The Credibility Revolution

Difference-in-Differences Designs: A Practitioner's Guide

Properties1
tagsliterature

2 min read

Difference-in-Differences Designs: A Practitioner’s Guide

Causal Question / Estimand

A hands-on guide to estimating the average treatment effect on the treated (and its group-time / event-study disaggregations, Causal-Estimand) across the full range of DiD designs — 2×2, multiple periods, staggered adoption, with covariates and weights.

Identification Strategy

Provides an organizing framework that classifies DiD designs by their structure and matches each to an appropriate estimator. Walks through the canonical 2×2 case (Parallel-Trends identifies the ATT), then the complications — covariates (Conditional-Parallel-Trends), weighting, multiple periods, and staggered timing — explaining when TWFE fails (Negative-Weighting) and which heterogeneity-robust estimator to use.

Key Assumptions

Parallel-Trends / Conditional-Parallel-Trends, No-Anticipation, valid (clean) comparison groups, and honest treatment of Treatment-Effect-Heterogeneity. SUTVA throughout.

Threats to Validity

n/a — practitioner guide. Emphasizes the standard failure modes (ad hoc TWFE under staggered timing, naive pre-trend testing, mishandled covariates/weights and clustering) and how to avoid each.

Setting / Data

n/a — methodological guide with empirical illustrations and pointers to software implementations.

Key Claims

  • A unifying taxonomy: identify your DiD design type, then choose the matching estimator.
  • Practical treatment of covariates, weights, multiple periods, and staggered adoption in one place.
  • Bridges the technical robust-DiD literature and everyday applied practice.

Connections

Citation

Baker, A., Callaway, B., Cunningham, S., Goodman-Bacon, A., & Sant’Anna, P. H. C. (2025). Difference-in-Differences Designs: A Practitioner’s Guide. Working paper, arXiv:2503.13323.

Graph View

Backlinks