At this point in the class, you should know the proper way to set up a title page, headings, etc. Correct APA format for the

JEIBI VOLUME 4 – NUMBER 3

532

Social skills interventions for preschoolers with Autism Spectrum Disorder: A description of single-

subject design studies

Jennifer M. Gillis, Ph.D., BCBA1 & Robert C. Butler, B.S.

Abstract Social skill development is one of the primary areas of intervention for young children with Autism Spectrum Disorders (ASD). The purpose of this article was to conduct a retrospective review of social skills intervention research for preschool children with ASD. A review of 17 single-subject design studies from twelve journals (1999-2006) was conducted. We assessed information concerning the features of social skills interventions, the elements of single subject designs utilized across studies, and the overall success of interventions for social behaviors. A comparison with a review conducted by Vaughn et al. (2003) revealed some important common elements of interventions such as reinforcement, modeling and prompting, providing converging evidence from group design studies and single subject studies. Keywords: Autism Spectrum Disorders, Social Skills, Intervention, Preschool

The term, Autism Spectrum Disorder (ASD), is commonly used to identify a family of disorders, including, Autistic Disorder, Asperger’s Disorder, or Pervasive Developmental Disorder Not Otherwise Specified, in the Diagnostic Statistical Manual–IV, Text Revision (DSM-IV TR; American Psychiatric Association, 2000). An individual with ASD exhibits deficits in core areas including social development and social skills, language and communication skills, and repetitive, ritualistic, stereotyped behaviors. Across the ASD spectrum, the expression of these deficits ranges from mild to severe. Social Skills Deficits in ASD

The primary deficit of ASD is social impairment (Romanczyk, White, & Gillis, 2005; Weiss, 2001; Weiss & Harris, 2001a). We use the term, social impairment to refer to any deficit or limitation or delay in social awareness, social competence, and social development. We define social skills as behaviors each person learns to facilitate awareness of his/her social environment and social contingencies, to be able to solve social problems (i.e., demonstrate social competence), and other behaviors that are developmentally appropriate.

Social skills deficits are common to all individuals with an ASD, are pervasive, and typically

manifest at a very young age. The DSM-IV-TR (APA, 2000) lists the impairments in social interaction as: 1) deficits in nonverbal behaviors (e.g., eye-to-eye gaze, recognition of facial expressions, use of gestures to regulate social interaction), 2) failure to develop peer relationships appropriate to the child’s developmental level, 3) lack of spontaneous seeking to share enjoyment and interests (e.g., failure to show, bring, or point out objects of interest), and 4) lack of social or emotional reciprocity (e.g., unable to or does not respond to emotions or social interactions from others).

ASD is a heterogeneous disorder, meaning that one individual with ASD will have differing areas

of social impairment with differing levels of severity from another individual with ASD.

1 Author Notes-The authors would like to thank Trina L. Butler for her assistance in data collection for this literature review. The authors wish to acknowledge the helpful feedback and comments of the two anonymous reviewers.

JEIBI VOLUME 4 – NUMBER 3

533

Additional examples of social impairment observed in children with ASD include difficulties initiating social interactions, responding to the social initiations of others, initiating or responding to joint attention, and recognizing emotions in self. Areas of social impairment frequently observed in preschool age children with ASD include social play, dramatic play, friendship-seeking behavior, and cooperative play. Poor motivation to learn social behaviors or to engage in social activities is also commonly observed in children with ASD. These comprise only a fraction of social skills observed in individuals with ASD.

The extant literature suggests that early social impairments greatly impact future relationships,

employment, independent living, and other mental health issues (e.g., anxiety, depression) (Bellini, 2004; Strain & Scwartz, 2001; Tantam, 2000). Thus, there should be an emphasis for intervention to occur early and to focus on the development of social skills in order to build social interactions and relationships. Unfortunately, it remains unclear as to which social skills are important to teach that will lead to better outcomes in areas related to social impairment for children with ASD. An issue that remains elusive and is of equal importance is to determine the specific types of interventions that effectively teach social skills. Normal Social Skills Development in Preschoolers

Typical social development in preschool age children (i.e., ages two through five) involves the development of social interactions with peers and development of play behaviors. For all children, play begins with nonsocial or solitary play, when a child is content to play alone with toys/objects with or without the engagement of others in the same play activity. Children then develop parallel play, which is when a child plays with the same toys as another child, but does not necessarily interact with the other child (Rubin, Fein, & Vandenberg, 1983).

