We first conducted a confirmatory factor analysis in Mplus (Muthén & Muthén, 2004) to determine whether the phenotypic structure of the internalizing spectrum disorders could be accounted for by one latent factor. Raw data, in the form of symptom counts for depression, generalized anxiety, and panic attacks, and the scale score for neuroticism, were fit to a one-factor model using a weighted least squares estimator with a mean- and variance-adjusted chi-square test statistic (WLSMV) to account for the non-normality of the data. No participants were missing data on the symptom count variables. For neuroticism, 36 participants were missing data; these missing values were handled by substituting the mean of the variable in SPSS before conducting the factor analysis in Mplus. The metric of the factor was defined by setting the loading for depression symptoms to 1.0. The factor loadings for the remaining indicators, the error variances, and the factor variance were all freely estimated, and all error covariances were fixed to 0.
