 |
 |
Hi Tali, For that first contrast, that test the mean of the 1st group (of 18 subjects), what happens is the following: the data are residualized with respect to the 3 other subjects, permuted, then regressed against that main EV. The number of permutations is computed with respect to that EV, which is seen as a "multiset", that comprises one set of 18 undistinguishable elements, and another set of 3 undistinguishable elements, for a grand total of 21. The formula is well known, and I've just used in the previous email. See this Wikipedia article: https://en.wikipedia.org/wiki/Permutation#Permutations_of_multisets You can do sign-flippings if you want to and if the assumptions are met. The number of possible sign flippings will be 2^21 = 2,097,152. If the assumptions are correct for your data, the results will be similar up to the resolution of p-values afforded by the one that offers fewer resamplings, in this case, the permutations. All the best, Anderson On Fri, 19 Oct 2018 at 01:48, Tali Weiss <[log in to unmask]> wrote: > In both - it is a one-sample test. Thus there are 2^N possible sign flips. > 2 ^ 18 =262,144 possibilities? > Why the number of permutation differ bw GROUP Main Effect and Group>single > subject? > > In the terminal: > Use the -1 option with randomise to indicate a one-sample t-test > The GUI will recognise it one-sample t-test by default? > > thank you > Tali > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1