This is a theoretical question.
I am wondering if multiplicity correction is applicable in the case of non-parametric tests.
I have performed some Mann-Whitney U rank test on two independent samples using this scipy function.
I was then thinking to apply fdr correction for multiplicity correction.
Does multiplicity correction makes sense in the case of non-parametric tests?
Best Answer
Corrections for multiple testing do not care what kinds of tests you run. When you reject a null hypothesis, you are making a decision, and corrections for multiple testing helps control the error rate of that decision.
Use it for parametric tests; use it for nonparametric tests; use it if you mix the two. The control has to do with the resulting decision.