Dear SPMers,
For my VBM study, I would like to compare gray matter between 2 groups
in correlation with differences in neuropsychological data. I read that
the multiple regression would be a good design to look at differences
between groups with covariates.
I would like to know if I can oppose neuropsychological data with a
design like: gp1 (111000) gp2 (000111) cov1 (xxx000) cov1 (000xxx) cov2
(xxx000) cov2 (000xxx) ... then I can compare the regression coef for
cov1 between group1 & 2 with a contrast 001-1. Do you think it is right ?
In addition, I think that I have a big pblm as several covariables are
correlated. The signal is explained several times by the same underlying
variable .. my idea is to work with partial correlation. Does anybody
know how to do that in SPM or have reference for this method?
Thank you
Cyril
PS: thank you to all for your previous e-mails on how to check the
quality of my segmented images
|