I am a bit puzzled because you mention *3* conditions at the first level
from which you compute differential contrasts and the use of an eye(*6*)
F-contrast at the second level.
One approach would be to compute differential contrasts [1 -1 0] and [0
1 -1] at the first level, enter these images in a two sample t-test (one
way anova between subject) and use an F-contrast eye(2) at the second level.
This corresponds to the procedure described in section 5.1 of this document:
On 04/05/17 14:31, Rik Sijben wrote:
> Dear SPM experts,
> I have, what I expected to be, a simple question yet the more I dive in the archive the more uncertain I get.
> I tested an fMRI paradigm with 19 subjects whom each participated in a single session containing 3 conditions (no baseline). In order to find regions of interest for a follow up PPI analysis I want to generate an activation map (at second level) which displays regions which show differential activation between any of the conditions (the type of contrast generating the difference is not relevant).
> My initial approach of generating differential contrasts on the first level and then inserting these into a second level one way anova (followed by an eye(6) F test) turned out to be wrong since I'm not comparing different groups.
> How do I approach this and which contrasts do I set (1st/2nd level). A within subject ANOVA comes to mind but from what I've been able to figure out these don't allow for simple F tests.
> Follow up: If I were interested in differential contrasts is it prefered to make these at the first or second level?
> My apologies for the basic nature of the problem but I can't seem to find a conclusive answer as most examples investigate either different groups or multiple sessions.
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG