Hi - no, for the second option I meant to use all 1s in the paired
design (in the EVs), and put the (demeaned) confound values into the
contrast.
Cheers.
On 28 Oct 2008, at 18:03, Jessica Sophie Damoiseaux wrote:
> Hi,
>
> Thanks for your quick reply Steve!
> I would prefer to use the second approach you mentioned but I'm not
> completely sure I understood it correctly. Should I put the
> individual covariate values (demeaned) where I would normally
> specify the same subject, like this:
> 1.0 -4.55 0 0 0 0 0 0 0
> 1.0 0 -0.55 0 0 0 0 0 0
> 1.0 0 0 5.45 0 0 0 0 0
> 1.0 0 0 0 -17.55 0 0 0 0
> 1.0 0 0 0 0 12.45 0 0 0
> 1.0 0.0 0 0 0 0 14.45 0 0
> 1.0 0 0.0 0 0 0 0 7.45 0
> 1.0 0 0 0.0 0 0 0 0 -7.55
> -1.0 -4.55 0 0 0.0 0 0 0 0
> -1.0 0 -0.55 0 0 0.0 0 0 0
> -1.0 0 0 5.45 0 0 0 0 0
> -1.0 0 0 0 -17.55 0 0 0 0
> -1.0 0 0 0 0 12.45 0 0 0
> -1.0 0 0 0 0 0 14.45 0 0
> -1.0 0 0 0 0 0 0 7.45 0
> -1.0 0 0 0 0 0 0 0 -7.55
>
> And if this is correct, how should I specify the contrast if I'm
> interested in the effect of the covariate? Like this:
> a) 0 1 1 1 1 1 1 1 1
> b) 0 .125 .125 .125 .125 .125 .125 .125 0.125
> or in another way?
>
> Thanks!
> Jeske
>
> ----- Original Message -----
> From: "Steve Smith" <[log in to unmask]>
> To: [log in to unmask]
> Sent: Saturday, October 25, 2008 6:04:13 AM GMT -08:00 US/Canada
> Pacific
> Subject: Re: [FSL] repeated measures with covariate
>
> Hi,
>
> There are a number of ways of resolving this rank deficiency. One way
> would be to replace the subject mean EVs with one constant height mean
> EV for all of them and one modulated by the covariate, orthogonalised
> wrt the constant height one. However you have now no longer modelled
> all subject means separately, so this is a less flexible model in
> terms of modelling subject mean variability. An alternative, which
> sounds a little odd but I think is ok, is to use the original basic
> paired design, and enter your new covariate as elements of a
> _contrast_ over the subject mean EVs (I guess you'd need to demean all
> the covariate values first). This should work ok.
>
> Cheers.
>
>
>
> On 24 Oct 2008, at 20:22, Jeske Damoiseaux wrote:
>
> > Hi,
> >
> > I would like to run a higher level repeated measures analysis on a
> > group of
> > 8 subjects scanned twice including a covariate (time between 1st and
> > 2nd
> > scan). I’m interested in the difference between the 1st and 2nd scan
> > and the
> > effect of the time between the 2 scans. I’m not sure how to set up
> the
> > design. If I use the design for a paired t-test from your example on
> > the
> > website and add the time difference (demeaned) as an extra EV the
> > design
> > becomes rank deficient. Is there a way to set up a design combining
> > the
> > repeated measures with the covariate or do I have to run these
> > separately?
> > And if I have to run them separately, how do I set up the design for
> > the
> > covariate? I’ve looked through the archives but couldn’t find a
> > suitable answer.
> >
> > Thanks,
> > Jeske
> >
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|