Dear Rachael,
>Dear Narender, Richard-
>aha yes!! I was using F contrasts rather than t. Suddenly these
>contrast maps look much more like what I expected them to... How do
>the F and t contrasts differ? (I can also look this one up on the
>archives..)
There's a whole lot of stuff to F contrasts and how to use them (only
some of which I understand). But briefly, in a situation like yours
the F contrast is the t contrast squared. Hence it doesn't matter
whether you use +1 or -1, the result is the same.
In general, F contrasts test whether the covariates that you specify
contribute significantly to modelling the variance, regardless of the
sign of the parameter estimate. Your contrast 0 1 (or 0 -1) was
pulling out all of the voxels within which the second covariate
explained a significant amount of the variance in the data.
>thanks!!
You are more than welcome.
Best wishes,
Richard.
--
from: Dr Richard Perry,
Clinical Lecturer, Wellcome Department of Cognitive Neurology, Darwin
Building, University College London, Gower Street, London WC1E 6BT.
Tel: 0207 679 2187; e mail: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|