Cheating with statistics

Huge quantities of medical papers are published every day in biomedical journals, but unfortunately, a high proportion of them have methodological errors that may question the validity of their results. These errors are usually due to the lack of knowledge about methodology by the authors, who are primarily clinical physicians, and the lack of adequately trained reviewers in scientific journals. Also, sometimes mistakes can be made deliberately to favor obtaining certain conclusions, as in the cases with conflict of interest. In this article we review the most common mistakes that can be observed in the use of statistical tests, either for lack of training of the authors, or to mask the data to show the desired conclusions.