Abstract: |
This article is concerned with the study of intraclass correlations in the mixed linear model. A brief account of the shortcomings of the existing methods (frequentist, likelihood and Bayesian) is followed by alternative Bayesian parametrizations involving intraclass correlations and variance ratios. Our prior specifications accommodate a priori dependencies as well as situations which involve little or no prior information. We give examples of interval estimation and hypothesis testing using data from an animal breeding study. Copyright © 2000 by Marcel Dekker, Inc. |