Hi Lin Nga,
Apologies if I have misunderstood your experimental setup - but do you
have an experiment where you have a number of subjects, each of which
is scanned _once_ in either a stress or no stress condition?
If so I am not clear why you need a three level analysis, you should
be able to do this in two levels.
First level would be as you specified. Second level could then contain
4 EVs. Containing:
EV1 that picks out stress/males,
EV2 that picks out stress/females,
EV3 that picks out no stress/males,
EV4 that picks out no stress/females
You could do this other ways, but this gives the most straightforward
contrasts, e.g:
stress/male>stress/female would then be a (1 -1 0 0) contrast.
This second level then gets run separately on each of your first level
contrasts.
Cheers, Mark.
----
Dr Mark Woolrich
EPSRC Advanced Research Fellow University Research Lecturer
Oxford University Centre for Functional MRI of the Brain (FMRIB),
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
Tel: (+44)1865-222782 Homepage: http://www.fmrib.ox.ac.uk/~woolrich
On 7 Jan 2009, at 19:20, Lin Nga wrote:
> Hi,
>
> My apologies if this is a re-posting; I wasn't sure it posted the
> first time.
> I’m working on setting up a group analysis using FSL and I have some
> questions about the higher-level analysis.
>
> For the first-level analysis I have:
> 4 EV's : neut_init, neut_repeat, angry_init, angry_repeat
> 5 contrasts: all>baseline, neutral>angry, initial>repeat,
> neut_init>neut_repeat, angry_init>angry_repeat
>
> So for the second-level analysis, from what I understand, I should
> have four
> separate group analysis (one for each parameter of interest):
> stress, no
> stress, male, female. I would have 1 EV and 2 contrasts, mean and
> (-) mean,
> for each of these groups.
>
> Then for a third-level analysis, I would feed into feat the .gfeat’s
> from
> the second-level and have 4 EVs also for stress, no stress, male,
> female.
> The setup here would look something like:
> Group EV1 EV2 EV3 EV4
> S NS M F
> Input 1 1 1 0 0 0
> Input 2 1 0 1 0 0
> Input 3 1 0 0 1 0
> Input 4 1 0 0 0 1
>
> Now for the contrasts, should I have a similar setup as above and
> include
> f-tests for each contrast or should I setup a contrast for each pair
> individually (that is: stress/male>stress/female, no stress/male>no
> stress/female, stress/male>no stress/male, stress/female>no stress/
> female,
> and each of their respective inverses)?
>
> Any comments or help would be greatly appreciated.
>
> Thanks,
> Lin Nga
>
|