SumW2 gives unexpected errors in RooMCStudy

Also, as confirmed in your answer #2 here, I thought SumW2Errors was not supposed affect the errors when fitting unweighted datasets…?