NCJ Number
85294
Date Published
1982
Length
27 pages
Annotation
Colorado's method of predicting prison populations, the commitment cohort model, uses historical data to predict future commitments by quarter.
Abstract
The primary input is a forecast of the number of people who are likely to be admitted to prison over the next few years, based on two factors -- the number of Colorado males between 18 and 24 and the unemployment rate. With a back substitution method, the model uses quarterly commitments and the quarterly prison population in a population propagation matrix to calculate the aggregate length of stay of each commitment cohort coming into prison. To determine the aggregate length of stay of a commitment cohort, one counts forward across the matrix, and one works the matrix in the opposite direction to calculate the future inmate population. There is a 9-quarter lag between the last cohort for which the aggregate length of stay can be calculated and the most recent commitment cohort. In Colorado, the intervening variable of unemployment affects commitments, while legislation and criminal justice system discretion influence the aggregate length of stay. The Colorado model has been criticized as insensitive to policy shifts, and its effectiveness could probably be improved by disaggregating the cohort into felony classes. Because of its reliance on economic trends, it is best utilized over a 2-year period. Forecasters must have input from State agencies and institutions whose decisions affect the key variables of unemployment levels, at-risk population, and lengths of stay. Practical suggestions for applying the commitment cohort model are presented. The population propagation matrix, formulas, tables, and nine references are included.