(PS1-54) Measurement Invariance of GAD-7 Across Four Ethnic Groups in a Nationally Representative Sample of US Adults

Abstract Text: The Generalized Anxiety Disorder-7 screener (GAD-7) is one of the most commonly used measures of anxiety in primary and specialty care settings, but it was originally developed in a primarily White sample. Less is known about its measurement invariance across diverse ethnic groups. In the current study, we examine the measurement invariance of the GAD-7 across a large nationally representative sample of White, Black, Hispanic, and Asian adults in the US. We utilized publicly available data from the 2022 administration of the National Health Interview Survey to examine the measurement invariance of the GAD-7 across White (n = 17,830), Black (n = 2,951), Asian (n = 1,618), and Hispanic (n = 3,833) adults in the US. We retained only participants who had complete GAD-7 data. The sample was 55% female, with no significant sex makeup differences between samples (Cramer’s v = .02). We conducted measurement invariance by first conducting confirmatory factor analyses (CFAs) of the one-factor model of GAD-7 separately for each group in Mplus version 8.11. We then conducted measurement invariance analyses by assessing configural invariance (equality of factor structure), then metric invariance (equality of factor loadings), then scalar invariance (equality of intercepts), and finally residual invariance (equality of residuals). Conducting follow-up tests were contingent upon a prior successful test (defined as a change in CFI < .01). Achieving configural invariance was necessary to conduct the test of metric invariance, metric invariance was required to conduct the test of scalar invariance, and scalar invariance was necessary to conduct the rest of residual invariance. We utilized White persons as the reference group, and we used robust maximum likelihood estimation. All groups achieved acceptable fit in their separate CFAs (all CFIs > .95, all RMSEAs < .08, all SRMRs < .04). The configural model demonstrated excellent fit (chi-square = 1100.40, p < .001, CFI = .962, TLI = .943, RMSEA = .053 [.051, .056], SRMR = .031). The groups achieved metric invariance (change in CFI from configural model = .001), scalar invariance (change in CFI from metric model = .009), and residual invariance (change in CFI from scalar model = .003). Findings indicate full measurement invariance of GAD-7 across White, Black, Asian, and Hispanic adults in the US. This supports the use of GAD-7 across these groups and the assumption that similar scores between these groups represent similar latent levels of anxiety. While the current data provide strong evidence of measurement invariance for these groups, future researchers should consider assessing for invariance across specific subgroups (e.g., Mexican, Honduran, Guatemalan for Hispanics), as well as incorporating other ethnic groups (e.g., Pacific Islanders, Native Americans).