The Estimation of Heritability with Unbalanced Data: II. Data Available on More Than Two Generations

Abstract

The maximum likelihood (ML) estimation of heritability is considered when data are available from 1 sex and for more than 2 generations and the progeny are non-inbred half-sibs. The covariances between relatives in different generations are derived. ML estimating equations are given using results derived in a companion paper (Thompson). Some of the formulae derived depend on the inversion of matrices and a convenient tabular form for deriving these inverse matrices is given that takes advantage of their special structure. Estimation when data are available on both sexes is briefly discussed.