Chunk #53 — Reasons to be Concerned about the Published cGxE Literature — Problems with the Recipe: Statistical Concerns in cGxE Research — cGxE versus gene-environment correlation (rGE)
be important; however, the presence of rGE can complicates the interpretation of cGxE. Behavioral genetic and twin models implement several statistical approaches to account for and even explicitly model rGE (e.g., L. Eaves & Erkanil, 2003; Purcell, 2002; van der Sluis, Posthuma, & Dolan, 2012). In measured gene studies, the first step in testing for potential rGE involves estimation of a correlation between genotype and environment. In the absence of such a correlation, as has been noted for 5-HTTLPR and stressful life events, cGxE testing can proceed without concern. If the correlation is solely attributable to outliers in the environmental measure, removal or winsorization may eliminate rGE (e.g., Bogdan, Williamson, & Hariri, 2012). A modest correlation between genotype and environment may require more careful consideration. For instance, in revisiting the interaction between a polymorphism in the monoamine oxidase A (MAOA) gene and exposure to childhood physical abuse in the development of antisocial behaviors (Caspi et al., 2002), Kim-Cohen and colleagues examined whether MAOA genotype was correlated with not only exposure to abuse (i.e. evocative rGE) but also, maternal antisocial behavior (i.e. passive rGE) with the latter being a key correlate of transmission of risk for antisociality and for increased likelihood
