Using data collected at the age 17y9mo clinic assessment (referred to as age 18 hereafter), we derived an alcohol problems factor score, as previously described [30]. Briefly, participants were administered (via computer) 10 items from the Alcohol Use Disorders Identification Test [31] along with 7 items aimed at assessing DSM-IV [32] symptoms of alcohol dependence. Three additional measures (getting into fights, police involvement, and drinking to alleviate withdrawal symptoms) were also included. To improve sample size, we used IVEware [33] to impute age 18 alcohol problems data for participants who completed the AUDIT at age 16y6mo, but not the age 18 assessment (N=1993). See Supplementary Material for additional information. Frequency and correlation checks after imputation showed that all imputations kept similar frequency distributions and that imputed and original variables were closely correlated. Factor scores were created for 5952 participants after imputation, using Mplus 6.11 [34]. Of these, genetic data were available for 4304 individuals after quality control screens were applied (see below).
