A Meta-Analysis of School-Based Social Skills Interventions for Children With Autism Spectrum Disorders

NICHCY’s Structured Abstract 75 describes the following:

Title | A Meta-Analysis of School-Based Social Skills Interventions for Children With Autism Spectrum Disorders

Authors | Bellini, S., Peters, J., Benner, L., & Hopf, A.

Source Journal of Remedial and Special Education, 28(3), 153-162.

Year Published | 2007

Social skills deficits are a central feature of autism spectrum disorders (ASD). This meta-analysis of 55 single-subject design studies examined the effectiveness of school-based social skills interventions for children and adolescents with ASD. Intervention, maintenance, and generalization effects were measured by computing the percentage of non-overlapping data points. The results suggest that social skills interventions have been minimally effective for children with ASD. Specific participant, setting, and procedural features that lead to the most effective intervention outcomes are highlighted, and implications for school personnel are discussed. Finally, the results are compared to the outcomes of similar meta-analyses involving social skills interventions with other populations of children.

Students with autism spectrum disorders (ASD) often struggle with a number of social skills deficits, including problems with initiating social interactions, interpreting body language and facial cues, taking another person’s perspective, maintaining reciprocity, inferring another person’s interests, and sharing enjoyment. These social skills deficits impact the ability of children with ASD to communicate, participate in new environments, and to establish and maintain relationships with others. Many students with ASD become socially isolated and withdraw because they are unable to establish and maintain meaningful social relationships. These students may experience anxiety, depression, poor academic achievement, and emotional disorders as a result of this social isolation. Since social skills are integral to successful cognitive, emotional, and social development, social skills interventions are an important part of educational programming for students with ASD.

Research Questions

  1. Are social skills interventions effective for students with ASD?
  2. If so, what are the effects?
  3. Are the effects maintained over time?
  4. Can the effects be generalized from one setting to another?

Research Design

  • Number of Studies Included | 55
  • Number of Subjects | 157
  • Years Spanned | 1986-2005

Research Subjects
Children with ASD taking part in social skills interventions.

Age/Grade of Subjects
Preschool-Secondary. Of the 55 studies:

  • 21 focused on preschool age children
  • 23 involved elementary school children
  • 5 involved students in middle school and high school
  • 6 studies had participants from more than one age group.

Specified Disability
Autism Spectrum Disorders (ASD)

Social Skills Interventions

Duration of Intervention
Social skills interventions were provided from between 2.5 and 28 hours, with a mean of 18 hours and a median of 7.25 hours. The number of sessions for each intervention ranged from 8-73, with a mean of 38 and a median of 25.5.


  1. The social skills interventions explored by this meta-analyses demonstrated only limited effectiveness for children with autism.
  2. Students with autism have difficulty generalizing the social skills they learn from one situation to another.
  3. The maintenance effects of social skills instruction were moderately strong. In other words, children with autism remember and use what they learn during social skills instruction relatively well.
  4. Social skills interventions were most effective for middel school and high school-age students. Students in this age group were also better at maintaining and generalizing what they learned in social skills training.
  5. Students in elementary school children with Autism Spectrum Disorder showed the lowest intervention and generalization effects.
  6. The lowest maintenance effects were observed in preschool-age children.
  7. Social skills interventions delivered in the general classroom showed significantly stronger intervention, maintenance, and generalization effects than social skills interventions delivered in pull-out programs.

Combined Effects Size
Instead of a combined effect size, the percentage of non-overlapping data (PND) was calculated and the magnitude of PND scores were interpreted. Like effect size, PND scores are a measure of the effectiveness of an intervention. However, PND scores are preferred over effect sizes for synthesizing single-subject studies.

The PND scores for:

  • Intervention Effects = 70% (Effective)
  • Generalization Effects = 53% (Questionable Effectiveness)
  • Maintenanace Effects = 80% (Effective)

Guidelines for interpreting PND scores:

  • above 90 are considered very effective
  • scores from 70 to 89 are considered effective
  • from 50 to 69 low or questionable effectiveness
  • below 50 are considered ineffective

The authors of this study recommend a number of changes to the way social skills interventions for children with ASD are implemented and research on social skills is conducted, including:

  1. Increase the intensity and frequency of social skills intervention delivery. Other researchers have suggested that for social skills interventions to be effective they should include at least 30 hours of instruction over 10-12 weeks. Bellini et al. suggest that one of the reasons for the low level intervention effects found in this meta-analysis may be that none of the studies examined implemented social skills instruction for the recommended 30 or more hours.
  2. Social skills instruction should take place in the general education classroom. Students with ASD maintain and generalize the social skills instruction they receive in typical classroom settings significantly better than instruction they receive in a pull-out program.
  3. Match social skills taught to the skill deficits of children in the program. If the student lacks a particular skill like initiating conversation, then an intervention that focuses on skill development would be appropriate. However, the child has the social skills needed to initiate conversations but fails to do so on a regular basis then a strategy that allows the student to practice and improve the performance of their current skill would be ideal.
  4. Research on social skills interventions should include intervention fidelity data. Without intervention fidelity data it is almost impossible to determine whether an intervention failed because it was poorly implemented or because the intervention was ineffective.


Back to top