Children begin to form their first friendships between three and four years of age. At this age,

children begin to increase cooperative and interactive play with peers and develop increased self- regulation (a necessary skill in order to play well with others). Children begin to understand turn-taking and simple rules to games in small groups. By the age of five, a child’s social communication skills are more complex, for example, children will tell jokes, share a variety of experiences with others, and take turns in conversations (American Academy of Pediatrics, 1999).

Social skills interventions for children in the preschool age range focus on the improvement of

behaviors related to social interactions with peers, including play skills (e.g., engaging in pretend play, sharing, turn-taking, etc). Other examples include, social communication skills (e.g., initiating and responding to others verbally or nonverbally), socio-emotional skills (e.g., affect recognition, empathy), and friendship skills. For children with ASD, focusing on age-appropriate skill development is important, but becomes difficult given the limited skill repertoires and difficulty in learning for children with ASD. Recent Review of Literature In a review of social interventions by Strain and Schwartz (2001), conclusions are made suggesting that there is plentiful research demonstrating the efficacy of teaching children with ASD specific social behaviors, including initiating and responding to social interactions. These authors identify the need to examine issues related to generalization and maintenance of social behaviors. In addition, the authors propose the use of eco-behavioral assessments (i.e., examining the behavior in applied contexts such as the classroom). Strain and Schwartz also emphasize the importance of social validity, in terms of evaluating the acceptability of an intervention by family members/caregivers and teachers and whether these interventions would be sustainable outside of the research setting.

JEIBI VOLUME 4 – NUMBER 3

534

There is a literature base to provide potential targets for social skills interventions (Terpstra, Higgins, & Pierce, 2002; Weiss & Harris, 2001a; Koegel & Koegel, 1995). However, the specific interventions that are successful for preschool children with ASD have not been fully confirmed nor endorsed by the majority of professionals across disciplines.

In an effort to examine key features of social skills interventions for preschool children with

developmental disabilities, including ASD, Vaughn, Kim, Morris Sloan, Hughes, Elbaum, and Sridhar (2003) reviewed 23 group design studies from 1975 to 1999. Vaughn et al. (2003) found 10 categories or features important in social skills interventions. These included prompting (i.e., verbal or physical prompting of a social skill), rehearsal/practice, play-related activities (to elicit social skills), free-play generalization, reinforcement of appropriate social skills, modeling of social skills, social skills-related storytelling, direct instruction of social skills, imitation, and time-out (to decrease inappropriate behaviors).

Vaughn et al. (2003) calculated effect sizes for the studies reviewed and found that the studies

with the greatest effect sizes included the following four features in the interventions: modeling, play- related activities, rehearsal/practice and/or, prompting procedures. It should be noted that the other categories listed in the previous paragraph were not necessarily associated with negative outcomes. For the 23 studies reviewed, there was an average of 3.4 categories of interventions included in each study. The authors stated that incorporating a free-play generalization phase is helpful in determining whether the above categories of intervention facilitated the generalization of the targeted social skills to new settings, peers, and social-related activities. Single Subject Research

Behavioral science not only aims to understand the nature of the determinants of behavior, but also how to apply these principles in order to change dysfunctional behavior through intervention. Interventions derived from empirical findings traditionally have been evaluated using two types of research designs. The two major types of research designs that meet the standards of scientific research methodology are single subject and group designs.

Single-subject and group designs provide valuable information. Both designs examine functional

or causal relationships between an independent variable (e.g., an intervention) and a dependent variable (e.g., behavior) and to control for threats to internal and external validity. Operational definitions of the intervention and the targeted behavior(s) are also fundamental aspects of both designs.

Despite some methodological similarities, single-subject and group designs posses several

important differences. First, group design studies measure dependent variables (DVs) before and after the intervention, whereas single subject design studies measure the DV(s), continuously and over time, including before (i.e., baseline) during, and after the intervention. This is especially helpful when a timely assessment of the effects of an intervention is warranted (e.g., determining the precise dosage of a medication).

