t-Statistic Cutoffs
Click a link below to see adjusted t-statistic critical values for four types of commonly used settings: (1) randomized control trials, (2) the staggered introduction of state-level changes, (3) instrumental variables, and (4) regression discontinuity designs. See Table A1 of our Journal of Finance article for details.(1)
For example, if you write a paper that examines a particular randomized control trial, and the existing literature has already examined 100 different dependent variables using this randomized control trial, our simulation results suggest you should use a t-statistic cutoff of 3.41 to determine statistical significance in order to have an interpretation similar to the typical 5% p-value.
(1) Suggested citation "Heath, Davidson, Matthew C. Ringgenberg, Mehrdad Samadi, and Ingrid M. Werner, 2022, Reusing Natural Experiments, Journal of Finance, forthcoming."