We used multiple-group CFA (MGCFA) to assess measurement equivalence across African American and European American families. Specifically, we compared the fit between models where loadings in the two samples were constrained to be equal to models where they were allowed to differ (unconstrained) using the likelihood ratio test (a χ2 difference test) and by examining the change in the comparative fit index (Δ CFI). Chi-square difference tests yielding nonsignificant p-values (p ≥ .05) were interpreted as support of measurement invariance (Bontempo, Hofer, & Lawrence, 2006). Since Δ χ2 is more likely to yield erroneous support for invariance in small samples, differences in the Δ CFI, which is not influenced by sample size, were also used. Differences less than or equal to −.01 were interpreted as evidence of noninvariance (Cheung & Rensvold, 2002). Results of these tests are provided in Table 3.
