Hi,
The problem is that the design is rank deficient - at least one EV is
a linear combination of others. I would use the following design:
EV1(control) EV2(patient) EV3(demeanedagecontrol)
EV4(demeanedagepatient)
1
0
3 0
1
0
-3 0
0
1
0 4
0
1
0 -4
and e.g. a contrast [0 0 1 -1] asks for (one direction of) the age
interation.
Cheers.
On 16 Jul 2008, at 20:20, Russell Chu wrote:
> Hi,
> I processed some patients and controls with TBSS and am now trying
> to find
> the age effect and age*group interaction in randomise. Would someone
> please
> check my design and contrast matrices below? Also, in the design
> matrix, I
> subtracted the mean age from each individual subject's age; is this
> the
> proper way to demean the ages?
>
> Design matrix:
> EV1(control) EV2(patient) EV3(age) EV4(age*control) EV5(age*patient)
> 1 0 4 4 0
> 1 0 -4 -4 0
> 0 1 -9.8 0 -9.8
> ...
> ...
> 0 1 -1.2 0 -1.2
> 0 1 -3 0 -3
> 1 0 -4 -4 0
> 1 0 4.3 4.3 0
>
> Contrast matrix:
> EV1(control) EV2(patient) EV3(age) EV4(age*control) EV5(age*patient)
> control>meth 1 -1 0 0 0
> meth>control -1 1 0 0 0
>
> randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d
> design.mat -t design.con -n 100 -c 3 -V
>
> Also, after all permutations (only 100 for quick initial results) are
> finished, the following error occurs:
>
> terminate called after throwing an instance of
> 'NEWMAT::ConvergenceException'
> Aborted
>
> Any input on these issues would be greatly appreciated.
>
> Thanks,
> Russell
>
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|