Learn Bayesian multilevel modelling

Graduate School workshop 2021

Online Skript

Workshop contents

This workshop is designed to provide a practical introduction to basic and advanced multilevel models. Participants will learn how to fit models using both maximum likelihood and Bayesian methods, although the focus will be on Bayesian parameter estimation and model comparison. We will start with a short introduction to multilevel modelling and to Bayesian statistics in general, followed by an introduction to Stan, a probabilistic programming language for fitting Bayesian models. We will then learn how to use the R package brms, which provides a user-friendly interface to Stan. The package supports a wide range of response distributions and modelling options, and allows us to fit multilevel generalized linear models. Depending on participants' wishes, we will take a closer look at modelling various types of data, such as choices, response times, ordinal or longitudinal data.

Specific topics include:

• Bayesian inference: an introduction

• Bayesian parameter estimation

• Model comparison & hypothesis testing

  • Bayes factors
  • Out-of-sample predictive accuracy (LOO)

• Specifying multilevel generalized linear models

• Understanding statistical models through data simulation

• A principled Bayesian workflow for data analysis

We will focus on learning new topics during the morning sessions, and participants should work through the assignments during the afternoon sessions. Participants are also encouraged to bring their own datasets.

Prerequisites

Basic knowledge of regression models and R is a necessity. I strongly recommend that you prepare for the workshop by working through this online script: https://methodenlehre.github.io/intro-to-rstats. Previous exposure to multilevel models and longitudinal models would be helpful, but is not strictly necessary. Knowledge of Bayesian statistics is not required.

Software

We will be using R and RStudio, as well a variety of R packages. It is advisable to ensure that you have a working R installation before the workshop starts, and that you install the two R packages rstan and brms. Detailed instructions for installing these packages on all platforms can be found at https://github.com/stan-dev/rstan/wiki/RStan-Getting-Started and https://paul-buerkner.github.io/brms.