Dear Yoram,
> I have a SPECT study with 34 patients and 2 conditions per patients and 1
> covariate.
> I want to find the regions where there is positive corelation between the
> rise in blood flow from the first scan to the second scan with the
> covariate.
> I centered the covariate around 0 and used a new covariate of +a/2,-a/2 as
> a new covariiate as recommended by Andrew Holmes.
>
> Could anyone please explain to me what would the difference be in this
> case, if I use the "Multi subject covariate only" design or a
> "Multi subject condition and covariate design" and use a [0 0 1] contrast.
If you use 'Multi subject condition and covariate design', the model
expresses your assumption that each series of observations in a voxel
(over subjects) can be explained by subject effects, condition effects
(which are the same for all subjects) and by your covariate.
If you choose 'Multi subject covariate only', you express your belief
that there is no need to model a condition effect, but that your
covariate alone times the estimated slope is a good explanation for your
observations.
So the difference between the two models is that in the first you model
some additive condition effect commonly observed over all subjects.
Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-(0)20-7833-7478
FAX : -7813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|