Dear SPM community,
After looking through the archives I'm still stuck and
am looking for some advice...
I have performed an optimised VBM analysis on a
data-set of 21 controls and 9 subjects. I have then
run an ANCOVA with age and grey matter volume as
covariates, but I am unsure whether I should be
correcting for non-sphericity when doing this.
My (limited) understanding is that SPM2 does some
correction for non-sphericity when it calculates the
t-statistic for each voxel using the ReML method. So,
1. Do I need to further correct for non-sphericity
(non-identical variances)?
2. If I make this correction when running the ANCOVA
then the values in my design matrix are altered
(pre-whitening??), especially the constant which
adopts values of approx 0.8 - 1.1 and stops being
constant...what does this mean and does it matter?
Thanks for any help,
Alastair
Dr Alastair Reid
Clinical Research Fellow and SpR Psychiatry
MRC Cyclotron Unit
Hammersmith Hospital
London W12 0NN
___________________________________________________________
Yahoo! Messenger - NEW crystal clear PC to PC calling worldwide with voicemail http://uk.messenger.yahoo.com
|