Peer review is letting through statistical errors : http://www.economist.com/science/displayStory.cfm?story_id=2724226