The cult of statistical significance 1 by Walter Krämer

This article takes issue with a recent book by Ziliak and McCloskey (2008) of the same title. Ziliak and McCloskey argue that statistical significance testing is a barrier rather than a booster for empirical research in many fields and should therefore be abandoned altogether. The present article argues that this is good advice in some research areas but not in others. Taking all issues which have appeared so far of the German Economic Review and a recent epidemiological meta-analysis as examples, it shows that there has indeed been a lot of misleading work in the context of significance testing, and that at the same time many promising avenues for fruitfully employing statistical significance tests, disregarded by Ziliak and McCloskey, have not been used. 1 Research supported by DFG under SFB 823. On purpose, the title is the same as that of a recent book by Ziliak and McClowskey (2008). I am grateful to Michael Bücker and Ronja Walter for excellent research assistance, and to Stephen Ziliak for comments on a book review in Statistical Papers on which this article is based.