You can't do a paired test because you have independent samples.
Either an unpaired T, or better yet a permutation test. This would be
your best bet and most valid approach, especially considering your
small sample size. check out Tom Nichols' great toolbox SnPM
(http://www.sph.umich.edu/ni-stat/SnPM/)
Cheers,
Michael
On Tue, Mar 2, 2010 at 8:06 PM, R. Matthew Hutchison <[log in to unmask]> wrote:
> Hi all,
>
> I apologize if this does not fit in the standard line of SPM/GIFT
> questioning.
>
> I was wondering if someone may offer me some stats advice on a seemingly
> simple problem when compared to fMRI statistics.
>
> I have 2 different anesthesia groups and I want to compare the number of
> individuals that have a functional network between groups across multiple
> networks.
>
> In a simplified example:
>
> For anesthesia type 1, 6 of 10 participants have Network A, 5/10 have
> network B, 3/10 have Network C.
> For anesthesia type 2, 8/10 have Network A, 4/10 have network B, and 5/10
> have Network C.
>
> My first instinct was to use a paired 2 tailed t-test, but I no longer feel
> this is correct.
>
> Any ideas would be greatly appreciated,
>
> Cheers,
>
> Matt Hutchison
>
> --
> R. Matthew Hutchison, PhD. Candidate
> Centre for Functional and Metabolic Mapping
> Robarts Research Institute
> Cuddy Wing - 9.4T Suite
> P.O. Box 5015, 100 Perth Drive
> London, Ontario, Canada N6A 5K8
>
--
Research Associate
Gazzaley Lab
Department of Neurology
University of California, San Francisco
|