Welcome to my one-day course on difference-in-differences at ISR at the University of Michigan.
This will be a four-hour workshop that will involve three main components: lectures, a coding exercise, and an "online office hour" for your unique DiD Q&A.
The exact timing will depend on how many questions people ask, but a rough outline of the planned schedule is as follows:
All times Eastern Time.
- 9-9:15 Introduction: Overview of Recent Advances in DiD
- 9:15-10:15 Lecture 1: Testing the Parallel Trends Assumption: Issues and Limitations
- 10:15-10:30 Break
- 10:30-11:45 Lecture 2: New Approaches When Parallel Trends May Be Violated
- 11:45-12:15 Coding Exercise
- 12:15-1 Open "Office Hours" (Q&A)
The coding exercise we will do today is Exercise 2 from my Advanced DiD class taught through MixTape Sessions. This coding exercise considers methods for sensitivity analysis for violations of the parallel trends assumption using the HonestDiD package. (You are welcome to try Exercise 1 on your own, which covers new methods for settings with staggered treatment timing.)
The data for Exercise 2 is the same as for Exercise 1, so you should first look at the Exercise 1 instructions for loading the data, then flip back to Exercise 2.
The instructions tell you what to do in both R and Stata. You can find the solutions available here: Stata R