Date:         Tue, 5 Apr 2011 11:35:47 -0400
Reply-To:     FRP3@PITT.EDU
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         SUBSCRIBE SAS-L Ameribrit <FRP3@PITT.EDU>
Subject:      Testing treatment means in GENMOD
Content-type: text/plain; charset=iso-8859-1

HI Guys:

I am running a marginal mean model in SAS using PROC GENMOD to account for
repeated measurements taken on a part.

If the model tells me that the mean outcome differs across treatments I
would like to know which ones differ but only compared to the control.

I know I can run dunnetts test in proc glm but I have corellation between
subsequent measurements that I need to account for.

What I would like to do is to run a dunnetts test in GENMOD but I dont
know if this is possoble. So how would I only test mean differences
between a treatment and a control accounting for correlated measurements.

Can this be done using the contrast statement any ideas.
thanks
Francis