I would like to compare 4 different groups for 10 different variables (the biodistribution of a material in different organs). Someone suggested me to use a Kruskal-Wallis test with Dunn's post-hoc test for each variable.

My questions are:

  1. I know that Kruskal-Wallis and Dunn's post-hoc tests have corrections for multiple comparisons, but should I use an extra correction because I would do 10 hypothesis tests for the same 4 groups (but different data)?

  2. How can I calculate the minimal group size to detect significance for this type of experiment (power analysis?)?

  3. Is there a better method of doing this?

Thank you in advance.

  • $\begingroup$ I'm afraid my comment might not suit your situation and I deleted it while rethinking. Anyway, I think it would be helpful to clarify what "10 hypothesis tests for the same 4 groups (but different data)" means. Do you perform your tests in different variables? $\endgroup$
    – Pere
    Feb 7 '17 at 10:17
  • $\begingroup$ I have also deleted my reply to your deleted comment. I'm going to measure the concentration of a material in all relevant organs of the 4 groups. I am going to consider these concentrations as different variables, but I do not know if they are independent. For example, the concentration in the liver and in the brain are two distinct variables, but if I do two statistical tests on these two variables, should these tests be considered independent?) The data, although not the same, come from the same animals and same experiment. $\endgroup$
    – David
    Feb 7 '17 at 10:58

Your Answer

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

Browse other questions tagged or ask your own question.