Inferences in Generalized Linear Mixed Models for Count Data

Lead Researcher and Department
Md. Rafiqul I Chowdhury, Ph.D. Student, Department of Mathematics & Statistics, Memorial University

Summary
Analyzing clustered count data is an important problem in economics and biomedical studies, among others. For example, in health economics studies, one may be interested to estimate the effects of certain covariates such as gender and education level on the number of visits to the physician paid by different members of a family. It is important to find the effects of the covariates on the responses after taking familial correlations into account.

So far, several estimation approaches are available in the literature, among which moment method (MM), penalized quasilikelihood (PQL), hierarchical likelihood (HL) and recently introduced generalized quasi-likelihood (GQL) approaches have proven to be effective in terms of consistency and efficiency. In this work, a comparison between GQL and HL approaches has been made mainly through an extensive simulation study involving the Poisson-normal mixed model by analyzing data on health care utilization in St. John’s, Newfoundland.

Compared to the HL approach it is noticed that the GQL approach always produces consistent estimates for all parameters of the model. Therefore, the study recommends the use of the GQL approach irrespective of family sizes and the magnitude of the over dispersion in the familial/cluster count data

Keywords
Clustered count data, Moment method, Generalized quasi-likelihood approach, Hierarchical likelihood approach, Health care utilization

Locations
St. John's
Avalon Peninsula

Industry Sectors
Offices of Physicians (Health Care and Social Assistance — Ambulatory Health Care Services)
Mathematics Research and Development Services (Professional, Scientific and Technical Services — Scientific Research and Development Services — Research and Development in the Physical, Engineering and Life Sciences)
Medical Research and Development Laboratories (Professional, Scientific and Technical Services — Scientific Research and Development Services — Research and Development in the Physical, Engineering and Life Sciences)
Economic Research and Development (Professional, Scientific and Technical Services — Scientific Research and Development Services — Research and Development in the Social Sciences and Humanities)

Thematic Categories
Health Professionals (Health)
Mathematics and Statistics (Science Research)

Departments
Mathematics & Statistics, Faculty of Science (STJ)