To understand the factor structure of DD and PU items, we first used exploratory factor analysis (EFA) on a random half of subjects and then conducted confirmatory factor analysis (CFA) on the remaining half of the sample. For all factor analyses, we used MPlus (Version 7) (30) using weighted least squares mean variance estimation. Parallel analysis (31) and scree plots were used to determine the number of factors, and standard EFA/CFA fit indices (e.g., root mean square error of approximation (RMSEA), comparable fit index (CFI), and Tucker-Lewis index (TLI)) to compare nested models (32, 33). EFA and CFA models indicated a single factor when using either the substance dependence diagnosis items or problem use items (see Supplemental Table S2 and Supplemental Figure S1 for detailed results). Based on the consensus between the EFA and CFA models, genetic analyses utilized factor scores (Mean (M) = 0, standard deviation (SD) = 1) extracted from CFA models using all individuals of European descent. Genetic analyses of DV utilized rank normalized scores (using the BLOM approximation method (34); M = 0, SD = 1).
