Statistical Modelling 5 (2005), 21–37
Two-component mixtures of generalized linear mixed effects models
for cluster correlated data
Daniel B. Hall
Department of Statistics,
University of Georgia,
Athens, Georgia 30602-1952
USA
eMail:
dhall@stat.uga.edu
Lihua Wang
Department of Statistics,
University of Georgia,
Athens, Georgia
USA
Abstract:
Finite mixtures of generalized linear mixed effect models are presented to
handle situations where within-cluster correlation and heterogeneity
(subpopulations) exist simultaneously. For this class of model, we consider
maximum likelihood (ML) as our main approach to estimation. Owing to the
complexity of the marginal loglikelihood of this model, the EM algorithm
is employed to facilitate computation. The major obstacle in this procedure
is to integrate over the random effects' distribution to evaluate the
expectation in the E step. When assuming normally distributed random
effects, we consider adaptive Gaussian quadrature to perform this
integration numerically. We also discuss nonparametric ML estimation
under a relaxation of the normality assumption on the random effects.
Two real data sets are analysed to compare our proposed model with other
existing models and illustrate our estimation methods.
Keywords:
ADAPTIVE GAUSSIAN QUADRATURE; EM ALGORITHM; FINITE MIXTURE OF DISTRIBUTIONS;
RANDOM EFFECTS
Downloads:
Example data sets,
Matlab and SAS code in zipped archive
back