In general you want to use the grand mean. See previous posts on this
topic in the archives, or Jeanette Mumford's web page on mean centering.
cheers,
-MH
On Wed, 2012-04-04 at 13:29 +0100, Seyed Sajjadi wrote:
> Hi,
>
>
>
> I have a question regarding designing matrices in fsl.
>
> I am trying to design a matrix for un-paired two groups comparison
> (patients vs. controls) with one nuisance factor (grey matter volume).
> I understand that I need two separate columns for patients and
> controls to enter the GMV values for each group in a separate column.
> Now, to demean the grey matter volumes do I have to use the grand mean
> (mean of both groups put together) or group specific means (demeaning
> patients value by patients’ mean and controls by controls’ mean)?
>
>
>
> Best wishes
>
> Seyed Sajjadi
>
>
|