Second, group designs require random or quasi-random assignment of participants to

experimental and control conditions (to help control for threats to internal validity). Single subject designs, however, do not require the random assignment of participants. In other words, there are no “participant requirements”, per se, in single subject design studies. Single subject designs typically include between three to eight participants, on average, with each participant serving as his/her own control. This latter characteristic of single-subject designs eliminates the need for experimental and control groups. This is a major advantage when a social skills training intervention occurs in a classroom

JEIBI VOLUME 4 – NUMBER 3

535

of preschool children with ASD. As these children present with a heterogeneous skill set and, most likely, varying levels of functioning, it is difficult to obtain an adequate control sample.

Third, group design studies allow for statistical analyses that yield estimates of the causal effects

of the intervention (e.g., Analysis of Variance). These analyses are not possible with single subject design studies. Rather, visual inspection of the data is utilized, which also permits inferences to be drawn about causal relationships between the independent IV(s) and DV(s). There are specific criteria established to guide the visual inspection process.

Single-subject designs can be particularly useful when conducting research in applied settings,

which is often the case in the evaluation of interventions for individuals with ASD. As mentioned above, single-subject research provides for the examination of the effects of an intervention at the individual level. Important aspects of single subject research designs include: a) the demonstration of a functional relationship between treatment and behavior change; b) DVs are operationally defined; c) repeated measurement of the DVs over time; d) the illustration of social validity of the intervention; e) the use of methods for evaluating procedural fidelity and external validity; f) a description of the baseline phase with sufficient data collected to evaluate effect of the IV; and g) a thorough description of the participants and setting, to allow for replication (Horner, Carr, Halle, McGee, Odom, & Wolery, 2005). For a more comprehensive description of single subject research designs used in special education research, please see Horner et al., (2005). Current Review In this paper, we utilized the categories identified by Vaughn et al. (2003) in their review of group design studies to determine whether these categories were also included in single subject design studies. Vaughn identified modeling, play-related activities, rehearsal/practice and prompting as the features associated with the largest effect sizes and, thus positive results. Effect sizes were not calculated for this review, as this statistic is not typically utilized in single subject research studies due to the low number of participants.

The primary purpose of this review was to examine the state of the single subject research over the past five years for social skills interventions for preschool age children with ASD. A second purpose of this review was to assess whether the features of social skills intervention packages were similar in Vaughn et al’s (2003) review of group design studies. A synthesis of the single subject research at this time is helpful in order to examine the status of interventions for social skills interventions and to identify areas that warrant attention for future research.

Method

We selected journal articles from PsychInfo and the search engine on the website for the Journal

of Applied Behavior Analysis. Studies included in this publication were published between June 1999 and December 2006. Keywords utilized for this search included autism, ASD, Asperger’s, handicapped, special education, language impairment, communication impairment, developmental delay, mentally retarded, social skills, joint attention, social referencing, imitation, friendship, and approach. Terms were entered alone and in conjunction with other terms, truncating the keywords when appropriate, i.e., autis* and dev*. We reviewed each article to determine if it met the inclusion criterion for this review, resulting in 17 studies. Articles came from many different journals (i.e., Behavioral Interventions, Early Education and Development, Education and Treatment of Children, Educational Psychology in Practice, Development and Education, Focus on Autism and Other Developmental Disabilities, International Journal of Disability, Journal of Applied Behavior Analysis, Journal of Autism

JEIBI VOLUME 4 – NUMBER 3

536

and Developmental Disorders, Journal of Positive Behavior Interventions, Topics in Early Childhood Special Education, and Topics in Language Disorders).

The criteria for inclusion used to select articles for review are presented below: 1. Preschool Age. At least half of the participants in each study were between the ages of

2 and 5. 1.2. Design. Only studies utilizing single subject designs were included. Group

designs were excluded. 1.3. Peer Review. We only used articles that had been peer reviewed for this study in

an attempt to include higher quality research. 1.4. Social Skills. Studies were included if they focused on increasing beneficial

social behaviors or skills (e.g., joint attention, communication, appropriate eye contact, cooperative play, sharing, sociodramatic play, imitation, self-management, complimenting, initiating play, expressing sympathy, showing disapproval appropriately, showing appreciation, pretend play, group interactions, verbalizing requests).

1.5. Diagnosis. The majority of participants were required to have a diagnosis of an autism spectrum disorder (e.g., Autistic Disorder/Autism, Asperger’s Disorder, PDD- NOS or pervasive developmental disorder.) Studies utilized various methods to diagnose the participants. Further information on diagnoses is available in Table 1.

