Dear SPMers-
I am sorry to say that previuosly posted messages, and the on-line SPM99
docs, have not clarified this issue for me...
I have run a two condition event-related experiment. After fitting the
model and plotting the adjusted and fitted values versus time, I export
the resulting betas (at significantly activated voxels) and get values
such as 0.1, 0.2, 156 (note the magnitude of the effects are 1000 times
smaller than the constant- this suggests very small signal changes, does
it not?). Also note that I did global normalization during my
modelling, and that my contrast vectors had hieghts of 1.
Am I interpreting the beta coefficients correctly as follows:
In the absence of events A and B, then the mean globally normalized
voxel intensity is 156.
If I have a "pure" A event and no B event, then the mean voxel intensity
if 156.1, corresponding to a percent change calculated by:
{(156.1-156)/156}*100= 0.0641%.
If I have a "pure" B event and no A event, then the mean voxel intensity
is 156.2 corresponding to a percent change calculation by
{(156.2-156)/156*100}= 0.128%.
Thanks for your help,
Michael
|