Dear Amirali,
The simplest way to do your group analysis would be to average the
trials for each subject using the 'Average' button and then export two
images per subject: one for standard and one for oddball and put them
in paired t-test design. There is no need to run statistics for each
subject and take contrast images to the second level. You will just
get exactly the same result.
Best,
Vladimir
On 17 Oct 2012, at 15:07, Amirali Shirazibeheshti wrote:
> Hi SPMs
>
> Please help,
>
> I have a set of EEG data from odd-ball paradigm, including 18
> subjects.
> I want to run a group level analysis across 18 subjects.
> How can I extract the subject effect or contrast image from each
> subject assuming that for each subject there are 400 “standard” and
> 100 “deviant” trials.
> Someone suggested me to apply two sample t-test for each subject
> separately, between deviant and standard, then SPM makes several
> images at the output directory including beta, con, ess, mask, ……
> We may use con images of two-sample t-test result from each subject
> to put them in another test, say like one sample t-test, to perform
> a group level analysis.
> Is this strategy true?
> Another method is to use all the trials from all the subjects and
> separate trials based on deviant and standard and put them in a two-
> sample t-test.
> Which one is true?
> Basically my question is that what is contrast image? How can I
> extract contrast image from trials to use them over a group level
> analysis like t-test or ANOVA.
>
> Thank you
|