Led by Dr Patricio Troncoso and Dr Ana Morales-Gomez
Overview
Multilevel modelling is an umbrella term for a wide range of statistical models appropriate for clustered data. Multilevel modelling can be thought of as an extension of the classical Multiple Regression Models that allows the researcher to assess the variation in an outcome of interest at different levels of a predefined hierarchy structure and simultaneously analyse the characteristics associated with that variation.
Multilevel modelling can be used for a variety of social science problems. Examples include: variation in educational outcomes of pupils nested within schools, or varying trajectories over time within pupils; variation in outcomes of prisoners nested within prisons; variation in income for individuals nested within geographical areas; variation in health outcomes of patients nested within GP practices and hospitals; amongst other applications.
In this short course, we will introduce the main concepts and theoretical issues around multilevel modelling. We will gain experience fitting this type of models with real-world multilevel data from the UK Data Service, using R and RStudio, but we will also discuss alternative software packages. This course will focus on the substantive interpretation of the outputs to gain an understanding of the type of research questions that multilevel modelling can tackle and the published research using multilevel modelling.
Mode of delivery
We will have a mixture of traditional lectures, where we introduce the concepts, and hands-on sessions, where we
put the concepts into practice.
Intended learning outcomes
By the end of this course, you will be able to:
• Understand the general concept of multilevel modelling and its applications in social science research questions.
• Understand a range of multilevel models and when to use them.
• Specify multilevel models for continuous and binary responses using R.
• Interpret the R output of standard multilevel models.
Prerequisites
• Prior knowledge of multilevel modelling is not assumed.
• Some experience using and/or interpreting regression models is necessary.
• Some experience using R and/or RStudio is necessary.
• Registering with the UK Data Service prior to the course is essential. To register with the UKDS, please visit:
• You can bring your own laptop, if you do, please install R and RStudio: https://posit.co/download/rstudio-desktop/
• Alternatively, you can sign up for RStudio Cloud: https://login.rstudio.cloud/
Recommended literature
This course is mostly based on these recommended readings. It is not essential to read any of these beforehand, but if you choose to prepare by doing some reading, you do not need to read them all. Choosing one of them should suffice as an introduction, as they cover most of the same topics. These books are ordered in a non-mutually exclusive order of ascending complexity (and/or social-science user-friendliness):
• Hox, J., Moerbeek, M., van de Schoot, R. (2017). Multilevel Analysis: Techniques and Applications (3rd Ed).
Routledge. (Basic to intermediate). Chapters 1 to 4.
• Snijders, T., Bosker, R. (2012). Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling
(2nd ed.). Sage. (Intermediate). Chapters 1 to 5.
• Goldstein, H. (2011). Multilevel statistical models (4th ed.). John Wiley and Sons. (Advanced). Chapters 1 and 2.
On the other hand, if you are planning to immerse yourself deeply in the world of multilevel modelling and apply it
in your own research, these titles are probably a must-have in your personal library.
Programme
This is an outline of the main topics to be covered. Flexibility around transition times is considered. Practical sessions are for individual or group work under the guidance of the facilitators and each session will include a brief demonstration and explanation of outputs.
• 10:00-10:30 – A brief overview of linear regression
• 10:30-11:00 – Multilevel data structures and examples
• 11:00-11:15 – Break
• 11:15-11:45 – Variance components and group-specific estimates
• 11:45-12:30 – Practical 1
• 12:30-13:30 – Lunch
• 13:30-13:50 – Accounting for individual and group characteristics: fixed effects
• 14:00-14:30 – Practical 2
• 14:30-15:00 – Multilevel modelling for binary responses
• 15:00-15:45 – Practical 3
• 15:45-16:00 – Break
• 16:00-16:20 – Differential processes between groups: random effects
• 16:20-16:50 – Practical 4
• 16:50-17:00 – Wrap up and Finish
Register for Summer School here!