Hi,
I'm wanting to set-up a paradigm whereby I'm using video stimuli of a
subject's own versus an unfamiliar parent displaying three types of emotion.
That is, there is 1 between-subjects factor with 2 levels (parent: own,
other), and 1 within-subjects factor with 3 levels (emotion: neutral,
aggressive, positive). The aim is to assess activity associated with
exposure to positive and negative parental emotions in depressed versus
non-depressed kids.
I'm thinking of a block design: video stimuli = 20 sec blocks + 10 sec
rest/fixation in between each block, with 2 presentations of each type of
video stimuli.
I've come up with 2 designs, both for which own and other alternate:
1. own-neutral, other-positive, own-aggressive, other-neutral, own-positive,
other-aggressive (repeated).
2. own-neutral, other-positive, own-neutral, other-positive, own-aggressive,
other-neutral, own-aggressive, other-neutral, own-positive,
other-aggressive, own-positive, other-aggressive.
The plan would probably be to counterbalance order across participants in
some way.
Although the first model is somewhat simpler, it's been suggested to me that
it may not be as efficient due to the fact that any two of the same
condition (eg, own-neutral and own-neutral) are quite far apart in time, and
task-related signal may therefore have interference from low-frequency
drift. However, efficiency values for the simple contrasts for each
condition indicate that the first model may actually be slightly more
efficient (eg effect required for one contrast 0.973 versus 1.084). These
models have been entered into FSL with a high-pass filter of 100 sec,
temporal filtering applied for each EV.
Any advice/help with this issue would be greatly appreciated!! I suspect
that the difference between models may be trivial, but I'd like to come up
with the most efficient design possible!
|