Bhismadev,
If you are using a 'Global Null', you need to use orthogonal
contrasts. If you are using a 'Conjunction Null', you don't need to
have orthogonal contrasts. (If you are using SPM2, make sure you have
all the updates to ensure you see these questions).
-Tom
-- Thomas Nichols -------------------- Department of Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
[log in to unmask] 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029
On Mon, 19 Sep 2005, Bhismadev Chakrabarti wrote:
> dear all,
>
> i am trying to perform two different conjunction analyses at the group level
> using spm2 and was wondering if it's necessary for the individual
> conjunctions to be orthogonal to each other.
> i have 4 conditions, A, B. C, D and two instances of the same baseline Base1
> and Base2.
> the design matrix looks as follows:
>
> A B C D base1 base2
> 1 0 0 0 -1 0 | this is conjunction 1 contrast 1
> 0 1 0 0 0 -1 | this is conjunction 1 contrast 2
> 0 0 1 0 -1 0 | this is conjunction 2 contrast 1
> 0 0 0 1 0 -1 | this is conjunction 2 contrast 2
>
> the two conjunctions i am trying to estimate are one between conditions A&B
> and the other between conditions C&D.
>
> while the contrasts within each conjunction analysis are orthogonal to each
> other, i am not sure if the two different conjunctions are orthogonal - and
> if i can report the results of these two conjunctions separately.
>
> many thanks for any suggestions!
>
> best regards,
>
> bhisma
>
> _______________________________
>
> Bhismadev Chakrabarti
> Department of Psychiatry
> University of Cambridge
> Cambridge, UK
>
|