforts when the child with autism does not respond in the desired fashion, Skill Set C was included so the sibling would learn to follow through on requests and overtures. In addition, possible escape-motivated behavior from the child with autism would not be reinforced by the sibling's withdrawal from the interaction.

Procedures Baseline. Baseline data were collected for 3 sessions for each of the 3 dyads. They also were collected with a second toy to assess generalization o f the sibling trainee's skills. In dyad 1, baseline data also were collected with a second sibling, Connie. At the start of the baseline condition, the sibling was instructed to "play with (child's name)," using the barn set. The sibling was asked to interact with the child with autism in the manner that was typical; no specific instructions or feedback were given during the baseline phase. However, when the child

BEHAVIORAL INTERVENTION

581

with autism left the play area and the sibling was unable to retrieve him or her within 10 seconds, aggressed against the sibling, or engaged in active noncompliance, the experimenter paused the data collection and intervened to return the child to the play area or help him or her regain self-controL Sibling training. During the initial training sessions for each new set o f skills, the trainer demonstrated the behavioral skills for a maximum of 5 minutes with the child with autism while the sibling trainee observed. Throughout this period, the trainer discussed the various skills and procedures to be acquired. For the next 10 minutes the sibling practiced the behavior with the child with autism, while the trainer offered feedback. If necessary, the sibling role-played with the trainer behaviors that were causing difficulty. If no particular deficit existed, the trainer demonstrated the delivery of a specific technique in a more sophisticated manner to facilitate the "fine-tuning" o f skills. The siblings received ample positive praise statements for their efforts, particularly given that during the initial phase of training all three children with autism were uncooperative and frustrating to their sibling. Initially, behavior-specific feedback regarding skill usage comprised a continuous reinforcement schedule which was gradually thinned to an intermittent schedule. The sibling was trained successively in each o f the three sets of skills until she reached the criterion o f 80% correct responding over two consecutive sessions for each set. The training phase for Set A was extended for Lauren and Jessica, because their overall rate of initiations was somewhat low. The trainer modeled all of the techniques for the three sets regardless o f condition; however, only the techniques associated with the given condition were emphasized and taught directly. Although the final phase of each training session involved only the barn set, the modeling and practice phases involved a larger set o f toys, including puzzles, blocks, a circus set, a Play Doh kit, and a toy doctor's kit. The 15-minute training session involved the same toys for all conditions. The last 10 minutes o f each 25-minute session were videotaped by the experimenter, who sat quietly on the floor. No training took place until the videotaping was complete. Additional feedback was offered briefly at the end o f each session. Later sessions were organized in the same way. Although the initial modeling component was phased out as the sibling demonstrated increasing competence at the respective set of skills, the training sessions remained 15 minutes in length. Each of the siblings was taken out for a special treat at the end of training; Susan was also taken out for lunch upon mastering Skill Set A, because Rick's behavior was initially very negative (e.g., he frequently left the area, threw toys, and verbally rejected his sister), and the experimenter wanted to acknowledge that she had had a particularly difficult time.

Data Collection Probes Performanceprobe. The final 10 minutes of the session involved videotaped data collection. The sibling's instructions were identical to that given during the baseline probes (i.e., "Play with (child's name), using the barn set). If necessary, the sibling was reminded to ignore the video camera. Follow-up probe. Follow-up data were collected 3 weeks, 6 weeks, and 16

582

CELIBERTI A N D HARRIS

weeks after the completion of training. No booster training, reminders, or feedback were given to the sibling trainee prior to or after the follow-up videotaping, other than general praise for participation. The follow-up sessions lasted 10 minutes. Generalization probe. Before and after formal training and at the 16-week follow-up, the sibling participated in a 10-minute generalization probe using a novel toy. Jessica also participated in 6 generalization probes with Connie, her handicapped sister, to assess the extent to which the skills acquired with Jimmy could be transferred to equivalent interactions with Connie, who did not participate in any training. Two probes were conducted immediately after the baseline probes with Jimmy, a third probe upon Jessica's mastery of the skills associated with Set A, and a fourth probe upon mastery of the last set of skills, Set C. Two final probes occurred during the 6-week and 16-week follow-up sessions. We had hoped to obtain generalization probes with Connie upon mastery of each skill set with Jimmy, but on several occasions Connie was ill and unable to participate. lnterobserver agreement. The observers separately viewed the videotapes and independently scored both children's behavior on a continuous 10-secondinterval observe-and-record basis. These observations were conducted for 10 minutes, yielding 60 intervals per session. Interobserver agreement was calculated by dividing the number of intervals in which both observers agreed that the behavior occurred by the number of agreements plus disagreements and multiplying this quotient by 100. In the event of a disagreement, data from the person designated as the primary observer were used. Interrater reliability was tabulated for 56°70 of the data-collection probes. These reliability checks were evenly distributed across all training conditions for all three dyads, with at least one reliability session for every condition for each dyad. Reliability. Reliability for each of the dependent variables shown in Table 1 was acceptable. Sessions in which the frequency of a behavior was low sometimes resulted in lower reliability, but the majority of sessions contained reliability of 80070 or higher.

