Hi Ben, Alle Meije
> Hi Ben,
> In SPM2 (and I think in 99 as well) the F-map `effects of interest'
> was always generated, I think it was part of clusterwise testing. The
> idea is to use an omnibus test for effects of interest first. This
> gives you a number of clusters which include the ones you're
> interested in. Starting with these omnibus clusters, you then use
> another threshold to select significant clusters. The reason that it
> has gone is probably (?) that the method for doing clusterwise testing
> has changed in SPM5.
I thought the omnibus F test was used (given a threshold) to determine
which voxels to use to estimate the error and sphericity ?? looking at
spm_spm both spm2 and 5 are using it but in spm5 it is not written down .. ?
>> Someone might have asked the question already, but I just noticed that
>> the "effects of interest" automatically generated in spm2 "Results" or
>> early versions of spm is not generated in SPM5. Why is that? And how is
>> it generated if I want to use adjusted data for DCM analysis?
Not sure about Ben question, the effect of interest correspond to a 1
per regressor and per line in the contrast manager (or used the reduced
model line it's faster), i.e. find any voxel that respond to any
conditions or combination of them. Now the adjustement for DCM ?? I
don't get it
Best
Cyril
|