Fitting Models to Your Experimental Data When They Are Counts or Proportions. An Introduction to Generalized Linear Mixed Models.

The general linear mixed model methodology has become commonplace in agricultural literature. However its application is appropriate when responses that are fitted to such a model can be considered to follow a normal distribution. Over the last decade computer software has become available that allows for a more general class of response variables to be fitted to a mixed model thus extending the mixed model methodology to what is now referred to as the generalized linear mixed model. This is a basic introduction on how to write and fit a model for statistical analysis of data gathered from an experiment in the context of a generalized linear mixed model when response data are counts and proportions. Various examples from agricultural experiments will be provided using both SAS and R software. The inter-relationship between experimental design and model will be emphasized.