Hi. Is anyone aware of a reason why it would be bad to
"rip" derivatives out of the design matrix just prior to
estimating a model? I have a mixed state and item-
related design in which I wish to use the Canonical HRF
with temp/dispersion derivatives for the item-related
regressors, but state-related regressors I wish only
to convolve with the Canonical HRF.
Since the SPM ui does not appear to support such a
model, we have incorporated a function into our batch
script which "post-hoc" removes the two derivatives
associated with the block regressors. This is done after
constructing and assigning data to the design matrix,
but prior to estimating the model.
The results look perfectly reasonable, but I fear I may
be doing something to the block regressors that I'm
unaware of.
Any thoughts appreciated.
Chad
***********************************************************
Chad Woodruff, PhD
Center for Neurobiology of Learning and Memory
Department of Neurobiology of Behavior
University of California, Irvine
Irvine, CA 92697-3800
phone: (949) 824-8861
fax: (949) 824-4807
http://fnim.bio.uci.edu/
************************************************************
|