Multivariate models of mixed assortment: phenotypic assortment and social homogamy for education and fluid ability.

Phenotypic assortment is assumed to be the principal mechanism of spouse similarity in most biometrical studies. Other assortment mechanisms, such as social homogamy, may be plausible. Two models are presented that consider phenotypic assortment and social homogamy simultaneously (i.e., mixed assortment), where selective associations between social background factors (Model I) versus selective associations between total environments (Model II) distinguish the models. A series of illustrative analyses was undertaken for education and fluid ability available on a sample of 116 Swedish twin pairs and their spouses. On the basis of several fit criteria Model I was preferred over Model II. Both social homogamy and phenotypic assortment may contribute to spouse similarity for educational attainment and fluid ability. Furthermore, spouse similarity for fluid ability may arise indirectly from social homogamy and phenotypic assortment for educational attainment. Power analyses indicated greater observed power for Model I than Model II. Additional power analyses indicated that considerably more twin-spouse sets would be needed for Model II than Model I, to resolve social homogamy and phenotypic assortment. Effects of misspecification of mechanisms of spouse similarity are also briefly discussed.