|
Sounds like a hierarchical linear model. A nice tutorial on how to do
these analyses using SAS PROC MIXED is in Statistics in Medicine
18:855-888.
FWIW,
Jay.
On 21 Apr 00 15:02:55 GMT, klemke@JHSPH.EDU (Klaus Lemke) wrote:
>Greetings,
>
>I used proc genmod with a repeated statement to account for clustering at a
>single level. Now i like to know how to expand this model to account for
>clustering at two levels. Can this be done with genmod and the repeated
>statement?
>
>In my example, patients are clustered within providers who in turn are
>assigned to clinics. My dependent variable is whether or not a patient
>self-referred (as opposed to being MD-referred).
>
>proc genmod;
>class provider;
>model selfref=... /d=binomial;
>repeated subject=provider /type=exch;
>run;
>
>My question is how to account for the correlation of providers within clinics.
>
>Klaus
| 
