Commentary on “To replicate or not to replicate, or when did we start to ignore the concept of statistical power?” by Meyners, Carr and Hasted

Abstract Studies should be planned with due attention to power and sample size considerations. Abandoning power computations, including analyses of the utility of replications, is a sure way to reduce the likelihood that an experiment yields insights. This commentary describes how screening for effects with low power leads to biased effect estimates and detrimentally high rates of false findings, and why aiming for, say, 90% power is not enough to ensure that power is on average 90%. It is also suggested that scenario-based evaluations of operating characteristics can expose the consequences of power policies. These discussions show that we are easily deceived by low powered studies and that power considerations are crucial for meaningful scientific experimentation.