Hi Brad,
If you want to test whether A is greater than the mean(B, C, D) then
your contrast will do it but note that it simplifies to [3 -1 -1
-1]. If you want to ensure that A is greater than each of B, C, and
D (at some threshold), then you can define three simple contrasts [1
-1 0 0], [1 0 -1 0], and [1 0 0 -1], threshold these at some level
and multiply the binary masks together to get the region(s) where A
is greater than B, C, and D (rather than their mean) at whatever
probability you used to define the masks. Note that if you used
cluster thresholds for the simple contrasts, then there are some
tricky issues about whether the remaining extent is meaningful
whereas the interpretation for voxel thresholds is more clear
cut. People often report the initial comparison of means and then
inclusively mask them by each of the simply contrasts so that they
can report Z-statistics for the comparison in individual areas.
Hope this helps.
- Joe
-------------
Joseph T. Devlin, Ph. D.
FMRIB Centre, Dept. of Clinical Neurology
University of Oxford
Headley Way
Oxford OX9 3DU
Phone: +44-(0)1865-222-494
Email: [log in to unmask]
|