Social Validation Component This phase of the study was conducted upon completion of the training for the three dyads. The stimulus videotape depicted 12 randomly ordered, one-minute vignettes, two baseline and two posttraining segments for each of the three dyads. Half of the segments were of the siblings playing with the toy barn set and half with the toy airport set. These segments comprised the ninth minute of the first baseline session, the first generalization probe, the last training session, and the posttraining generalization probe for all three dyads. The "ninth" minute was randomly selected from a pool of 10 possible minutes. The Social Validation Raters used a 5-point scale to rate the vignettes, based on the behaviors of the siblings and the appropriateness of play. Higher scores signify more positive perceptions. These data were collected in a group format, although the raters made their judgments independently. After ratings were completed, the group was debriefed. These procedures were identical for the replication group.

BEHAVIORAL INTERVENTION

583

TABLE 1 RELIABILITY

Eliciting responses Mean °70 Range 070 # Sessions below 80070 Attempts to elicit that met criteria Mean 070 Range 070 # Sessions below 80070 Praising Mean 070 Range 070 # Sessions below 80070 Prompting Mean 070 Range 070 # Sessions below 80070 Child with antism's desired responses Mean 070 Range 070 # Sessions below 80070

Jessica and Jimmy

Jessica and Connie

Lauren and Annie

Susan and Rick

80.16 67-100 8

78.5 67-90 1

88.82 82-100 0

84.13 74-100 7

95.87 89-100 0

94.5 89-100 0

98.00 91-100 0

93.19 85-100 0

82.07 60-100 4

100 n.a. 0

91.38 85-100 0

86.89 64-100 2

86.46 55-100 4

93.00 n.a. 0

84.13 50-100 4

90.46 57-100 3

80.90 56-100 5

n.a. n.a. n.a.

87.50 71-100 3

78.20 54-100 6

The Social Validation Rating Scale comprised 10 items. Some of these items addressed affective factors in the sibling's style of interaction with the child with autism, including the raters' perception of the sibling's level of frustration during the episode (sibling's frustration), her level of confidence toward the child with autism (sibling's confidence), the level of interest toward the child with autism (sibling's interest), and the level of pleasure of the nonhandicapped child as a result of the interaction (sibling's pleasure). Sibling behaviors that were addressed included her enthusiasm toward the child with autism (sibling's enthusiasm) and her effectiveness during the interaction (sibling's effectiveness). The degree to which the interaction was beneficial to the child with autism (benefit to child) as well as the child with autism's perceived interest in the sibling (child's interest in play) were also measured. The child's behavior was evaluated in terms of cooperation with the sibling (child's cooperation). An overall rating of how well the child with autism played during the interaction was also obtained (overall).

Results Performance data. Figure 1 displays Jessica's performance data. She participated in 3 baseline and then 21 training sessions, showing rapid acquisition of all three skills. Following 3 sessions of baseline in which there were no opportunities to deliver praise, Jessica participated in 5 sessions of training for this skill, and rapidly rose to a mean rate of 90% praise for Jimmy's ap-

584

CELIBERTI AND

Base,me IOO

Training

Maintenance

::'.~

\/\! •

i~/: ~

80

Follow-up

IX.: .,----. ~"\,* ',.,":.-~+

:\

9o ~1

HARRIS

~./

,:

\

:

+ : 0

~

no'vii child

50

g~

noveltoy ~',

i° im* r eo

~-

o

~ ~o

,

I

io I\~o-~ °

o Oo

o',

~'

,.,,o,

*.\11

A

:.

~,

i

,o

~

\

so

._~.

°w .~

i

/\./

so

'g 3o

lo

0

= . ~ . ~ - ' '

2

4

"

6

J

8

"

'

"

lO

'

12



'

14

"

'

16

"''

*

18

'

?o

"

'

2Z



"'

~

24

~6

'

J

28

Sessions

~G. 1. Reflects percentage of target behaviors across baseline, training, maintenance, and follow-up sessions for Jessica (filled circle) and percentage of elicitations that met criteria with Jimmy (open circle) and Connie (open triangle). Also reflects percentage of target behaviors across generalization probes with novel child (plus sign) and novel toy (square) and percentage of elicitations with novel toy that met criteria (open square).

