Reproduce “Difference-in-Differences Designs: A Practitioner's Guide” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Andrew Baker, Brantly Callaway, Scott Cunningham, Andrew Goodman-Bacon, Pedro H. C. Sant'Anna
This practitioner's guide clarifies the complexities of Difference-in-Differences (DiD) designs, particularly in multi-period and staggered treatment adoption settings, by advocating for a 'forward-engineering' approach to explicitly define causal parameters and their identification assumptions. The authors demonstrate how commonly used Two-Way Fixed Effects (TWFE) estimators can produce biased results in these complex scenarios and provide robust, transparent alternatives for applied researchers.
Key insights
- All complex Difference-in-Differences (DiD) designs can be understood as aggregations of fundamental 2x2 building blocks, clarifying target parameters and identification assumptions.
- The commonly used Two-Way Fixed Effects (TWFE) estimator is problematic in multi-period and staggered adoption DiD settings, as it can yield biased estimates due to 'bad comparisons' and implicit negative weighting when treatment effects are heterogeneous.
- Adopting a 'forward-engineering' approach—first defining causal parameters and then choosing estimators—enhances transparency and interpretability over starting with a regression specification.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.