A chi-square test to distinguish allelic association from other causes of phenotypic association between two loci

In unrelated individuals, their phenotypes at two gene loci may appear correlated. The two major reasons for such a deviation from independence are linkage disequilibrium (allelic association) and an interaction between the phenotypes. In this paper, it is shown how the usual chi-square test for association of phenotypes can be partitioned into two components, one being due to allelic association and the other being due to other causes of phenotypic association. The test is demonstrated with an example.