Thank you for offering to help. There are attached examples for 3 groups. The first model is the factor model. It has both sessions so all data is included in this model and the F test is for time. The second model is for the subtraction image (post scan - pre scan is the input) and is looking for a positive or negative correlation with score across all groups. I find that the first model does not return anything in region X but the second model does return results in region X. I am taking this to mean that the change is region X is correlated with memory score. Why do I not see a change in region X in the first model, when the second model makes it look like there is?
Thanks a lot. Seasons greetings.