Hi,

I have probably a trivial question but I just want to be sure.

I have rfMRI data from 19 subjects before and after treatment as well as one cognitive test parameters.

To know the difference between my two sessions, I realized a paired t-test.
To find correlation with my test parameters I made a 3 columns design ( one, age, parameter) with a design [0,0,1] or [0,0,-1]

Now I would like to see the improvement between my two sessions and so maybe see a correlation with the difference between run1 and run2 of my test parameters..... but statistically, am I allowed to subtract correlation matrices? (I mean I am planning to fisher transform them but is it right?)

Thanks a lot for your help

Cc

Hi Cecile,

yes - it is possible to subtract the matrices in an element-wise fashion. I assume that you mean to subtract time point 1 from time point 2 for each subject. You would then test whether the difference in connectivity between the two time points is associated with the difference in cognition.

Note that an alternative approach is to include both time points in the design and treat the analysis as a repeated measures design.

Andrew