Dear Irina,
>I am doing PPI analysis in SPM2 and having a hard time interpreting the
>group data, i.e. individual PPI con images combined in one sample
>t-test. What does it mean when a significant PPI is demonstrated in an
>area that doesn't show up in the original contrast? All the articles
>that I have seen so far only mention PPI in the area with significant
>main effects.
There is no requirement for a PPI to be expressed in regions showing a
main effect of the psychological factor (although this can happen). The
PPI is simply an interaction. If you have a factorial design with factors
A and B, one usually chooses an area expressing the main effect of A to
provide the physiological variable. The PPI effect is approximately
AxB. This is the interaction. In short, you would expect to see PPIs
in the regions showing an interaction between A and B, if you choose
the index region based on a main effect of A.
>Also, what would be an acceptable cut-off threshold for the group PPI
>analysis?
This threshold is exactly the same for any SPM. Usually people render
SPM at p=0.001 uncorrected and then report (in the main text and tables)
maxima with a random field corrected p-value of p <0.05 adjusted for the
search volume (which may be the entire brain or some smaller volume
constrained by anatomy or an orthogonal contrast).
I hope this helps - Karl
|