1 Contribution of Genes and Environment to the Correlation between Variables

The question of what causes variables to correlate is the usual entry point to multivariate genetic analysis. Students of genetics have long been familiar with the concept of pleiotropy, i.e., that one genetic factor can affect several different phenotypes. Obviously, we can imagine environmental advantages and insults that affect many traits in a similar way. Students of the psychology of individual differences, and especially of factor analysis, will be aware that Spearman introduced the concept of the ``general intelligence factor'' as a latent variable accounting for the pattern of correlations observed between multiple abilities. He also introduced an empirical test (the method of tetrad differences) of the consistency between his general factor theory and the empirical data on the correlations between abilities. Such factor models however, only operate at the descriptive phenotypic level. They aggregate into a single model genetic and environmental processes which might be quite separate and heterogeneous if only the genetic and environmental causes of inter-variable correlation could be analyzed separately. Cattell recognized this when he put forward the notion of ``fluid'' and ``crystallized'' intelligence. The former was dependent primarily on genetic processes and would tend to increase the correlation between measures that index genetic abilities. The latter was determined more by the content of the environment (an ``environmental mold'' trait) and would thus appear as loading more on traits that reflect the cultural environment. An analysis of multiple symptoms of anxiety and depression by Kendler et al. (1986) illustrates very nicely the point that the pattern of genetic and environmental effects on multiple measures may differ very markedly. They showed that twins' responses to a checklist of symptoms reflected a single underlying genetic dimension which influenced symptoms of both anxiety and depression. By contrast, the effects of the environment were organized along two dimensions (``group factors'') -- one affecting only symptoms of anxiety and the other symptoms of depression. More recently, this finding has been replicated with psychiatric diagnoses [,], which suggests that the liability to either disorder is due to a single common set of genes, while the specific expression of that liability as either anxiety or depression is a function of what kind of environmental event triggers the disorder in the vulnerable person. Such insights are impossible without methods that can analyze the correlations between multiple measures into their genetic and environmental components.