Cluster analysis of autism spectrum disorder symptomatology: Qualitatively distinct subtypes or quantitative degrees of severity of a single disorder?

Bitsika et al. (2018) fed lots of autism symptom scores into a computer. The computer looked for natural groupings in school-age kids with ASD.

They wanted to know: do kids fall into separate 'types' of autism, or just mild versus severe?

The numbers lined up in two big blobs: high-severity and low-severity. No special blob for, say, 'chatty but odd' or 'quiet plus behavior storms'.

Challenging behavior showed up in both blobs, so it did not help split the groups.

Rivard et al. (2026) also used cluster analysis but saw three clear sub-types in preschoolers, not a simple mild-to-severe line. The kids were younger, so age may change how clusters look.

Erickson et al. (2016) had guessed that quick social-timing scores could form tidy clusters. Vicki’s real data backs a simpler severity split instead.

Older papers like Weiss et al. (2001) already doubted neat autism ‘sub-species.’ Vicki gives fresh numbers supporting that old idea.

Stop hunting for fancy sub-labels. Just rate overall symptom strength and pick goals from there. A short severity cut (high/low) is enough to guide intensity of instruction, staffing ratio, and prompt levels for most learners on your caseload.