Dear Norbert,
2009/5/14 Norbert <[log in to unmask]>:
> Dear Vladimir,
>
> Just a short question regarding the contrasts in SPM t-tests for EEG data
>
> If I would like to test the hypotheshis-1:
> that females have higher activation than males; than a t-test assigning to
> the female images a +1 and male images a -1 would be correct?
>
> The opposite assumption (hypothsis-2):
> would be that females have lower activation than males resulting in a -1
> (female) +1 (male) t-test would be correct?
>
Yes, that is correct.
> Because in some brain areas a female>male relation of activity; and in
> others the opposite relation (female<male) may exists simultaneously, it
> would be necessary to test both options - is this OK and not violating the
> multiple comparison probability level, so that an additional Bonferroni
> correction is needed?
>
If you want to be rigorous and don't have any prior hypothesis you
should use Bonferroni corrections between those two tests, basically
divide your threshold by 2. Alternatively you can use an F-test that
will test for both directions together and then just look at the
t-values at significant peaks to see which way they go.
Best,
Vladimir
|