Another alternative would be to use PROC NPAR1WAY to perform a Kruskal-Wallis test on all three groups for each outcome. Then only perform pairwise tests when the K-W test reaches statistical significance (5% ?).

This all assumes that it is not possible to apply transformations to the original data to approximate Normal distributions and homoscedasticity and then apply ANOVA/GLM.

Either way I feel it is better to report all of you findings and let others interpret them in the light of you analysis (don't just report the statistically significant ones).

Brad

bm18@le.ac.uk

-----Original Message----- From: Christian F.G. Schendera [mailto:schendera@NIKOCITY.DE] Sent: 21 February 2001 10:26 To: SAS-L@LISTSERV.UGA.EDU Subject: multiple comparisons of not normally distributed data

Hi, Data situation: 3 independent groups, several continuous dependent/response vars. Continuous vars are not normal distributed (only a third of the data reach ShapiroWilks >0.1). Problem: Collegues compared the three groups pairwise with simple Mann Whitneys at alpha 0.05. Journal reviewers criticized this proceeding for not having used p-adjusting procedures like Bonferroni. Question: Are reviewers right? How could one apply p-adjusting procedures when conditions for ANOVA/GLM are not met? Could MULTTEST be used to perform multiple comparisons on the described data? Or adjust the p in the pairwise comparisons? Whta would you recommend in this situation? Thanks in advance, Chris