TIME SERIES ANALYSIS OF REPORTED CRIME - A METHODOLOGICAL STUDY

• NCJ Number: 58805
• Author(s): C MARSHALL
• Date Published: 1978
• Length: 222 pages
• Corporate Author: Police Foundation; The Urban Institute
• Sale Source: Police Foundation
• Publication Type: Statistics
• Language: English
• Country: United States of America

Annotation

TIME-SERIES ANALYSIS OF REPORTED CRIME WAS CARRIED OUT AS PART OF THE EVALUATION OF THE COMMUNITY SECTOR TEAM POLICING PROGRAM IN CINCINNATI, OHIO. THIS REPORT PRESENTS THE METHODOLOGY.

Abstract

THE ANALYSIS USED MONTHLY OFFENSE DATA FOR ONE POLICE DISTRICT IN CINCINNATI OVER A 26-WEEK PERIOD. FOUR CRIME TYPES WERE SELECTED (ROBBERY, BURGLARY, AGGRAVATED ASSAULT, AND RAPE) AND SUBJECTED TO THREE DIFFERENT ANALYTIC METHODS. THE FIRST, SIMPLE STATISTICS (MEAN AND STANDARD DEVIATION), IS SPECIFIC AND COMES DIRECTLY FROM DATA. SIMPLE STATISTICS ARE UNCOMPLICATED ENOUGH TO YIELD TO DIRECT INTERPRETATION, BUT MAY NOT BRING OUT ALL THE INFORMTION CONTAINED IN THE DATA. DETAILED STATISTICS, THE SECOND METHOD USED, BRING OUT MOST INFORMATION CONTAINED IN DATA BUT ARE TOO NUMEROUS AND DIFFICULT TO INTERPRET DIRECTLY. STOCHASTIC MODELS, THE THIRD METHOD, USE DETAILED STATISTICS AND REPRESENT MOST DATA INFORMATION IN A RELATIVELY SIMPLE WAY, ALLOWING REASONABLY DIRECT INTERPRETATION. AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (ARIMA) WAS THE STOCHASTIC APPROACH SELECTED FOR THIS APPLICATION OVER BOTH COMPONENTS AND EXPONENTIAL MODELS, SINCE IT CONSIDERS THE DESIRABILITY OF TRANSFORMING THE SERIES DATA INTO A FORM MORE AMENABLE TO STOCHASTIC MODEL FORMULATION--USUALLY LOGARITHMIC AND SQUARE ROOT FORMS. WHEN APPLIED TO POLICE DATA, THE STOCHASTIC MODEL DEMONSTRATES THE BENEFITS OF THESE MODELS AS EXTENSIONS OF OTHER STATISTICAL STUDIES. IT INTRODUCES A NUMBER OF CRIME-RELATED ISSUES: THE POSSIBLITY OF DEVELOPING MEANINGFUL DESCRIPTIONS OF CRIME LEVELS, THE ROLE OF FORM AND NUMERICAL PARAMETERS OF STOCHASTIC MODELS IN INDICATING BASIC ASPECTS OF CRIME, AND THE EFFECT OF AGGREGATION ON TIME-SERIES ANALYSIS. ARIMA, HOWEVER, CANNOT BE USED AS A FORECASTING TOOL TO PROVIDE RAPID DYNAMIC EVALUATION, AND THE STOCHASTIC MODEL SOFTWARE AND METHODOLOGY NEED TO BE IMPROVED TO INCREASE RELIABILITY AND REDUCE COST. THESE ISSUES AND OTHERS ARE IDENTIFIED FOR FUTURE RESEARCH. A BIBLIOGRAPHY IS PROVIDED. (DAG)