G × E interactions in studies of human characteristics may be due to a number of factors. One likely explanation is related to statistical power. In general, it is difficult to detect G × E effects due to their low statistical power (Rowe, 2003). For example, behavioral genetic studies rely on genetic relatedness for groups of individuals, rather than on sharing of specific alleles between pairs of relatives. Studies relying on variance partitioning often do not find significant G × E effects, or find that they explain a very small portion of the total variance, and are thus dropped from further analysis. When G × E is not taken into account in behavioral genetic studies, heritability estimates will tend to be biased, although the direction of the bias depends on whether the moderating environmental influences are of the shared or nonshared variety.
