We employed multiple regression analyses to test additive and interactive associations of genetic factors and implicit drinking motives with drinking behavior. Main effects for genetic factor and IAT score were entered, as was the interaction between the two. All IAT scores were grand-mean centered, and interaction terms were created by multiplying participants’ grand-mean centered IAT scores by genotype score. Ethnicity and gender were entered as covariates. All predictors were entered in a single step. Significant main effects or interaction terms are denoted by a significant t-score, indicating that the variable predicted unique variance in a given drinking outcome. Analyses were conducted for each combination of IAT, genetic and drinking variable. We did not implement strict alpha correction for multiple testing given power considerations, including the small sample size, the small effect sizes typical for genetic associations with complex behaviors, and the relatively small number of statistical tests. Additionally, drinking variables were highly correlated, which reduces concerns that multiple statistical tests reflect independent hypothesis tests (e.g., Sankoh et al., 1997).
