Hi to All,
I might miss the posts or the documentation which answer that question.
I ran a first-level FEAT analyses for many runs and for all subjects, now I need to run higher-level at each subject's session using a group-mean model, then to run a multi-subject higher-level analysis.
For the first-level analyses, I configured the GUI to estimate the parameters for linear normalization into the standard space, because this level concerns each subject individually. However, the multi-subject analysis is supposed to use the non-linear registration parameters to compare the subjects.
If I need to run multi-subject analysis, should I select non-linear registration at the first-level ?
My concern is that selecting non-linear registration at the first level changes the results from GLM. Why ?
I probably misunderstood something.