Behaviorally-Based Interventions for Improving Social Interaction Skills of Children with ASD in Inclusive Settings: A Systematic Review and Meta-Analysis

Students with autism spectrum disorders present deficits in social interaction skills that may prevent their successful inclusion in general education placements. Considering the increasing number of children with autism spectrum disorders (ASD) being educated in inclusive settings and recent requirements on the use of research-based interventions in schools, the purposes of this study were (1) to evaluate the quality of single-case research and determine whether behaviorally-based interventions to improve social interaction skills of children with ASD in inclusive settings can be considered evidence-based practices and (2) to conduct a meta-analysis investigating whether specific factors such as participants' age, behavioral components used in the intervention, target social interaction skills, intervention implementer, and peer training moderate effectiveness of the interventions.

Specific criteria for quality of single-case research were used to classify studies according to their certainty of evidence. Tau-U, a non-parametric index of effect size in single-case research, was used to measure the intervention's magnitude of change on target outcomes. Differences between levels of the moderators were analyzed using statistical significance test (p = .05) through the use of 83.4% confidence intervals. Results indicate that the use of behaviorally-based interventions to improve social interaction skills of students with ASD in inclusive settings can be considered evidence based practices. In addition, the interventions produced overall high effect size, indicating their effectiveness based on studies meeting minimum standards of methodological quality. More specifically, the interventions are demonstrated to be effective for preschool and elementary school children between the ages of 2 and 10 years. Studies targeting social interaction initiations or responses in isolation were more effective than studies focusing on both skills. While interventions using planned reinforcement were shown to be more effective, no differential effects were found regarding the use of planned modeling. No differences were found regarding intervention implementer. Finally, the use of peer training did not appear to increase effectiveness of the behaviorally-based social skill interventions. The results and their implications for practice and future research are discussed.