Multilevel Modeling


  1. Multilevel Modeling: Principles and Applications with R
  2. Applied Multilevel Regression Modeling

In the final stages of my doctoral studies, and after this period, I was involved in teaching a host of classes on multilevel modeling in different formats: week-long or two-week-long intensive workshops, and even block courses.

Though all materials were developed from the ground up, a great deal of insight was gained from past experiences TAing for similar classes and workshops taught over the years by Zoltan Fazekas and Levente Littvay. I am grateful to both for the opportunity to learn.

Each class listed below has a short description, and points to a GitHub repository where all materials are made available. I hope you find them useful.

Multilevel Modeling: Principles and Applications with R

This 3-day workshop at the University of Bamberg in 2021 exposed participants to the rigorous application of multilevel models (MLMs) in the R statistical environment. Over the course of 3 days, we covered the fundamentals of such hierarchical linear models, starting from simple random-intercept specifications and advancing to more complex ones, which allow us to understand how an effect varies between contexts. During this progression we touched on estimation of MLMs, sample size considerations, and model fit criteria. In the last day of the class we discussed how MLMs can be used to model change over time, under the form of the longitudinal growth model, and practice estimating such a model.

All materials (data sets, slides, and code files) are made available here.

Applied Multilevel Regression Modeling

This 2-week course at the 2019 ECPR Summer School, at the Central European University, Budapest, took participants from the basics of multilevel specifications to advanced applications.

The first week set the foundations. We started from basic hierarchical linear models (HLMs), with only random intercepts, to more complex specifications, that allow us to understand how an effect varies across contexts. As part of this progression we covered estimation, 2- and 3-level configurations, what sample size considerations apply to HLMs, and how to assess models’ adequacy. In the second week we explored alterations to this fundamental framework introduced the week prior. We covered the use of dichotomous outcomes, applying a multilevel specification to assess change over time (growth curve modeling), as well as how to analyze non-hierarchical data configurations. The sessions were conducted entirely in R.

All materials (data sets, slides, and code files) are made available here. For the first half of the workshop the GitHub repository also includes Python code for all specifications.