As described in Shi et al.,(Shi et al., 2009) GAIN controls were recruited from the general population using random digit dialing sampling and administered questionnaires to ascertain their psychiatric health histories, which included DSM-IV assessment for drug abuse and dependence. Exclusions were made for reporting any psychotic or bipolar disorder, leaving a control set representative of the general population. We used these data to define illicit drug abuse cases and controls, who according to the GAIN recruitment criteria had no co-morbid psychiatric disease. We defined illicit drug abuse cases as follows: (1) ever using at least one drug (heroin/opioids, cocaine, marijuana, amphetamines/stimulants, or other drugs) and (2) reporting at least one DSM-IV symptom of drug dependence. Controls were defined as those who reported never using any drug. Replication analyses included 755 AAs (259 cases and 496 controls) and 1,131 EAs (273 cases and 858 controls), who passed quality control, had complete DSM-IV and covariate data available, and met our case/control definitions: see Table 1. We also conducted a sensitivity analysis in which we increased the number of dependence symptoms from
