Generalized Linear Mixed Models
2013; Linguagem: Inglês
10.2134/2012.generalized-linear-mixed-models.c5
ISSN2691-2341
AutoresEdward E. Gbur, Walter W. Stroup, Kevin S. McCarter, Susan L. Durham, Linda J. Young, Mary C. Christman, Mark J. West, Matthew Kramer,
Tópico(s)Optimal Experimental Design Methods
ResumoGeneralized linear mixed models (GLMMs) combine the generalized linear models with the linear mixed models. As an extension of generalized linear models, they incorporate random effects into the linear predictor. As a mixed model, they contain at least one fixed effect and at least one random effect. Inference in GLMMs involves estimation and testing of the unknown parameters in β, R, and G as well as prediction of the random effects. As with linear mixed models, GLMMs can be formulated as conditional or marginal models. Various conditional and marginal GLMMs are used to account for or adjust for over-dispersion, an issue unique to models for non-normally distributed data. Understanding the distinction between conditional and marginal GLMMs and the issues that arise is best accomplished by revisiting the linear mixed model conditional and marginal examples but working through them with a non-normal response variable.
Referência(s)