The aims of the current study were to identify and adjust for measurement non-equivalence (i.e., bias) with respect to sex (as a proxy for gender), race/ethnicity, and intersectionality (sex by race/ethnicity) in the Perceived Discrimination Scale (PDS) for middle-school-aged Black, Latinx, and White youth. Data were drawn from Follow-up 2 of the Adolescent Brain Cognitive Development Study (n = 9,360; mean age = 12.03, SD = 0.10; 47.53% female; 15.90% Black, 22.42% Latinx, and 61.68% White). Moderated nonlinear factor analysis was applied to the 7-item scale. Measurement non-equivalence was observed exclusively with respect to sex for two items: unfair or negative treatment because of their ethnic background (1) by teachers and (2) by other students. Measurement equivalence with respect to race/ethnicity was observed at both the factor and item levels. Adjusting PDS scores for the modest degree of measurement non-equivalence shifted the effect size from the small to medium range for differences between Latinx and White youth (0.40 to 0.52) and from the medium to large range for differences between Black and White youth (0.770 to 0.893) - a meaningful change in how these group differences would be characterized. Findings indicate that the PDS captures racial discrimination similarly across Black, Latinx, and White middle school-aged youth. Importantly, they also demonstrate the impact of modest measurement bias on the validity of comparisons across racial/ethnic groups, underscoring the importance of assessing and adjusting for measurement non-equivalence as a standard practice.
Sartor et al. (Mon,) studied this question.