BEHAVIORAL I N T E R V E N T I O N

585

propriate responses. This behavior persisted at a mean rate of 90% during the remaining 16 sessions of training. As demonstrated in Figure 1, concurrent with training in praise delivery there was an increase in the rate of Jessica's attempts to elicit responses from her brother. Thus, although her first 3 days o f baseline evidenced no efforts to elicit responses, during the next 5 days of baseline, her efforts increased to a mean of 29%. Following the 8-day baseline in eliciting play responses (M = 18%0), 9 sessions of training produced a mean rate of 61%. This dropped slightly (M = 51%) when prompting strategies were introduced. The percentage of Jessica's attempt to elicit a response that met the training criteria rose slightly from a mean o f 77% during the 8-session baseline to 79% during the 9-session training phase. During maintenance, when prompting was being trained, the percentage of attempts to elicit that met criteria rose to 96%. Prompting remained in baseline for 17 sessions ( M = 31%) and rose to a mean of 67% during 7 training sessions. Figure 2 displays Lauren's performance data. She quickly acquired the three targeted skills over the course o f 14 training sessions. With 6 sessions of training in how to elicit play responses from her sister, her performance rose from a baseline mean of 4% to 42%. This rate was maintained ( M = 42%) as the next two sets of skills were introduced. In addition, the rate of Lauren's attempts to elicit a response that met the criteria was unstable and then rose to a stable m e a n o f 99% during the training phase and a mean o f 96% over the remainder of the sessions. The percentage o f Annie's appropriate play responses that were praised by Lauren rose from a mean of 17% during 9 sessions o f baseline to 73% over 4 training sessions, and averaged 91% over the remainder of the sessions. When prompting was introduced, Lauren's performance rose from a mean of 22% during 13 sessions of baseline to 66% during 4 sessions o f training. Figure 3 displays Susan's performance data. She required 3 baseline, then 28 sessions to complete training. After a 3-session baseline in which she made no attempts to elicit play from her brother, her performance rose slowly and sporadically to a mean o f 42°70 over 17 sessions. This rate was maintained ( M = 47%) as the next two sets o f skills were introduced. The percentage o f Susan's attempts to elicit a response that met criteria rose from 0% during baseline to a mean o f 82% during the training phase and was maintained ( M = 86%) over the remainder of the training sessions. With the introduction o f training in praise delivery, the percentage o f Rick's appropriate play responses that were praised by Susan rose from a mean of 9% during 20 sessions o f baseline to 70% over 4 training sessions and to a mean o f 86% during the remainder of the study. Prompting was the final skill to be taught. The percentage of instances in which Susan prompted Rick rose from a mean of 13% during 24 sessions of baseline to 53% during 7 sessions o f training. Follow-up data. Performance data from the 3-week, 6-week, and 16-week follow-up probes demonstrated that all three siblings maintained most of their newly acquired skills at rates that roughly matched or exceeded that of training. The percentage of intervals in which Jessica attempted to elicit a response from J i m m y was 77% at the 3-week follow-up, 52% at the 6-week follow-up, and 63% at the 16-week follow-up. The percentage of these attempts that met criteria

586

CELIBERTI

Baseline 100

~

0 I0--0

0

90

I

HARRIS

Training Maintenance 0-01 0 '~0 " 0 ~ I0~ /0~0~ ~'0

e~tc II'

/' --o al met criteria

tilria

|g-

AND

Follow-up /

GI

0

I0~' •

X0/ '

I i

(novel toy)

I I I I

I

•1

~'0

40

i•

I~*a M ~

30

i "\J

°

:.,/,...,/ !\

x I \ /

iI

I

novel toYii

P..!

i '

IO0

1

I

/-:I

,

90 v) ~

I

/ .-'i/

80

U ro

I~. '° .=_~ 6o



\•/

I • Ill

0 ~

50

0)'13

40

\,.,!

le

100

i

90

I'i/

~a ~e ~ •

I\

,J

:

30

I0 0o

2

4

6

8

I0

17

14

10

18

ZO

Sessions

FiG. 2. Reflects percentage of target behaviors across baseline, training, maintenance, and follow-up sessions for Lauren (filled circle) and percentage of el•citations that met criteria with Annie (open circle). Also reflects percentage of target behaviors across generalization probes with novel toy (square) and percentage of el•citations with novel toy that met criteria (open square).

587

BEHAVIORAL INTERVENTION BaseJine

~:

ioo 90

Training

Maintenance

Follow-up

/,~ }~ ,o: .o.O.oO.o.o0:O,,o :r,,.,,,o.o/~ I~ "', ~'\ I Oo

'

~7o ~:-,,o.o- //,! °V ~/"I I 8o

met criteria

\o IP I~

SO

.~ "~

40

• /t

l~'~

i,~ll,~ "

;

/

,

/

~

./P/'*

V/l\ ,I ",: v :,~ .;'

i~ ~°

O.



nov~

V?"! :i

toy

'.'

10

o.O

IO0

---I

...............

elI

.

.,.o_

~,o 11 ~.



40

~|



I q

ID I1~ 3 0

~.

20 10

o



/, ,, ? / I""~ l Il .\, ~, .le

90

J

I //

,

,,

~

' 'I



.e,l\

,e.

' I

I_ . . . . .

100

' - - -- "II

411

,

/:

90

I

~ ~

oo

,

~!o

70

!

~'o

r,o

I •

~]°~-'° /!

II

~o



,o

i 4

6



I I



•il ~

~k" Q'} 3 0

Z

l

,.

r\/

40

~



8

10

/\

1Z t4

/ .

o,

%.:%../\:. 16

~8 ZO ZZ Z4



Z6 Z8 30

' ,

: 3Z

34

Sessions

Fio. 3. Reflects percentage of target behaviors across baseline, training, maintenance, and follow-up sessions for Susan (filled circle) and percentage of elicitations that met criteria with Rick (open circle). Also reflects percentage of target behaviors across generalization probes with novel toy (square) and percentage of elicitations with novel toy that met criteria (open square).

