I am fairly new to RooFit and am learning to use RooMCStudy to do a large number of toy MC experiments. I have a specific question regarding the pull distribution calculated by RooMCStudy.
From a presentation by the RooFit author, I found the definition of pull, of say number of signal event (nsig) as:
pull(nsig) = ( Nsig_fit - Nsig_true ) / error_Nfit
My question is: what is Nsig_true? If I use extended maximum likelihood fit, is it Nsig_true varies from one toy MC to another? Or is Nsig_true a fixed number equals to the input expected mean nsig?