Assessment & Research

Subtypes of autism by cluster analysis.

Eaves et al. (1994) · Journal of autism and developmental disorders 1994
★ The Verdict

Four data-driven autism subtypes give you a quick map for picking goals that fit each learner.

✓ Read this if BCBAs who write assessment reports or treatment plans for autistic clients.
✗ Skip if RBTs who only run already-written protocols and never help with assessment.

01Research in Context

01

What this study did

The team fed many kinds of data into a computer. Test scores, language samples, and behavior notes all went in.

The computer grouped children with autism into four clusters. Each cluster had its own pattern of strengths and needs.

02

What they found

Four clear autism subtypes showed up. One group might talk well but struggle with play. Another might have strong self-care skills yet poor eye contact.

The clusters were not based on guesswork. They came straight from the data.

03

How this fits with other research

Pickering et al. (1985) tried a smaller version of this idea. They also found cognitive subtypes, but their sample mixed many disabilities. C et al. kept the focus only on autism.

Matson et al. (2013) zoomed in on just repetitive behaviors. They found two tight sub-groups inside that one area. C et al. looked at the whole child, not just one behavior.

McCarron et al. (2002) later showed the Autism Behavior Checklist does not split into the five parts it claims. C et al. warn us: even broad subtypes need fresh data checks.

04

Why it matters

Stop using "one-size-fits-all" goals. Look at the four subtype patterns and pick the one that best matches your learner. Then write targets that fit that profile. Your session plan becomes sharper and progress often speeds up.

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

Open your last five assessments. See which of the four subtype patterns each child matches best. Adjust one goal to fit that pattern.

02At a glance

Intervention
not applicable
Design
other
Sample size
166
Population
autism spectrum disorder
Finding
not reported

03Original abstract

Multidisciplinary data from 166 children with autistic spectrum disorders were subjected to cluster analysis. Cross-validation between random halves of the sample showed acceptable consistency of the clustering method. Four clinically meaningful subtypes emerged from the analysis. They did not differ in demographic characteristics but did show, on average, distinct differences in behavioral and cognitive areas. Over half of the sample fell into a subtype described as typically autistic with abnormal verbal and nonverbal communication, aloofness, impaired social skills, and sensory disturbances. Another 19% were similarly autistic but with moderate to severe mental handicap. The remaining children formed two subtypes: a high-functioning Asperger-like group who were overactive and aggressive, and a small group who were impaired in social and language skills, had restricted interests, and a family history of learning problems. This study highlights important differences among children with autism and emphasizes relationships between cognitive functioning and subtypes of the disorder.

Journal of autism and developmental disorders, 1994 · doi:10.1007/BF02172209