when using RooMCStudy and the gen parameters are randomized for each toy through a module derived from RooAbsMCStudyModule (for instance RooRandomizeParamMCSModule), the pulls stored in the fitParDataSet are computed from the fixed initial parameter values, and not from the new varname_gen values.
In fact, having a look here:
it seems that the calcPulls() method only computes the pulls from the randomized gen values if also the _perExptGenParams bool variable is set to true. This happens only if the constraint are provided when building the RooMCStudy, and not when using a toy study module for the randomization.
Is this the case? If so, is it possible to change the behaviour in order to have the pulls computed from the randomized gen variables also when using a toy study module?