Hello Steve,
thanks a lot for your reply, that has been really helpful.
Although my RESEL size is below 1 for individual runs, it is in the range of 6 for my FE analyses and therefore I probably could apply GRF but the FDR approach also works also works well with your suggestions.
However, I still have some questions left.
Mainly, I still do not understand how the dof are calculated.
For my single run with 194 images and 8 predictors (incl. the derivatives) I get dof=188.
For the within-subject FE analysis consisting of 4 runs as above I get dof=751 (with dof, the intensity value obtained from ...stats/tdof_t1 is meant here and in the following, as you suggested in your email)
Trying a third level analysis with 7 subjects I get dof=999, if I use a FE analysis again, and dof=6, if I do a mixed (FLAME1) effects analysis.
Is there any brief explanation of how these dofs are calculated? And is this one of the reasons why ME is more conservative than FE?
I would be happy to understand these issues.
Thanks again for your reply,
best, Meike
|