Measurement invariance in the Youth Electronic Cigarette Outcome Expectancies Questionnaire across sex and race/ethnicity in pre- to early adolescence.

Abstract

Introduction: This study examined whether the Youth E-cigarette Outcome Expectancies Questionnaire (YEOEQ) measures e-cigarette beliefs consistently with respect to sex, race/ethnicity, and intersectional identity among Black, Latino, and White youth. Inconsistent measurement, or non-invariance, can lead to misleading conclusions about which groups are at higher risk of e-cigarette use. By identifying and adjusting for this bias, we aimed to improve the accuracy of group comparisons.

Methods: Data were drawn from follow-up 3 of the Adolescent Brain Cognitive Development Study (n = 8976; Mage = 12.91, 47.27% female; 15.02% Black, 22.94% Latino, 62.03% White). Moderated nonlinear factor analysis was applied to the Positive and Negative Expectancies YEOEQ subscales. Mean subscale scores were compared across groups pre- and post-adjustment for non-invariance.

Results: Non-invariance by sex, race/ethnicity, and intersectional identity was observed at the item level for both subscales, and at the factor level for Positive Expectancies. For Positive Expectancies, adjusted comparisons revealed previously undetected lower scores among Black vs Latino (gadjusted = 0.15) and White youth (gadjusted = 0.16) and reduction to non-significance of previously observed higher scores for Latino vs White youth. For Negative Expectancies, observed lower scores for Black vs Latino youth (gunadjusted = 0.15) and Latino vs White youth (gunadjusted = 0.15) remained significant but effect sizes decreased when adjusted.

Conclusions: Findings demonstrate measurement bias in the YEOEQ by sex and race/ethnicity in early adolescents. Unadjusted scores suggested riskier (more positive) e-cigarette expectancies among Latino than White youth and missed lower risk (less positive) among Black youth, potentially misguiding attention to these marginalized groups.

Implications: This is the first known study to assess measurement invariance jointly across sex and race/ethnicity in the Youth Electronic Cigarette Outcome Expectancies Questionnaire and the only known to produce scores adjusted for measurement non-invariance. Results revealed non-invariance across sex and race/ethnicity, implying that valid group comparisons may be compromised without adjusting for this measurement error and that non-adjusted scores may improperly identify group differences that are artifacts of measurement error. The resulting improper assignment of relative risk level (eg, higher positive expectancies in certain groups) could perpetuate harmful stereotypes or exacerbate health disparities with respect to intervention referrals.