Examining Socialization Improvement Growth Trends of Adolescents Participating in an RCT of the Social Tools and Rules for Teens (START) Program for ASD: A Multi-Level Modeling Study

Background: The interest in effective interventions to target the social vulnerabilities of adolescents with ASD has significantly increased in recent years (Miller et al, 2015). Such intervention efforts are not administered in a single therapeutic dose, but rather are conceptualized as operating through a series of interactions that, over time, aim to build self-sustaining interpersonal capabilities and alter the social trajectory of participating individuals. In recent years, researchers have become increasingly focused on the accumulating effects of an intervention on the unfolding behaviors and skills of a treatment recipient over time (e.g. Lerner et al, 2011). Such knowledge is crucial for further developing promising interventions, as it informs the relationship between dosage, time, accumulating therapy exposure, and eventual outcome. When data is serially collected at multiple time points along the course of an intervention timeline, valuable information about social growth trajectories can be gathered and analyzed.

Objectives: This investigation evaluated the effectiveness of the Social Tools And Rules for Teens (START) intervention program for adolescents with ASD using (a) serial data collection and (b) multilevel modeling to thoroughly understand the intercept and slope of social improvement data.

Methods: The participants of the project were 35 adolescents with ASD. All participants were randomly assigned to (a) the START program group or (b) a waitlist group for 20 weeks. The START program is a motivation-based curriculum that combines experiential and didactic elements into a comprehensive intervention package that includes high school peer models, individualized and group social targets, and self-management of selected skills. Participants assigned to the START condition participated in a weekly two-hour session consisting of: individual check-in sessions, free socialization periods, social activities, videos, social topic discussion and practice, and checkout sessions with parents. Dependent measures included parent and adolescent survey measures (Social Skills Improvement System, Social Responsiveness Scale - 2, Social Competence & Motivation Scale) and regular social conversation probes collected repeatedly at 5-week intervals. Multilevel modeling was used to analyze social competence change both within and between individuals. Unconditional, time-only, and full HLM models were specified.

Results: In the full HLM model, Time alone produced a significant slope effect on the adolescent SSIS scores (p < .05) and parent SCMS scores (p < 0.1). Treatment Group (i.e. START program participation) created a significant slope effect on adolescent SSIS scores (p < .05), SRS scores (p < .01), and both parent (p < .001) and adolescent (p < 0.5) SCMS scores, providing evidence of significant therapeutic gains unique to START program participation.

Conclusions: The full HLM model provides strong evidence that START participants experience markedly improved social growth trends when compared to waitlist controls. Specifically, significant positive trends in social skill (SSIS) and social competence/motivation (SCMS) were noted, along with reductions in ASD-related social impairments (SRS-2). This investigation is one of the first social skills studies to use MLM techniques to model robust changes on multiple measures of social competence. Overall, this study presents promising findings regarding the efficacy of an experiential social intervention for adolescents with ASD.