Chu,
One comment on Will's great list of refs...
> 3. Details of the principles underlying how the T and F statistics
> are derived are given in the paper
>
> http://www.fil.ion.ucl.ac.uk/spm/doc/papers/sjk_heuristic.pdf
The the T & F derivations are relevant, but note that the actual
degrees-of-freedom result of that paper isn't used in SPM2 (Will or
Stephan: Same in SPM5?).
Since the lambda's are estimated from many voxels and then pooled over
voxels, the variance V is assumed to be noiselessly estimated by
sum_j \hat{\lambda}_j Q_j
and hence the DF are the usual n-p, and not those of the heuristic
paper.
-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
|