We used the PROCESS macro v2.15 (Hayes, 2013) in SPSS v22.0 (IBM Corp., 2013) to test the conceptual conditional process model found in Fig. 1. Specifically, we tested whether the mediated effect of X (5-HTTLPR genotype; LL vs. LS/SS) on Y (alcohol outcomes) through M (level of response to alcohol [SRE-5]) is contingent on V (parental monitoring), controlling for drinks before SRE-5, sex, age at SRE-5 assessment, and BMI. This model (PROCESS model 14) was run separately for each of the four alcohol outcomes. The two variables involved in the interaction, SRE-5 and parental monitoring, were mean-centered prior to entry into the model in order to avoid problems with multicollinearity (Aiken & West, 1991). Bootstrapping with 10,000 resamples was performed to determine bias-corrected (asymmetric) 95% confidence intervals for the conditional indirect effects (Preacher & Hayes, 2004), which are the focus of these analyses. Bootstrapping is based on random sampling of the data and does not make assumptions about the shape of the sampling distribution. Bootstrapping also allows for the construction of more accurate confidence intervals than those derived from normal
