Did I ask a question that is unanswerable or was I unclear in the problem
we're facing? I suspect it's the latter, so please let me know if I can
clarify anything in an attempt to receive consulation.
Thanks,
Jeff
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Jeff Browndyke
Sent: Wednesday, May 10, 2006 11:20
To: [log in to unmask]
Subject: [SPM] Problems w/ Mixed Model Design Construction
SPMers,
We're attempting to model both event-related signal and block-related
signal in a three run experiment. I've attached the current model
matrix as we have scripted it so far, but I'm wondering if we are
combining "apples and oranges" in the model. The first two columns
relate to specific response types (hits and misses), but as you can
see these columns also include periods of no activity that reflect
long off-block fixation periods. I suspect that this is wrong because
any comparison of column one (hits) or two (misses) with the constant
also includes off-block fixation activity, when really what we should
be comparing for event data is event signal vs. on-block ISI signal.
Similarly, a direct constrast of column one and two (hits vs. misses)
also includes the off-block fixation, which I guess would cancel each
other out but may be diluting the event contrast signal.
Creating a block comparison model isn't a problem (i.e., hits, misses,
and on-block ISI averaged vs. off-block fixation averaged signal).
What I'm trying to figure out is how to model just the event-related
signals relative to the on-block ISI, while excluding the signal
associated with the long off-block fixation periods.
Thanks for any pointers or guidance,
Jeff
|