73

Abstract of thesis entitled

Use of GIS in Campus Crime Analysis: A Case Study of the University of Hong Kong Submitted by

Chi Pun Chung, Edward for the Degree of Master of Geographic Information Systems at the University of Hong Kong in June 2005

Accurate forecasting of crime can have immense benefits to crime prevention, and if acted upon appropriately, may lead to apprehension of criminals. From the perspective of campus security of the University of Hong Kong, this paper studies the crime situation of the university campus by adopting Geographic Information Systems (GIS) to manage, visualise and analyse the spatial data of crimes. A special technique of GIS, hot-spot analysis, will be employed to reveal the trend, spreading pattern and temporal changes of crime and also help forecast potential future crime spots. GIS, if developed appropriately, can help stipulate effective crime prevention measures and patrol strategies in the campus.

Use of GIS in Campus Crime Analysis: A Case Study of the University of Hong Kong By Chi Pun Chung, Edward

A thesis submitted in partial fulfilment of the requirements for the Degree of Master of Geographic Information Systems at the University of Hong Kong June 2005

Declaration I declare that this thesis represents my own work, except where due acknowledge is made, and that it has not been previously included in a thesis, dissertation or report submitted to this University or to any other institution for a degree, diploma or other qualification.

Signed Chi Pun Chung, Edward June 2005

i

Acknowledgements I would like to express my sincere gratitude to Dr. Ann Mak for her wholeheartedly guidance, knowledge and assistance in this research, particularly her profound influence and helpful comments in the preparation of the study. I am indebted to Miss Kawin Chan for her frankly technical support throughout the research period. Her generous encouragement helped overcome the crucial moment in this research. I would also like to extend my appreciation to Mr. Teddy Wong who provided the crime data of the campus and generously shared his knowledge, expertise and experience to help shape my understanding of the security of the campus.

ii

TABLE OF CONTENTS DECLARATION

i

ACKNOWLEDGEMENTS

ii

TABLE OF CONTENTS

iii

LIST OF FIGURES

vii

LIST OF TABLES

ix

CHAPTER ONE

INTRODUCTION

1

1.1

Research Background

2

1.2

Scope and Objectives of the Research

3

1.3

Chapter Organisation

4

CHAPTER TWO

LITERATURE REVIEW

6

2.1

Introduction

6

2.2

Crime Mapping

7

2.2.1

Crime Analysis

8

2.2.2

Crime Place Research

9

2.2.3

Crime Analysis by Manual Pin Maps

13

2.2.4

Crime Analysis by Geographic Information Systems (GIS)

