via nytimes.com
If only the original journal had made them submit their data and source code for peer inspection. It wouldn’t stop this kind of thing happening but it would make it less likely and mistakes could be spotted more quickly.