I am a bit confused about Dunn's Multiple Comparisons test. My data are non-normal, and I'm comparing the sizes of fish in different years. Here's what I'm not sure about:
1) What is compared? Dunn (1964) shows it as a way of comparing means. However, I've seen it stated (inc. on CrossValidated and Stata) that it is for medians, basically a Mann-Whitney for more than two sets of observations.
2) How do you report the results of Dunn's?
Thank you,
Best Answer
Dunne's test compares the Z-test which is comparison of the means when the variances are known and the sample size is large. In addition to the pairwise comparison, it gives you P-value that tells you if you would reject the null hypothesis at a particular significant level or not.