hello all,
i have a question regarding the second stage t-test performed
using the random effects model. as i understand it, this stage
uses the contrast images formed for each individual subject
in the study. in the simplest case, my understanding is that
these images are essentially the parameter estimates of the
height estimate of the synthetic hrf for a given condition.
the second stage then involves asking does the 95 percent
confidence interval of the mean of these estimates
encompass zero.
my question is related to the information actually contained
in these contrast images. is it the case that these images
do NOT contain any information regarding how well the
data were fit for each subject? that is, during the second
stage t-test are we in fact treating each subject's values in
the contrast image as though each were equally reliable?
on a related note, a friend was under the impression that one
should use the spm T images in the second stage model.
presumably because these have information regarding
not only the height estimate but also its reliability in
relation to each subject's error variance...does this make
sense?
thanks in advance.
ian.
--
Ian G. Dobbins
MGH-NMR Center
Bldg. 149, 13th Street
Charlestown, MA 02129
(ph) 617-724-9989 (fax) 617-726-7422
[log in to unmask]
www.nmr.mgh.harvard.edu/people_final.htm
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|