Hi - this looks fine - if C1 is supposed to ask about the difference
of the means then its third entry would be 0.
To find out about correlation with EV3 then use [0 0 1].
To look for interactions, i.e. where does group 2 correlate with the
extra regressor more than group 1, split EV3 into two halves; demean
each half separately, then pad with zeros, then use a differential [0
0 1 -1] etc contrast.
Note that if you are using any of the above in randomise then
strictly you should consider the model as two separate submodels (EV1
& EV2) and (EV3 or EV3 & EV4) and when you are asking a contrast of
one submodel, split the other off and enter it as a "confound" matrix
with the -x option.
Cheers.
On 14 Nov 2006, at 06:00, Stephen Rose wrote:
> Hi
>
> I would like to perform a grouped analysis on some FA data using
> randomise. The FA data has
> been generated using tbss. The question I have relates to setting
> up the design matrix using the
> GLM GUI. I would like to investigate the differences in WM tracts
> between two groups of subjects
> given additional information (covariate), such as a cognitive score
> (eg a unpaired two-group
> analysis with a confounding variable). If for argument sake the
> scores for 6 subjects (3 in each
> group) were 21, 22, 23, 23, 22, 21, using the GLM GUI, I am not
> sure how to best set up the
> contrasts. In the example below I have demeaned the scores (EV3).
>
>
> Group EV1 EV2 EV3
> Input1 1 1 0 -1
> Input2 1 1 0 0
> Input3 1 1 0 1
> Input4 2 0 1 1
> Input5 2 0 1 0
> Input6 2 0 1 -1
>
> Contrasts
> EV1 EV2 EV3
> C1 1 -1 ?
> C2 -1 1 ?
>
> Any help would be much appreciated.
>
> Regards
> Stephen
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|