Hi,
All of these contrasts are t-test contrasts. If you want an F-test that looks at any differences between A, B or C, then you just need to include the A>B and A>C contrasts in your F-test (when you add an F-test in FSL it gives you a little button next to each t-contrast and you select the ones that you want to include in the F-test). There are other, completely equivalent, ways of setting up the same F-test, but this one is easy here.
As for post-hoc vs not - due to the flexibility of the setup this is not prescribed and you need to decide what is appropriate for what you are doing. If you are setting up the F-contrast above and then looking at the results of the individual t-contrasts in the regions where the F-contrast is significant, then this is treating the t-contrasts as post-hoc tests. However, this would not cover the first three contrasts (A, B or C) as they are not covered in the F-test, but it would cover the rest of the contrasts (as any between-group difference, even B>C, is included in the F-test described above). So if you only use the differential contrasts (number 4 onwards) as post-hoc contrasts, to dissect the F-test results, then you don't need to do any other multiple comparison correction.
All the best,
Mark
On 11 Apr 2014, at 08:37, Sanne Detiger <[log in to unmask]> wrote:
> Dear FSL users,
>
> I'm quite new to FSL and I stumbled upon a problem that I can not seem to figure out. I'm doing an analysis with three groups, let's say A, B and C. Over these three groups I want to do an F-test. In the same analysis I want to do a between-group analysis. So my matrix looks like this:
> A 1 0 0
> B 0 1 0
> C 0 0 1
> A>B 1 -1 0
> A>C 1 0 -1
> B>A -1 1 0
> B>C 0 1 -1
> C>A -1 0 1
> C>B 0 -1 1
>
> Now my question is this: are these t-contrasts t-tests or post-hoc tests? Since if these are t-tests, I probably would have to correct the p-value and if they are post-hoc tests the p-value would already have been corrected right?
|