Methodological Updates for Equivalence Tests in Randomized Controlled Trials

• 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

Researchers may verify valid randomization by testing the equivalence of means for baseline variables and the equivalence of means for post-treatment variables when other dimensions (e.g., costs) differ. However, the analytical and design methods for equivalence tests are limited (e.g., the lack of covariate adjustments and the absence of correct power formula development). The present study develops and validates a statistical power formula for randomized controlled trials to detect statistical equivalence, both with and without covariate adjustment. It further presents a Monte Carlo confidence interval method for convenient testing and a framework to identify the optimal sample allocation that maximizes statistical power under a fixed budget. The proposed methods have been implemented in the R package XXX.

Author

• Zuchao Shen, University of Georgia