Hello Chris,
I think your design below might have some rank-deficiency issues, one way to represt the scanners could be as two difference EVs ( e.g. the difference between 1 and 2, 2 and 3, all demeaned ).
Kind Regards
Matthew
> Dear FSL experts,
>
> I am trying to setup a design to compare three groups of subjects (two patient groups and controls) adjusting for different scanners (three) used during data acquisition. I want to include scanners as a covariates in a model as suggested earlier in few posts .
>
> I found this solution but it deals with only two groups scenario:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1501&L=FSL&P=R97017&1=FSL&9=A&J=on&d=No+Match%3BMatch%3BMatches&z=4
>
> So I am thinking about following design:
>
> EV1: 1 to code 1st group
> EV2: 1 to code 2nd group
> EV3: 1 to code 3rd group
> EV4: 1 to code scanner 1 and 0 to code other scanners
> EV5: 1 to code scanner 2 and 0 to code other scanners
> EV6: 1 to code scanner 3 and 0 to code other scanners
>
> Now, the questions are:
> 1) Is this design is ok?
> 2) Should I mean-center each of the “scanner” EVs?
>
> Help will be greatly appreciated.
>
> Best regards,
> Chris
|