Finally, we conducted two sets of supplemental sensitivity analyses to further evaluate our findings. First, although the GSCAN GWAS sample is ideal for creating alc-PRS for our EA subsample given that it is the largest GWAS of alcohol phenotype (drinks per week) which also matched the alcohol phenotype of the present study, we recognize that the GSCAN sample is much larger than the MVP sample that we used to creating alc-PRS for AAs, which may be contributing to our finding of significant effects of alc-PRS in EAs but not AAs. Thus, we created alc-PRS for EAs using GWAS estimates from the MVP EA sample and ran analyses using this alc-PRS for EAs. Second, we evaluated developmental differences in G×E effects across emerging, young, and middle adulthood by creating three age bands that represent theoretically defined developmental periods (ages 18–29, 30–44, and 45–65). However, we recognize that these age bands are relatively large, and that there are vast differences within each age band concerning alcohol use. For example, alcohol use at age 18 is riskier and more problematic than at age
