Dear all,
I encountered a weird situation in my research. I have 3 groups, they are all non-parametric and no equal variance as well. I ran a Kruskal-wallis test to compare the difference and my results are significant, p Group 3;
In Jamovi, only DSCF (Dwass-Steel-Critchlow-Fligner test) showed me Group 1
See: https://www.reddit.com/r/AskStatistics/comments/1ak65ub/nonparametric_anova_post_hoc_test_dunns_or/
You can get this for a few reasons One is lack of transitivity. Here is the handout I give my class (I have made intransitive dice as a class prop ... cheaper to buy blank dice and a pen).
Daniel Wright , I suspect it's just that the different functions are testing the groups in different orders, A-B vs. B-A. That's why one function shows a negative difference and one shows a positive difference. But who can say without a, say it together now, minimal reproducible example.