Simultaneous Estimation of Variance and Covariance Components from Multitrait Mixed Model Equations

Abstract

A procedure is described for estimating variance and covariance components when different variables are observed on different experimental units. The procedure allows for different linear models in the different traits. The method consists of maximizing the likelihood of a set of error contrasts. It is an extension of a procedure presented by Thompson (1973) for the case when all variables are measured on all traits. The method is also an iterative version of Rao''s (1971) minimum norm quadratic unbiased estimation procedure (MINQUE). The calculations are described in terms of solutions to Henderson''s (1973) mixed model equations. The procedure was developed to estimate sire components of variance and covariance for yearling weights of male and female progeny in beef cattle, and a sample of such data is used as an example.