588

CELIBERTI AND HARRIS

was 96% at the 3-week follow-up, 94% at the 6-week follow-up, and 93% at the 16-week follow-up. Jessica praised 91% o f her brother's appropriate responses at the 3-week follow-up, 95% of the responses at the 6-week followup, and 82% at the 16-week follow-up. At the 3-week follow-up, Jessica prompted 100% o f Jimmy's failures to respond appropriately; this behavior dropped to 88% at the 6-week follow-up and 63% at the 16-week follow-up. Lauren attempted to elicit a response from Annie during 48% o f the intervals at the 3-week follow-up, 62% at the 6-week follow-up, and 42% at the 16-week probe. The percentage of these attempts that met criteria was 97% at the 3-week follow-up, 1000/0 at the 6-week follow-up, and 96% at the 16week follow-up. At the 3-week follow-up, Lauren praised 90% o f her sister's appropriate responses, 83% of the responses at the 6-week follow-up, and 88% at the 16-week follow-up. Lauren prompted 64% of Annie's failures to respond appropriately at the 3-week follow-up. This behavior rose to 89% at the 6-week foUow-up and decreased to 7507o at the 16-week follow-up. The percentage of intervals in which Susan attempted to elicit a response from Rick was 40% at the 3-week follow-up, 43% at the 6-week follow-up, and 62% at the 16-week follow-up. The percentage of these attempts that met criteria was 100% at the 3-week follow-up, 92% at the 6-week follow-up, and 95 % at the 16-week follow-up. The percentage of Rick's appropriate responses that were praised by Susan was 89% at the 3-week follow-up, 67% at the 6-week follow-up, and 90% at the 16-week follow-up. At the 3-week follow-up, Susan prompted 80% of Jimmy's failure to respond appropriately. This behavior dropped to 72% at the 6-week follow-up and 50% at the 16-week follow-up. Generalization data. All three siblings transferred skills to a novel toy. When using the generalization toy during baseline, Jessica attempted to elicit a response from her brother during only 2% of the intervals. Upon completion o f training, this behavior rose to 37% and was 53% at the 16-week follow-up. The percentage of attempts to elicit a response that met criteria using the novel toy rose from zero before training to 92% after training and 94% at the 16week follow-up. Jessica praised none of her brother's appropriate responses before training, 87% of the appropriate responses after training, and 71% of the appropriate responses at the 16-week follow-up. Similarly, the rate at which Jessica prompted Jimmy when he failed to deliver an appropriate response rose from zero at pretraining to 78% after training and 71% at the 16-week follow-up. At baseline using the novel toy, Lauren attempted to elicit play behavior from her sister during 2% o f the intervals. After training, these behaviors increased to 33%. The percentage of attempts to interact using the novel toy that met criteria was 100% before training, based on one attempt. The posttraining probe revealed that all of Lauren's attempts to elicit play behaviors met criteria, based on interactions that took place during 33 % o f the intervals. At the 16-week follow-up probe, Lauren attempted to elicit a response from her sister during 43% o f the intervals; 96°7o of these attempts at the 16-week follow-up probe met criteria. Lauren praised none of her sister's appropriate responses before training, 100% of the appropriate responses after training, and 91% at the 16-week follow-up probe. The rate at which Lauren prompted

BEHAVIORAL INTERVENTION

589

Annie when she failed to deliver an appropriate response rose from zero at pretraining to 75% after training and 71% at the 16-week follow-up probe. Susan attempted to elicit play behavior from her brother using a novel toy during 2% of the intervals at baseline. These behaviors rose to 48% on completion of training and 57% at the 16-week follow-up. As was the case with Lauren, the percentage of attempts to interact using the novel toy that met criteria was 100% before training, based on one attempt. The posttraining probe revealed that 100% of Susan's attempts to elicit play behaviors met criteria, based on interactions that took place during 48% of the intervals. At the 16-week follow-up probe, this behavior decreased slightly, to 88%. Susan praised none of her brother's appropriate responses before training, 67% of his appropriate responses after training, and 72% of his appropriate responses at the 16-week follow-up. The rate at which Susan prompted Rick when he failed to deliver an appropriate response rose from zero at pretraining to 69% after training but declined to 40% at the 16-week follow-up. Figure 1 reflects Jessica's generalization probes with Connie using the toy barn. During the initial two baseline probes, Jessica attempted to elicit a play response during 4% of the intervals, and none of these attempts met criteria. Jessica praised a mean of 33% of Connie's appropriate responses and prompted none of her failures to respond appropriately. During a probe with Connie after Jessica completed the first stage of training, the percentage of interactions rose to 43% of the intervals, with 68% of these attempts meeting criteria. Jessica praised 83 % of Connie's appropriate responses during this probe, which was consistent with the concurrent increase in her use of praise with her brother. Jessica's prompting increased to 23% during this probe. The next set of generalization probes with Jessica and Connie were upon Jessica's mastery of the final set of training, at 6 weeks, and at the 16-week follow-up. The percentage of intervals in which Jessica attempted to elicit a response was 45% at the end of training, 63% during the 6-week probe, and 65% during the 16-week follow-up. The percentage of attempts that met criteria was 86% during the last session. For the two follow-up probes, the percentage of attempts that met criteria was 100% and 86% for the 6-week follow-up and the 16-week follow-up probe, respectively. Jessica praised all of Connie's appropriate responses at the end of the training and at the 6-week follow-up probe. This decreased slightly to 92% at the 16-week follow-up probe. The percentage of Connie's failures to respond after a prompt from Jessica decreased from 58% during the last session of training to 23% during the 6-week follow-up. This increased to 58% at the 16-week follow-up probe. Figure 4 displays the frequency of desired responses during all baseline, performance, and follow-up probes for Jimmy, Annie, and Rick. For Jimmy, there were no opportunities to emit desired responses during baseline. However, the mean number of desired responses rose steadily over the course of training, with a decrease toward the end as Jessica was learning prompting strategies (M = 18%). During the follow-up probes, this frequency increased (M = 24%). For Annie, there were no opportunities to emit desired responses during two of the three baseline sessions. During one baseline session she emitted 3 desired responses. However, the mean number of desired responses

