Dear SPM-ers,
Last week I posted the message below. Unfortunately nobody responded.
Perhaps it is a very silly question, but nevertheless I'm still trying to
figure this out. I hope somebody is able to help me.
-------
By including covariates (and so more parameters for the linear regression
model) in our PET study, we try to better model the variance in the data and
'clean up' our condition effect. Looking at the F map, voxels can be found
in which variance can be explained by the covariate included.
We would like to find how much extra variance is explained by including a
covariate. If a covariate does not contribute much explained variance then
we would like to exclude it, to increase the degrees of freedom for the
statistics. Is there a simple way to find the explained variance of the
model (total and/or per parameter) or in which matlab variables can we find
it? Is there an indication when we can eliminate a covariate?
Thanks a lot for any advice/help given!
Simone.
--
A.A.T.S. Reinders
University Hospital of Groningen
The Netherlands
|