Estimation of parameters in Belkin-Blumstein-Glass Models of the criminal justice system

摘要

Several mathematical models of the criminal justice system have been described by Belkin, Blumstein and Glass and by others. We show that such models can be fitted to data by use of the computational tool, quasilinearization. Initial estimates of recidivism parameters can be refined, and better predictions can be generated. The method extends to the estimation of virgin arrest rates on the basis of data on total arrests, as well as to more general feedback models and other differential- and integral-equation models.