Should We Adjust for the Test for Pre-trends in Difference-in-Difference Designs?

Jonathan Roth

The common practice in difference-in-difference (DiD) designs is to check for parallel trends prior to treatment assignment, yet typical estimation and inference does not account f…