15

2.3

Crime Analysis Process

19

2.3.1

Analysis Process

20

2.3.1.1

Tactical Crime Analysis

21

2.3.1.2

Strategic Crime Analysis

22

2.3.1.3

Administrative Crime Analysis

23

iii

2.3.2 2.4

2.3.1.4

Criminal Intelligence Analysis

23

2.3.1.5

Police Operations Analysis

24

Crime Analysis Model

24

Problem Solving Philosophy in Crime Analysis

25

2.4.1

Problem-oriented Policing

26

2.4.2

SARA Model

29

2.5

CompStat

32

2.6

Literatures on Campus Crime

37

2.7

Summary

38

CHAPTER THREE

METHODOLOGY

40

3.1

Introduction

40

3.2

The Research Data

40

3.2.1

Study Area – the University of Hong Kong

41

3.2.2

Campus Security

45

3.2.3

Research Tools

48

3.3

Methodological Framework

49

3.3.1

Data Management

52

3.3.1.1

Data Collection

52

3.3.1.2

Data Set in this Research

54

3.3.1.3

Data Classification

56

3.3.1.4

Temporal Data Conversion

59

3.3.1.5

Geocoding

59

3.3.1.6

Data Re-classification

60

3.3.4

Data Analysis

61

iv

3.4

3.3.4.1

Visual Inspection

61

3.3.4.2

Hot Spot Identification

62

3.3.4.3

Hot Spot Analysis Techniques

63

Summary

CHAPTER FOUR

65

CAMPUS CRIME ANALYSIS

66

4.1

Introduction

66

4.2

Campus Crime – the University of Hong Kong

66

4.3

Spatial Analysis

69

4.3.1

Spatial Analysis – Theft

73

4.3.2

Spatial Analysis – Loitering & Peeping Tom

81

4.3.3

Spatial Analysis - Burglary

83

4.4

Temporal Analysis

84

4.4.1

Temporal Identification – Changes in Months

85

4.4.2

Temporal Identification – Changes in Weekdays

87

4.4.3

Temporal Identification – Changes in Hours

88

4.4.4

Temporal Analysis – Theft

95

4.4.5

Temporal Analysis – Loitering & Peeping Tom

97

4.4.6

Temporal Analysis - Burglary

98

4.5

Campus Crime outside Hong Kong

100

4.6

Summary

102

CHAPTER FIVE

103

CONCLUSION

5.1

Introduction

103

5.2

Limitations of Study

103

v

5.3

Analysis Findings

104

5.3.1

Crime in General

104

5.3.2

Prevailing Crimes

105

5.3.3

Forecast of Prevailing Crimes

107

5.4

Recommendations

110

5.5

Suggestions for Future Research

114

5.6

Conclusion

114 116

REFERENCES

vi

LIST OF FIGURES Figure 2.1

Interaction of Crime Pattern Theory

11

Figure 2.2

Problem Analysis Triangle

12

Figure 2.3

Crime Analysis Model

25

Figure 2.4

Incident-driven Policing

27

Figure 2.5

Problem-oriented Policing

29

Figure 2.6

SARA Model

31

Figure 2.7

Principles of CompStat Model

36

Figure 3.1

Sub-division of the study area of HKU

42

Figure 3.2

Population Density around HKU Campus

44

Figure 3.3

Income Groups around HKU Campus

44

Figure 3.4

Sub-division of the campus security of HKU

47

Figure 3.5

Sample of the monthly crime report of HKU

48

Figure 3.6

Methodological framework adopted for this research

52

Figure 3.7

Re-classification process of crime event data

60

Figure 4.1

Crime Statistics of HKU between 2002 and 2004

68

Figure 4.2

Spatial distribution of crimes of HKU between 2002 and 2004

71

Figure 4.3

Hot spots of theft in the Main Campus in 2002

78

Figure 4.4

Hot spots of theft in the Main Campus and the Western Region

79

in 2003 Figure 4.5

Hot spots of theft in the Main Campus and the Western Region

81

in 2004 Figure 4.6

Hot spots of loitering and suspected peeping tom activities in the Main Campus in 2004

vii

83

Figure 4.7

Spatial distribution of burglary in the Main Campus between

84

2003 and 2004 Figure 4.8

Comparison of frequency of crime by months between 2002

86

and 2004 Figure 4.9

Comparison of accumulative frequency of crime by month

87

between 2002 and 2004 Figure 4.10 Frequency of crime by weekday between 2002 and 2004

88

Figure 4.11 Frequency of crime by hours between 2002 and 2004

91

Figure 4.12 Temporal distribution of crime in the Main Campus between

92

0800 and 1800hrs in 2002 Figure 4.13 Temporal distribution of crime in the Main Campus between

92

1800 and 2400hrs in 2002 Figure 4.14 Temporal distribution of crime in the Main Campus between

93

0800 and 1800hrs in 2003 Figure 4.15 Temporal distribution of crime in the Main Campus between

93

1800 and 2400hrs in 2003 Figure 4.16 Temporal distribution of crime in the Main Campus between

94

0800 and 1800hrs in 2004 Figure 4.17 Temporal distribution of crime in the Main Campus between

94

1800 and 2400hrs in 2004 Figure 4.18 Temporal variation of theft between 2002 and 2004

