Hello,
I would like to solicit advice from the list about whether (both theoretically and in practical terms)
doing any of the three things below will improve the significance or beta values in a group-level
random effects model. My task is an event-related task in which images are shown for 1s and
subjects are asked to make a straight-forward decision about the stimulus, which they do within
800-900ms of the stimulus coming on.
1. At the single-subject level adding the 6 motion parameters regressors.
2. At the single-subject level adding a global mean (per TR) regressor.
3. In setting up the single-subject model, using an "epoch" for each stimulus of 1s, rather than
simply a delta function of 0s duration at each stimulus onset time.
thanks!!
Amit
|