Dear SPM-experts,
I use SPM5 for a VBM-study, looking at WM-volume differences between 21 patients and 21 controls
with the following design:
PET / two-sample t-test
Independence: yes
Variance: unequal
Grand mean scaling: no
Ancova: yes
1 Covariate: white matter integrals of the 42 images (Centering: overall mean)
Absolute threshold masking 0.05
Implicit mask: yes
Global calculation: mean
Global normalisation: no overall grand mean scaling; no normalisation
(For the global calculation, I first tried to choose "omit" and use WM integrals instead. However, SPM5 overrides this omission and includes global means.)
Now I have two questions:
1.) Is this design o.k. ?
2.) Does it still make sense to use integrals as a covariate for VBM in SPM5 (as it was frequently done in SPM2-VBM-studies)?
Looking at the Design (see attached bmp-file), the information in the integrals (third column) may be included in the two columns for the globals (two right column), and thus might not only be unnecessary but also make me lose one df.
Or should I use another design instead that allows me to omit the global calculation and use the integrals?
I'd be very grateful for any help,
many thanks and best wishes
Nicolas
|