Longitudinal Data Analysis Using R and LLMs - A Short Course

The vast majority of what you will learn in this course can be applied in any software package. However, this seminar will mostly use R for empirical examples and exercises. To replicate the instructor’s workflow in the course, you should have R and RStudio already installed on your computer when the course begins.

For LLM support, the instructor will use the most recent paid version of Claude. However, most modern LLMs (e.g., ChatGPT and Gemini) will be useful for understanding, modifying, and interpreting longitudinal models.

Stata notes and syntax are available upon request.

Basic familiarity with R is highly desirable. If you are new to R, check out Professor Vaisey’s one-hour Introduction to R video to get up to speed. You may also want to consider taking a short introductory seminar on R such as Introduction to R for Data Analysis, R for SPSS Users, R for SAS Users, or R for Stata Users.

If you’d like to take this course but are concerned that you don’t know enough R, there are excellent online resources for learning the basics. Here are our recommendations.

This course is for anyone who wants to learn how to analyze repeated measures panel data. You should have a basic foundation in linear regression.

Module 1. Foundations of Panel Data

• What is panel data?
• Long form and wide form
• Within and between variation
• LLMs: giving the context of your panel data

Module 2. Growth Curves and Early Causal Designs

• Mixed models (growth curves) for time-constant treatments
• LLMs: prompting for growth-curve model specification and interpretation
• Introduction to fixed effects (“within” estimation)
• Pre-test/post-test designs
• Two-period difference-in-differences

Module 3. Within, Between, and Identification

• Two-way fixed effects
• Between-within models for time-varying treatments
• LLMs: diagnosing identification assumptions and interpreting within vs. between effects

Module 4. Dynamic Models and Robustness

• Introduction to linear structural equations for panel data
• Introduction to dynamic panel models
• LLMs: common pitfalls and robustness checks for dynamic panel models

“Professor Vaisey is fabulous at explaining things and emphasizes a deeper understanding of the statistics he is teaching … Through this approach, he fits a wide range of longitudinal models into a single conceptual framework, helping students connect their research questions and data structure to the modeling approaches they should consider. What’s more, the class is fun.”
  Nick Huntington, Brandeis University

“Dr. Vaisey was great. He had clear explanations, was engaging, very knowledgeable, yet also very chill and approachable … He is the only teacher I’ve ever had that acknowledged formulas are intimidating and that our brain wants to skid over them, which was really validating. I would definitely take a course with him again.”
  Mariah Wood, Kaiser Permanente

“The course was very well organized, with strong conceptual instruction, separate coding files, exercises, and other helpful materials. Stephen is very good at communicating the conceptual parts of the models. I would recommend the course to anyone who needs a good overview of different longitudinal models and wants to understand the underlying mechanisms of them.”
  Pernille Melander-Nyboe, Odense University Hospital

“The lectures were excellent, but the most useful takeaway is the additional course materials. The working R scripts for each of the methods presented, with sample exercises, were critical to help reinforce understanding and give us the ability to tinker a bit to understand how the code works and why it does what it does.”
  Andrew Althouse, University of Pittsburgh

“Dr. Vaisey was amazing! He broke everything down well, provided all of the R code in the slides and R Markdown file, and kept the class entertaining and lighthearted.”
  Nick Hollman, University of Oklahoma