The structure of autism symptoms as measured by the autism diagnostic observation schedule.

Norris et al. (2012) looked at how autism symptoms group together on the ADOS. They tested Modules 1 and 3, which cover non-verbal kids and fluent speakers.

The team ran many factor models on the checklist scores. They wanted one clear map of social, communication, and repetitive traits.

No single model won. Every factor structure fit the data almost equally well.

The best pattern changed with age and IQ. Younger or lower-functioning kids gave different clusters than older or higher-functioning kids.

Albert et al. (2024) updates this picture. Using the newer ADOS-2, they show a bifactor model fits best: one broad autism trait plus two small factors for social affect and repetitive behaviors. Their clearer result supersedes the 2012 muddle.

Sipes et al. (2014) extends the same question to toddlers. With the BISCUIT screener, a three-factor model fit best, hinting that age and tool both shape the answer.

Greer et al. (2014) found a similar social-versus-repetitive split in kids with ADHD using the SCQ. The split holds even outside pure autism samples.

If you give the ADOS-2, use the bifactor structure from Albert et al. (2024) to guide your report: note a general autism score plus separate social and RRB scores. When you read older files that used Modules 1 or 3, remember the factor picture was fuzzy—look at age and IQ before comparing scores. For toddlers, keep the three BISCUIT factors in mind. Clear factors help you pick targets and show parents where their child stands.