Hi,
On 10 Apr 2007, at 20:46, Martin M Monti wrote:
> Dear FSLers,
>
> In a first-level FEAT analysis I have the choice of breaking down
> one of my tasks (it's a slow event related paradigm) into 2
> different EV as opposed to keeping all in 1 EV.
>
> So, say I have 8 trials, which can be further grouped in 2
> categories (4 trials each).
>
> I envision 2 options (at a later point I'll be interested in
> looking at differences between the two categories, but not for now):
>
> a) create 1 EV file (with all 8 trials) and then create a contrast
> with +1 in this EV only
> b) create 2 EV files (with 4 trials each) and then create a
> contrast with + 1 in both EVs
>
> Would (a) and (b) yield the same results (I'm thinking esp of the
> variance of estimation in the voxel-wise regression)?
In general they will be similar. b) reduces the degrees-of-freedom
slightly, which reduces zstats, but also reduces the unmodelled
variance, which increases zstats. In general the latter factor would
be expected to win out.
Cheers.
>
> cheers
>
> martin
>
> --
> --------------------------------------------
>
> “Well! I’ve often seen a cat without a grin,” thought Alice, “but a
> grin without a cat! It’s the most curious thing I ever saw in my
> life!” [Lewis Carroll, Alice in Wonderland]
>
> --------------------------------------------
> Martin M. Monti
> Princeton University
> Department of Psychology
> Green Hall 3-S-8
> (609) 258-5679
> www.webmartin.net
> --------------------------------------------
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|