advanced methodologies such as principal component analysis of measured cofounders as a method to reduce collider bias (Thomas et al., 2022). Finally, we note that the effect sizes for the significant genetic main and G × E effects are quite small. While these small effect sizes appear to run counter to the heritability estimates from quantitative genetic studies, they are consistent with findings from other studies using a polygenic score approach. These likely reflect methodological differences between the polygenic score approach and quantitative genetic approaches (which estimate the overall latent genetic influences on a phenotype). These findings also indicate limitations of the current polygenic score approach in predicting alcohol use outcomes, and highlight the need for improving predictability of polygenic scores with methodology advancements (e.g., larger discovery GWAS particularly among AAs, functional polygenic scores).
