Hi there,
>
> Dear SPM-ers,
> I would like to set up an analysis using VBM.
>
> I have two groups (patient vs. control) and use their total brain
> volume (TBV) as described in different sources.
> I have entered TVB as a confounding covariate, so I have three columns
> in design matrix (Patient, Contol, TBV).
>
> Which t-contrast should I use to take account for TBV? [-1 1 1] or [-1
> 1 -1] for testing whether Patients have atrophy?
>
For atrophy -1 1 0 ; the effects of the 2 columns, patients and
controls, are adjusted for the TBV, i.e. their part of variance is not
correlated to the TVB - this last column models this confound but is of
no use in the analysis ..
Cyril
|