590

CELIBERTI AND HARRIS Bssellm 35'

]j

Training

Fodlow-up

J

Io

30' •

25-

i

2015. 10" 5" 0

3S



2

i

u

i

i

,

i

i

i

u

~

6 8 10121416182022242628

8sse~Ine

Folow-ep

Training

30 ]i

2s

~E

is

j!lO

5

A 0

2

,

4

,

6

,

,

8

,

,

,i

,

Training

k s ~ e

Fallov,~up

35

1

30 25 20

~

10 12 14 16 18 20 22

i i





i\ i,

i

,

15

|.'i

10 5 0

i j',, __.J . . . . .

.

.

.

.

.

.

I

,

i

la

i

0 2 4 6 810121416182022242628303234 Sessions

Fzo. 4. Reflects frequency of desired responses by the child with autism across baseline, training, maintenance, and follow-up sessions (filled circle).

increased immediately and remained high over the course of training, with no decrease toward the end as Lauren was learning prompting (M = 1707o). During the follow-up probes, this frequency increased slightly (M = 21°70). For Rick, there were no opportunities to emit desired responses during baseline. However, the mean number of desired responses increased sporadically

BEHAVIORAL INTERVENTION

591

over the course of training, with a decrease toward the end as Susan was learning prompting (M = 13070).During the follow-up probes, this frequency remained stable (M = 11070). Social validation. Ratings for both baseline vignettes were combined to form an aggregate score, as were the ratings from the two posttraining vignettes. Means and standard deviations were obtained for ratings of baseline and posttraining vignettes. These statistics were computed separately for the three sibling dyads for all 10 factors assessed in the Social Validation Rating Scale. Table 2 reflects paired t statistics a n d p values of comparison between baseline and posttraining vignettes for both the original sample of 17 Social Validation Raters (Samp. #1) and the replication sample of 45 Social Validation Raters (Samp. #2). As shown in Table 2, means and standard deviations for the replication group closely approximate the original group of raters. All of the t test comparisons of reactions to baseline and posttraining vignettes achieved significance in the predicted direction at p = < .001, except for the item assessing both Jessica's and Susan's level of frustration. For Susan, significance was obtained; however, it was not in the predicted direction; she was perceived as more frustrated after training than at baseline. At posttraining, all three siblings were rated as appearing more confident, more interested in the child with autism, and more pleased by their interactions with the child than they were at baseline. The three siblings were also rated as acting more enthusiastically and effectively while playing with the child; and the play interaction at posttraining, compared to baseline, was rated as more beneficial to the child. All three handicapped children were perceived to be more interested in the sibling and playing more cooperatively at posttraining than at baseline. Finally, for all three dyads, the child with autism appeared to play "better" at posttraining than during baseline.

Discussion Skill Acquisition Each of the girls in the present study showed relatively rapid mastery o f all three target skills and sustained maintenance of skills as training shifted to other targeted areas and during the follow-up probes. They also demonstrated generalization to a novel toy upon completion o f training and during the 16-week follow-up. The results o f the present study provided support for the hypothesis that siblings can acquire behavioral skills and use them during a play interaction with their brother or sister with autism. These findings are consistent with previous research documenting that siblings can learn such skills (Coe et al., 1991; Colletti & Harris, 1977; James & Egel, 1986; Powell et al., 1983; Schreibman et al., 1983). Although the acquisition rate for each sibling varied, some skills were easier to master. Across all three siblings, the number o f sessions necessary to become proficient in eliciting appropriate responding was higher (M = 10.7 sessions) than the number necessary for reinforcing appropriate responding (M = 4.3 sessions) and for prompting (M = 6 sessions). Susan needed the

592

CELIBERTI AND HARRIS

V

V

NV

V

--V

--V

V

V

--V

~V

e~V

V

--V

-- V

--V

--V

--V

0 Z

0 Z

m~

V

V

0

~

~

"

0 ~.~.~~

0

~

~ ~-

~

V

~

~

o

o

.

~.~

0

.

~

--V

.~

.

