Power Analysis for Difference-in-Differences Studies with Staggered Treatment Adoption

• In Event: AERA Roundtable Session Friday 3:45 pm - Gold 4
  In Roundtable Session: RT5: Advances and Implications for Causal Inference and Power in Randomized Treatment Designs

Abstract

This study develops a simulation-based framework for statistical power analysis in Difference-in-Differences (DID) designs with staggered treatment adoption. While staggered DID estimators improve unbiasedness under treatment effect heterogeneity and variable adoption timing, little guidance exists on their statistical power. We systematically evaluate eight leading estimators using realistic data-generating processes calibrated to national education datasets. Results reveal substantial variation in power across estimators, with many failing to detect moderate effects under typical sample sizes. We also identify implementation challenges in software and provide recommendations for estimator selection and study design. Findings offer practical guidance for researchers planning well-powered longitudinal evaluations using staggered DID designs in educational settings.

Authors

• Wei Li, University of Florida

• Katherine Strickland, University of Florida

• Jing Huang, University of Florida

• Shuyuan Li, University of Florida