Dear Alex,
> I'm puzzled about this: I have a global volume effect in a PET study
> despite using global normalization of the data. My design is multi-subject
> (n=10), 2 conditions, 1 scan per condition (the scan is an average scan of
> two replications of each condition), no covariates. I use Grand Mean
> scaling to 50 and 0.8 as the threshold for masking. I tried global
> normalization using AnCOVA or proportional scaling: in both cases, SPM96-
> and SPM99- analyses, respectively, result in a global decrease (only one
> big blob at p=0,001 uncorrected) for one contrast, while there is no effect
> for the opposite contrast.
>
> I always assumed that global normalization should make this very case
> impossible. Any explanations or suggestions?
Do you mean that you are surprised that you can't see any increases with
p(peak height) < 0.001? Although I can't really come up with any
intelligent idea (maybe Jesper could?) under which configurations you
could see something as you observed, I would look at the t-map in the
display tool. Here you see the negative and positive peaks at one glance
and if there is some spatial structure, you'll see it. Displaying the
t-map and the functional or structural image using 'checkreg' might also
be quite useful.
Hope this helps a tiny bit, 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]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|