Now I see that you asked both about the variability of the FIR estimates and about the related efficiency. I have only answered the former.
Regarding the efficiency, it is not straightforward. With narrower bins used in FIR modelling more subtle hemodynamic response differences might be modelled, but at the same time you lose power/degrees of freedom. Once again, I think that there is virtually no real difference between 2 and 2.5s, but there might be a troublesome difference if you compare FIR bin widths of 1s and 2s.
|