Results & Discussion

Seventeen studies met criteria for inclusion in this study. Each of these studies used single subject

research designs, however, the specific type of design used varied. The median age was 50 months. There were 41 males and 9 females in this study. This reflects a similar ratio to that of male to females in ASD. The total number of participants included in all of the reviewed studies was 50. Table 1 shows the participant demographics across the studies, including diagnoses, gender, and age.

TABLE 1, NEXT PAGE

JEIBI VOLUME 4 – NUMBER 3

537

Social Skills Interventions in Single Subject Design Studies

We examined aspects of interventions included in single subject design studies and compared them to the data obtained in Vaughn et al.’s review (2003) of group design studies. Specifically, we examined the following aspects of social skills interventions: 1) who implemented interventions; 2) the length of the interventions; and 3) the categories of interventions most commonly used.

Implementation of the interventions. Vaughn et al. (2003) reported that mostly teachers

implemented interventions, followed by researchers, therapists, parents, and one study with a peer who implemented the intervention. The studies we reviewed utilized several sources to implement the intervention as well. These include: peers (n=5), video (n=5), researcher/research assistant (n=4), teacher/tutor (n=3), parent (n=2), and therapist (n=1). In some studies, more than one person (or source) implemented an aspect of the intervention, hence the total number of sources that implemented the intervention is more than 17. In the studies reviewed, the interventions were mostly implemented by peers and with a video, which is different from Vaughn et al.’s findings. Based on our review of these studies, there did not appear to be a relationship between the person/source that implemented the intervention and the results of an intervention.

Length of the interventions. The length or duration of an intervention was measured differently

across the reviewed articles including: number of sessions, number of observations, and total time commitment (i.e., hours, months). Thirteen of the 17 studies reviewed reported the length of intervention in terms of sessions conducted. In this review we utilized the authors report on number of sessions in the intervention phase. When the authors did not report the number of sessions in the intervention phase, we assumed that each data point reported in the X-Y graphs of the studies correlated with one session, unless the article indicated that this was an inappropriate interpretation. The number of sessions utilized in the intervention phase ranged between 1 and 80 with an average of 34.19 sessions per intervention. In two of the studies the length of intervention was reported in total number of hours required to implement the intervention. The intervention described by Dauphin, Kinney, and Stromer (2004) required 8-10 hours. The intervention by Symon (2005) required 25 hours. Two studies measured the length of intervention in terms of numbers of observations recorded. Arntzen, Halstadtrø, and Halstadtrø (2003) reported 118-120 trials across fifteen days. Tarbox, Ghezzi, and Wilson (2006) measured 300 observations in a one-day period. Across the reviewed studies, the overall length of intervention varied greatly, ranging from one day to 7 months.

Table 1. Participant demographics across studies.

Reported Diagnoses n

Autism 33 AspergerÕs Disorder 5 PDD-NOS 3 PDD 3 Developmental Disability/Delay 3 Comorbid Autism and PDD 2 Comorbid ASD with ADHD 1

Gender n

Male 41 Female 9

Age months

Mean ( Standard Deviation ) 53 (15) Median 50 Range 29-90 Note. Two of the participants were diagnosed as having comorbid Autism and PDD. It is important to note that according to the DSM-IV TR the diagnosis of autism excludes the diagnosis of comorbid PDD and Autism (American Psychiatric Association, 2000). It is unclear as to why one article utilized this diagnosis.

JEIBI VOLUME 4 – NUMBER 3

538

In comparison to the review by Vaughn et al. (2003), similar results are found. We acknowledge

that a the comparison of our findings to those of Vaughn et al.’s is an approximate and conservative comparison given that the authors presented the data differently. Considering this limitation, it appeared that the group design studies reviewed included for some social skills interventions, a range from 5 to 60 sessions was reported, others reported between 10-60 minutes for each session length. The group design studies reported intervention length by different time periods (e.g., weeks, months, etc).

Number of categories used in each intervention study. Unlike the group design studies reviewed

by Vaughn et al. (2003), which averaged 3.4 categories of interventions in each study, the review of single subject design studies indicated that an average of 2.6 categories of interventions were used in each study. This difference may be due to the differences of how we defined categories of intervention compared to Vaughn et al. (2003).

