Hello spm experts,

 

I have a question about the way a PPI design is set. If I have two tasks A & B (P), VOI timeseries (Y) and while setting up the PPI, the contrast I used was P = +1 for A and -1 for B. Then as usual, I enter these 3 outputs - PPI, P & Y in a GLM. My question is that the results I see for a positive PPI interaction are 1. the voxels which have increased connectivity with the VOI for task A greater than task B or 2. voxels which have increased connectivity with the VOI because of differences between tasks A & B.

To make it less confusing, does specifying P = -1 for A and +1 for B, be the same as P = +1 for A and -1 for B? Is negative PPI interaction for the case above, same as P = -1 for A and +1 for B?

 

Second question: I have run PPI analyses on two groups of subjects. Inorder to compare how the PPI interaction differs between groups, I would simply enter the con*.images into a second level analysis - 2 sample t-tests - with +1 giving me greater functional connectivity for group1 than 2 and -1 for group2>group1. Is this valid or an ANOVA on individual PPI analyses done for each task for each group ( i.e. 2 tasks PPI done for each group) would be a better approach? 

 

Thanks in advance,

Jane