Hi,
> I am using the marsbar toolbox to extract contrast values for my ROI's. I
> noticed something interesting about the output in the MATLAB window.
>
> It seems when I have a positive contrast value of say 1.99 the p-value
> associated with this contrast is acorrdingly p = .0001. However, when the
> contrast value is of equal significance in the negative direction, say
> -1.99, the p-value does not adjust accordingly. I have a p-value =.95.
>
> Can any please explain why spm marsbar toolbox does not take into account
> the contrast value is significant even though it is in a negative direction.
I've Cc'ed the marsbar list - it's the best place for marsbar questions.
The t test in SPM and in marsbar, by convention, is a one-tailed
t-test - that is, you are only testing for significance in a
particular direction (positive). For a two tailed test, you would
have to double the p value (it's twice as likely you'll find something
that is (> thresh or < -thresh), than something that is just (>
thresh)). If you want a two-tailed test, you can do an F test with the
same contrast weights. Does that answer your question?
Best,
Matthew
|