~

.

o

.

~:~.~.~

V

~

~ '2

~, ~ . , . ,

593

BEHAVIORAL I N T E R V E N T I O N

NV

~V

~V

--V

--V

--V

e¢,o

~,o",¢-o',..~o

'~,o ~ V

t'4O'~l-,O~,o "

~V

--V

¢~V

V

~o"~l-oee,o

V

~o"~t'oee,o

V

~o'~'o~"~o

~V

V

V

V

~V

~o'~1-o¢,,Io ~

~V

--V

V

V

e,l o ' , ~ , o e o o

~V

"-' V

~V

~l,o"~to~o

V

~o,,::l..o.,:~ro

--V

-- V

V

V

V

V

,t',lo'.~l-ol'-,,.o

~V

V



g

£

.=. #

~V

8 0

594

CELIBERTI AND HARRIS

most sessions (17) to reach mastery in eliciting a response from her brother; her rate of initiations was a direct function of her brother's noncompliant behavior. When Rick was compliant and cooperative, Susan made more overtures that were reinforced by his response; when he was noncompliant, she became less interactive. Rick's intrusive behaviors dissipated over the course o f training as he began to find cooperative play a more rewarding activity; and his increased participation provided Susan with better opportunities to practice and master the two remaining skills, reinforcing and prompting. Jessica and Lauren had similar encounters with their siblings' noncompliance, but the interactions were not sufficiently problematic to hinder skill acquisition. The training phase that focused on eliciting responses from the child with autism was made up o f two components that were addressed simultaneously. The first component involved increasing the frequency of the sibling's attempts to initiate an interaction. The other component was included to "fine-tune" these attempts so that they would be more likely to result in a favorable response. With training, all three siblings demonstrated greater consistency in attempting interactions that met the criteria. Those baseline sessions in which Susan and Lauren earned 100°70 for their attempts to elicit a response from the child with autism were based on only one attempt. On the other hand, their high proficiencies demonstrated later in training are much more reliable indices of competency because these were based on a much larger sample of attempts. Maintenance probes collected 3 weeks, 6 weeks, and 16 weeks after the end of training indicated that the siblings maintained their skills at levels approximating the final sessions of training. Given the intensity of training and the emphasis placed on consistency, it was encouraging that the targeted skills endured over time without booster training sessions or direct parental involvement. The levels o f interaction and rates o f praising and prompting also generalized to a novel toy. Generalization probes with Jessica and Connie revealed that Jessica's use of the target skills paralleled the acquisition of these skills with Jimmy. This transfer of skills occurred without Connie's participation in any of the training sessions. Furthermore, Jessica was not given any specific feedback or instructions (other than general praise) on how to interact with Connie. The generalization persisted at the 16-week follow-up, indicating that not only was Jessica able to use the targeted skills with another handicapped sibling, but that the transfer maintained over extended periods of time.

Anecdotal Information The three siblings were interviewed before and after training. After training, each described being more comfortable interacting with her brother or sister with autism and attributed this change to participation in training. The range of activities in which the siblings interacted with their brother or sister with autism broadened as a function of training. According to the siblings and their parents, the transfer of the targeted skills also generalized to interactions without toys, including such contexts as dinnertime interactions, outdoor activities, and other types of sibling contacts (e.g., Lauren reported that she and

BEHAVIORAL INTERVENTION

595

her sister had begun to do each other's hair and Susan described how she and her brother were starting to do arts and crafts together). All three siblings indicated that before training their decision to stop playing with their brother or sister with autism was related to the child's behavior. After training, two of the siblings said it was often her decision when to stop playing and that it was rare that the child with autism would misbehave sufficiently that the sibling would want to stop playing.

Social Validation Information The results of social validation assessment provided strong support for the notion that trained siblings exhibit behaviors that are detectable by naive observers and are perceived to be positive in nature. For both the original sample and the replication sample, every pre-post comparison reached statistical significance except for the item addressing frustration in the sibling. Interestingly, Jessica and Susan were rated as being slightly more frustrated after training, which is not what one would expect when considering the social validation findings as a whole. However, during the 1-minute baseline vignettes of Jessica with Jimmy and Susan with Rick, there were no notable baseline interactions that could have been perceived to be frustrating to the siblings. Lauren, on the other hand, attempted to interact with Annie but was rejected. For the three siblings, the baseline sessions typified "parallel play." The absence of interactions during baseline may explain the apparent increase in frustration. Clinical Issues and Implications During the baseline probes before training, the three siblings made few attempts to initiate interactions with their brother or sister with autism. The siblings' use of prompts and reinforcing statements was consistently low across baseline sessions despite ample opportunities to utilize such skills. The baseline videotapes document that the pretraining play o f the dyads comprised parallel play; the children were positioned adjacent to one another but were playing independently. There were few episodes o f joint attentive behaviors and a relative absence of positive affect, particularly in the child with autism. The parents and siblings regarded these baseline interactions as accurate representations of the children's play. In fact after a baseline probe, one sibling (Jessica) asked the experimenter if she had done a very good job playing with her brother when in fact there were no interactions at all. Early in training, all three siblings were often discouraged and frustrated by their brother's or sister's behavior. The children with autism were characteristically uncooperative and attempted to exert a punitive effect on the sibling's overtures. Nevertheless, the siblings were generously reinforced for their efforts regardless o f the cooperation of the child with autism. This reinforcement was directed toward the "correctness" of the sibling's application of the targeted skills. This feedback may have assumed a function of temporarily shifting the locus o f reinforcement so that the experimenter's praise and encouragement overrode the punishing nature of the child's behavior. This shift

