Hi everybody!
I feel that this is a rather basic question, but I cannot really find out a good answer.
Let's say that I have acquired a task-related fMRI in two different samples before and after a certain treatment. One of the sample is having a treatment and the other is having a different one.
My main hypotheses are:
1) the two treatments lead to different modifications of the cerebral activation associated with the task
2) these modifications are related to a certain behavioral index measured at time 2
I can see two ways of testing this:
1) For each subject, create a first-level comprising fMRI at time one AND fMRI at time two, and then create a contrast (Experimental Task T1 - Control Task T1) - (Experimental Task T2 - Control Task T2) and use this contrasts in a one sample t-test to check for differences between and after WITHIN group, in a two sample t-test to check for differences BETWEEN group, and a simple regression model to test for association between the modifications and my behavioral variable
2)For each subject AND each section, create a first level and then create a contrast (Experimental Task - Control Task), use these contrasts in a flexible factorial design to check for interaction between time and group, and the interaction between time and the behavioral variable.
Is one of these two preferable to the other ?
Or are there both wrong and I need to rethink this ?
Thank in advance for any tip or suggestion
Alain
|