BAYESIAN COMPARISON OF MEANS OF A MIXED MODEL WITH APPLICATION TO REGRESSION ANALYSIS

摘要

In the analysis of variance, the mathematical models employed are usually classified into three main categories: fixed-effect, random-effect and mixed models. From the 'traditional' sampling theory point of view, the analysis of fixed-effect models leads of course to such familiar pro¬cedures as the t-test for individual comparison and F-test for overall comparison.nIn the present paper, we direct our attention to the third category of models. Specifically, we adopt a Bayesian approach to the problem of making inferences about the means of a two way mixed model. Although only a rather simple model is considered here, the method of analysis can be readily extended to more complicated ones.