Using and Understanding Power in Psychological Research: A Survey Study

Abstract

Statistical power is key to planning studies if understood and used correctly. Power is the probability of obtaining a statistically significant p-value, given a set alpha, sample size, and population effect size. The literature suggests that psychology studies are underpowered due to small sample sizes, and that researchers do not hold accurate intuitions about sensible sample sizes and associated levels of power. In this study, we surveyed 214 psychological researchers, and asked them about their experiences of using a priori power analysis, effect size estimation methods, post hoc power, and their understanding of what the term “power” actually means. Power analysis use was high, although participants reported difficulties with complex research designs, and effect size estimation. Participants also typically could not accurately define power. If psychological researchers are expected to compute a priori power analyses to plan their research, clearer educational material and guidelines should be made available.