Dear Jack,
> Karl mentions many times the possibility to correct for a space
> constrained hypothesis. Could it apply, and how if it can to the
> following case : Lets say we have a the result of a categorical
> design. Can it be used as a spatially constrained model for a factorial
> design with the question of what parameters influence the activation of
> one area or the other ? In a way it would be inclusive masking but with
> correction because of the limited volume explored. If it sounds
> reasonable, can it be done differently than by setting each regions
> manually using SPM99b ? Thanks for advises
Yes. Indeed that is the motivation for having the small volume
correction option accept a cluster as the volume definition. It would
typically be used in a factorail design to test for a main effect [or
an interaction] in a cluster defined by the other [orthogonal] main
effect. To do this the contrast about which you want to make an
inference is thresholded at a very low level (to get all voxels) and is
masked with the volume-defining contrast.
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|