Hello Hannah,
> I have a question abt. how to do group effects analyses with
> parametric modulation. I ran a gambling study lately. Participants
> make a decision, see the outcome, and then on the next screen, they
> are asked to rate their feelings abt. "how much i wish i chose a
> different gamble" (5 pt. scale). There are 4 different types of
> outcome conditions; one of them is regret. For regret trials, i want
> to see what areas of regret are modulated by this rating.
> In SPM2, for each subject, what I did is that I ran a model with 11
> regressors (including baseline), and for the regret condition, I added
> the ratings as its parametric modulator, yielding 12 regressors in the
> model.
maybe you can also try without modeling the baseline if possible... :-)
> My question would be what should I do next? Do I use the contrast
> images of the regret rating regressor for the random effects? Do I
> create a different contrast test? Or do use the beta images for
> regret modulated with the ratings in the random effects model?
I think you have already done the right thing... i.e. image the
parametric regressor only
> Do I run a 1-sample t-test in the random effects model?
yes
> Once I see the results, how can I see how the ratings modulated the
> regret areas, e.g., higher ratings are correlated with which areas for
> the regret condition?
It is by definition the result you get with a one-sample [1] contrast.
You can then select a cluster of interest and plot the data related to
the modulated regressor regret vs scan or time and plot the data related
to the regressor regret vs your rating.
Best
Cyril
|