97

Figure 5.1

108

Forecast of crimes in the Main Campus and the Western Region

Figure 5.2

Forecast of crimes in the Southern Region

108

Figure 5.3

Crime Statistics of campus crime between 2002 and 2004

109

viii

LIST OF TABLES Table 2.1

Historical Statistics of CompStat between 1993 and 2003

33

Table 3.1

Extract of raw crime data of HKU between 2002 and 2004

53

Table 3.2

Statistics of crime data with reported and non-reported time

55

between 2002 and 2004 Table 4.1

Crime statistics of HKU between 2002 and 2004

67

Table 4.2

Comparison of Overall Crime Rate between HK and HKU

68

between 2002 and 2004 Table 4.3

Comparison of Overall Violent Crime Rate between HK and

69

HKU between 2002 and 2004 Table 4.4

Geographical distribution of crimes of HKU between the year

70

2002 and 2004 Table 4.5

Crime statistics of Main Campus (M), Western Region (W),

71

and Southern Region (S) of HKU between 2002 and 2004 Table 4.6

Victimisation rate of theft of notebook computer between 2002

72

and 2004 Table 4.7

Victimisation rate of theft of wallet and cash between 2002

73

and 2004 Table 4.8

Prominent locations of theft between 2002 and 2004

73

Table 4.9

Prominent locations of theft of notebook computer between

74

2002 and 2004 Table 4.10

Prominent locations of theft of wallet and cash between 2002

75

and 2004 Table 4.11

Prominent locations of theft of projector between 2002 and

ix

76

2004 Table 4.12

Classroom facilities in the Main Campus buildings

77

Table 4.13

Statistics of known and unknown crime records between 2002

89

and 2004 Table 4.14

Statistics of known crime records by hours between 2002 and

90

2004 Table 4.15

Temporal distribution of theft between 2002 and 2004

95

Table 4.16

Temporal distribution of loitering, peeping tom and indecent

98

exposure between 2002 and 2004 Table 4.17

Temporal distribution of burglary between 2002 and 2004

98

Table 4.18

Statistics of theft of notebook computer and projector between

100

2002 and 2004 Table 4.19

Comparison of campus crime statistics between HKU and US Universities in 2003

x

101

CHAPTER ONE INTRODUCTION

In the recent decades, computerised mapping through the use of Geographic Information Systems (GIS) has become a valuable and popular tool for spatial analysis. Law enforcement agencies in Western countries have utilised this advanced technology to its full extent in identifying spatial pattern as well as exploring temporal relationship as an aid in problem-solving and decision-making processes to reduce and control crimes, and improve efficiency in the allocation of police resources.

Criminologists have long been interested in the study of criminal behaviour and its motivation. Some have studied the environmental influences of criminals and have found that their offending behaviour often appears to form discernible patterns and is not randomly distributed. This spatial characteristic has been carefully examined and studied to suggest that crime occurrences can be very informative. Schools of environmental criminologists later developed the crime place theory with various supporting observations and arguments that crime can be profiled from a different perspective. Many of the analytical techniques developed together with well-structured crime databases allow the practitioners to test and experiment various spatiotemporal hypotheses. The provision of information generated by computerised mapping has been extremely valuable in real situations in providing a better picture of patterns and trends in crime. The success of its application in crime control and crime reduction increases the confidence of the public and secures better co-operation between law enforcement agencies and the public. 1

1.1

Research Background

In the eyes of the public, the campus of the University of Hong Kong has always presented a peaceful and harmonic impression of an elegant tertiary institution educating elites of the society of Hong Kong. Crimes occurring on the campus have never caught much attention from the members of the public based on a simple perception that well educated people do not commit crime, or even any misdemeanour act. This belief understates the seriousness of crime and it can be quite threatening.

In actual fact, the crime rate of the campus is really low in terms of crime per population of Hong Kong and almost all of the crimes reported were minor crimes. Yet minor crime can always turn into serious crime if it is left unchecked, especially as the campus is not usually patrolled by the police but by the campus security guards. It is for this reason the study of campus crime necessitates the management have a genuine appreciation of the overall crime situation to secure the long-enjoyed reputation and to ensure that the frontline security personnel have a tactical approach to prevent crimes from prevailing.

Indeed, there are many ways of studying crime trends and crime pattern. Through the history of past incidents, it is possible to see the trend of crime and suggest ways to control its spread. Through the records of previous incidents, it is also possible to identify the pattern of crime and increase the chance of detection and apprehension. Cartographic presentation of crime clusters is always the most direct and simplest way to promote one’s knowledge and understanding, it collaborates with an old saying “a picture is better than a thousand words”. 2

Nevertheless, this is not to suggest GIS computerised mapping is a panacea that can cure all problems of crime, it is merely an analytical tool, which is able to assist the management and frontline security personnel in reviewing crimes away from their traditional angle.

