Bob Reed

• November 02, 2021
• Speaker: Bob Reed
  
• Topic: Yes, You Can Calculate Ex Post Power! by Alex Tian, Tom Coupé, Sayak Khatua, Bob Reed, and Ben Wood
  

Statistical power is the probability of obtaining a statistically significant estimate, conditional on the true size of the underlying effect. Research funders commonly ask for ex ante power calculations to determine whether the sample size of the proposed research is sufficient to demonstrate an effect. However, ex ante power calculations often involve subjective judgments. The purpose of our study is to present a simple estimator of power that can be used after a study is completed. We first provide Monte Carlo evidence of the reliability of this estimator. We then apply it to a set of studies that were funded by 3ie (International Initiative for Impact Evaluation). We compare our ex post calculations of power with the ex-ante calculations that were submitted to the funder. Finally, we discuss how ex post power calculations can be useful in a number of settings, including current debates in the “replication crisis”.

Bob Reed is a Professor in the Department of Economics and Finance at the University of Canterbury, New Zealand. His research interests include health economics, public economics, applied econometrics, development economics, and labour economics.

The presentation slides are available here.