> data^2 is all pure statistics, while I would redo the fit with no MC
> statistics in the fit to get the pure data statistics effect in the MC
> subtraction (and quote the difference in quadrature of the rest as MC
> stat error).
but the central values with and without MC statistics in could be not
the same, right? then in principle, this method is not completely
correct. Anyway, I must admit, this is a small effect. I like this
approach, I vote for it.
Daniele
|