Hi, I have a quick modeling question.
I have a finger-tapping efMRI design designed to look at the effect of tapping speed (e.g.,
slow, fast, very fast). The problem is that individuals also differ in how long they are
tapping the button for (duration of 10 secs for one person in the slow condition, 15 for
another).
I’m interested in the three tapping speeds, but want to ignore how long the participants
tapped for on each trial. What’s the best way to model the effects of this duration? I’ve
been told it’s good to use a parametric regressor; i.e. model the duration of each
pressing event as 0, and then include a parametric regressor–same onset–which takes
the event’s real duration.
When I do this (each pressing speed has 1 regressor + 1 parametric regressor) the
results look the same as if I don’t include the parametric regressor. I checked the
parametric regressor alone, and it doesn’t seem to be picking up much variance.
Would using the same model with each regressor having time modulation rather than a
parametric regressor make more sense?
I’m using SPM8, and specifying the parametric/time modulation regressors within each
condition... I’ve tried it with and without a conditions batch script, the results are the
same.
Thanks!
Chris
|