STATISTICAL APPROACHES TO GENETIC EVALUATION FOR THRESHOLD BINARY TRAITS

This article describes 3 statistical methods of inference about location and dispersion parameters of the threshold model applied to binary traits under a mixed model structure of variation. These methods are : 1), the linear approach of Grizzle, Starmer and Koch (1969) and its Bayesian extension to a mixed model; 2), the quasi-likelihood approach to the generalized linear model as proposed by Gilmour et al (1985-1987), and 3), the Bayesian method (joint mode a posteriori-MAP) of Gianola and Foulley (1983). Different aspects of comparison among these procedures are discussed.