Hello
I am new to hierarchical models and will be grateful for the tips.
Here is some background info.
The study question is: Is there a difference between and within hospitals in number of referrals to certain social services?
Referral is coded as YES/NO (outcome). Individual characteristics are either kept as continuous or as yes/no, similar with hospital characteristics. There are 53 hospitals grouped into 3 levels.
I am running PROC GLIMMIX using the following code:
/*Random effect of hospital*/
PROC GLIMMIX Data=
CLASS Hosp_Disch;
MODEL referral (DESCENDING)= WGTGRMS Sex /dist=binary link=logit ddfm=bw solution;
random intercept/subject=Hosp_Disch type=cs G;
random _residual_/subject=Hosp_Disch ;
Title 'GLIMMIX: Birth weight continuous plus NEWPLURAL and random effect of hospitals';
/*Random and fixed effect of hospital*/
PROC GLIMMIX Data=;
CLASS Hosp_Disch Hosp_Level;
MODEL referral (DESCENDING)= WGTGRMS NEWPLURAL Hosp_Level/dist=binary link=logit ddfm=bw solution;
random intercept/subject=Hosp_Disch type=cs G;
random _residual_/subject=Hosp_Disch ;
Title 'GLIMMIX: Birth weight continous plus NEWPLURAL and random/fixed effect of hospitals';
Results from these two codes look exactly the same, and I do not have p-value for hospital variance. What am I missing?
Gabi
------------------------------
Gabriela Kernan
Epidemiologist
University of Massachussets Lowell
------------------------------