comparison to evaluate if the groups with AUD and/or SUD diagnoses differed from Group 4 with no diagnosis in order to identify relevant items to test regarding association with Groups 1, 2, and 3 separately. Those significant items were then entered into a series of backward elimination logistic regressions to evaluate if LR, externalizing, and/or internalizing items were still significantly related to each of the three groups with diagnoses when considered in the context of other significant items. Comparisons regarding Hypothesis 5 (relating to the severity of the clinical course of AUDs and/or SUDs) were limited to Groups 1–3 who had relevant diagnoses, where an overall significant difference across groups was followed by planned comparisons of how Groups 2 and 3 differed from the combined diagnosis Group 1. For all evaluations, missing data were handled through a maximum likelihood procedure (Collins et al., 2001), including 6.8% who needed correction for one item and 0.8% for two items.
