Hiroshima Mathematical Journal

A class of multivariate discrete distributions based on an approximate density in {GLMM}

Tetsuji Tonda

Full-text: Open access

Abstract

It is well known that the generalized linear mixed model is useful for analyzing the overdispersion and correlation structure for multivariate discrete data. In this paper, we derive an approximation of the density function for the generalized linear mixed model. This approximation is found to satisfy the properties of probability density function under some conditions. Therefore, this approximation can be regarded as a class of multivariate distributions. Estimation of the parameters in this class can be carried out by the maximum likelihood method. We give the likelihood ratio criteria for testing several covariance structures. Several simulation studies were also conducted for the Poisson log-normal model when the proposed density function is regarded as an approximate likelihood of the generalized linear mixed model.

Article information

Source
Hiroshima Math. J. Volume 35, Number 2 (2005), 327-349.

Dates
First available in Project Euclid: 22 June 2006

Permanent link to this document
https://projecteuclid.org/euclid.hmj/1150998277

Mathematical Reviews number (MathSciNet)
MR2176056

Zentralblatt MATH identifier
1082.62050

Subjects
Primary: 62J12: Generalized linear models
Secondary: 62F10: Point estimation 62H12: Estimation 62H15: Hypothesis testing

Citation

Tonda, Tetsuji. A class of multivariate discrete distributions based on an approximate density in {GLMM}. Hiroshima Math. J. 35 (2005), no. 2, 327--349. https://projecteuclid.org/euclid.hmj/1150998277.


Export citation