Improving the reporting of P-values generated by randomization methods



  1. We increasingly calculate P-values using randomization approaches.
  2. Often, it is not possible to consider full evaluation of all possible samples (e.g. permutations or bootstraps) in order to get an exact P-value, so an approximate P-value is calculated across a random sample of these possible samples.
  3. We provide methods that allow the uncertainty introduced by this approximation to be compactly and clearly presented.
  4. If the recommendations proposed here are adopted, then reviewers and readers of the increasing number of publications using these approaches will be better informed.