The Effects of a Randomized Controlled Social Skills Intervention on Peer Relationships and Social Networks of Children with Autism in the School Setting

Background: Several peer intervention models have been employed to remediate social skills in children with autism; however, these models have not been subjected to systematic comparison, nor have they been implemented in regular school programs for school-aged children.

Objectives: This study examined the effects of intervention (child- or peer-mediated, combined, or no intervention) on the peer relationships and social networks of children with autism.

Methods: Participants included 60 fully-included children with autism (54 male, 6 female) and 1786 typically-developing children. All were recruited from grades 1-5 from 56 classrooms in 30 different schools across the Los Angeles area. They were an average of 8.14 years old (SD=1.56), with an average IQ of 90.97 (SD=16.33). Children with autism and their peers completed a friendship survey at the beginning and end of a 12 session, 6-week social skills intervention that was coded following the methods outlined in Cairns and Cairns (1994). All intervention sessions occurred at the target child’s school during recess periods twice a week.  Each child with autism was randomly assigned to the peer-mediated, child-mediated, combined, or no intervention (control) and received a follow-up visit 3 months after the intervention.

Results: Social network ratings, number of received friendship nominations, and the percentage of reciprocal three best friendships were analyzed using a multilevel random coefficient model in SAS Proc Mixed 9.1. Results indicated a significant treatment group by time interaction, F(8, 56) = 2.16, p<.05. Children with autism who received the combination treatment had higher social network ratings compared to children in the control condition t(8, 56)=2.71, p<.01.  Children with autism in the child-mediated condition did not significantly differ in social network ratings after treatment from children in the control condition. A treatment group by time interaction was also significant for children’s number of received friendship nominations, F(8, 56) = 2.42, p<.05. Children with autism in the peer-mediated treatment had more friendship nominations after treatment as compared to children in the control condition, t(8, 56)=2.10, p<.05. Lastly, results indicated that the percentage of children’s three reciprocal best friendships was significantly different across groups, F(3, 53) = 3.39, p<.05, such that children with autism in the child-mediated condition had significantly fewer reciprocal three best friendships in comparison to children in the peer-mediated t(3, 53)=-2.37, p<.05, combined t(3, 53)=-2.28, p<.05, and control t(3, 53)=-2.94, p<.01 conditions.

Conclusions: This study reports changes in children’s social network ratings and friendships in 12 school-based intervention sessions. Social skills can be taught to children with autism and the best avenue to do so is through a multi-agent model that involves the target child and typically-developing peers. Targeting only the child with autism did not improve the child’s social position in the class or reciprocated friendships suggesting that an adult-mediated one-on-one approach at school may be more stigmatizing to the child, setting him/her apart from his/her classmates. Treatment effects faded somewhat over time for all children suggesting that children likely need continued support in the school setting.