1.2

Scope and Objectives of the Research

This research represents the first attempt of its kind in adopting GIS crime mapping techniques to visualise crime incidents which occurred on the campus of the University of Hong Kong between the year of 2002 and 2004. The primary concern of this research is to introduce the techniques of GIS crime mapping to the estate management and frontline patrols of the University of Hong Kong who are mostly unfamiliar with GIS. The demonstration of the capability of GIS in identifying spatial and temporal distribution of crimes and the methodology of analytical functions and crime prevention strategies employed in this study will be explicable to them. Subsequently, they would be able to appreciate the crime situation by visualising the aggregating concentration of crime clusters in space and time and react immediately to allocate sufficient resources effectively in the area required for their attention.

The research will attempt to present a balanced view from a technical and management perspective with a view to proposing some feasible real world solutions. In summary, the focus of the current study is to: (1)

explore data management and data visualisation techniques in presenting crime data;

(2)

examine the feasibility of applying spatial and temporal analysis in 3

the campus environment; and (3)

recommend practical crime control and crime reduction strategies in the campus environment.

1.3

Chapter Organisation

Chapter one provides an introduction to the current research by discussing the background to the study and outlining the research foci. The significance of the current study is highlighted through a discussion of the needs to appreciate the past in order to understand the crimes in the future.

Chapter two presents a review of the literature. This chapter highlights the development of computerised mapping and the centrality of work in crime place theory strengthening the relationship between them in the application of crime mapping. In addition the functionalities of various crime analysis are discussed, and a considerable body of empirical research draws the attention to the problem-solving technique and the citation of a successful example on crime mapping. Toward the end, it also touches on the lack of literature on the use of GIS in campus crime analysis.

Chapter three describes the methodology adopted for the current research. Information pertaining to the validity and reliability in the data collection stage is presented. The benefits of using various techniques in data management, such as data integrity and data reclassification, and the procedures are also described. The chapter also describes the visual identification method to locate the hot spots of crime. 4

Chapter four presents the findings of this research from spatial analysis and temporal analysis. The analysis focuses on some of the prevailing crimes and intends to predict the future crime trend as well.

Chapter five presents a summary of the research and discusses some of the limitations raised in the study. Finally, it provides some recommendations to improve crime control and crime prevention measures, and future areas for research are also discussed.

5

CHAPTER TWO LITERATURE REVIEW

2.1

Introduction

This chapter will first review the classical studies of crime mapping from the perspective of sociology and criminology in section 2.2. This spatial analysis of crime shows how a close association between people and place and crime location soon becomes a focus of studying crimes occurrence. In the early days, crime analysis by means of pin mapping did produce fruitful results but technology and philosophy at that time restricted crime mapping from further development. The crime place research in Section 2.2.3 will summarise the principle of crime mapping stemming from the major criminological theories.

In Section 2.2.4 will show that the introduction of Geographic Information Systems (GIS) has evolved from a traditional crime analysis to a new era in crime mapping. The automation of crime mapping is not only capable of visualising instant high quality geographic distribution of crime incidents, it is also capable of making enquires from a series of complicated variables to perform spatiotemporal crime analysis. Section 2.3 will discuss various crime analysis techniques developed in support of the growth of GIS, which in turn enable law enforcement agencies at various levels to devise effective tactics and strategies in crime control and crime reduction programmes.

Section 2.4 will illustrate the transformation of conventional policing 6

philosophy to Problem-Oriented Policing (POP) philosophy, which when later supplemented by the SARA model gives a further boost to the development of qualitative and quantitative crime analysis. Research has proved that a proactive approach in policing together with detailed crime analysis can eradicate the root problem of a crime and may even assist in predicting future crime. Finally, Section 2.5 will examine the structure of crime analysis conducted by tertiary institutions in the US as part of their current crime prevention strategy.

