Triple Differences Research Designs at Causal Solutions

Difference-in-Differences (DiD) is one of the most popular methods in the social sciences for estimating causal effects in non-experimental settings. Its primary identifying assumption is a so-called parallel trends assumption that states that, in the absence of the treatment/intervention, the outcomes would evolve in parallel across different treatment groups/cohorts. In some applications, however, one may be concerned about the plausibility of this assumption. In this Lecture, Pedro Sant'Anna (https://psantanna.com/) explains how one can potentially relax the parallel trends assumption by leveraging a Triple-Differences strategy. He discusses the case with two time periods, and also the case of several time periods and staggered treatment adoption. He also discusses how to incorporate covariates into the setup in a statistically-sounded manner. ##-------------------------------------------------------------------------------------------------------------## This lecture is part of the Causal Solutions course on Difference-in-Differences - https://www.causal-solutions.com It builds on material previously discussed in the course that is restricted to the students of the course. ##-------------------------------------------------------------------------------------------------------------## Slides are available at https://psantanna.com/files/DDD.pdf