Premium
Model comparison of generalized linear mixed models
Author(s) -
Song XinYuan,
Lee SikYum
Publication year - 2005
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2318
Subject(s) - generalized linear model , computer science , generalized linear mixed model , likelihood function , bayesian information criterion , bayesian probability , statistics , path (computing) , sampling (signal processing) , maximum likelihood , linear model , econometrics , mathematics , data mining , algorithm , filter (signal processing) , computer vision , programming language
Abstract Generalized linear mixed models (GLMMs) have been widely appreciated in biological and medical research. Maximum likelihood estimation has received a great deal of attention. Comparatively, not much has been done on model comparison or hypotheses testing. In this article, we propose a path sampling procedure to compute the observed‐data log‐likelihood function, so that the Bayesian information criterion (BIC) can be applied to model comparison or hypothesis testing. Advantages of the proposed path sampling procedure are discussed. Two medical data sets are analysed for providing illustrative examples of the proposed methodology. Copyright © 2005 John Wiley & Sons, Ltd.