Non Parametric Confidence Intervals more than 2 groups

[MikeAAC]

Hi all ,

I have a population of 3 variables  rate1, rate2, group. The variable group has 3 levels (group1, group2, group3) and the distributions of the rates are not  normal. How in SAS, can I determine which group has higher means vs another group? I would like for example to build a non parametric confidence intervals between different groups for each rate: CI (group1 - group2), CI (group1 - group3) and CI(group2 - group3). Thanks for the help.  

[PGStats]

The expression CI(Group1-Group2) can mean many different things.

Have you considered transforming rate1 and rate2 to normality (within groups)? Or finding what distribution the rates may be from? GLIMs (proc genmod, proc glimmix) can be fitted to many distributions and provide you with CIs for group mean differences.

[MikeAAC]

Thanks PG