Dear SPM stats experts,
I've got a couple of questions on Bayes stats.
.
1. in SPM, the book (chp 22), one can read 'A 3rd prblm is the
correction or adjustment applied to the p-values to resolve the multiple
comparisons prblm' ; does that mean that once I estimated the parameters
and used them as priors (empirical Bayes) I do not need to correct? or
that the correction does not depends anymore on the search volume (FWE
but not FDR?) ? I must say this confuses me ...
2. Also if I have an effect testing H0 and the same after Bayes estimate
I can be confident, right? what about if I'm interested in the absence
of effect? testing H0 in a VBM study I didn't get any effect (and happy
about it), now if I use empirical Bayes and still get nothing, can I
then conclude that there is no effect (whereas before I couldn't really
conclude)
Thanks for your help
Cyril
|