I have a group of twelve subjects I scanned twice (early learning and late
learning)... In every session, subjects underwent ten scanning runs, for a
total of twenty scans per subject. Subjects performed two motor tasks
(experimental and control) in every run. When doing separate fixed effect
analysis for the two learning stages (early or late), the analysis gives me
meaningful results. But when I combine the two groups to contrast the two
learning stages the matrix is too big and matlab crashes... I was told
about the 32 bit application limit of matlab... So it was suggested that I
do a random effect analysis My problem is as follows:
When I did a random effect analysis (to circumvent the 32 bit matlab
problem) for the early learning stage there was no activations reported,
but there was for the late learning stage... The question is this:
Can I combine the all the runs so that I contrast between the two learning
stages (early and late) even though the early learning runs give no
results????
Also, why don't I get any results for the early learning random effect,
while the fixed effect gives me meaningful results??? What can I do to fix
this???
thanks for your help,
RB
|