Date:         Wed, 12 Jun 2002 19:27:20 +0200
Reply-To:     Annette <anne5432@UNI.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Annette <anne5432@UNI.DE>
Subject:      Re: linear-by-linear association in log-linear model
Comments: To: "CUI,JINRUI" <jcui@UCLA.EDU>, SASL <SAS-L@UGA.CC.UGA.EDU>
Content-Type: text/plain; charset="iso-8859-1"

You want to do two things at one time:
1) try
PROC FREQ data= YOURDATA ;
table A*B*C /cmh;
run;
CMH stands for Cochran Mantel Haenszel.

2) then
PROC CATMOD
Now model your data.
CATMOD is from the loglinear family, basing on gen. odds. Maybe an other
approach fits better.
Cheers, Annette