各位同仁，近来看到国外网友的一则提问，不解，请各位同仁指教！多谢！
The following is SAS output when I fit a random effects model.
All factors are random but I don't understand the covariance estimates.
How come they could be zero??
It is impossible that there is no variance between subjects or operators....
The SAS statement that I used is
proc mixed data=mixed method=reml covtest;
by type; class subject operator image;
model L=; random subject operator image(operator);
run;
Anyone ever came across such a situation?
Convergence criteria met.

Covariance Parameter Estimates
          Standard      Z
Cov Parm  Estimate Error Value Pr Z
subject   0         .      .    .
operator  0         .      .    .
image(operator) 0.000293 0.000301 0.97 0.1650
Residual 0.005291 0.000617 8.58 <.0001