Hi,
I've started playing around with the Bayesian estimation option at the first
level. During the estimation, I get a warning: "Matrix is close to singular"
from spm_vb_w. What does this indicate and should I worry? It doesn't happen
in all the subjects, but then I use a different design matrix for each
subject. It doesn't happen however if I use classical estimation.
All the best,
Hubert
|