NCJ Number
103158
Journal
Journal of Quantitative Criminology Volume: 2 Issue: 3 Dated: (September 1986) Pages: 251-264
Date Published
1986
Length
14 pages
Annotation
A time-series model for predicting Louisiana's prison population was developed using the iterative Box-Jenkins modeling methodology.
Abstract
Data were taken from the Louisiana Department of Public Safety and Corrections data base, a computerized file of inmates' records. Data for the time-series models were the number of inmates for each month from January 1975 through December 1983. The first step in developing the Box-Jenkins model was to examine the time-sequence plot of the Louisiana prison population from 1975 to 1983. The second step estimated the unknown parameters in the tentatively identified model. The final step was a diagnostic check to determine whether or not the selected model adequately represented the given time series. The time-series forecasts were contrasted with the results of regression models and an exponential smoothing model. This was done by examining the deviations of the various model predictions from the actual prison population for January 1984 to April 1985. The time-series model was superior to the others according to the usual measures of predictive accuracy. The time-series predictions were sufficient for short-term correctional planning. A disadvantage of the Box-Jenkins models are their cost, since they require experts for updating and advanced diagnostic checking of the model. 4 tables, 2 figures, and 27 references.