Hi Zrinka,
Your design matrix is not quite right. The most obvious problem, apart
from it not achieving the modelling that you want, is that you have as
many EVs as data points, so all your variances will be zero!
I suspect that you are effectively wanting to do paired t-test
comparisons. have a look at:
http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html#PairedTwoGroupDifference
so you probably want a slightly more complex version:
group EV1 EV2 EV3 EV4 EV5 ......... EV23
S1pre 1 1 0 1 0 0 ......... 0
S1post 1 -1 0 1 0 0 ......... 0
S2pre 1 1 0 0 1 0 ......... 0
S2post 1 -1 0 0 1 0 ......... 0
...
S21pre 2 0 1 0 0 0 ......... 1
S21post 2 0 -1 0 0 0 ......... 1
con1 1 0 0 0 0 ......... 0 = group1 pre-post
con2 -1 0 0 0 0 ......... 0 = group1 post-pre
con3 0 1 0 0 0 ......... 0 = group2 pre-post
con4 0 -1 0 0 0 ......... 0 = group2 post-pre
con5 1 -1 0 0 0 ......... 0
con6 -1 1 0 0 0 ......... 0
con5 = group1(pre-post)-group2(pre-post) OR can alternatively be seen as
group2(post-pre)-group1(post-pre); you can disambiguate these choices by
using contrast masking.
Good luck!
Steve.
On Wed, 19 Mar 2003, Zrinka Bilusic wrote:
> Hello to everyone... I have tried running a second-level across-session
> analysis in FEAT yesterday, and have run into some problems... The model
> is as follows: I have 21 subjects in two groups (10 in one and 11 in the
> other group) that were scanned twice with the same stimuli but under
> different conditions, pre and post. I want to see whether there are
> differences between the sessions for each subject and then eventually
> across subjects (third-level analysis). So, in the end, I have 42 EVs
> and 21 contrasts:
>
> group EV1 EV2 EV3 EV4 EV41 EV42
> 1. subject1 pre 1 1 0 0 0..... 0 0
> 2. subject1 post 1 0 1 0 0..... 0 0
> 3. subject2 pre 1 0 0 1 0.... 0 0
> 4. subject2 post 1 0 0 0 1..... 0 0
> .....
> 41.subjec21 pre 2 0 0 0 0..... 1 0
> 42.subject21 post 2 0 0 0 0..... 0 1
>
> contrasts:
> EV1 EV2 EV3 EV4... EV41 EV42
> 1. subject1 1 -1 0 0.... 0 0
> 2. subject2 0 0 1 -1... 0 0
> ...
> 21. subject 0 0 0 0... 1 -1
>
>
> I have run first-level anlysis on all these before, and there were 2 EV for each subject/session.
Stephen M. Smith MA DPhil CEng MIEE
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|