dear list.
i designed a 2 factorial flexible model in spm for one group of subjects
according gläscher and gitelman very helpful note 'contrast weights in
flexible factorial design with multiple groups of subjects' (page 5). Both
main factors have 2 levels, factor 1 (A) is the same session for each
subject either under hypnosis or non-hypnosis. i got a design with the first
4 columns representing the levels of each factor, the last 4 columns the
interaction between the factors. the design looks like design 2 in gläschers
and gitelmanns note (page 5), with the exeption of only 2 levels of factor B
and only 4 (instead of 6) interactions.
I tested for the interaction between both factors and got reasonable
results. Now i want to test for differential effects between 2 of the
interaction regressors (A1B1 >A2B1 = hypnosis vs non-hypnosis effect on B1)
comparing to an two-sample test with dependency (because of within-subject
repetition of the session).
i tried to test this in the flexible factorial design as: -1 1 0 0 -1 0 1 0
and i tested this in a separate 2-sample t-test with dependency (A1B2 > A2B1)
Both ways showed different results. from my perspective they should be very
similiar to each other.
where is my mistake (wrong contrast weights in the factorial design ?) and
which would be the correct approach?
thanks
markus
|