I think it might benefit from an explanation why your regression residuals differ from the government’s. For example, how do you treat cases which resulted in probation? When the sentence is over 60 months, do any effects maintain their significance?