Psychological Statistics

There has been quite a bit of buzz recently about the Button et al. Nature Reviews Neuroscience paper on statistical power. Several similar reviews have been published in psychology and other disciplines and come to broadly the same conclusion - that most studies are underpowered. The main difference with the Button et al. study is that they don't just find that typical studies are underpowered to  detect the average size of effect in a field, but they find extremely low power in neuroscience research (around 20%, and below 10% for some subfields). Contrast this with a typical review from psychology and related disciplines. Sedlmeier and Gigerenzer (1989, Table) report power to detect a medium effect size ranging from 37% to 89%.