4-level mixed effects model with binary outcome

[AprilS]

Hi All,

Does anyone know how to model 4-level clustered data with binary outcome using proc glimmix or any other proc?

The data has 3 nested clusters like this：A(B(C(level 1 events))).

Any help is much appreciated!

[SteveDenham · Accepted solution]

This should be as simple as:

proc glimmix data=yourdata;
class A B C <fixed effects> ;
model <response> = <fixed effects>/dist=binomial;
random intercept B B*C/subject=A;
run;

Here everything in angle brackets is up for grabs, without greater detail for the design.  The RANDOM statement is written in intercept form to improve convergence properties.

Try filling in the angle brackets, and seeing how this works.

Steve Denham