The purpose of this study was to evaluate the effectiveness of predictive policing strategies used by the Shreveport, Louisiana Police Department that were developed by crime estimation and predictive models.
This collection is part of a larger two-phase study funded by the National Institute of Justice (NIJ). Following a year long planning phase grant from the NIJ, the Shreveport, Louisiana Police Department (SPD) developed its own statistical models for predicting crime based on leading indicators using multivariate logistic regression analysis for grid city-blocks, 400 feet by 400 feet.
Following the development of a prediction model, the SPD implemented a field experiment of Predictive Intelligence Led Operational Targeting (PILOT), using the predictive model in order to deploy overtime labor to the predicted hot spots. The specific resource allocation decisions and specific strategies to be employed by the officers in the experimental districts was determined by the SPD in monthly planning meetings.
Following the SPD's PILOT trial, RAND researchers evaluated the PILOT trial. RAND researchers conducted multiple interviews and focus groups with the SPD throughout the course of the trial to document the implementation of the predictive and prevention models. Furthermore, RAND evaluated the impact of the PILOT strategies on levels of property crime as a whole and by type of crime by testing differences between treatment and control districts and applying advanced program evaluation techniques (e.g., difference-in-difference methods). Finally, RAND evaluated the expenses of PILOT using a cost savings analysis (CSA) using administrative and supplemental secondary data. The data in this collection only focus on the property crime aspect of the study.
Shreveport Police Department staff developed and estimated a statistical model of the likelihood of property crimes occurring within block-sized areas. Then, using a blocked randomized approach to identify three treatment and control district pairs, districts assigned to the treatment group were given maps that highlighted blocks predicted to be at higher risk of property crime. These three treatment districts were also provided with overtime resources to conduct special operations. The three Control districts conducted property crime-related special operations using overtime resources as well, just targeting areas that had recently seen property crimes (hot spots).
Longitudinal: Trend / Repeated Cross-section
Monthly property crime counts by police district in Shreveport, Louisiana
administrative records data
The dataset (n=91) consists of 8 variables.
The dataset contains time variables that are related to the experimental and control periods within the study. These variables include Month, Year, and Time.
The dataset contains variables related to the experimental and control districts within the study. These variables include District, Chief, and Treat.
The dataset contains variables that were used to measure a reduction in crime as a result of predictive policing tactics. These variables include Crime_total and Manhours.