596

CELIBERTI AND HARRIS

allowed for two important changes: (1) The siblings were able to acquire behavioral skills and begin to feel efficacious in the use of these skills; and (2) the brother or sister with autism became desensitized to the perceived aversiveness of the play sessions, more comfortable with the nature and extent of demands placed upon him or her, and more intrinsically interested in the play activities. Fortunately, the behavior of the children with autism shifted, and they became more interested in the play sessions. As interactions with the child with autism became less problematic, the experimenter's praise statements were directed more toward highlighting the child with autism's increased responsiveness, interest, and appropriate behaviors as a function of the sibling's efforts rather than being directed to the "correctness" of the sibling's use of the skills. Sensitizing siblings to how their skills were impacting upon the child with autism and influencing the child's behavior may have facilitated their taking "ownership" of the appropriate responses and feeling a greater degree of pride. This sensitivity may positively influence the maintenance of skills over time. Limitations A multiple-baseline design across skills becomes somewhat less rigorous if the skills are related and the subjects have opportunities to observe targeted skills prematurely and partially incorporate the skills into their repertoires. Overall, the siblings' data support a functional relationship between training and skill development; this relationship is most clearly evidenced by Susan's skill-acquisition profile and least evidenced by Jessica's and Lauren's acquisition of prompting skills. Given that the three sets of skills were interwoven and that the experimenter demonstrated all three sets of skills when he modeled play for the sibling during the initial segment of the lesson, it was not surprising that there was some spontaneous acquisition of skills during the extended baseline phases. Furthermore, performance data collected across the various phases of training suggest that the siblings may acquire some skills through observation but that, in order for the skills to become well learned and an integral part of the sibling's repertoire, direct and systematic training is necessary. This suggests that the presence of "models" of behavior skills (e.g., parents) and opportunities to observe these models is not sufficient for siblings to acquire complex behavioral skills, and behavioral training targeting the sibling is warranted. Although observational coding was geared toward the skills associated with the siblings' training, the data for the children with autism do show a functional relationship between the sibling's training and corresponding increases in frequency of desired responses. It might have been interesting to track other changes in the play behavior of the child with autism, such as the percentage of appropriate responses as a function of training. However, two findings lessened the degree to which the child's percentage of appropriate responses was meaningful. First, as the siblings learned the skills and became more confident in their interactions with the child with autism, they stepped up the complexity and sophistication of commands (e.g., "Make the cow jump over the fence

BEHAVIORAL I N T E R V E N T I O N

597

twice and have him give the farmer a lick."). Therefore, it would be hard to compare percentage of appropriate responses when the level of commands became so markedly different over the course of training. Second, the siblings were purposely asking more difficult commands during the last phase of training because they were attempting to practice their new prompting/correction skills. Given these findings, a dichotomous coding of "appropriately responded" and "did not appropriately respond" was largely inadequate for understanding changes in the child with autism's play. The benefits of involving siblings are not without drawbacks. One concern is that, as a function of the sibling's participation in training and acquisition of relatively sophisticated behavioral skills, parents might increase their expectations of the sibling, as might the sibling. Parents may place excessive demands upon the sibling, or increase their expectations, believing that they are acting in the best interests of the child with autism. Directions for future research. Future research in the area of sibling training needs to address ways to enhance the generalization and maintenance of the sibling's behavioral skills. The present study was sensitive to issues of limited generalization and maintenance over time by using multiple exemplars (i.e., a broad set of training toys); teaching simple, yet effective, strategies versus highly sophisticated techniques; incorporating a high criterion to reflect proficiency so that the skills become highly internalized and automatic; and endeavoring to shift the locus of reinforcement for the sibling's efforts to the brother's or sister's appropriate responses, thereby promoting the sibling's intrinsic motivation to play with the child with autism. However, follow-up booster sessions or homework assignments were not included in this study and might lead to an even more favorable outcome. It might also be useful to incorporate supplemental training in unstructured settings so as to help the siblings generalize their skills. Future research should also demonstrate that siblings, as a function of behavioral training, can actually enhance the functioning of their brothers and sisters with autism. To this end, more sophisticated measures of the play behavior of the child with autism can be included as dependent measures. The siblings who participated in this study were all normally developing girls who were three to six years older than the child with autism. Given the homogeneity of the subject sample and the small sample size, the external validity of this study is limited. The moderating role of intellectual functioning, gender, age, siblings' birth order, and child characteristics could not be addressed in the present study. Future studies are warranted to tease out the influence of these variables and to determine how training interventions and goals might be altered to enhance outcome. Other research efforts could explore the role that parents can assume in teaching, assessing, and maintaining these skills in their nonhandicapped children. CoUetti and Harris (1977) alluded to the potential long-term benefits of involving parents by training them to be reliable observers of the sibling's skills. Powell et al. (1983) recognized the important role of parents in facilitating interactions between their mentally retarded and normally developing children and demonstrated the efficacy of having parents coordinate social interactions.

