Date:         Tue, 13 Nov 2007 14:52:58 -0500
Reply-To:     Sudip Chatterjee <sudip.memphis@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Sudip Chatterjee <sudip.memphis@GMAIL.COM>
Subject:      GLIMMIX DOUBT

Dear All

 I am new to multilevel Modeling and GLIMMIX. I am trying to model infant
mortality.I consider mothers are nested into ZIPCODES. I have 6 years data
with census information and crime information of every zipcodes. I am
trying to build a 3 level model keeping poverty as level-3 ( as this
remain constant for all 6 years in all the ZIPCODES) then crime as level-2
( Chages with years) and all maternal information as level-1.

My model looks like this :

 proc glimmix data = Mylib.model abspconv=0.001 ;
nloptions maxiter = 100  ;
class ZIP cluster LBW RACE NEWMOMED status FAC_TYPE;
model IMR(event='1') = poverty Crime_num NLBW RACE NEWMOMED FAC_TYPE
poverty*Crime_num Crime_num*LBW Crime_num*poverty/solution dist = binary
link = logit  ;
random intercept LBW /subject = ZIP solution  ;
random intercept LBW / subject = cluster (ZIP) solution ;
run ;

I am not sure of this model. Any Suggestion will be appreciated.


 Thanks.