FORECASTING REPORTED CRIME DATA - SOME TESTS OF ALTERNATIVE MODELS (FROM CRIMINAL JUSTICE STATISTICS ASSOCIATION, INC. - PROCEEDINGS..., 1979 - SEE NCJ-63318)

• NCJ Number: 63321
• Author(s): L R MACPHETERS; D E SCHLAGENHAUF
• Date Published: 1979
• Length: 10 pages
• Corporate Author: Justice Research and Statistics Association
• Sale Source: National Institute of Justice/
• Publication Type: Statistics
• Language: English
• Country: United States of America

Annotation

THIS PAPER DESCRIBES TESTS OF ALTERNATIVE MODELS FOR FORECASTING REPORTED CRIME DATA FOR A MEDIUM-SIZED CITY, TAMPA, FLA.

Abstract

RESOURCE ALLOCATION FOR CRIME CONTROL WITHIN A METROPOLITAN AREA MAY BE IMPROVED THROUGH SUCCESSFUL APPLICATION OF CRIME FORECASTING METHODS, YET IT IS PRIMARILY THE LARGER CITIES WHICH HAVE EXPERIMENTED WITH SUCH TECHNIQUES. A NUMBER OF ALTERNATIVE FORECASTING MODELS AND TECHNIQUES WHICH MIGHT REASONABLY BE APPLIED BY A SMALLER RESEARCH UNIT ARE COMPARED, AND THE APPROXIMATE COSTS OF THE METHODS ARE THEN WEIGHED AGAINST A LISTING OF POTENTIAL BENEFITS FROM IMPROVED ESTIMATES OF FUTURE CRIME LEVELS, MADE POSSIBLE THROUGH FORECASTING. THE FORECASTING METHODS TESTED INCLUDE A NAIVE MODEL, VARIATIONS OF SMOOTHING MODELS, AND BOX-JENKINS TECHNIQUES. MODELS ARE RATED ACCORDING TO FORECAST ERRORS. THE RELATIVE COSTS OF THE MODELS ARE REVIEWED, AND SOME BENEFITS OF CRIME FORECASTING ARE DISCUSSED. RESULTS OF THE STUDY SUGGEST THAT (1) ELABORATE MODELS ARE NOT ALWAYS NECESSARY FOR A USEFUL FORECAST AND (2) THE MOST LIKELY TYPE OF INTUITIVE FORECAST UNDERTAKEN BY AN EXPERIENCED POLICE ADMINISTRATOR WOULD PROBABLY YIELD A COST-EFFECTIVE FORECAST FOR A MEDIUM-SIZED DEPARTMENT. TABULAR DATA AND REFERENCES ARE GIVEN. (MJW)