This is a tutorial on how to conduct group-level statistical analyses of fMRI data using mixed effect linear models in R. It arose out of an extended Evlab meeting and reflects our practices at the time. We expect to develop and improve upon them as time goes by.

This tutorial assumes that you have already conducted subject-level analyses and extracted average response values for each brain region of interest (ROI). It assumes no previous knowledge of R and only a bit of stats (e.g. we will remind you the basics of linear regression but will assume that you know what p-values are for). Our goal is to:

  1. Explain the basic logic behind linear mixed effect models.

  2. Show how to run them in R using the lme4 package.

  3. Discuss how to interpret the results you get.

  4. Highlight the major decision points you as a modeler will have to make along the way.