Alcohol use was measured as the number of drinks consumed in a typical week in the past 6 months (Bucholz et al., 1994). Participants reported the number of drinks of different kinds of alcoholic beverages (beer, wine, liquor, other) they consumed on a typical day of the week (Monday, Tuesday, … Sunday) over the past 6 months. Scores were summed across days and kinds of drinks to derive a composite score that indicated standard drinks per week. A total of 966 (24.1%) EA participants and 340 (26.7%) AA participants indicated that they consumed no alcohol (zero drinks) during the past 6 months and they were coded as zero. Extreme values (> 140 drinks per week; n = 24 (0.6%) for EAs and n = 12 (0.9%) for AAs) were winsorized (Keselman et al., 2008). Preliminary analysis indicated that the alcohol use variable was positively skewed (skewness = 4.19 and 3.42 for EAs and AAs, respectively), and thus was log transformed for subsequent regression analyses. The skewness for the log-transformed alcohol use variable was .65 and .69 for EAs and AAs, respectively.
