Date:         Tue, 13 May 2003 21:11:58 +0000
Reply-To:     mbabyakspamaway@DUKE.EDU
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         mbabyakspamaway@DUKE.EDU
Organization: Duke University, Durham, NC, USA
Subject:      ordinal response in genmid

I'm trying to fit a predictive model with a response variable that
contains three ordered categories.  One of my choices would be to use the
multinomial option in genmod, however it isn't clear to me whether this
model requires the proportional odds assumption to be met.  Can anyone
clarify this or point me to some relevant literature?

Thanks,

Mike Babyak