Informant agreement for youth with autism spectrum disorder or intellectual disability: a meta-analysis.

Boudreau et al. (2015) pooled 49 studies that compared parent, teacher, and self-ratings of kids with autism or intellectual disability. They asked one simple question: how often do the adults and the child agree on the child’s behavior?

The team weighted every study by sample size and computed a mean correlation. They also split results by problem type and by who was doing the rating.

The average agreement was only moderate, r = .36. That means if a parent says a behavior is a big problem, a teacher will say the same only about one-third of the time.

Agreement was strongest for externalizing problems like hitting or yelling. It was also higher when two similar adults rated the child (parent-teacher) than when a child rated himself.

Bitsika et al. (2015) zoomed in on anxiety in boys with ASD. They also saw parent-child gaps, and the gap got wider as kids got older. Their finding extends the meta-analysis by showing the disagreement is not random—it grows with age.

Greene et al. (2019) asked only teachers about anxiety. Nearly half of students with ASD were rated as clinically anxious. This classroom picture fills in the teacher slice of the multi-informant puzzle flagged by A et al.

Buse et al. (2014) used parent and teacher data to predict bullying risk, but they did not report agreement size. Their study is topically related yet shows that even when agreement numbers are missing, multi-source data still matter for risk screening.

Low agreement means one viewpoint is not enough. Before you write a behavior plan, collect at least two reports—parent plus teacher or self plus adult—and weight externalizing items more heavily. If you see a big gap, probe context: maybe the behavior only happens at recess, or only at home. Use the .36 benchmark to set realistic expectations for team meetings and avoid over-relying on a single informant.