Hi,
I have pre and post design and two behavioral variables (A - B). I would like to run a seed to voxel analysis.
In the SETUP tab, I set the interaction between A and B.
In the RESULTS (2nd-level) tab, in the "subject effects" section, I selected "AllSubjects" and "A*B interaction" (Between-subjects contrast: 1 0). In the "Conditions" section, I tried to set both "Post > Pre" (between-conditions contrast: -1 1) and "any effects (F-test)" options (between-conditions contrast: 1 0; 0 1). I found significant clusters only in this latter, but I struggle to figure out how this significance can be interpretated.
Any insight would be extremely helpful!
Bianca
Hi Bianca,
Typically when testing an interaction one also includes the main effects of the individual terms and you need to indicate in your between-subjects contrast that you want to evaluate the interaction term , e.g. you would define your model selecting "AllSubjects", "A", "B", and "A*B interaction" with Between-subjects contrast [0 0 0 1] (so the 1 is applied to the "A*B interaction" term)
Then, regarding the pre/post conditions, if you want to evaluate cross-sectionally whether the above interaction is present in the "pre" condition, you would select in the conditions list "pre" with between-conditions contrast 1. If you want to extend that previous test to evaluate whether the interaction was present any of the two timepoints (pre and post), then you would select in the conditions list "pre" and "post" with between conditions contrast [1 0; 0 1] (any effect). Last, if you want to evaluate the three-way A*B*intervention interaction (e.g. to test whether the change in connectivity with the intervention shows an A*B interaction across subjects), then you would select in the conditions list "pre" and "post" with between conditions contrast [-1 1].
Hope this helps
Alfonso
Originally posted by Bianca Monachesi:
Hi,
I have pre and post design and two behavioral variables (A - B). I would like to run a seed to voxel analysis.
In the SETUP tab, I set the interaction between A and B.
In the RESULTS (2nd-level) tab, in the "subject effects" section, I selected "AllSubjects" and "A*B interaction" (Between-subjects contrast: 1 0). In the "Conditions" section, I tried to set both "Post > Pre" (between-conditions contrast: -1 1) and "any effects (F-test)" options (between-conditions contrast: 1 0; 0 1). I found significant clusters only in this latter, but I struggle to figure out how this significance can be interpretated.
Any insight would be extremely helpful!
Bianca