Five per group may not be enough to get a reasonable power in your random
effects analysis. I would suggest roughly twice that per group. Would others
agree?
The procedure however is straightforward. Hit the basic models button and
select 2 sample t-test. It will ask you for images and these will be the
individual contrast images for condition A. It wont ask for groups
separately so be sure your first five images are group 1 and second 5 are
group 2. The vector for the contrast is as you have specified below.
Good luck,
Sterling C. Johnson, Ph.D.
Neuropsychology
Barrow Neurological Institute
222 W. Thomas Rd. Ste 315
Phoenix AZ 85013
ph: 602-406-3671
fx: 602-406-6115
email [log in to unmask]
-----Original Message-----
From: Stefan Golaszewski [mailto:[log in to unmask]]
Sent: Thursday, August 24, 2000 3:23 PM
To: [log in to unmask]
Subject: group comparison
Dear Hotline,
we have 2 groups (1 and 2) of fmri datasets, 5 subjects
each group. Paradigm is R A R A R A R A R. Now we want
to test for a difference between both groups with
respect to condition A (contrast 1 1 1 1 1 -1 -1 -1 -1
-1) using random effects model. We have already
preprocessed all datasets and specified and estimated
the model for each subject. What should be the next
step?
Thank you, Christian
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|