Hello,
We run a higher-level analysis with performance as a regressor (orthogonalized to the group mean and another regressor). We find nice and plausible results in the whole brain analysis. To check for possible outliers etc. we extracted individual PEs from spheric ROIs around the peak activations of the significant clusters and plotted them against performance (the original scores and the demeaned regressor from the design.mat file). To our surprise, we find no significant correlations between performance and PEs (rs around zero). Why might this be?
Thanks for your help!
|