2
$\begingroup$

I have two sets of data, one for males and females, and I'm analysing the neuronal spike intervals. I want to see if there's a significant difference in the mean interval between males and females. But I also want to see if there's a significant difference in the distribution of these intervals. One could imagine a scenario in which the mean is not different but the distribution is different. So my question is, is it good practice to use both the Mann-whitney to analyse the means and also KS test for the distributions on the same dataset? Thank you.

$\endgroup$
0
$\begingroup$

Notice that, if you show that the means are different,$^{\dagger}$ you've already shown that the distributions are different. Even so, since you planned to do two tests and simply stopped after the first test gave a desirable result, you would have to adjust for two p-values as if you calculated both p-values. If you do the adjustment, then I see no harm in doing both tests.

$^{\dagger}$That's not quite what Wilcoxon-Mann-Whitney U shows; it's actually a test of distribution equality that has particularly good properties for looking for mean differences.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.