JMP for Mixed Models

Chapter 9Generalized Linear Mixed Models

Throughout this book we have thus far considered examples for which the response is continuous, assuming that the random effects and residual errors follow a normal (Gaussian) distribution. While this is arguably the most common type of mixed model, in many cases the response is discrete. In this chapter, we address mixed models for non-normally distributed data.

9.1 Motivating Examples

The three most common examples of non-normal data are:

Binomial — the number of successes, y, out of n trials. The Shrub Coverage example from Parks Canada measures the number of sampling points out of 100 that contain species from a plant functional group during a given year. A "success" in this case is the presence of ...