Hi,

On 30 Oct 2006, at 17:47, Sang Lee wrote:

> Dear FSL list users and staff,
>
> I am a novice in using FSL as well as MRI imaging analysis. I am  
> trying t
> o figure out a design
> matrix where I am looking at two-group average with a covariate.
>
> So, I have two groups (one control group and one patient group),  
> each sub
> jects have completed a
> MRI experiment where the subject was looking at 6 different visual  
> stimul
> i. Each subject has also
> completed a survey questionnaire that has a total score.
>
> With out a covariate, my design matrix is as follows:
>
> EV1 - EV12
>
> EV1 : 1st visual stimulus - group1
> EV2 : 2nd visual stimulus - group 1
> EV3 : 3rd visual stimulus - group 1
> EV4 : 4th visual stimulus - group 1
> EV5 : 5th visual stimulus - group 1
> EV6 : 6th visual stimulus - group 1
> EV7 : 1st visual stimulus - group 2
> EV8 : 2nd visual stimulus - group 2
> EV9 : 3rd visual stimulus - group 2
> EV10 : 4th visual stimulus - group 2
> EV11 : 5th visual stimulus - group 2
> EV12 : 6th visual stimulus -group 2
>
> Each subject will have 6 cope files that will be used as input for  
> each E
> Vs.
> i.e. for the subject’s 1st stimulus cope file (who is in group1),  
> EV1 w
> ill have a value of 1 and the
> rest of the EVs will have a values of 0 (zero).
>
> 6 contrasts and 1 F-test
> 6 contrast = 6 different visual stimulus
> With all F-test turned on for all 6 contrasts
>
> 	EV1	EV2	EV3	EV4	EV5	EV6	EV7	EV8	EV9	EV10 EV11 EV12
> C1	1	0	0	0	0	0	-1	0	0	0	 0	  0
> 	
> C2	0	1	0	0	0	0	0	-1	0	0	 0	  0
> C3	0	0	1	0	0	0	0	0	-1	0	 0	  0
> 	
> C4	0	0	0	1	0	0	0	0	0	-1	 0	  0
> C5	0	0	0	0	1	0	0	0	0	0	 -1	  0
> C6	0	0	0	0	0	1	0	0	0	0	 0	  -1
>

This looks ok - it's the 6-level equivalent of the unpaired t-test,  
but would you not gain extra sensitivity by using the equivalent of  
the paired t-test - i.e. also modelling all the subject mean effects  
as well? See recent postings, particularly on the "qudruple" t-test,  
for examples.

> Now, I have this survey questionnaire that I would like to use it  
> as a co
> variate that has a numerical
> total score. But, I am not sure how to incorporate this covariate  
> into th
> e design matrix. I have gone
> through the FSL online course, but only thing I can find was  
> Unpaired Two
> -group difference
> without covariate or one group with covariate.
>
> I would appreciate it if anyone can suggest how to add in the  
> covariate t
> o my existing design
> matrix or suggest alternative design matrix that would fit what I  
> am tryi
> ng to accomplish.

It depends what you want. If you simply want to include this  
covariate as a confound then you can demean it and add it in as a new  
covariate which you ignore. Or did you want to see how it related to  
modulation of the differential contrasts? What did you want to use  
this extra information for?

Cheers.


>
> Thank you for all your help in advance.
>
> Yours Sincerely,
>
> Sang Lee


------------------------------------------------------------------------ 
---
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
------------------------------------------------------------------------ 
---