598

CELIBERTI AND HARRIS

Training packages that do not incorporate parents as the actual "trainers" should at least educate them in how to function as adequate "maintainers" of the sibling's newly acquired skills. However, parents who train their own children would presumably be in a better position both to monitor and assess the effect of the intervention and to ensure that the skills maintain over time. Furthermore, the siblings would likely have in their parents very suitable "role models" of the effective use of behavioral skills. The family and the child with autism could benefit from the cohesiveness and consistency, and this could translate into the potential of a better outcome for the child with autism.

References American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. Bernstein, G. S. (1989). Social validity and the debate over the use of aversive/intrusive procedures, the Behavior Therapist, 12, 123-125. Carr, E. G., Taylor, J. C., & Robinson, S. (1991). The effects of severe behavior problems on the teaching behavior of adults. Journal of Applied Behavior Analysis, 24, 523-535. Coe, D. A., Matson, J. L., Craigie, C. J., & Gossen, M. A. (1991). Play skills of autistic children: Assessment and instruction. ChiM and Family Behavior Therapy, 13, 13-40. Colletti, G., & Harris, S. L. (1977). Behavior modification in the home: Siblings as behavior modifiers, parents as observers. Journal of Abnormal Child Psychology, 5, 21-30. Egel, A. L., & Gradel, K. (1988). Social integration of autistic children: Evaluation and recommendations, the Behavior Therapist, 11, 7-11. Handleman, J. S., & Harris, S. L. (1986). Educating the developmentally disabled." Meeting the needs of children and families. San Diego, CA: College Hill Press. Hawkins, R. P. (1991). Is social validity what we are interested in? Argument for a functional approach. Journal of Applied Behavior Analysis, 24, 205-213. Irvin, L. K., & Lundervoid, D. A. (1988). Social validation of decelerative procedures by special educators of severely handicapped students. Research in Developmental Disorders, 9, 331-350. James, S. D., & Egel, A. L. (1986). A direct prompting strategy for increasing reciprocal interactions between handicapped and non-handicapped siblings. Journal of Applied Behavior Anal),sis, 19, 173-186. Koegel, R. L., Schreibman, L., Britten, K. R., Burke, J. C., & O'Neill, R. C. (1982). A comparison of parent training of direct clinic treatment. In R. L. Koegel, A. Rincover, & A. L. Egel (Eds.), Educating and understanding autistic children (pp. 260-279). San Diego, CA: College Hill Press. Lobato, D. (1983). Siblings of handicapped children: A review. Journal of Autism and Developmental Disorders, 13, 347-364. Lobato, D., & Tlaker, A. (1985). Sibling intervention with a retarded child. Education and Treatment of Children, 8, 221-228. Meyer, L. H., Fox, A., Schermer, A., Keteisen, D., Mortan, N., Maley, K., & Cole, D. (1987). The effects of teacher intrusion on social play interactions between children with autism and their non-handicapped peers. Journal of Autism and Developmental Disorders, 17, 315-332. Odom, S. L., & Strain, P. S. (1984). Peer-mediated approaches to promoting children's social interaction: A review. American Journal of Orthopsychiatry, 54, 544-557. Powell, T. H., Salzberg, C. L., Rule, S., Levy, S., & Itzkowitz, J. S. (1983). Teaching mentally retarded children to play with their siblings using parents as trainers. Education and Treatment of Children, 6, 343-362.

BEHAVIORAL INTERVENTION

599

Schreibman, L., O'Neill, R. E., & Koegel, R. L. (1983). Behavioral training for siblings of autistic children. Journal of Applied Behavior Analysis, 16, 129-138. Schwartz, I. S., & Baer, D. M. (1991). Social validity assessments: Is current practice state of the art? Journal of Applied Behavior Analysis, 24, 189-204. Strain, P. S., Hoyson, M., & Jamieson, B. (1985). Normally developing preschoolers as intervention agents for autistic-like children: Effects on class deportment and social interaction. Journal of the Division of Early Childhood, 9, 105-115. Swenson-Pierce, A., Kohl, E L., & Egel, A. L. (1987). Siblings as home trainers: A strategy for teaching domestic skills to children. Journal of the Association for Persons with Severe Handicaps, 12, 53-60. Thorndike, R. L., Hagen, E. R., & Sattler, J. M. (1986). The Stanford-Binet Intelligence Test: 4th Edition. Chicago: The Riverside Publishing Company. Waiters, A. S., Barrett, R. P., & Feinstein, C. (1990). Social relatedness and autism: Current research, issues, directions. Research in Developmental Disabilities, H, 303-326. Wulff, S. B. (1985). The symbolic and object play of children with autism: A review. Journal of Autism and Developmental Disorders, 15, 139-148. RECEIVED: April 19, 1993 ACCEPTED: July 13, 1993