I want to do multiple comparisons of medians using kruskal wallis test. How to do?

[MaryA_Marion]

Here is what I have tried:

PROC NPAR1WAY DATA=trout4 wilcoxon median;
CLASS dose;
VAR hemoglobin;
RUN; quit; title;

It is not giving me individual contrasts? 

[StatDave]

Always check the list of Frequently Asked-for Statistics (FASTats) in the Important Links section of the Statistical Procedures Community page. See the "Nonparametric methods" item which mentions the DSCF option.

[MaryA_Marion · Accepted solution]

Remaining questions have been:

I have run a Kruskal Wallis test using proc npar1way.  Pairwise two-sided multiple comparison analysis is output.

What is the alpha significance level here?  Are we using alpha/12 in case of 4 four dose groups to obtain P>DSCF?

What if I wanted to run a one-tailed test? How would sas syntax change?

PROC NPAR1WAY DATA=trout4 wilcoxon dscf;
CLASS dose;
VAR hemoglobin;
RUN; quit; title;

To interested readers, checking out this reference gave the answers.

Juneau, P. (2004), "Simultaneous Nonparametric Inference in a One-Way Layout Using the SAS System," Proceedings of the PharmaSUG 2004 Annual Conference, Paper SP04.

[MaryA_Marion]

Ok. I found dcsf.  

dcsf requests a pairwise multiple comparison analysis and specifically Wilcoxon multiple comparisons. Are these the same?

I am working with Gibbons Nonparametric Statistical Inference equality of k distributions in nonparametric section of the text. See eq 10.4.8.