0

I am running one-way ANOVA for my research data, and I'm meeting all the assumptions except Homogeneity of variances.

Am I allowed to run the Kruskall-Wallis test, as it's the non-parametric test for one-way ANOVA, or are there any other statistical methods I should use?

Hari
  • 1
  • In real life, violation of that assumption is almost inevitable. KW is always an option to use. But you have to assess how much discrepancy there is re the violation of the homogeneity of variance. – User1865345 Oct 01 '22 at 07:37
  • The question is what made you give priority to a parametric test in the first place, i.e. a test that makes many more assumptions without gaining power? And beware of hypothesis testing for assumptions, as they don't have a power of 1.0. – Frank Harrell Oct 01 '22 at 12:53

0 Answers0