Vaughn et al. (2003) described play-related activities and storytelling as categories of

intervention because these types of activities might have solicited social behavior and social skills. We chose to not include these intervention categories because it is difficult to separate the effects of a context that includes play-related activities from another intervention category (e.g., prompting or imitation in a play activity context). There were no comparative studies that addressed this issue. We did find, however, that 5 of the studies included play-related activities as part of the intervention. We also excluded rehearsal/practice, because all but one of the single subject studies reviewed included multiple sessions that provided opportunities for the participants to practice social skills.

Figure 1. The frequency of different categories of interventions used to teach social skills across the reviewed articles are displayed.

Figure 1 shows the most commonly utilized interventions in the articles reviewed. The most frequently used categories of interventions were modeling, prompting, and reinforcement. For studies

JEIBI VOLUME 4 – NUMBER 3

539

utilizing modeling, there were three major types of models used. These included peer modeling (n=4), adults (n=1), and video modeling (n=4). The next most frequently used categories of interventions included imitation and peer training. Other categories of interventions used in at least one study included, parent training, priming, social communication acquisition, fading, naturalistic teaching strategies, and token economies. None of the studies reviewed implemented time-out.

The categories of reinforcement, modeling and prompting were used in both reviews and were

present in many studies. The ubiquitous nature of these categories provides convergent evidence of the potential necessity of including these categories of intervention in social skills interventions for preschool children with ASD. In this review, we make a distinction between the different types of modeling that are used in interventions. These include adult models, peer models and video models for teaching social behavior. State of Single Subject Design Studies

As part of our review, we examined the single subject designs of the reviewed studies in order to assess the quality of the social skills interventions across studies. An excellent article by Horner et al. (2005) provides descriptions of quality indicators of single subject design research and provides guidelines for evaluating such designs. The quality indicators that we evaluated and summarized below included demonstration of a stable baseline, description of participant characteristics and intervention setting, behavior that was measured, measurement of social validity, demonstration of procedural fidelity (which is related to evaluation of the independent variable or intervention), and external validity (i.e., maintenance and generalization). Please see Horner et al., (2005) for a thorough review of these and other quality indicators of single subject design research.

Baseline. One of the quality indicators is the presence of baseline data. In single subject designs,

there needs to be a presence of stable baseline data. Horner et al., (2005) suggest five data points or more. According to Kazdin (2003), two or more baseline data points are sufficient prior to implementing an intervention. Considering these differing reports and other issues related to the stability of the data in the baseline phase (i.e., slope, variability, steady state), we decided that a baseline phase needed to consist of at least three data points and the baseline needed to be described as stable.

It is our opinion that in 38% of the studies reviewed stable baseline data was not included for at

one or more participants prior to implementation of the intervention. Variability of baseline data or less than three baseline data points were present in these studies. Specifically, 22% of studies included at least one participant with no baseline data or less than three points of data in the baseline phase. Therefore, without stable baseline data, it is difficult to determine whether the intervention had a functional relationship with the change in behavior.

Participants and setting. A description of participants is an important characteristic of single

subject research with respect to determining if an intervention is appropriate and might be effective for other participants with similar characteristics. The manner in which the participants were described in the reviewed studies greatly varied. All of the studies reviewed indicated that the majority of participants were diagnosed with or stated as having an ASD (see Table 1). Information related to how participants received a diagnosis varied from participants receiving a diagnosis from an independent psychologist (47% of studies), to the authors only stating that the participants met the criteria listed in the DSM-IV TR as a basis for diagnosis (12% of studies), to a diagnosis based on a review of records (12% of studies), to simply stating that the participants had an ASD, but not reporting how the diagnosis was obtained (29% of studies). Standardized assessments that measure domains of functioning, such as social skills, cognitive

JEIBI VOLUME 4 – NUMBER 3

540

abilities, language/communication skills, adaptive skills, etc, were greatly underutilized. Many of the reviewed studies contained descriptions of participant characteristics by stating strengths and weaknesses of participants or reporting observations of the participants’ abilities and levels of functioning in different domains. Some studies reported a combination of standardized assessment scores and descriptions of characteristics. In some studies, there was not enough information about participants to determine the level of functioning and skill set of the participant(s), presenting a challenge for the reader in determining whether the intervention is applicable to other children with ASD

From our review of the description of …