On Thu, 16 Jun 2005 11:45:09 -0400, Gang Chen <[log in to unmask]>
wrote:
Gang,
>Hi SPMers,
>
>What is the option "GMsca: grand mean scaling" exactly for in one-way
ANOVA
>within-subject design? And how about the option "Global calculation"?
I don't think GMsca has much to do with the statistical model directly.
Grand mean scaling just means "take a run (in SPM-speak, a session) of
data, take the mean over all in-cerebral voxels, and scale all voxels by
that mean" (possibly multiplying by 100 after; can't remember).
IIRC "global calculation" has to do with how the global mean is computed
(since there are various ways to attempt to ensure the mean is only over
in-cerebral voxels). What's done with that mean is another matter.
>Since I would like to test whether some of the conditions are 0's, I am
>interested in a general linear model without constant (grand mean), just
as
>the model of one-way ANOVA between-subjects (without constant). Is this
>design in SPM2 automatically modeled without constant? Or is it taken care
>of by one of the options somehow?
I don't know the answer, but the way I usually find out is to run the
model, load SPM.mat into matlab, and look at the design matrix (SPM.xX.X,
if you're looking at SPM2; xX.X for SPM99).
>I apologize if these questions were discussed before in the mail list.
>
>Many thanks,
>Gang
Best,
S
|