imple calculations seem to show that larger studies should have higherstatistical power, but empirical meta-analyses of published work in criminology havefound zero or weak correlations between sample size and estimated statistical power. Thisis ‘‘Weisburd’s paradox’’ and has been attributed by Weisburd et al. (in Crime Justice17:337–379,1993) to a difficulty in maintaining quality control as studies get larger, andattributed by Nelson et al. (in J Exp Criminol 11:141–163,2015) to a negative correlationbetween sample sizes and the underlying sizes of the effects being measured. We argueagainst the necessity of both these explanations, instead suggesting that the apparentWeisburd paradox might be explainable as an artifact of systematic overestimationinherent in post-hoc power calculations, a bias that is large with smallN.MethodsWe discuss Weisburd’s paradox in light of the concepts of type S and type Merrors, and re-examine the publications used in previous studies of the so-called paradox.ResultsWe suggest that the apparent Weisburd paradox might be explainable as anartifact of systematic overestimation inherent in post-hoc power calculations, a bias that islarge with smallN.ConclusionsSpeaking more generally, we recommend abandoning the use of statisticalpower as a measure of the strength of a study, because implicit in the definition of power isthe bad idea of statistical significance as a research goa

• Type M Error Might Explain Weisburd’s Paradox