Hi,
I would like to use bayesian methods (because I've had problems with the estimation procedure using SPM5 classical methods so far) to analyse two groups on a cognitive task.
It seems ok to use bayesian stats for single subjects with the use of spatial priors and use un-smoothed images for the analysis. This makes it better I believe for anatomical localisation and for further DCM analysis.
But for group analysis, one would need to smooth the data to take into account differences in anatomy. Does this then defeat the point of using bayesian stats, or weaken the benefit of the spatial prior and therefore is it still better?
Nick
|