THE MISUSE OF STATISTICS AMONG BIOLOGISTS

This paper deals with some common misunderstandings about the statistical evaluation of data by biologists. Neglecting assumptions, twisting models, misusing GLM and GLMs, non-representativeness of samples and the use of exoteric-looking statistical manipulation of data to demonstrate obvious results are relevant examples. Often, the “best” methods are the simplest ones, as long as they are appropriate. One should never forget that, to a biologist, the statistical manipulation of data should just be a way to evaluate the significance of his conclusions: it ought to never become his main goal, e.g. because statisticians can do it better.