Hello,
Recently we have begun to question what is global normalisation and when
should we be normalizing our data? We are collecting event-related fMRI data
from two groups. We have modeled the groups separately and now want to
make group comparisions. It seems clear that global normalisation should be
conducted at the first level, but should it also be done when comparing the
two groups (who have only been normalized to others within the same group).
If there are overall global differences between the two groups, how is this
handled? And at what times is grand mean scaling more appropriate than
normalisation? Should one be used solely for first level modeling and the other
for second level comparisions? Any guidance on these matters would be very
much appreciated.
Thank you,
Erin Shannon
Wake Forest School of Medicine
Winston-Salem, NC
|