Stop Reproducing the Reproducibility Crisis
Christophe Bernard
Abstract
Crisis?What crisis?Is the lack of reproducibility a supertramp that disperses widely across science fields?Is it carelessness, or is it also intrinsic to life sciences as I shall argue?Lack of reproducibility is a big flag brandished by many, but what do they mean by it?Where does it originate from?These were the first questions that crossed my mind as, while in Lige, Belgium for a PhD defense, I was asked to give a talk on the reproducibility crisis.Rather than a formal presentation, we had an open discussion.This editorial is its outcome.According to Wikipedia's definition of reproducibility, "results obtained by an experiment or an observational study or in a statistical analysis of a data set should be achieved again with a high degree of reliability when the study is replicated" (https://en.wikipedia.org/wiki/Reproducibility).The definition is thus very strict: the results must be replicable, which means that the same experimental conditions or data analysis method must be used.Finding similar results with different approaches is not replicability stricto sensu; it just gives more weight to a study and equates to generalizability.Most often there is a confusion between reproducibility of the conclusions and reproducibility of results.