2.2

Crime Mapping

Mapping has a long history. Maps have served for a wide range of purposes in human civilisation but it was not until two centuries ago that maps were used for showing crimes occurrence (Weisburd and McEven 1998; Boba 2001). According to the book “Mapping Out Crime: Providing 21st Century Tools for Safer Communities” published by the United States Department of Justice in 1999 reveals that it could be traced back to 19th century in France where cartographers first analysed national patterns of crime (Quetlet 1835 as reported in Weisburd and McEven 1998). Another source cites that spatial studies of crime and delinquency were written by sociologists and criminologists dating back to about 1830 (Harries 1999). A classical example by Shaw and McKay (1942) who conducted spatial analysis on juvenile delinquency in Chicago showed that there was a substantial correlation between delinquency and various social conditions in Chicago (Weisburd and McEwen 1998).

Maps are static and compiled for designated purposes and users. In the early days of crime mapping, it encountered a great deal of limitations, as it could neither 7

be manipulated or queried, nor could it be produced or amended in a short period of time. However, cartographers pioneering in the field of crime mapping already recognised it as an integral part of the process of crime analysis (Harries 1999).

Crime mapping is not a new term coined, however, until the emergence of Geographic Information Systems (GIS) – a computerised spatial analysis of crime incidents becomes widely accepted and through its continuously development GIS has gradually obtained more weight than before in crime analysis.

2.2.1

Crime Analysis

Crime happens everywhere and anytime – both space and time. Crime has never been continuously distributed and the distribution of criminal activities can be clustered, dispersed or sparse and some activities can occur more in daytime than in night time, or more in summer than in winter. Spatial analysis allows patterns to be measured by means of simple reference systems and temporal analysis can be measured in many different ways, such as time interval: day, week, month and season.

Many psychologists and criminologists have tried to study these criminal acts by focusing closely upon the offender’s internal motivation. As early as in 1830’s, French social ecologists Guerry and Quetelet were already interested in examining the concentration of crime in distinct types of communities ((Weisburd and McEven 1998; Anselin 2000). Over the years psychologists and criminologists have studied various criminal behaviours, which were attributed to many causes, such as genetic theories, anti-social theories and personality dimensions theories 8

(Ainsworth 2001). Recent researches have shown that environment tends to suggest attractiveness of a place for crime and disorder to take place (Harries 1990). Nevertheless, analysing crime from the perspectives of sociologist, psychologists and criminologists requires long-term observation and study, but this can assist the law enforcement agencies in gaining a more comprehensive understanding of criminal intent.

For crimes to occur (except today’s electronic crimes), offenders and victims, and perhaps their properties must exist at the same location. The act committed by an offender against a victim, a criminal act, constitutes a criminal offence. In order to study the occurrence of a specific crime and before establishing a hypothesis or identifying a crime pattern, it is necessary to understand the attributes of a crime. Indeed, there are many factors which influence offenders in committing crimes, such as mens rea (guilt intent), modus operandi (methods of operation) and properties involved to develop a correlation of a crime incident. Interestingly enough, crime patterns are not limited to offender and victim attributes. The location and time of a crime may have a unique characteristic explaining why people choose to break the law. Therefore, it is important to understand where and why crimes take place.

2.2.2

Crime Place Research

There are many other criminological theories providing a basic explanation for constructing a theory of crime places. Three recent and prominent theories: crime pattern theory, rational choice theory and routine activity theory have influenced the basic understanding of the relationships between crime and place.

9

Brantingham P.L. and Brantingham P.J., who are the founders of environmental criminology, developed the Crime Pattern Theory. They hypothesise that crime is the result of people’s (both offenders and potential victims) interaction and movement in the urban landscape in space or time (Brantingham and Brantingham 1984). So, for a crime to take place it must consist of four essential elements: a law, an offender, a target and a place (Brantingham and Brantingham 1981). The theory they developed suggests that criminal opportunities at a location open to the attention of offenders have an increased risk to become targets (Eck and Weisburd 1995). Their further research on location quotients and crime hot spots reiterated that crime occurs in spatial patterns and therefore some parts of a city experience more crimes than others, and that some crimes seem to congregate near certain types of locations. They also affirm that crime has never occurred randomly in time and in space and there are temporal patterns, for example, bar assaults are evening events (Brantingham and Brantingham, 1995). From their observation, there is always a cognitive connection or potential link between certain crimes and designated places. Likewise, certain crimes occur at a particular time.

