Hi Georgios,
Sorry for forgetting about this.
The maps of beta weights/contrasts should be basically
the same, whether the steps of spatial normalization
and smoothing happen before the first level regression
or after it.
While the significance volumes (t- or F-stats, say) will
_not_ be the same, it is probably not a concern if the
results are just based on group analysis of betas.
But keep in mind the estimates of the smoothness of the
nosie. If those come from the errts time series, then
you would want to warp and smooth the errts similarly,
before estimating its smoothness.
On the other hand, it may be more convenient to map
everything to standard space to begin with, unless there
is some particular need for orig space results.
- rick