GenStat Generalised Linear Mixed Models (GLMM) Workshop

Generalized linear models with random effects (GLMMs & HGLMs)

Roger Payne & David Baird
VSN International

This workshop will describe some of the methods that are available to analyse non-Normal data that are subject to several sources of random variation. These occur in many application areas, including medical trials, biological experiments, reliability studies etc. So, for example, binomial data may arise from counting numbers of infected plants in a field experiment, or from seeing whether patients have been treated successfully in a medical experiment. Poisson data may arise from counting numbers of aphids on leaves, or recording numbers of accident victims in a hospital casualty department etc. Many of the methods have been developed by statisticians working in agriculture and biology, but they are very widely applicable.

The methods to be covered will include:

• generalized linear models (brief recap),
• generalized linear mixed models,
• generalized estimating equations,
• hierarchical generalized linear models, and
• modelling of dispersion in generalized linear models,

Extension topics may include hierarchical generalized nonlinear models and double
hierarchical generalized linear models. 

The methods are all available in Genstat for Windows, many through the use of simple menus. These will be illustrated using real examples, and there will be practicals to allow the participants to try out the methods (and the relevant Genstat commands and menus). A download of a developmental Genstat edition, with new procedures and menus for GLMMs, will be provided for installation prior to the workshop.