 |
 |
Hi, I'm reposting my message from last week... haven't gotten a response and didn't want it to get overlooked! Thank you and happy holidays! I was reading some earlier posts about issues with PPI, notably the necessity of reduced power from including the task regressor, the seed region time-course, and the interaction. "In theory, PPI can be used with any experimental design. However, in practice you are unlikely to see a significant effect in most (!) experimental designs. The reason for this is that, to avoid the confounds described above, you have to include the main effects from which you derived your interaction variable in the GLM. But because the psychological, physiological and interaction variables are now strongly correlated, the design lacks power. You can see this from the covariance matrix below – we would need a signal change of over 2% associated with the interaction term to see a significant effect." I think I am having this issue with a dataset i'm working on- I don't see much significance in the interaction term. However, when I do a group analysis and look at EV2 (my seed region time-course), I do see a significant difference between groups. How can I interpret this difference? Can I say that regardless of task condition, Group A has greater overall connectivity with the seed region than Group B? Thank you, Jodi