ANOVA or Kruskall Wallace Test?

I am looking at 3 groups of data and not sure whether to use an ANOVA or Kruskall Wallace Test to compare them. 2 sets of the data have normal distribution according to the Shapiro-Wilks test and one is just significantly skewed at P=0.04. Does this mean that I need to use Kruskall Wallace when examining this one group, using it for all 3 groups, or is the skew so slight that I can just use ANOVA? Also, if I use the Kruskall Wallace is a Tukey post-hoc test still appropriate to use. Not sure what to do! Thanks for any help :slight_smile:

Welcome to the community!

How are the variances? Can we assume homoscedasticity? That is because KW also needs equal shapes, or at least to comparable level.

If you use Kruskal Wallis, I don't think Tukey is appropriate. A non parametric test should be used. I'd have suggested Wilcoxon, but as per Google and Stack Overflow, Dunn Test is the most standard post hoc after Kruskal. I actually haven't used this test myself, so can't say much.

Fab thank you! Dunn Test was what I needed to do just needed to be pointed in the right direction.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.