>does anyone out there know of a good intro to general GLM and contrast issues
christensen: plane answers to complex questions, the theory of linear models (2nd ed. springer, 1996)
is quite good, but definitely beyond a simple intro.
Von: Stephen Smith [mailto:[log in to unmask]]
Gesendet: Mi 13.10.2004 19:57
An: [log in to unmask]
Betreff: Re: [FSL] design matrix rank deficiency
Hi Eyleen - these are quite general stats design questions - probably the
best place to start is a look at the intro slides at
http://www.fmrib.ox.ac.uk/fslcourse in particular the FMRI section - and
then possibly an introductory stats textbook - does anyone out there know
of a good intro to general GLM and contrast issues? I'm not aware of one
On Mon, 11 Oct 2004, Yi zhang wrote:
> Thanks Christian,
> As is my understanding of the FAQ: when the design matrix or number of
> contrasts is large enough, it will be highly likely to be rank deficient?
> But for the higher level analysis, I can carry on whatever combinations of
> contrasts and the EVs won't add up to zero? If that's right, I think our
> problem is solved.
> Another question is whether the sum of all EVs has anything to do with my
> experimental design?
> If my control conditions are not optimally distributed, would that render my
> design to be rank deficient?
> Thanks a lot!
> >From: "Christian F. Beckmann" <[log in to unmask]>
> >Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
> >To: [log in to unmask]
> >Subject: Re: [FSL] design matrix rank deficiency
> >Date: Mon, 11 Oct 2004 03:46:06 +0100
> >this is covered in the FAQ, see
> >http://www.fmrib.ox.ac.uk/fslfaq/#feat_rank for details. In order to
> >give more specific advice we'd need to know details of your design -
> >the feat directory should have design.mat and design.gif in it, maybe
> >you want to post those?
> >On 11 Oct 2004, at 01:56, Adam Gries wrote:
> >>1) I've been trying to run a lower level analysis for a large number of
> >>contrasts in different combinations and have received the following
> >>At least one EV is (close to) a linear combination of the others. You
> >>should probably alter your design. (Design matrix is rank deficient -
> >>ratio of min:max eigenvalues in SVD of matrix is 0.000000e+00)
> >>2) I could technically run the analysis later but received no results.
> >>3) The EV's are expected to have very different values and I do not
> >>understand how one could be a combination of another.
> >>Any explanation and help would be appreciated,
> >>Adam Gries
> > Christian F. Beckmann
> > Oxford University Centre for Functional
> > Magnetic Resonance Imaging of the Brain,
> > John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
> > Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
> > Phone: +44(0)1865 222782 Fax: +44(0)1865 222717
Stephen M. Smith DPhil
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