Hi,
On 26 Nov 2007, at 05:56, Darren G wrote:
> Hi
>
> I am trying to get my head around setting up 2nd level stats in FSL.
> I have
> 2 subject groups. Each subject performs an nback working memory task
> under 3
> conditions. I've analyzed all my subjects at the first level and
> made 3
> copes, 1 for each task condition.
>
> At the second level would I select the COPE images or the feat
> directories?
It sounds like the 3 COPES are from the same session, so you will need
to use the "select COPE images" option.
> I think it would have to be the cope images. I tried selecting the
> FEAT
> directories but the events don't seem correct then.
>
> Anyway, assuming that I select the copes I selected them in the order
> subj1-cope1, subj1-cope2, subj1-cope3, subj2-cope1 and so on.
>
> Here is what I think the model would be
>
> grp ev1 ev2 ev3 ev4 ev5 ev6 ev7 ev8 ev9 ev10
> subj1a-cope1 1 1 0 0 0 0 0 1 0 0 0
> subj1a-cope2 1 0 0 1 0 0 0 1 0 0 0
> subj1a-cope3 1 0 0 0 0 1 0 1 0 0 0
> subj2a-cope1 1 1 0 0 0 0 0 0 1 0 0
> subj2a-cope2 1 0 0 1 0 0 0 0 1 0 0
> subj2a-cope3 1 0 0 0 0 1 0 0 1 0 0
> subj3b-cope1 2 0 1 0 0 0 0 0 0 1 0
> subj3b-cope2 2 0 0 0 1 0 0 0 0 1 0
> subj3b-cope3 2 0 0 0 0 0 1 0 0 1 0
> subj4b-cope1 2 0 1 0 0 0 0 0 0 0 1
> subj4b-cope2 2 0 0 0 1 0 0 0 0 0 1
> subj4b-cope3 2 0 0 0 0 0 1 0 0 0 1
>
> ev1 = group a cond 1
> ev2 = group b cond 1
> ev3 = group a cond 2
> ev4 = group b cond 2
> ev5 = group a cond 3
> ev6 = group b cond 3
> ev7-ev10 subject effects
>
> Do I need to have a mean column in this analysis?
No - you already have subject-specific means in the right-most EVs.
But I fear that this may be rank deficient - in which case you would
probably drop EVs5-6 (see the example at
http://fsl.fmrib.ox.ac.uk/fsl/feat5/detail.html#ANOVA1factor4levels)
> Now for the contrasts, which is where I get very confused
> Main effect of group (F-test)
>
> ev 1 2 3 4 5 6 7 8 9 10 F
> 1 0 1 0 1 0 0 0 0 0 on
> 0 1 0 1 0 1 0 0 0 0 on
I think instead you want:
1 -1 0 0 0 0 0....
0 0 1 -1 0 0 0.....
0 0 0 0 1 -1 0....
with an F across these, to get the main group effect?
though this would need amending in the light of my above comment
regarding the exact contrasts...
>
>
> Main effect of working memory (I think I only have to test 2/3
> contrasts for
> this?) (F-test)
> ev 1 2 3 4 5 6 7 8 9 10 F
> 1-2 1 1 -1 -1 0 0 0 0 0 0 on
> 2-3 0 0 1 1 -1 -1 0 0 0 0 on
Yes - I think so.....but probably best to get the above right first.....
Cheers.
> Interaction of group with working memory (I think I only have to
> test 2/3
> contrasts for this?) (F-test)
> ev 1 2 3 4 5 6 7 8 9 10 F
> a1-2 1 0 -1 0 0 0 0 0 0 0 on off
> b2-1 0 -1 0 -1 0 0 0 0 0 0 on off
> a2-3 0 0 1 0 -1 0 0 0 0 0 off on
> b3-2 0 0 0 -1 0 1 0 0 0 0 off on
>
> What if I wanted to look at a simple t test of group differences for
> cond 1
> -3 would that be the following
> ev 1 2 3 4 5 6 7 8 9 10
> (a1-a3)-(b1-b3) 1 -1 0 0 -1 1 0 0 0 0
>
>
> How would I get the overall standard anova result?
> thanks,
> Darren
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|