proc surveyreg and post-hoc comparisons

[cc]

I have previously used proc glm to compare means of 2+ groups (1-way anova) and then requested a multiple comparisons test (sheffe's or tukeys) for pairwise comparisons. However, the current data I am analyzing requires the use of proc surveyreg (sample survey data) and I'm unsure how to go about getting the pairwise comparisons I need (there are 3 levels to the variable)--is there a similar option, or how do I go about testing which pairwise comparisons are significant, if I have an overall significant F test?

Thanks very much in advance.

[SteveDenham]

Take a look at PROC MULTTEST, especially at the example adjusting p values.

Steve Denham