Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!psuvax1!psuvm!umab.bitnet!rao
From: RAO@UMAB.BITNET (Malla Rao)
Newsgroups: bit.listserv.sas-l
Subject: Conditional logits.
Message-ID: <SAS-L.90011011113507@VTVM1.BITNET>
Date: 10 Jan 90 15:08:35 GMT
Sender: "SAS(r) Discussion" <SAS-L@OHSTVMA>
Reply-To: Malla Rao <RAO@UMAB>
Lines: 12
Approved: Since they are moderated

From: Malla Rao

Does any one have any experience with PROC MCSTRAT in the SAS Supplemental
Library? It does conditional logistic regression.
I tried a model with very closely matched clusters and got coefficient
estimates that seemed implausible. An unconditional model using PROC LOGIST
with the matching variable entered directly as one of the independant variables
provided coefficients which were way off. I do not think it could be residual
confounding that could attribute the difference.
Did any of you get erroneous results with this PROC??

---  Malla Rao  -----