10

Figure 2.1: Interaction of Crime Pattern Theory Source: Compiled by author based on Crime Pattern Theory, Brantingham and Brantingham, 1995

Rational choice theory constructed by Cornish and Clarke (1986) suggests that offenders will weigh the risk involved in committing a criminal act before selecting target and the basic rationale for selecting a place is important to achieve their goals. In other words, successful crime prevention measures will increase the cost of offending and at the same time reduce the likelihood of rewards. However, this theory may well be able to explain target selection for some types of crime and for some types of offenders but is less helpful in explaining target selection in other forms of crime (Ainsworth 2001). For example, some burglars, the steal-to-order type burglars in particular, may ignore any preventive measures, such as CCTV, or any other forms of risk and still attack the targeted premises which appears to offer them fruitful results.

11

Another well-known theory is routine activity theory (or crime triangle) formulated by Cohen and Felson (1979) in which they explain that predatory crime occurs when a likely offender and potential target come together in time and place without the presence or effectiveness of other types of controllers. An intimate handler, such as parents or friends; or a capable guardian which includes human actors, such as police or security guards, or physical devices, such as CCTV monitors; or a place manager, such as bartenders or shop managers, who is capable of protecting their belongings and security devices (Eck and Weisburd 1995). That is to say, in order for a crime to develop, there must be a motivated offender and a desirable target at the same place and at the same time while the controller is absent or ineffective (Felson 1995).

Figure 2.2: Problem Analysis Triangle Source: The Problem Analysis Triangle, Goldstein, H (2001) at http://www.popcenter.org/about-triangle.htm accessed in Jan 2005

12

Obviously, all of these three theories can be valid in their contexts and situations. But of greater importance is that all three theories conclude that place has a direct connection to crime. Only through comprehension of these theories will analyst be able to investigate the interaction of the physical and social environment with the choice of targeting.

2.2.3

Crime Analysis by Manual Pin Maps

Traditional crime analysis was always conducted by deduction method and investigation of crime with emphasis on who committed the crime and why the crime was committed, leading to the end purpose of apprehending the criminal and bring closure to the criminal case (Vellani and Nahoun, 2001). This philosophy of policing before 1980’s or before the development of Problem-Oriented Policing (POP) by Goldstein in 1979 was considered to be reactive and ineffective (Goldstein 1979, http://popcenter.org/about-whatisPOP.htm).

All along, police tend to incline heavily on criminal intelligence to detect crimes and even today the reality remains much the same. That was particularly true before 1980’s or even much earlier because the police perception of crime analysis was mostly limited by their abilities, knowledge and skills, particularly computer skills (Spelman 1988). Mindset of many officers remained focused on their short-term problems rather than long-term problems (Grinder 2000). No wonder many of them relied very much on their police instinct and very often their common sense too, and put less emphasis on crime analysis. Nevertheless, there were conventional techniques developed to assist them in understanding crimes and 13

mapping crimes. One of the most common techniques used in analysing crimes was to examine the frequency of specific crimes, the temporal analysis, in order to derive a time sequence model for preventing any future occurrence of crime in their crime prevention strategy. Calculation could be done easily with simple formulae and manual mathematics. Yet, this approach only solved part of the puzzles, the prediction of a crime as to where it was going to happen was far beyond the technology capabilities at that time.

