 |
Acknowledgement. Chapter contents. Lyria Bennett Moses and Janet Chan. Introduction
|
|
|
|
Acknowledgement
Most of the work reported here has been funded by the Economic and Social Research Council, especially under the Secondary Data Analysis Initiative (Grants ES/F015186/1, ES/K003771/1& 2 and ES/L014971/1 & 2).
CHAPTER CONTENTS
· Introduction 252
· What Is ‘Big Data’ and how Is It Used in Research? 252
· Uses of Data Analytics in Crime Prediction 253
¡ Predictive policing 253
¡ Offender risk assessment 255
· Limitations of These Techniques 258
¡ Limitations in data 258
¡ Limitations of approaches based on correlation 259
¡ Assumptions embedded in techniques 260
¡ Differential impact and discrimination 260
¡ Individual harm 262
· Challenges for Researchers 262
¡ Non-transparency 262
¡ Evaluation 263
· Summary and Review 264
· Study Questions and Activities for Students 265
· Suggestions for Further Reading 265
· References 266
GLOSSARY TERMS
Big Data prediction
intelligence led policing predictive policing
hot spots
offender risk assessment
machine learning algorithm training set
|
|
|
data analytics correlation
USING BIG DATA AND DATA ANALYTICS IN CRIMINOLOGICAL RESEARCH
LyrIa Bennett Moses and Janet Chan
INTRODUCTION
This chapter explains and critiques approaches to policing and crime prediction based on Big Data and data analytics, particularly predictive policing and offender risk assessment. While limitations of these approaches are discussed, this chapter does not take a position on the comparative question, namely whether these approaches are better or worse than alternatives. Rather, it focuses on understanding the kinds of tools and techniques used, the ways in which they are used and the limitations inherent in their use.
The chapter begins by explaining what ‘ Big Data ’ is and how it is used in crimi- nological research. It then examines two uses of data analytics for crime prediction. The first, predictive policing, is a tool used by police to target particular locations and times, or occasionally individuals, where the likelihood of criminal activity is heightened. The second, offender risk assessment, attempts to quantify the risk that a person charged with or convicted of a crime will commit an offence while on bail, comply with bail conditions, re-offend once released on bail or parole, and/or com- mit a violent offence while on bail or parole. These assessments can be carried out in order to assess a bail application, during sentencing or in an application for parole. Similar tools are also used in prisoner management.
In the following section, the tools and their current uses are described. After that, the limitations of these kinds of techniques are described at a general level. Because of the diversity of tools available, and the variety of data on which they operate, it is not possible to explain the advantages and disadvantages of each dif- ferent technique, or compare one to another. However, there are some general matters that need to be considered when assessing predictive accuracy, effectiveness and negative impacts of data-analytic tools in policing and criminal justice. Finally, the chapter describes some difficulties that researchers face in assessing and evalu- ating these kinds of tools. In particular, we discuss the lack of transparency and the difficulties in obtaining data as well as difficulties in conducting a proper evalua- tion of effectiveness and impact.
|
|
|
