Dear Stephan,
> For conjunctions and the 2nd level, would inclusive masking using
> Imcalc also be a valid option?
>
> For example, we have conducted this analysis with two contrasts
> [aversive - neutral], one using word stimuli [Word(A) - Word(N)] and
> the other using picture stimuli [Pict(A) - Pict(N)] with 14 subjects.
> The aim was to characterize regions jointly activated by aversive
> emotion regardless of stimulus modality.
>
> After conducting separate RFX analyses for word and picture data, we
> "write filtered" the two SPMS at .07 and then used Imcalc to
> inclusive-mask these images (i1.*(i2>0)), the idea being to end with a
> joint p of approx. .005, (.07)^2. This seems to follow the spirit of
> SPM99-style conjunctions.
Absolutely, and should give identical results to the approach outlined
in my previous email. The advantage of doing the formal conjunction in
SPM is that the corrected p values are calculated using GRF theory for
SPM{Tmin} (i.e. conjunction or intersection SPMs). Doing it 'by hand'
means you can only compute the uncorrected p values as you outline
above.
> Perhaps a new section of the RFX page could be added, in which FAQs
> regarding conjunctions could be referred to? Several people have told
> me they are trying to work out how to do similar things and it would be
> much appreciated.
I will see if we can do that.
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|