Given the pervasiveness of emotional and behavioral deficits in individuals with autism spectrum disorders, there is a pressing need for effective interventions to address their difficulties in social communication and interactions. This study examines the effectiveness of a software application-based intervention (ABI) program that embeds several widely recognized evidence-based practices into an adaptive training system to directly address the challenges in core social skills faced by children with autism. This research expands the current body of studies by incorporating puppet role-play, conversation partner role-play activities with video modeling, Social Stories, and question answering via the correction staircase approach in the program.

A sequential explanatory mixed methods design was adopted, and this divided the research into two phases. The statistical portion of the study compared the ABI group with the treatment as usual (TAU) group using a randomized controlled trial pretest-posttest design. Nineteen participants were examined in this phase. Four measures of functioning – Social Responsiveness Scale-2, Gilliam Autism Rating Scale-3, Vineland Adaptive Behavior Scales-3, and Social Communication Questionnaire – were utilized to compare the treatment approaches. Nonparametric tests (random permutation and Mann-Whitney U) were used to compare the treatments; the results demonstrated that participants who received ABI functioned at a significantly higher level at posttest than those who received TAU.

A multiple probe across participant design, replicated for four groups in phase two, was used to collect quantitative data across the baseline, intervention, maintenance, and generalization portions of the study. Participants significantly improved their social greeting, self-introduction, and play-initiation skills through the ABI program. The effectiveness of the intervention was evaluated based on student engagement and performance in discriminating, understanding, and expressing the target behaviors through their role-play and adaptive training sessions. Procedural reliability, interobserver agreement, and effectiveness data demonstrated that the procedures and the teachers were successful in imparting the target skills. The performance of nine out of 11 participants remained at very high accuracy levels in the maintenance probes, demonstrating stable or upward trends in target behaviors compared to the intervention sessions. The generalization probes showed that most of the students were able to generalize the skills learned to new conversation partners, and the participants performed better when interacting with new adults than with novel peers.