Dear Christian
thank you for your answer! However, in the paired t-test (opposed to the
unpaired t-test), each predictor represents already a pair of contrasts
and therefore a comparison in itself. When I add n = 5 subjects with n =
10 contrasts, I get n = 5 predictors. Therefore, they should all be
treated equally. I don't quite understand this.
Do you know where I am wrong?
Best regards Dominik
Mohr, Christian wrote:
> Hello Dominik!
> If i understand you correctly (problem with the two-sample t-test),
> you entered ones for all contrast-images in your t-test independant of
> group-affiliation. Correct?
> This does not work because SPM asked you for a string which sorts your
> contrast-images in the two groups to compare, e.g. for 2 groups (each
> n = 5) with 2 conditions to compare [a b a b a b a b a b] or [1 2 1 2
> 1 2 1 2 1 2]...
> So you should give SPM first the contrasts (relevantstimulus1 -
> controlstimulus1), (relevantstimulus2 - controlstimulus2) for each
> subject and than the group / condition variable, e.g. [a b...].
> This should work.
>
> Hope this will help!
> Best,
> Chris
>
> Christian Mohr
> Klinik für Neurologie
>
> Universitätsklinikum Schleswig-Holstein
> Campus Lübeck
> Ratzeburger Allee 160
> 23538 Lübeck
>
> Telefon: +49 451-500-3709
> FAX: + 49 451-500-2489
> email1: [log in to unmask]
> email2: [log in to unmask]
>
> Dominik Bach schrieb:
> > Dear SPM experts
> >
> > I have a rather basic question on comparing two sessions with paired
> > t-tests. I have an event-related fMRI design with two sessions that
> use the
> > same types of stimuli but different instructions. Now I want to
> assess the
> > effect of the instruction by comparing (relevantstimulus1 -
> > controlstimulus1) versus (relevantstimulus2 - controlstimulus2). On
> a first
> > level, I have fitted models for each session separately. Now I want to
> > compare the resulting contrasts on a second level by paired t-tests.
> >
> > First, I wonder whether this is correct, because
> >
> > secondly, when I try to do so, I get a separate predictor for each
> pair of
> > scans. Now when I set all predictors to 1 (except constants), SPM
> gives an
> > error message.
> >
> > Where am I wrong?
> >
> > Thank you for your input
> > Dominik R Bach
> > University Hospital of Psychiatry
> > Berne, Switzerland
> >
|