Presented by Sandro Leidi & James Gallagher
This course is running online, via Zoom.
Mixed models have become increasingly popular, as they have many practical applications. However, the traditional linear mixed model with normally distributed errors may not always be appropriate for modelling discrete response variables, such as binary data and counts. Typically these types of responses are analysed using generalised linear models such as logistic regression and Poisson regression.
Commonly-used generalised linear models will be extended to deal with multiple error structures, using a variety of examples, generally drawn from medical and health related fields. Specific applications, such as repeated measurements and multi-centre trials will also be considered. For example, investigating the presence or absence of adverse events collected in a multi-centre clinical trial.
The emphasis will be on practical understanding, although an outline of the theory will be presented. Practical examples will be used to illustrate the methods, and participants will have the opportunity to fit and interpret models themselves in hands-on computer based practicals.
Note this course does not cover marginal or GEE type models, but will consider population-averaged effect measures derived from a generalised linear mixed model.