Dear SPM:
I noticed that the DCM software does not save any information related to
convergence. Thus unless one writes this down there is no way to know later
whether a DCM happened to converge or not. My understanding is that
convergence provides some indication of identification of a model. Thus
wouldn't it be the case that if a DCM didn't converge better than let's
say 1e00 it would be suspect?
Darren
|