Dear Ce,
it's a little confusing, in the first post you said you had conducted "a one-way within-subject ANOVA analysis and performed an F contrast to test the condition effects (i.e. main effect)". Now you mention three groups and talk about an F-contrast of [1 0 0; 0 1 0; 0 0 1].
In any case, the contrast [1 0 0; 0 1 0; 0 0 1] does NOT test for group differences in a between-subject ANOVA with one factor group. If you're interested in the main effect of group you have to set up a contrast like [1 -1 0; 0 1 -1]. Same if you're interested in the main effect condition in a within-subject ANOVA. If you're interested in the average activation across conditions / across groups set up a contrast [1 1 1] in case of three columns.
F-contrasts like [1 0 0; 0 1 0; 0 0 1] are called "effects of interest". They test whether the beta value of a particular voxel is significantly different from zero for any of the regressors (be it conditions or groups). In most cases this contrast itself does not make much sense for interpretation of your data. But you can use this contrast to have a closer look at your data. Say you have three groups, and the contrast [1 -1 0; 0 1 -1] shows some significant voxels. But you don't know whether it's due to Group 1 > Group 2 (or the other way round), or due to Group 2 > Group 3, ... So go to one of the voxels, then "plot", "Contrast estimates" and select the contrast [1 0 0; 0 1 0; 0 0 1]. Now you get the contrast estimates for the three groups.
Best,
Helmut
it is not really useful
See here http://mindhive.mit.edu/node/60 :
|