Hi,
I'm just after a bit of clarification please.
I have some repeated measures data which I've run VBM on (3 groups, scanned at two time points). I've used randomise to compare differences in GMD between time 1 and time 2 in our subjects.
Now that randomise can support repeated measures with exchangeability blocks (EB), I have run the analyses as per attached design matrix, using the approach outlined at https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;150cb6af.1407
I have compared the EB result to the older method, where I've subtracted t2 from t1, and computed each group's mean as outlined here https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;d6651f48.1008
For the contrasts looking at the effect of time per group (i.e. c1 in attached), the TFCE corrected results are largely similar, although the results using EB are less extensive in terms of changes in GMD.
Is this difference because of within subject permutation in the EB approach? Is the recommendation now to follow the EB pipeline for RM/longitudinal VBM?
Many thanks in advance.
Charlotte
|