Hello everyone,
I have been working the past weeks on a fMRI-dataset using SPM 12 and have used the wisdom of this mailing archive extensively. So let me say thank you first of all!:-)
Now to my current problem. My data is preprocessed and the contrasts for the first level have been set and I have already tried some second level analysis using the t-Tests, the Full Factorial Model and the Flexibel Factorial Model. I think the last is most appropriate for my experimental design, but I am encountering an unforeseen problem with the second level contrasts. My experimental design is a 2(group)x2(first within-subject factor)x2(second within-subject factor) design. I have been following the advice from Glascher and Gitelman (2008) on contrast weighting and everything seemed to work out fine. But when I am looking at the group main effect the whole brain seems to be more or less one huge significant cluster (approximately 75 000 voxels). That seems not very plausible to me and when I look at the same contrast with the Full Factorial Model only a very small cluster reaches significance at the cluster level. As I said I think the Flexible Factorial is probably more appropriate with my data, but these results seem very strange. Does anyone have an idea what has happened here?
The following main and interaction terms were included in the model:
(1) Main Effect of Group
(2) Main Effect of First Within Factor
(3) Main Effect of Second Within Factor
(4) Interaction between Group and First Within Factor
My contrast for the group main effect looks like this:
Main Effect of Group (MEg): 1 -1
Main Effect of First Within Factor (Mec1) : 0 0
Main Effect of Second Within Factor (Mec2) : 0 0
Interaction between Group and First Within Factor (INT): ones(1,2)/2 -ones(1,2)/2
Subjects of Group 1 (G1): ones(1,18)/18
Subjects of Group 2 (G2): -ones(1,20)/20
So the contrast looks like this:
MEg Mec1 Mec2 INT G1 G2
or alternatively:
1 -1 zeros(1,4) ones(1,2)/2 -ones(1,2)/2 ones(1,18)/18 -ones(1,20)/20
Is this correct?
Any advice would be greatly appreciated!
Stay safe and positive!
Paul
|