February 2007
Beginner to intermediate
464 pages
16h
English
For non-Gaussian outcomes, there is no single, broadly applicable counterpart to the multivariate normal model, within which the linear mixed model is positioned. Therefore, it is important to carefully distinguish between three model families: marginal models, random-effects models, and conditional models. A comprehensive introduction of these and comparison between them has been provided in Dmitrienko et al. (2005, Chapter 5). Here, we provide a brief overview of the marginal family, with focus on generalized estimating equations (GEE), and put some emphasis on the random-effects family. In particular, we emphasize the generalized linear mixed model (GLMM).
Marginal models describe the outcomes ...
Read now
Unlock full access