Multilevel Models

about

I’ve designed this as an applied introductory course about multilevel models for students with nested (e.g., repeated measures), clustered (e.g., observations in schools), or longitudinal data. For students who don’t have data, I’ll work to find them a data set on a topic they are interested in. Ultimately, I want students to leave the course with a reference document, written in their own words, for doing basic multilevel models. The last week will introduce alternative methods for analyzing clustered data. The course is intended to provide enough exposure to, and practice with, core issues in multilevel models so that students can learn how to do more complicated models on their own (e.g., moderation or interactions between levels). I assume that students have had at least one graduate course in statistics that covered multiple regression.

format

This is a hybrid course to allow maximum flexibility so that

• you can learn the material at times that best suit you

• you can count on weekly class meetings to cover the tough stuff, edge cases, and issues that arise when working with real data

Assignments

There are 3 graded assignments that cover

(a) describing your data [e.g., research question, variables, data structure, and multilevel model in both math notation and words]

(b) writing the code/syntax to analyze your data, making sure it runs, and detailing the values you will attend to and why [e.g., from model, the values in the estimate will tell me… If they are positive, then … If they are negative…]

(c) analyze your data, explain what you found in everyday language and formally in an APA-formatted results section

Lectures

I will provide recorded lectures that connect information from the textbook to a worked data example (or two). I’ll also provide example R scripts for you to modify to fit your data.

Class meetings

Weekly class meetings will provide you with the opportunity to ask questions and practice applying the textbook and lecture material to their own data. Each week, you'll be provided with a list of questions that you should be able to answer from the course material. During class we will discuss the hard-to-answer questions, clarify why some questions have multiple correct answers, and apply the material to our own research questions. Class meetings will last as long as you have questions or we reach the end of the scheduled meeting--which ever comes first.

Required materials

We will be using the open access textbook

Leyland, A. H., Groenewegen, P. P. (2020). Multilevel Modelling for Public Health and Health Services Research: Health in Context. Germany: Springer International Publishing. https://www.springer.com/gp/book/9783030347994

Software

• Course examples will be provided in R

• Textbook examples are in MLwiN