 |
 |
Hi Geoff, On 20 September 2015 at 16:32, Geoff Rosen <[log in to unmask]> wrote: > Can one take F-tests, from example from basis functions, to the second > level? If so, are there any considerations relative to taking up t-tests? > The F-tests can't go higher directly but there is a workaround. Please, see this post <https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;1f8aa5a8.1406> by MJ. > > And a related question. When setting up models with multiple basis > functions (e.g., FIR) versus the standard gamma HRF plus the temporal > derivative term, why when setting the contrasts, does the former > automatically create the appropriate F-tests with a 1 in each basis > function column, while the latter does not? Is that because the temporal > derivative term is normally an "unwanted component"? > I think the time and dispersion derivatives are indeed treated as nuisance. However, if you want, you are free to include them in the F-test, along with the main effect. Cheers, Anderson > > Best wishes > > Geoff > BGU, Beer Sheva, Israel >