[running SPM5 with fMRI functions]
Colleagues have used a 2nd Level factorial model for an ASL study, with
"classical" estimation. This provides automatic contrasts with "Main effect"
and "Interactions" F-tests and "Positive effect" T-tests.
This looks like it would be useful for our parallel PET study as SPM PET
doesn't set any automatic contrasts. I have run it and it gives the same
results but...
Why are only contrasts for one-side of the T-tests produced?
E.G. We have a 2x2 factorial model - Age & Disease. There is a main effect
of Age so why aren't both the contrasts automatically produced i.e. [-1 1 -1
1] as well as [1 -1 1 -1]? Is there a cunning way to explore the negative?
Has SPM worked out which side is the best?
Is there a reason SPM PET doesn't produce these automatic contrasts?
Or why SPM fMRI produces only these "positive" effects?
I understand that a traditional Anova analysis would be to do the F-tests
then use only relevant T-tests for post-hoc analysis.
Regards,
Joel
|