Hi Andrew,

There have been many earlier questions with this exact same design, 2x2 mixed. Have a look in the mailing list for the "design_ekaterina.ods" example.

If you compute differences there is no need for subject-specific intercepts, but the design should nonetheless include an intercept running across all subjects. This can either be a column full of ones, in which case the groups would be coded as +1/-1, or this column can be omitted but then you need 2 EVs for the groups, each coded using 1 or 0.

About subtracting or not IQ, it depends. Do you consider changes in IQ a potential nuisance on changes possibly seen on imaging data, or the actual value of IQ (averaged over the two timepoints) a nuisance? Depending on which way you go, you'd subtract or average IQ for the design in which you subtract thicknesses.

For the design in which you don't subtract, IQ can go "as is" (differences on IQ being then the nuisance), or with the signs alternating, i.e., multiply the sign of, say, visit 2, by -1.

Check the earlier emails and if something isn't clear, please feel free to post again.

All the best,

Anderson

On Thu, 4 Oct 2018 at 15:59, Andrew Gordon <[log in to unmask]> wrote:

Hi Anderson,

Thanks for the quick reply!

I will actually be running it both with and without subtracted images. Apologies if this is a rookie question, but I assume that if I was to run it with 'difference' images I would not require the intercept? And for the other analysis (i.e. the actual maps without subtraction) I would require an extra column of 1's for the intercept...? Is this right? I am not sure I understand why I would need to subtract the IQ measurements though?

I take your point about the values range for EVs 5 and 6 - I will re-scale them appropriately.

########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1