Mary,
> SPM Users:
> I completed one-sample t-tests on two groups of patients (n = 17; n = 18)
> that show differences in extent of activation when one visually examines
> the group maps. However, there are no appreciable differences between the
> groups when I complete a 2 sample t-test.
>
> I have 2 questions for the group:
> 1. Is it statistically appropriate to probe group differences using ROIs
> based on areas of activation from the one-sample t-test?
No, the p-values will not be valid because there is a non-zero covariance
between those two estimators (i.e., the estimator of the effect in one group
and the estimator of the difference between the two groups).
Eric
> 2. If so, how is this accomplished in SPM?
>
> Thanks.
> Mary Machulda
> Mayo Clinic
> MR Research
>
|