Racial-Ethnic Comparisons of ADOS-2 Algorithms for Young Verbal Children

Autism Res. 2026 Apr;19(4):e70203. doi: 10.1002/aur.70203. Epub 2026 Mar 10.

Abstract

The current study reports comparative analyses of the ADOS-2 Module 1 (Some Words) (n = 918) and Module 2 (Phrase Speech) (n = 881) algorithmic items between Black, Hispanic, and White children aged 3-5 on data from the Study to Explore Early Development (SEED). Significant Differential Item Functioning (DIF) was identified on ADOS-2 Social Affect and Restrictive Repetitive Behavior (RRB) items from both modules, but differential test functioning (DTF) was close to zero for each subscale (range = -0.07 to 0.08). No discernible patterns were identified when comparing these results with other published studies conducted with older populations. Item level scoring differences may reflect unique study sample variance, and existing data suggests DIF is unlikely to impact scale level ADOS-2 interpretations for clinicians assessing preschool age children from different racial and ethnic backgrounds.

Keywords: autism assessment; differential item functioning; early identification; item response theorymeasurement.

Plain language summary

This study used advanced statistical analyses to determine if ADOS‐2 items and scales differed between White, Black, and Hispanic children in a manner that might impact clinical interpretation. Analyses indicate that some items do differ between groups, but these differences are minimal at the scale level clinicians use. Findings suggest that ADOS‐2 subscale scores are minimally impacted when used with Black, Hispanic, and White children.

Publication types

  • Comparative Study

MeSH terms

  • Algorithms*
  • Black or African American / psychology
  • Black or African American / statistics & numerical data
  • Child, Preschool
  • Ethnicity*
  • Female
  • Hispanic or Latino* / psychology
  • Hispanic or Latino* / statistics & numerical data
  • Humans
  • Male
  • Reproducibility of Results
  • White People* / psychology
  • White People* / statistics & numerical data