Dear Amy,
1) You can do this kind of t-test if you set the variance to be equal. This is one of the options when specifying the design.
2) There is no simple way to threshold a t-image computed outside SPM. You need additional information (e.g. the residuals) to use RFT correction.
Best,
Vladimir
On 28 Dec 2013, at 00:03, Amy Kuceyeski <[log in to unmask]> wrote:
> Hi all!
>
> I am performing an unpaired t-test in SPM where one of the "groups" is a single individual. It seems as though this is equivalent to performing the Crawford modified t-test, which is exactly what I want. However, when I try to implement this in SPM8, it throws an error when I enter only one subject for group 2. In particular, the contrast definition in the Results section that fails to complete. Has anyone out there had success with doing this sort of analysis? I would eventually like to be able to do this command line or with batch script.
>
> On a related note, is there a way for me to use the results section on a t-map that I create myself? For example, I could calculate the t-map using the Crawford modified t-test formula and then use SPM for the thresholding. Or does the thresholding involve the other images that are created along the way (beta_0001/beta_0002, etc)?
>
> Thanks for any help you can give and Happy new year!
>
> Amy
|