Minimal Race, Sex Bias Identified in the ADOS-2 Autism Measure

autism-concept
Differential item functioning (DIF) is minimal across race and sex in the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2).

HealthDay News Differential item functioning (DIF) is minimal across race and sex in the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2), according to a study published online April 26 in JAMA Network Open.

Luther G. Kalb, PhD, from the Kennedy Krieger Institute in Baltimore, and colleagues conducted a cross-sectional study of children who were evaluated for autism spectrum disorder between 2014 and 2020 to examine DIF of ADOS-2 across child race and sex. Item-level biases across ADOS-2 harmonized algorithm items were examined across three modules. Data were included for 6269 children (25.9% Black/African American, 50.3% White, and 79.4% male).

The researchers found that 16 of 140 (11%) ADOS-2 diagnostic items had a significant DIF. Eight items had a significant DIF for race, including 6 that involved social affect. No single item demonstrated DIF consistently across all 3 modules examined. Most of the items with DIF had greater difficulty and poorer discrimination in Black/African American versus White children. Five items showed significant DIF for sex; DIF was split between social affect and repetitive/restricted behaviors. Across all 5 algorithms, hand mannerisms had evidence of DIF, with generally greater difficulty. Two items had moderate-to-large magnitude of DIF: hand mannerisms for female children and repetitive interests among Black/African American children.

“The findings of this cross-sectional study suggest minimal overall item-level bias of the ADOS-2,” the authors write.

Abstract/Full Text

Editorial