Analyzing over-dispersed count data in two-way cross-classification problems using generalized linear models

Abstract

Several methods of testing factor effects in a two-way cross-classification analysis are compared in a Monte Carlo simulation study, using over-dispersed count data generated from the family of negative binomial distributions. Tests compared are based on general linear models, using raw and transformed data, and on generalized linear models specifying either the Poisson or negative binomial distribution. The general linear model is recommended, especially in the case of small means and small numbers of replications.