Dear SPMers,
I have a question regarding defining ROIs for multiple regression.
I have two groups in my study: a control group and a patient group.
Basically, I am examining 1) whether a patient group reveals reduced brain activity for numerosity in a magnitude estimation task, and if so, 2) whether the reduced brain activity is accounted for by patient's genetic factors.
When conducting a multiple regression analysis, I'd like to define an ROI region for numerosity to use it as a dependent variable. More specifically, I'd like to define an ROI in the IPS/IPL to conduct the regression analysis. Is it reasonable to define the ROI based on brain activity in control participants during the magnitude comparison task to use it for the patient group only? In other words, is it going to violate the assumption of independency if I use the ROI defined by brain activation in the control group to use it in the multiple regression analysis for the patient group only? I am trying to avoid the so-called "double dipping" problem here, so any advise will be appreciated! Also, if any of you have any other suggestions for defining functionally defined ROIs, such as IPS, please let me know.
I appreciate your help in advance.
Best,
So-yeon
|