I'm analyzing an event-related fMRI experiment in SPM, and I have a
question about the use of the informed basis set (IBS) for task
parameter convolution. My design has one group, two repeated measures,
two levels for each repeated measure. My read of the use of the IBS
from prior posts (and the example data provided by SPM) indicates that,
if one were to look for overall experimental effects across all the
basis functions, one would typically do so by calculating contrasts
separately for each basis function at the first level, then collapsing
across them via contrasts at the second level. My question: is there
anything inherently wrong with collapsing them at the first level
instead, when generating condition-wise contrast maps of interest, then
submitting those contrast maps to the second level? I realize that
doing so would inherently limit some types of analyses (namely, those
that look at only one or a subset of the IBS). But I was wondering if
it was feasible.
Thanks - John
|