|
|
||||||||
Department of Animal Science, Cornell University, Ithaca, NY 14853
ABSTRACT
Computation of variances of quadratic unbiased estimators of variance components is tedious and tells nothing about the form of the distribution of the estimators and nothing about the bias in or accuracy of estimation of the ratios of variance. In most applications to animal breeding, ratios of variances are wanted, and little is known about the merits of alternative methods for obtaining such estimates. Simulation is the only apparent method for examining properties of estimators of variances and of their ratios. This paper presents efficient methods for generating samples that can be used in examining properties of estimators that utilize least squares reductions, for example, Methods 1, 2, and 3.
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |