Evaluation of the Shreveport Predictive Policing Experiment

Evaluation of the Shreveport Predictive Policing Experiment

Priscillia Hunt
Jessica Saunders
John S. Hollywood
Copyright Date: 2014
Published by: RAND Corporation
Pages: 84
https://www.jstor.org/stable/10.7249/j.ctt14bs27t
  • Cite this Item
  • Book Info
    Evaluation of the Shreveport Predictive Policing Experiment
    Book Description:

    This report documents an assessment of a predictive policing effort that was conducted to evaluate the crime reduction effects of policing guided by statistical predictions. In addition to a basic appraisal of the process, the report shows the crime impacts and costs directly attributable to the strategy in order to provide a fuller picture for police departments considering if and how a predictive policing strategy should be adopted.

    eISBN: 978-0-8330-8990-8
    Subjects: Sociology, Political Science, Law, General Science, Statistics

Table of Contents

  1. Front Matter
    (pp. i-ii)
  2. Preface
    (pp. iii-iv)
  3. Table of Contents
    (pp. v-vi)
  4. Figures and Tables
    (pp. vii-viii)
  5. Summary
    (pp. ix-xviii)
  6. Acknowledgments
    (pp. xix-xx)
  7. Abbreviations
    (pp. xxi-xxii)
  8. CHAPTER ONE Introduction
    (pp. 1-6)

    Predictive policingis the use of statistical models to anticipate increased risks of crime, followed by interventions to prevent those risks from being realized. The process of anticipating where crimes will occur and placing police officers in those areas has been deemed by some as the new future of policing.¹ Since the early 2000s, the complexity of the statistical models to predict changes in crime rates has grown, but there is limited evidence on whether policing interventions informed by predictions have any effect on crime when compared with other policing strategies.

    In 2012, the Shreveport Police Department (SPD) in Louisiana,²...

  9. CHAPTER TWO PILOT Process Evaluation
    (pp. 7-28)

    This process evaluation describes the PILOT program implementation in Shreveport. Understanding how a proposed program was actually implemented is a critical part of evaluations, as it allows for understanding the source of effects on outcomes. Notably, if a program is not implemented as planned, it would be incorrect to ascribe any effects, whether positive or negative, to the planned program.

    Note that the process evaluation takes no position on correct implementation, and rather describes what was planned, what actually happened in the field, and where there were opportunities and challenges. We present the results of the analysis for the prediction...

  10. CHAPTER THREE PILOT Impact Evaluation
    (pp. 29-40)

    The PILOT strategy is to identify places at increased risk of experiencing crime and to both improve information gathering about criminal activity (leading to arrests for past property crimes and inconveniencing and deterring those considering future crimes) and maintain presence in those places.

    In this chapter, we provide details of the methodology and results estimating the impact of PILOT on property crime.

    Although this evaluation is based on activities during the implementation of the trial, the reliability of results depends on the robust selection of control and treatment groups preimplementation. Therefore, in addition to the estimation strategy and crime data,...

  11. CHAPTER FOUR PILOT Cost Evaluation
    (pp. 41-48)

    Special operations funding in Shreveport is based on a criminogenic need—a district has to state the crime problem to acquire overtime funding. It is not the case that districts can conduct a property crime special operation because there is overtime budget available to do so; in the current system, a district can only access overtime funding for a special operation after a crime surge has occurred. One exception is for special events (e.g., Mardi Gras, winter holiday markets), in which case districts can conduct special operations for the expected increase in criminal activity rather than recent past crime surges....

  12. CHAPTER FIVE Conclusions
    (pp. 49-52)

    This report presents evidence from a process, outcome, and cost evaluation of a seven-month field trial implemented in Shreveport, Louisiana, in 2012. The policing strategy—PILOT—included a prediction and a prevention model aimed to reduce residential, auto-related, and business property crimes.

    Overall, the PILOT program did not generate a statistically significant reduction in property crime. There are several possibilities to explain the null results.

    In part, the finding of no statistical impact may be due to a statistical issue—there were few participating districts over a limited duration, thus providing low statistical power to detect any true effect of...

  13. APPENDIX Technical Details
    (pp. 53-60)
  14. References
    (pp. 61-62)