Print

Print
Hi Navot,

Please see below:


On 19 February 2018 at 17:03, Navot Naor <[log in to unmask]> wrote:
Hi all,

I have a question about the way to design the group level in PPI.
In the example at the PPIHowToRun wiki there are 3 EV's PSY, PHYS and the interaction.

At the end of the manual it says "This completes the set-up for the first-level analysis. You can then compare between groups at the second level as normal"
But what kind of group level design, as there are no contrasts to do here? So do i jut average the interaction EV for all the participants?

Any contrast that is relevant to you at the 2nd level. The simplest could be the average (1 sample t-test), but could as well be a group comparison or correlation with some continuous variable.
 

And if i have more EVs, Say PSY1, PSY2 and PHYS.

Yes. However, more EVs mean more possibilities for interactions. For example PSY1*PSY2, and PSY1*PSY2*PHYS, etc.
 
Can i do interaction between PSY1 and PHYS, interaction between PSY2 and PHYS and then compere the interactions in group level? can this be done in a paire-t design?

Yes and no... as far as the stats are concerned, yes, it's possible. However, spatial variability of the BOLD response may mean that different regions of the brain could yield signals that aren't comparable. I would avoid this contrast. A three-way interaction (that has a different meaning) would be fine, though.

All the best,

Anderson

 

Thanks,

Navot