Establishing measurement equivalence across sex, race/ethnicity, and intersectional identity for the Alcohol Expectancy Questionnaire-Adolescent, Brief: findings from the ABCD Study.

Abstract

Aims: The current study aimed to identify possible measurement non-equivalence (i.e., bias) with respect to sex (proxy for gender), race/ethnicity, and intersectional identity (sex by race/ethnicity) and generate adjusted scores for the Alcohol Expectancy Questionnaire-Adolescent, Brief (AEQ-AB) in a sample of middle-school-aged Black, Latinx, and White youth.

Methods: Data were drawn from Follow-up 3 of the Adolescent Brain Cognitive Development Study, conducted from 2020-2022 (n = 8981; mean age = 12.91, SD = .65; 47.29% female; 15.02% Black, 22.95% Latinx, and 62.03% White). Moderated nonlinear factor analysis was applied to the 4-item Positive Expectancies and 3-item Negative Expectancies subscales.

Results: Analyses revealed non-equivalence with respect to both sex and race/ethnicity on the Positive Expectancies subscale and race/ethnicity on the Negative Expectancies subscale. AEQ-AB mean subscale scores were compared across groups pre- and post-adjustment for measurement non-equivalence. Adjusted scores revealed previously undetected higher positive expectancies in girls versus boys, a lower magnitude of differences in positive expectancies between racial/ethnic groups, previously observed higher negative expectancies among girls versus boys as non-significant, and even higher relative negative expectancies for Black and Latinx than White youth.

Conclusions: Collectively, our findings highlight the importance of establishing measurement equivalence broadly, and specifically for the AEQ-AB, and demonstrate the potential for measurement bias to underestimate or even mask the relative risk in girls and youth of color, which can impact assumptions about and attention to these marginalized groups.