P-Curve: A Tool for Detecting Publication Bias

Author: Uri Simonsohn, Leif D. Nelson, and Joseph P. Simmons.

Publisher: Berkeley Initiative for Transparency in the Social Sciences

Publication Year: 2014

Summary: The following article overviews the concept of the “P-curve,” a tool that can help detect if published research is likely due to p-hacking. P-hacking is the practice of data scientists mining their data until they yield a low p-value, a sign of statistical significance, even if no relationship between variables really exists. P-hacking as a practice is a problem in academic research the incentive structure adds pressure to publish research whereas journals tend to publish only significant results. The authors hope this method can help the public analyze published research results to detect p-hacking, an effort to improve academic research.