Abstract

Generalized linear models (GLMs); represent a class of regression models for several types of dependent variables where the linear predictor includes only fixed effects. Incorporation of random effects into GLMs yields the class of models known as generalized linear mixed models (GLMMs). Random effects are typically included for analysis of clustered and/or longitudinal data to account for the correlation of the data. GLMMs are especially useful for analysis of correlated nonnormal data, and the term GLMMs often refers to models for these kinds of data.

Keywords: generalized linear models; multilevel models; hierarchical linear models; logistic regression; probit regression; Poisson regression; maximum likelihood estimation