Multilevel Modelling

About

"Learn how to specify, estimate and interpret multilevel models, ranging from simple two-level random intercept models to more complex ones (including three-level, logistic or longitudinal models).

Need to know
Familiarity with linear regression analysis is required.

In depth
Key topics covered
Over the course of five modules (one per day), a variety of topics that are fundamental to multilevel modelling are discussed.

Module 1
Covers the multilevel modelling. You will have the opportunity to join in discussions to do with basic concepts and will be introduced to the idea of nested data. In this module, you will study our very first multilevel model -the two-level random intercepts model- and discover how this approach decomposes the total variation into level 1 and level 2 components.

Module 2
It takes the two-level regressions a couple of steps further, and adds random slopes as well as cross-levels interactions to the model. You will learn how clever ways of entering predictors can help people to disentangle within- from between-effects - a challenging topic!

Module 3
The assumptions made by the multilevel regression model are the point of departure of Module 3. The Generalized Linear Model is introduced as a way to deal with data that does not follow a normal distribution. Discussions will take on one such a GLM -the two-level logistic regression model- in greater detail.

Module 4
It focuses on a particular but very useful application of multilevel models, namely for analysing longitudinal data. The Growth Curve Model is introduced, and you will acquire knowledge about various functional forms of time and covariance structures.

Module 5
It applies our multilevel toolkit to more complex and challenging data structures. The three-level multilevel model and the cross-classified model are discussed in greater detail.

How the course will work online
Each of the modules combines a number of pedagogical tools and resources. You will be required to prepare for the module by watching pre-recorded short lectures (typically 3 lectures of 15 minutes each per module), process the essential readings, and prepare a short hands-on exercises analysing real data using R (or another software package of your choice). During a daily interactive session (3 hours), examples and additional topics are discussed in greater detail, and you will get ample opportunity to ask questions and receive feedback. The instructor and TA will organize office hours, so that you can seek advice for your personal research projects."

Course leader

Bart Meuleman is a Full Professor at the Centre for Sociological Research, KU Leuven (Belgium). Bart is the National Coordinator of ESS Belgium, co-supervisor of the Belgian National Elections Study and the Belgian Ethnic Minority Elections Study.

Course aim

By the end of this course you will be able to specify, estimate and interpret multilevel models, ranging from simple two-level random intercept models to more complex ones (including three-level, logistic or longitudinal models).

Fee info