Analysis of Variance in the Service of Interactionism

Article: Vreeke (this issue) asserts: ‘If you hold that reality is interactive, but use methods which presuppose the separateness of genes and the environment, your are stuck with a problem of interpretation.’ A similar point was made recently by a colleague in behavior genetics who chastised me, a critic of heritability analysis [Wahlsten, 1990, 1994, 1999a, 1999b], for employing the analysis of variance (ANOVA) in a recent study [Crabbe, Wahlsten and Dudek, 1999]. In reply, I argue that (a) ANOVA does not presuppose the separateness of genes and environment; (b) devotées of interactionism often employ ANOVA to examine results of factorial experiments precisely because the method can reveal the presence of interactions; and (c) all users of ANOVA must struggle with problems of interpretation, but these problems are especially severe for those who apply correlational methods to study human populations.