Assessment & Research

Social relationship difficulties in autism and reactive attachment disorder: Improving diagnostic validity through structured assessment.

Davidson et al. (2015) · Research in developmental disabilities 2015
★ The Verdict

Live observation, not parent checklists, keeps you from calling autism RAD.

✓ Read this if BCBAs doing differential diagnosis for autism in clinic or school teams.
✗ Skip if Practitioners who only treat known autism cases with no RAD question on the table.

01Research in Context

01

What this study did

Davidson et al. (2015) watched kids with autism and kids with reactive attachment disorder during a set play routine. They scored eye contact, comfort seeking, and response to strangers.

Parents also filled out a checklist about the child’s social habits at home.

02

What they found

The live scores told the two groups apart almost every time. Parent checklists often said an autistic child had RAD signs when the child really did not.

In short, watching beats asking when you need to rule out RAD.

03

How this fits with other research

Allen (2018) warned that sloppy tools keep mixing up RAD and conduct disorder. Claire’s team answered with a clear, step-by-step observation plan.

Salley et al. (2015) used the same kind of design to split ASD from ADHD with ADOS scores. Claire’s work widens the lens to RAD, showing the method works across look-alike labels.

Kalinyak et al. (2025) found self-report surveys can hide true social anxiety in autism. Claire saw the same risk with parent surveys, so both papers push you to add live coding.

04

Why it matters

If you only use parent forms, you may add an RAD label that blocks an autism diagnosis and wastes time on the wrong training. Next time you assess a socially awkward kid, run a short, structured play session and score it with Claire’s checklist. You will spot the difference faster and start the right intervention sooner.

Free CEUs

Want CEUs on This Topic?

The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.

Join Free →
→ Action — try this Monday

Add a five-minute structured play probe to your assessment and score eye contact, comfort seeking, and stranger response before you write RAD in the report.

02At a glance

Intervention
not applicable
Design
quasi experimental
Sample size
125
Population
autism spectrum disorder, mixed clinical
Finding
positive
Magnitude
medium

03Original abstract

BACKGROUND: Autism Spectrum Disorder (ASD) versus Reactive Attachment Disorder (RAD) is a common diagnostic challenge for clinicians due to overlapping difficulties with social relationships. RAD is associated with neglect or maltreatment whereas ASD is not: accurate differential diagnosis is therefore critical. Very little research has investigated the relationship between the two, and it is unknown if standardised measures are able to discriminate between ASD and RAD. The current study aimed to address these issues. METHODS: Fifty eight children with ASD, and no history of maltreatment, were group matched on age with 67 children with RAD. Group profiles on multi-informant measures of RAD were investigated and group differences explored. Discriminant function analysis determined assessment features that best discriminated between the two groups. RESULTS: Although, according to parent report, children with ASD presented with significantly fewer indiscriminate friendliness behaviours compared to the RAD group (p<0.001), 36 children with ASD appeared to meet core RAD criteria. However, structured observation clearly demonstrated that features were indicative of ASD and not RAD for all but 1 of these 36 children. CONCLUSIONS: Children with RAD and children with ASD may demonstrate similar social relationship difficulties but there appears to be a difference in the social quality of the interactions between the groups. In most cases it was possible to differentiate between children with ASD and children with RAD via structured observation. Nevertheless, for a small proportion of children with ASD, particularly those whose difficulties may be more subtle, our current standardised measures, including structured observation, may not be effective in differentiating RAD from ASD.

Research in developmental disabilities, 2015 · doi:10.1016/j.ridd.2015.01.007