Hi Karin,
That's correct. If you were testing multiple components, you'd also divide the 0.025 by the number of components you're testing to maintain a FWER of 5%.
Cheers,
David
> On Aug 2, 2015, at 8:03 PM, Karin Cox <[log in to unmask]> wrote:
>
> Hello all,
>
> I was wondering if I could confirm that I am correctly understanding the randomise outputs generated by the dual regression script (i.e., the *corrp* files). I am invoking dual regression with a command of the following general format:
>
> dual_regression <melodic_IC file> 1 <design.mat file> <design.con file> 5000 <output directory> <input filelist>
>
> -- where the design.mat / design.con files were produced using design_ttest2.
>
> As expected, dual regression outputs *corrp* files for all combinations of ICA component # and group contrast direction. My question: To evaluate the results under the assumption of a two-tailed t-test and an alpha of .05, I need to threshold each of these images at p < .025, correct?
>
> Thanks very much in advance --
>
> Karin
|