Introduction to Generalised Linear Mixed Models for Ecologists

Course Description

This 5-day course offers a practical and theoretical introduction to statistical models, through the most fundamental and versatile statistical procedure in ecology: the linear model. We’ll begin by looking at simple linear models with both continuous and discrete predictors – this includes regression, ANOVA, ANCOVA, and all classic ANOVA designs. We’ll also look at extending these linear models to include counts and proportions as responses, and to include data that naturally falls into groups – such as repeated measures or spatial plots – which require a multilevel or hierarchical model.

Participants will learn to work with ecological data from both experiments and designs, where factors we’re interested in may be nested, crossed or in some way structured. We’ll cover ways to add flexibility into models, such as adding group-level predictors and overdispersion. We’ll learn and contrast both Bayesian and frequentist approaches to modelling. Along the way we’ll practice essential skills such as diagnostic plots, model selection, and predictions. We’ll do this by applying the models we’re studying to real-world data.

This course is ideal for ecologists and applied scientists working with grouped, hierarchical, or repeated-measures data.

The course will be taught using R and Rstudio, and the packages lme4 (for frequentist statistics) and rstanarm for Bayesian statistics.

What You’ll Learn