Dear all, I am using the Kruskal-Wallis test in R (kruskal.test()) to compare non-normally distributed observations for 5 different groups. I now want to perform multiple comparisons to identify the groups with significant differences in the mean ranks. On searching the forum I found a number of suggestions of approaches to try and decided to use kruskalmc() from the pgirmess package. However, I appear to get contradicting results as I have a significant p-value of 0.0007 from the Kruskal-Wallis test indicating a difference between at least two groups but no apparent differences in the 5 groups. Why is this the case? Apologies for my ignorance if there is an obvious answer to this question! Cheers, Karen -- View this message in context: http://www.nabble.com/Kruskal-Wallis-test-and-kruskalmc-tp26094625p26094625.html Sent from the R help mailing list archive at Nabble.com.