Thanks Sten for a useful article. Sadly, this should surprise no one. Some years ago I worked at an NIH-funded research institute that brought in a top-level statistician from UC Berkeley to review a few dozen past papers from the institute that had all been published in prestigious journals. The idea was to see if any of the statistical analyses needed improving for future papers. In fact, it turned out that every single one of the published papers contained huge problems with how the data had been analyzed. Of the 50+ papers reviewed by the expert statistician, only a handful turned out to be valid in reality. Yet the prestigious publications hadn’t caught any of the errors during the peer review phase, and all the papers were being cited. If researchers can’t be relied upon to know how to use regular statistical techniques properly it’s no surprise that they don’t know how to perform more complex data analysis routines.