Can you estimate nested random effects in PROC MIXED?

[Melk]

I have some nested data on a surgical intervention on patients with multiple fingers - some with both limbs. So I have a random effect of finger within limb within patient in a mixed model. My question is, if I want to produce an estimate for say just the "thumbs" - could I use an estimate statement for that? 

[Rick_SAS]

Sure. If you use the simulated nested data in the PROC NLMIXED documentation,

the MIXED code might look like this:

proc mixed data = nested method=ML;
   class A B;
   model resp = / s;
   random a b(a);
run;

[Melk]

So lets say I have 2 levels for hand and 5 levels for finger, I could use:
estimate a 0 1 b(a) ...
to get the estimated mean Y for a specific finger for both hands?