G × E interactions can be tested or modeled in behavioral genetic studies using several different study designs (e.g., twin or adoption). The two most frequently used methods testing for G × E interactions in twin and adoption studies include: (1) a mean levels approach, testing whether mean values of a phenotype differ across different combinations of genetic risk and environmental settings and (2) a moderated variance components approach, examining whether genetic and environmental variance for a trait varies across different measured environmental settings. These two different methods stem from the same conceptual idea, namely, that genetic effects vary across environments or vice versa. Their interpretations and meanings can be rather different, since one is based on means and the other is based on variances. The mean levels G × E is perhaps a more traditional approach, is typically presented as a statistical interaction in an ANOVA, and indicates whether particular experiences, exposures, or other conditions may (on average) ameliorate or exacerbate specific genetic effects in groups of individuals with similar genetic and environmental risks. In comparison, the moderated effects
