Chi2 in RooFit: why no error on prediction?

Hello,

I wonder why a chi2 fit in RooFit does not take the error on prediction into account. AFAIK the denominator, eg. the error, is that of the data only.
Is there a reason not to consider the error on prediction as well (eg. finite statistics if using MC, other sources of errors if using data-driven models etc.)?

Thanks and cheers,

jl

Hello,

Any piece of info from the authors or someone knowledgeable would be greatly appreciated.

Thanks,

jl