Thanks for the quick reply Eugene.
That is true that I only have one behavioural measure
for each subject and so do need to use it for both sessions.
Basically I have found differences between pre and post and want to
see if this survives the behavioural measure being used as a covariant of no interest.
However are you saying that ev2-ev4 already does this?
thanks,
Joe
>>> Eugene Duff <[log in to unmask]> 02/03/09 6:19 PM >>>
Hi Joe,
As you have it here, you have only have one behavioural measure for each
subject (they are the same for both sessions for each subject). Ev2-ev4
model any differences in the mean signal levels of subjects, and so together
could fit the effect modelled in ev5, if it were present.
Eugene
2009/2/3 Joe Galea <[log in to unmask]>
> I am currently running a higher-level analysis comparing
> pre and post sessions using a paired t-test design matrix.
> Between these sessions the subjects performed some training.
> I would like to add a covariate (of no interest) which is a demeaned
> average of their movement amplitude during training.
>
> 1) Is ev5 the correct way to add my demeaned behavioral covar?
> 2)Is this design matrix correct (shortened version), as at the moment I get
> a rank deficient warning?
>
> group ev1 ev2 ev3 ev4 ev5
> 1 1 1 0.85
> 1 1 1 -0.64
> 1 1 1 0.15
> 1 -1 1 0.85
> 1 -1 1 -0.64
> 1 -1 1 0.15
>
> thanks for any information,
> Joe
>
--
Eugene Duff
FMRIB Centre,
University of Oxford
John Radcliffe Hospital, Headington OX3 9DU Oxford UK
Ph: +44 (0) 1865 222 739 Fax: +44 (0) 1865 222 717
----------------------------------------------------------------
|