Greetings Folks,
I am running some non-parametric tests on some endocrinology data that violated the assumptions of even distributions severely for a typical ANOVA. Thus I turned to a Kruskal-Wallis One-way Analysis of Variance.
I performed the analysis in SYSTAT 13.0, and have been given some output that I am having trouble interpreting given my lack of familiarity in non-parametric methods. Systat provides the option for two pairwise comparison tests to accompany the K-W test.
The Dwass-Steel-Chritchlow-Fligner, and the Conover-Inman.
Are these to simply be interpreted like a typical pairwise comparison by the accompanied P-value? Some of the category comparisons don't even register p values which also has me questioning the proper interpretation (though this might stem from one of my categories having sample size of 1).
Thank you for any help you may provide!
Yes, they can be interpreted like pairwise tests. And, yes, having a sample size of 1 is likely to give you funky results for at least some comparisons.
I'm familiar with the Conover-Inman test, and not familiar with the other. The Conover-Inman test is probably less common for this purpose than the Dunn test (1964), but is relatively common.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude