However, when he modified his statistical methods to take into account the fact that he was testing 24 hypotheses, not one, the boundary of significance dropped dramatically.

ECONOMIST: Why so much medical research is rot