I would like to run a Kruskal Wallis test. There are 5 groups that I need to compare.

Group 1 Sample Size: 37
Group 2 Sample Size: 33
Group 3 Sample Size: 36
Group 4 Sample Size: 16
Group 5 Sample Size: 172

Im concerned with the sample size of the last group.Does it have any effect with the results? A post hoc analysis will be ran if the difference is found significant.


  • $\begingroup$ Your sample number 4 seems quite small, wouldn't that be a concern, too?.. $\endgroup$ – Richard Hardy Feb 7 '15 at 15:49
  • $\begingroup$ I thought that is okay. Anyway, what is the effect of having these small/big sample size? $\endgroup$ – dixi Feb 7 '15 at 15:51
  • $\begingroup$ I have not encountered this before. Maybe you could try simulations under the null to see how the test works for the particular sample sizes? You could start from just a couple of samples like 37 and 172 or 16 and 172 if dimensionality is a problem. $\endgroup$ – Richard Hardy Feb 7 '15 at 15:59

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.