status. In these models, factor means were set to zero for women without DUIs and were freely estimated in the other groups. As there was evidence for unequal variances (X2(4) = 128.54, p < .0001), variances were freely estimated across groups and the discrimination and difficulty parameters for role failure, which showed no evidence of differential criterion functioning (DCF), were constrained across groups to identify the model. The baseline model against which subsequent models were compared constrained the discrimination and difficulty parameters for all criteria across groups. For each criterion individually (except role failure), a model was fit which freed the discrimination and difficulty parameters for individuals with DUIs (male and female DUI groups were equated) and a likelihood ratio test on 2 degrees of freedom was computed using the baseline model for comparison. If this test yielded a statistically significant improvement in fit (based on a p-value of ≤ .01, and calculated using equations available at statmodel.com), 2 additional tests were conducted to assess DCF by gender for individuals with and without DUIs, using the previous model for comparison. If no DCF by DUI status was found, then a single test of DCF by gender was performed, using the
