Random statement in mixed model for repeated measures

[fatemeh]

Hello all,

I appreciate you to help me understand what is the difference between these two models. In the first model, in random statement trt1 and trt2 are entered, while in the second model it is empty. (these codes and data are from sas tutorial video)

proc mixed data=test;
class patient trt1 trt2 time;
model resp=trt1 trt2 trt1*trt2 time;
random int  trt1 trt2 / subject=patient;
repeated time/ subject=patient*trt1*trt2 type=ar(1);
run;

proc mixed data=test;
class patient trt1 trt2 time;
model resp=trt1 trt2 trt1*trt2 time;
random int / subject=patient;
repeated time/ subject=patient*trt1*trt2 type=ar(1);
run;

patient	trt1	trt2	time	resp
1	1	1	1	4.3
1	1	1	2	3.8
1	1	1	3	3.1
1	1	1	4	2.7
1	1	2	1	4.3
1	1	2	2	3.4
1	1	2	3	3.2
1	1	2	4	3.5
1	2	1	1	3
1	2	1	2	3.3
1	2	1	3	2.5
1	2	1	4	3.4
1	2	2	1	3.3
1	2	2	2	3.2
1	2	2	3	3
1	2	2	4	3.2

[jiltao · Accepted solution]

Model 1 essentially models two more random effects compared to model 2: trt1*patient and trt2*patient. Because each patient gets multiple trt1's and multiple trt2's, modeling these two random effects is reasonable. Both models are valid models. Which one is better for your data might be evaluated by examining the Fit Statistics table from PROC MIXED output -- the smaller the fit statistic such as AICC or BIC, the better fit of the model.

[jiltao]

I forgot to mention that you might also want to examine the log to be sure there is no message such as G matrix not positive definite, before interpreting the output.