The self-fulfilling prophecy of post-hoc power calculations
Researchers often resort to post-hoc power calculations, and unfortunately, this practice is often encouraged by reviewers, editors, and journals. In this short article, we attempt to clarify the shortcomings of post-hoc power calculations by demonstrating its irrelevance at the analysis stage, as opposed to power calculations at the design stage of a study. We discuss the false but sometimes convenient and promising interpretation that an effect went undetected because of low power. We advise against using post-hoc power as it contains no more information than that carried by the observed P value.