Hi Andreas
> Hello Donald and all,
>
> It always struck me as odd that F-contrasts in SPM for a (one-way) ANOVA are specified this way (e.g. [1 -1 0; 0 1 -1]), as these two contrast vectors are non-orthogonal, wheras typical ANOVA's would use a full set of orthogonal contrasts to divide the "treatment" SS, e.g. [1 -1 0; 1 1 -2]. What am I missing here?
the full set of contrasts for all combinations is [1 -1 0; 0 1 -1; 1 0
-1] - test for any differences against error
however since [1 -1 0] + [0 1 -1] = [1 0 -1] that is the last contrast
is nothing else but the combinations of the two other ones there is no
point setting it up (this gives indeed the same result)
Cyril
> Keep you eyes on the list to see if anyone responds about how to
> correct for the fact that you are doing 3 post-hoc tests, which
> inflates the type 1 error.
To also comment on Donald's answer, you could set your threshold at
alpha/3 for your post-hoc tests (although it is a bit conservative)
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|