This article in Medical Xpress addresses a huge pet peeve of mine. Scientists largely suck at statistics! Especially for experimental scientists, this is insane. Yet I cannot express how frustrating it has been to me throughout my career as an experimental scientist to see study after study proven invalid because the statistics were botched. And I’m not just talking about bad scientists; even the very best are constantly caught in this trap. Many a brilliant scientist has treated statistics as an afterthought and been burned by it. In fact, I’d bet many non-scientists can name multiple times that some big breakthrough was announced in the media, only to fade away into nothingness. Virtually always this was due to the responsible scientists not understanding their statistics.
This article delves into possible causes for this. I myself was drawn into the fray a few years back, to try and stem the tide of poor statistics in my own field, nuclear physics. Yes, you heard that right: even nuclear physicists constantly botch statistics. I was asked by a government agency to write a paper concerning the right way to handle nuclear statistics, and the result was so often requested that I eventually posted it on this website for easy access.
My hope is that this problem will be taken more and more seriously, so that a larger emphasis is put on statistics when training scientists. Then, maybe, my head can finally stop exploding!