Hello,
I would suggest to model the sessions only separately if the sessions were scanned on different days and use a flexible factorial design in the second level and model there a non independent session (may be treatment) factor to identify between session effects.
The advantage to model two sessions in two designs separately on the first level is in my understanding that they don’t have to be realigned between the different sessions.
Kind regards
Christoph
> You only need 1 model. When you run it the session number should be
> reflected in the covariate for each task. that is usually averaged, so
> just
> specify the specific task/session to use when you make your contrasts.
>
> Cheers,
> Michael
>
> On Fri, Dec 31, 2010 at 1:09 PM, Jung Eun Han <[log in to unmask]>
> wrote:
>
>> Hello,
>>
>> Our study is a repeated-measures design (each subject had two
>> sessions). Event types (conditions or different kinds of stimuli) were
>> ordered differently for all subjects (all subjects have different
>> onset times for each event type). Ultimately, we would like to compare
>> brain activations among different event types (1) WITHIN each session
>> and (2) AMONG the two sessions.
>>
>> Am I to create a design matrix(spm.mat file) for each subject (in each
>> session) or
>> two design matrices (for two sessions) for all subjects?
>>
>> I hope my question is clear
>> Thanks,
>> Claire
>>
>> Sent from my BlackBerry® smartphone
>
>
>
>
> --
> Research Associate
> Gazzaley Lab
> Department of Neurology
> University of California, San Francisco
>
|