Another way of analysing crime is by way of plotting crime incidents on a map. This manual approach of pin mapping was useful for visualising where crime occurred but there were certain limitations: (1) when updating the maps the previous details were lost; (2) the maps could not be archived except by photographing them; (3) maps were static and therefore they could not be manipulated and queried; (4) maps could be difficult to read when several types of crime were mixed, especially when coloured pins were used (Harries 1999); and (5) locations of crime were approximate and not exact (Vann and Garson, 2003). Having said that, pin mapping is still regarded as a simple method for analysing and presenting tactical crime pattern data though limited from combining other data for an in-depth analysis (Velasco and Boba 2000). It sometimes serves its purpose adequately and successfully, particularly in smaller communities with no major crimes. For instance, an example quoted by Vann and Garson (2003) about mapping crime in a small tourist town in western North Carolina with approximately 13,000 residents and the most serious incidents ever reported were vandalism, auto accidents and alcohol-related incidents. That is to say, using manual pin maps to visualise crime patterns remains practical but rather preliminary, rough-cut at problem identification, and very much dictated by the size of an area and the amount of crime. 14

Even though pin mapping has been practiced for some time and up till now in many law enforcement agencies it is more commonly used for informative purpose than for analytical purpose. In fact, before GIS came into operation not much spatial analysis could be done by pin mapping because of its aforementioned limitations. There was also little sense of organising theories or perspectives developed to integrate criminal activity maps into police operations (Weisburd and McEven 1998).

No doubt, mapping crime is capable of assisting law enforcers in uncovering spatial relationships of a crime. The definition of crime mapping is a study which involves manipulation and processing of spatially referenced crime data in order to have it visually displayed in an output that is informative to the particular user (Bowers and Hirschfield 2001). The obvious distinction between crime mapping and GIS can be described that the former is a software while the latter is a hardware. A combination of the two allows crime analysts to understand the occurrence of crime more effectively from the perspective of spatial and temporal analysis.

2.2.4

Crime Analysis by Geographic Information Systems (GIS)

The earliest applications of using GIS in crime analysis conducted by Pauly, Finch and McEwen happened in 1967 (Wiesburd and McEwen 1998). They used mainframe computers and punch cards to produce black and white shaded choropleth maps from a line printer outlining the distribution of a type of crime in St. Louis. This pioneering work conducted by St. Louis Police Department was done with the intention of establishing a Resource Allocation Research Unit to improve 15

the efficiency of patrol operations (Harries 1999). Certainly, these maps proved to be pragmatic in their use and soon they were observed to possess a great deal of potential for understanding spatial distribution of criminal activities and for assisting the management in better deployment of police resources in problematic areas. Research eventually expanded by plotting the boundaries on these maps and since then maps could be used for crime mapping purpose. All in all, the system developed in this way was basically an informative crime mapping system that produced maps of crime distribution and yet it was still lacking analytical capability.

The development of micro-computers in the late 1970’s, faster speed of processing power of micro-computer in the 1980’s (Boba 2001), larger storage and networking capability of micro-computers in the 1990’s (Rich 1999) together with sophisticated mapping software (Rich 1995), has meant that adopting GIS in crime mapping on desktop computer has created a new age of the history of crime mapping. Not only does it become an affordable analytical device, it also provides user-friendly operations in handling complicated queries, higher compatibility in data exchanging (Weisburd and McEven 1998) and better connectivity in sharing of crime information with other agencies (Vann and Garson, 2003).

Mapping crime data is a scientific process and without explicit theory in crime analysis, the value of crime assessment can only rely on the proficiency of the analyst’s personal understanding of relations between crime and space (Eck 1998). However, criminological theories of crime and places have been developed along with the growth of computer technology and practical experiences have been shared and integrated crime mapping into law enforcement operations. Crime mapping has become a well-established discipline of science in crime prevention. 16

The versatility of computer technology and the advance of criminological theory in the past two decades brought the automated crime mapping into being Geographic Information Systems (GIS). GIS is a computerised mapping system that permits stacking of information layers to produce detailed descriptions of conditions and analysis of relationships among variables (Harries 1999). It also provides a digital representation that enables the user to map crimes analytically, not just descriptively (La Vigne 1999). Mapping crime incidents on GIS turns out to be no more a time-consuming and tedious task as it is often confused by a series of possible variables in solving a crime. Its output provides instant analysis for immediate police action that improves the overall efficiency of modern policing. Moreover, GIS is capable of combining data, including temporal data, from various databases, which share common geographic features, and many others from sources outside the police, to perform layering operations for complicated crime analysis. In this way, spatiotemporal study of a crime analysis becomes more viable than before.

