Dear Anat,
What does SPM do with the contrast weights which you enter? The
answer is, it multiplies each one by the relevant parameter estimate,
sums these, and sees whether the result is significantly different
from zero. This is a useful test if, under the null hypothesis, this
weighted average is expected to be zero.
In your case, under the null hypothesis the parameter estimates for
the first three bins are expected to be equal. The appropriate
weights are therefore /-.5/-.5/+1, because this weighted average
would sum to zero under the null hypothesis (whereas /-1/-1/+1 would
not if the parameter estimates are all equal).
Good luck,
Richard.
>Greetings,
>
>the following question concerns weights given when creating contrasts.
>Having 4 hypothetical bins, and contrasting the first two with the last 2,
>the weights might be /-1/-1/+1/+1. However, if one subject has no
>observations in the last bin, SHOULD they still sum up to zero, that is,
>should the weights for this subject be /-.5/-.5/+1 or should they continue
>to be 1's and -1's (thus not summing to zero)?
>
>thanks,
>- anat
>Anat Maril
>Harvard University
>Department of Psychology
>William James Hall
>33 Kirkland St., Cambridge, MA 01238
>(617) 496-5909
>[log in to unmask]
--
from: Dr Richard Perry,
Clinical Lecturer, Wellcome Department of Cognitive Neurology,
Institute of Neurology, Darwin Building, University College London,
Gower Street, London WC1E 6BT.
Tel: 0207 679 2187; e mail: [log in to unmask]
|