Decomposition of multivariate phenotypic means in multigroup genetic covarinace structure analysis

Uses D. Sorbom's (1974) method to study differences in latent means in multivariate twin data. By restricting the analysis to a comparison between groups, the results pertain only to the additive contributions of common genetic and environmental factors to the deviation of the group means from what can be considered a vector-valued grand mean. Assumptions made in the decomposition of the phenotypic means relate to the constitution of the phenotypic means and to the equality of the variances in each group. The latter assumption is testable in the standard analysis of covariance (ANCOVA) structure by means of equality constraints on the relevant parameters across the groups. The former is testable in the illustration involving simulated data.