Is Crime Predictable? A Test of Methodology for Forecasting Criminal Offenses

• NCJ Number: 108790
• Author(s): C R Block; S L Knight
• Date Published: 1987
• Length: 165 pages
• Corporate Author: Ottawa Police Planning and Research Section
• Sale Source: National Institute of Justice/; NCJRS Photocopy Services
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This study examined the effectiveness of a standard forecasting method, called ARIMA, in predicting the number of Index robberies, aggravated asaults, burglaries, and larceny/thefts known to police in 1982-1983 in selected Illinois jurisdictions.

Abstract

Results indicate that a standard time-series methodology made it possible to predict the number of criminal offenses in some jurisdictions for some crime types. Crime was more predictable in some jurisdictions regardless of crime type. Serendipitous interventions and unexpected sharp increases and decreases in crime levels repeatedly were found. Larceny/theft was the most predictable crime type assessed. Analysis suggests that degree of predictability may be related to the quality of data collection and recordkeeping. An analysis of the effects of a 1983 change in recordkeeping showed that the amount of increase in robbery offenses associated with the change varied as a function of offense seriousness, while the number of Index aggravated assaults generally increased more as a result of recordkeeping changes than did Index robberies. An analysis of actual change in crime rates found that firearm and knife robbery and other-weapon assault increased, while other-weapon robbery and body-as-weapon assault stayed the same. 55 figures, 16 tables, and 35 references.