Abstract
This paper deals with theory, methods and problems in the analysis of categorical data in animal breeding when a threshold model is postulated for an underlying normal distribution of phenotypes. Linear and nonlinear statistical models potentially usedufl for estimating parameters of the underlying and outward distributions are reviewed and extended. Methods for evaluating the genetic value of candidates for selection from single or multiple populations are discussed comparatively. Copyright © 1982. American Society of Animal Science . Copyright 1982 by American Society of Animal Science