Research Review: Conceptualizing and measuring ‘problem behavior’ in early intervention autism research – a project AIM secondary systematic review
Most autism intervention studies target behaviors for reduction without defining them or explaining why, and rely on parent-report tools with limited validation.
01Research in Context
What this study did
The team read 102 early-intervention studies that tried to reduce behaviors in autistic kids.
They asked two simple questions: Did the paper say what the behavior was? Did it say why that behavior needed to go?
They also checked if the study used a real measure or just a parent form.
What they found
Most papers never defined the behavior they targeted.
Many called harmless autistic traits, like hand-flapping, problem behavior.
Hardly any used a checklist that had been tested for accuracy.
How this fits with other research
Bottema-Beutel et al. (2024) looked at teens and found the same holes: no clear definitions, no functional check, no side-effect watch.
The new review simply widens the lens from teens to toddlers.
Wynne et al. (1988) warned us decades ago that aggression studies lacked rigor.
Today’s autism trials repeat those same flaws, so the field has not moved forward.
Thom et al. (2026) show that labels keep changing, but sloppy measurement stays the same.
Why it matters
Before you write a behavior plan, state the exact topography and social impact you want to change. Pick a tool that has data backing it, not just a parent survey. If the literature you cite can’t meet this bar, treat it as weak evidence and probe for function yourself.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Write one clear, observable definition and a brief social-validity reason before you paste any behavior into the plan.
02At a glance
03Original abstract
Some autistic children exhibit behavior that caregivers, clinicians, and researchers consider problematic. However, there is little consensus about the types of behaviors that should be treated as a problem and reduced via intervention. In autism intervention research, problem behaviors range from inherently harmful behaviors such as aggression and self‐injury to nonnormative but not harmful behaviors associated with autism such as repetitive movements. Likewise, there are a variety of conceptualizations and measurement practices used to assess these behaviors. In this secondary systematic review of group‐design, nonpharmacological intervention studies for autistic children up to age eight, we explore researchers' conceptualizations of problem behavior and measurement systems to assess problem behavior. We defined problem behavior as any outcome where behaviors were targeted for reduction or elimination. A coding scheme was applied to 102 studies that met inclusion criteria for the secondary review. All studies were double coded by two independent coders. Sixty‐two percent of studies described reducing behavior as a primary or secondary purpose of the study and/or intervention, 33% gave a rationale for targeting behaviors for reduction, and 28% offered a conceptualization of the behavior(s) they targeted. Only 8% offered a conceptual definition. The most common measures were ‘off‐the‐shelf’ measures that had undergone at least some previous validation beyond interrater reliability and that involved parent reports. For the 10 most common assessment measures, two were validated along six different validation dimensions in autistic populations. All but one full scale or subscale measured behaviors that were nonnormative but not inherently harmful, or a mix of behaviors that were inherently harmful and that were nonnormative but not inherently harmful. Intervention researchers should provide clear definitions and rationales for targeting behaviors for reduction via intervention and should develop refined measurement tools for assessing these behaviors in collaboration with the autistic community.
Journal of Child Psychology and Psychiatry, and Allied Disciplines, 2025 · doi:10.1111/jcpp.14177