| Prison population forecasters confront two problems. First, they must produce valid and reliable forecasts of future prison populations, where forecasts are seen as conditional statements about future conditions if the assumptions on which they are based hold. To project populations, forecasters use four general approaches: Microsimulation models are used by 23 States and the Federal Bureau of Prisons; 15 States and the District of Columbia use disaggregated flow models; and 12 States use either statistical or mathematical models. The challenges of producing reliable forecasts are compounded during periods of policy change, as historical relationships and patterns between variables may change, and forecasters must decide what to use to base forecasts on in the absence of history. Second, forecasters have to convince decision makers of the reliability and validity of their forecasts. Accuracy is an important, but not the only, standard for gauging the quality of forecasts, as the independence of the forecaster from departmental or other political pressures is an important consideration for the users of forecasts. In some States, steps have been taken to organize forecasting in ways that are consistent with the principles of objectivity and independence. Policy changes also compound problems associated with the use of forecasts, as policy makers also must confront departures from history.
|
Updated 05/20/2006