Genotype-Environment Covariance for Multiple Phenotypes: A Multivariate Test Using Adopted and Nonadopted Children

A bivariate generalization of the genotype-environment (GE) covariation is presented. This biometrical parameter measures the relation between genotypic influences in one attribute with environmental influences in another attribute. A multivariate procedure for detecting the presence of both univariate and bivariate GE covariation is also described, based on a test of homogeneity of observed variance-covariance matrices in adopted and nonadopted individuals. An illustrative application to measures of specific cognitive abilities in four-year-old children in the Colorado Adoption Project (CAP) is provided. The analyses provided evidence of bivariate but not univariate GE covariance among perceptual speed, memory, verbal and spatial abilities in these children. A comparison of the results from the presently described and other more elaborate model-fitting procedures is made, with a discussion of the merits and disadvantages of each approach.