Systematic Review of Problem Behavior Interventions: Outcomes, Demographics, and Settings.

Stewart et al. (2018) read every school-based problem-behavior study they could find for students with autism. They kept only single-case experiments and ended up with 84 of them. They then coded where each study happened and what the students looked like.

Most of the 84 studies took place in separate, special-ed rooms instead of regular classrooms. The reports rarely said the child’s race or ethnicity. In short, the evidence we trust comes mostly from segregated spaces and invisible kids.

Campbell (2003) and Heyvaert et al. (2014) already showed that behavioral interventions work better when you do a functional analysis first. Stewart et al. (2018) pulled many of the same studies, so their pool lines up with those older meta-analyses.

Kasari et al. (2013) and Parsons et al. (2013) warned that autism research stays stuck in clinics. Stewart et al. (2018) proves those warnings are still true: most “school” studies still happen in pull-out rooms.

McHugh et al. (2023) extends the story by showing self-management also works for adults. The gap E et al. found in schools repeats across the lifespan—we just don’t test interventions where people actually live and work.

If you write a behavior plan for an autistic student, the research you rely on probably came from a separate room with kids whose race we don’t even know. That makes it hard to say the plan will work in a noisy, mixed third-grade class. Push your next study—or your next grant—to collect data in inclusive classrooms and to report who the child is, not just what the child does.