Online Program Home
My Program

Abstract Details

Activity Number: 560 - Using Large Healthcare Databases and Modern Statistical Methods to Impact Health Policy
Type: Topic Contributed
Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
Sponsor: Health Policy Statistics Section
Abstract #304716 Presentation
Title: A Fresh Look at Models, Assumptions, and Confounders in Diff-In-Diff
Author(s): Bret Zeldow* and Laura A Hatfield
Companies: Harvard Medical School and Harvard Medical School
Keywords: Difference-in-Differences; Confounding; Parallel Trends; Identifying Assumptions; Adjustment
Abstract:

In difference-in-differences studies, the so-called parallel trends assumption carries a heavy burden. The pursuit of parallel trends can sometimes mask an important discussion on model choice, assumptions, and confounders and their effects on our diff-in-diff estimates. In this talk, we focus on the definition of a confounder in diff-in-diff and develop strategies to adjust for it. Further, we discuss how to leverage covariates to achieve efficiency gains (smaller standard errors) in our diff-in-diff estimates. We conclude by discussing causal quantities and identifying assumptions for diff-in-diff when you have many post-treatment time points.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program