NCJ Number
82349
Date Published
Unknown
Length
0 pages
Annotation
The use of time series analysis to forecast the future course of crime is explained, and the principles of representing time changes with graphs are illustrated. Trends that can be extrapolated from these data and used by policymakers are explained.
Abstract
Time series models are the most viable means of predicting the future from past information; the prediction may take various forms (e.g., amount increases or exponential increases). The time series technique is especially useful when nothing is known about crime causes or the period of future prediction is fairly short. Graphs illustrating time changes tend to be more accurate when data cover a fairly long time period, and measurement intervals are short. Linear estimates may be made for predictive purposes to avoid the problems of data lag. Graphs may show a variety of patterns such as linear increases combined with cyclical patterns of recurrence. Graphs may plot data algebraically or logarithmically, resulting in different extrapolations and varying degrees of divergence from actual observed facts. Sudden drops in observed crime rates over predicted crime rates may be indicative of major technological changes, as in the case of the reduction of auto thefts resulting from improved steering column auto locks. Policymakers can give immediacy to efforts to predict the future. At the same time, the objective estimates of crime trends facilitate accurate and subjective policymaking decisions.