Statistical Analysis of Timeseries Data on Problem Solving


So Young Kim and Wesley G. Skogan

During the summer of 2002, the CAPS evaluation team conducted a study examining how Chicago police tackle neighborhood problems. The study focused on the problems most often identified by the police as local priorities. The findings of the study can be found in the main project report: Community Policing in Chicago, Years Eight and Nine.

The fieldwork component of the study was supplemented by statistical analyses of quantitative time series data on crime and 911 calls. This paper describes in detail the research design and statistical methods that were employed for the study. Time series trends in appropriate categories of calls for service and recorded crime data were created for each problem site. Comparable time series data were assembled for matched sets of beats in which the sample problem was not identified as a priority. Crime data were aggregated from information on 3.9 million individual crime incidents. 911 call data were aggregated from 23.4 million calls to the City of Chicago’s Office of Emergency Management and Communications. There is an extensive discussion of the use of Box-Jenkins Intervention Analysis to distinguish between gradual and immediate changes in crime, whether those changes were temporary or permanent in nature, and whether trends in the study beats were unique or matched trends in similar areas of the city.

PDF icon Download working paper PDF