Biologist Joachim Goedhart of the Swammerdam Institute for Life Sciences questions the habitual use of P-values in science. Goedhart has written a ‘Correspondence’ on this subject that appeared in Nature, entitled 'Dispense with redundant P values'.
P-values appear very often in figures of scientific papers in the life sciences. Statistics is routinely used to facilitate the interpretation of experimental results. The statistical tests often return a p-value. The exact meaning of p-values, however, is difficult to grasp. As a consequence, p-values are often misused and misinterpreted. Notwithstanding these issues, the calculation and presentation of p-values in scientific reports is a common practice. The proper use of p-values is a topic of intense and continuous discussion.
In many instances, the differences between conditions (e.g. a treated group versus a control group) is immediately clear from the data. Still, p-values are presented, since 'everybody does it'. Dr Goedhart argues that this parroting of p-value reporting results in a widespread use of p-values that have no added value. Especially when effects are large and obvious, p-values should be omitted. This will reduce the use of meaningless p-values and it will return the attention to the actual observations.
On the occasion of his Nature correspondence, Goedhart has posted a blog on the subject.