 |
 |
Hi Chris, hi Matthew, I'm not sure if I understood you correctly but could it be that you, Chris, are referring to the 1st-level option "Global normalisation" which can either be set to "None" or to "Scaling" while you, Matthew, are referring to the 2nd-level option "Global normalisation" → "Normalisation" which can be set to "None", "Proportional" or "ANCOVA"? I assume that these options are not identical. Is that correct? I was also wondering if the 1st-level option "Global normalisation" actually implements some sort of global signal regression. I think is doesn't but I am somewhat lost here and I would be really thankful for some clarification! My question is related to the discussion whether global signal regression introduces anti-correlated networks in resting state analyses – which one might want to avoid (see e.g. http://www.ncbi.nlm.nih.gov/pubmed/18976716). If I set "Global normalisation" to "None" does SPM still do some kind of implicit global signal regression? If not, do I have to enter the global signal as a regressor into my model in order to do global signal regression? And, if I decide to do global signal regression, are there any scripts or toolboxes available? (Probably this one: http://www.fil.ion.ucl.ac.uk/spm/ext/#LMGS) Best, Jan-Peter