 |
 |
Dear all,
I recently installed the SwE toolbox to conduct longitudinal
analysis of structural imaging data (VBM).
(http://www2.warwick.ac.uk/fac/sci/statistics/staff/academic-research/nichols/software/swe/
V1.2.5, release 1st Sep 2016; running on Matlab R2014a)
I have two groups (g1 with 15 subjects, g2 with 11 subjects) with
two measurements for each subject (t1, t2). Despite having checked
the example provided on the SwE homepage, I still have
difficulties figuring out in which way to prepare the data (i.e.
select the scans). Is the scheme below correct?
g1_sub1_t1
g1_sub1_t2
g1_sub2_t1
g1_sub2_t2
...
g1_sub15_t1
g1_sub15_t2
g2_sub16_t1
g2_sub16_t2
g2_sub17_t1
g2_sub17_t2
...
g2_sub26_t1
g2_sub26_t2
Another question is related to setting up the design matrix. The
design matrix below was constructed according to the above scheme.
Basically it is the same as could be used in a SPM flexible
factorial model (I didn't include all subjects though as it is
only supposed to be an illustration). I wonder however, if
designing the matrix in this way would also be the correct
approach for the SwE-analysis. Do I have to additionally define
any subject specific intercepts or maybe even construct the design
matrix in a completely different way?
Tim