Dear Steven,
> I need some assitance with the SPM99 statistics section. I have an
FDG-PET
> study with 2 scans per subject.
...
> I do not understand why proportional scaliing is not an option under
paired
> t-tests. Is proportional scaling implicit in the paired t-test design ?
If so,
> then why does the Design display indicate no proportional scaling. Is the
lack
> of proportional scaling also true for the other designs under Basic
Models.
>
> I think I need a simple explanation for the relative usage of the
PET/SPECT
> models vs. Basic Models.
There is a simple explanation... I think the primary use of 'Basic
models' is for second-level analyses. There you don't need a
proportional scaling/global normalization, because you would already
have taken care of that in the first-level analysis. For a first-level
analysis of PET data, the best option (as you did) is to choose from
PET/SPECT models.
Furthermore, if you have 2 scans per subject and choose 'Multi-subject -
Conditions and Covariates' in PET models (1 group, 2 scans per subject),
the resulting general linear model with respect to the design matrix
should be identical to the model you get from the 'Basic models/paired
t-test'.
Best wishes, Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-171-833-7478
FAX : -813-1420
email: [log in to unmask]
|