Hi all,

I just wanted to kindly ask whether somebody could give me some advice on my FEAT demeaning question? Any help is greatly appreciated!

I want to do a correlation analysis of a resting state data set and some clinical measures and therefore put these clinical variables as covariates in the design matrix. As I understand from the postings here and from the FEAT page, I have to demean this variable. On the FEAT page it says that "The additional effect of the extra measures can be found by entering this as an extra EV which has been orthogonalised wrt the group mean EV - so in this case simply demeaned"

What I do not understand is: do I only have to demean the clinical variable and then it is already orthogonalized or do I additionally have to check the Orthogonalize box in FEAT for this EV?


And if I have two groups of patients and one control group, do I have to split the clinical variable EVs? And do I have to demean across all subjects then, before splitting into single EVs?

I also found this post:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;7be40cbc.1106
Here it looks if the covariates are not demeaned?


Many thanks for any help!

Max