Evaluating Changes in Dynamic Social Interaction Skills Following a Randomized Controlled Trial of the START Socialization Intervention for Adolescents with ASD

Background: Treatment outcomes of interventions for Autism Spectrum Disorder (ASD), as well as social interventions in particular, have been primarily evaluated using parent report measures (Miller et al. 2014). These measures, while serving as valuable indicators of improvement, have notable limitations related to social desirability and other reporting biases (Moskowitz, 2006). Consequently, researchers are calling for more rigorous measurement tools to assess the effects of interventions (e.g. Lord et al., 2005; McMahon et al., 2013). Behavioral observations provide an objective and quantifiable outcome measure that can augment traditionally used survey measures. Systematic coding of observed behaviors during live social exchanges may offer a rigorous method for assessing changes in social competence.

Objectives: This study used direct observation of dynamic social interaction skills during video-recorded conversations to assess the efficacy of a randomized controlled trial (RCT) of the Social Tools and Rules for Teens (START) socialization intervention for adolescents with ASD.

Methods: Thirty-five adolescents with high-functioning ASD, ages 12-17 years, were enrolled in an RCT of the START program. Adolescents were randomized to either a treatment or waitlist group. The 20-week group intervention took place once a week for 90-minutes per session. Five-minute “get to know you” conversations between participants and similar-aged students, whom they have never met, were video-recorded at pre-intervention and post-intervention. Trained and reliable research assistants, but blind to condition and time point, systematically coded the videos for three types of social behaviors targeted during the intervention: question asking, mutual engagement, and positive facial expressions. An ANCOVA, using pre-intervention values of each social behavior as the covariate, compared the outcomes of the treatment and waitlist groups on the target social behaviors to examine if there was a difference. Paired samples t-tests were performed to examine the whole cohort of adolescents who completed the START program, and assess differences between pre-intervention and post-intervention.

Results: Results of the ANCOVA revealed a significant treatment effect for both question asking and positive facial expressions (p < 0.05). These skills additionally exhibited significant moderate to strong positive correlations (r values ranging from 0.47 to 0.61) with peer ratings of social ability (p < 0.01), which were made by blind observers after watching the video-recorded conversations. Examining the whole cohort collectively revealed significant increases in all three social behaviors from pre-intervention to post-intervention (p < 0.05).

Conclusions: To our knowledge, this study presents the first RCT utilizing behavioral coding of dynamic social exchanges to assess the efficacy of a social intervention for adolescents with ASD. Overall, the START program appears to positively impact the use of several verbal and nonverbal conversational strategies that are crucial for both social skill development and positive social impressions. Behavioral coding of specific target behaviors allows an unbiased, directly observable method for measuring social change that cannot be adequately captured by survey measures alone but serves to supplement the results from these measures. These findings provide evidence that these specific social behaviors may be linked to social desirability in every day interactions.