The contrast (-1 1 1 1) checks whether the sum of EVs 2-4 minus EV 1 is
significantly greater than zero. This would be a rather unusual
contrast, and I doubt it is what you really want to test.
If you were using QDEC in Freesurfer then I'm almost certain it didn't
generate a (-1 1 1 1) contrast as one of its automatic hypotheses
either.
cheers,
-MH
On Wed, 2010-05-05 at 05:39 +0100, Jay Ives wrote:
> Thanks Matthew,
>
> It did run after removing the first contrast as you suggested. I was surprised with the results however.
>
> The (original) second contrast (-1 1 1 1) showed complete saturation of significance (bright yellow) over the whole brain.
>
> The other contrasts showed very little. This might have been the true result, however I have also run the same data through Freesurfer which showed several significant clusters of cortical thickness difference, especially with the last 2 contrasts.
>
> Any comment on this?
>
> Jay
|