P-hacking and Significance Stars

In mid-2016, all journals of the American Economic Association stopped including significance stars in their regression tables. This policy aimed to reduce the emphasis on statistical significance and shift the focus to the broader economic importance of research findings. This study examines the impact of this change on p-hacking and publication bias. I employ a difference-in-difference design, as well as the battery of tests proposed by Elliott et al. (2022b). I find no statistically significant effect of this policy.