Like many computer systems, GIS is a combination of hardware and software. The configuration of GIS basically consists of four major sub-systems: (1) the data input sub-system for creating, importing and accessing data; (2) the data storage and retrieval sub-system for storing and retrieving data; (3) the manipulation and analysis sub-system for performing database management function and analysing geographical data; and (4) the reporting sub-system for producing visual representation of the data on a computer screen or a map printout. When comparing the advantages of a GIS map with the limitations of a pin map, it is not difficult to find out that a GIS map can overcome all the shortcomings of a pin map and that is the main reason for technological replacement. 17

The momentum of the use of automated crime mapping continues to grow. In 1997, National Institute of Justice (NIJ) determined to create its Crime Mapping Research Centre (CMRC) in United States to advance the use of analytic mapping in research and practice (Travis 1998).

In 1998, Vice President of the United States of America, Al Gore established a Task Force on Crime Mapping and Data-Driven Management to further the efforts of the Clinton-Gore Administration to reduce and prevent crime (Rich 1999). Quoted from a book “Mapping Out Crime” published by the United States Department of Justice in 1999, Gore appraised the use of crime mapping that “Maps can represent every dimension of a community …. They can show how healthy a community’s children are, where social services are most needed and most effective, and ways to protect the safety of each citizen …. Innovative communities are using maps to mobilise resources to solve their toughest problem.” Utilising GIS in crime mapping once again has been proven to provide the communities with an innovative approach in targeting crime and supporting decision-making in crime management.

Many police forces, particularly in Western countries, have quickly adopted GIS in a variety of police operational situations and crime prevention initiatives. CMRC conducted the nationwide Crime Mapping Survey in 1997 indicating that over 94% of the surveyed police departments with crime mapping capability used it to inform officers and investigators of crime incident locations; 56% to make resource allocation decisions; 49% to evaluate interventions (Mamalian et al 1999).

Griffin (2001) quoted from the Police Foundation survey of American police 18

departments showed that in 1998 almost 70% of large departments (100+ sworn personnel), and 40% of smaller departments (between 55 – 99 sworn personnel) have engaged in some form of crime mapping (Weisburd, Greenspan and Mastrofski 1998). In 1999, 44% of police forces in the United Kingdom have a crime mapping facility (Ratcliffe 1999). Canada, Australia and other nations have embarked on similar programme in the field as well.

That the concept of crime mapping has been widely accepted by many law enforcement agencies nowadays can be evidenced by a massive number of successful analysis reports published in Crime Mapping Case Studies: Successes in the Field (La Vigne and Wartell (Ed) (1998 and 2000). The strengthening of computer literacy among police officers also boosts their proficiency in handling spatial analysis (Griffin 2001). More and more officers begin to realise the significant advantages of crime mapping in police deployment to tackle prevailing crimes. The growth of these practices and efforts promote the availability of spatial data from diverse sources and the accumulation of information and knowledge strengthen the crime mapping capability for future prediction of crimes. Apparently, over the past few years after the said surveys, all these favourable factors encourage the utilisation of GIS in crime mapping worldwide. GIS has become a universal and effective tool in analysing and preventing crime. More importantly, GIS contributes greatly in decision support management where law enforcement resources are allocated more efficiently.

2.3

Crime Analysis Process

Crime analysis is an important process to law enforcement in understanding 19

the occurrence of a crime. It involves the collection and analysis of data relating to a criminal incident, offender and victim, and develops information of use for crime prevention and detection activities. Crime analysis is defined as a set of systematic, analytical processes directed at providing timely and pertinent information relative to crime patterns and trend correlations to assist operational and administrative personnel in planning the deployment of resources for the prevention and suppression of criminal activities, aiding the investigative process, and increasing apprehension of offenders and the clearance of outstanding investigation (Gottlieb 1994). Therefore, the ultimate goal of crime analysis is to identify and generate the information required for making appropriate decisions in deploying a suitable amount of resources to prevent and control crimes. In addition, crime analysis can be used to evaluate the effectiveness of crime prevention programmes, develop policy through research and help identify or define a problem (Canter 2000). It can also inform policy and decision makers about the actual or anticipated impact of interventions, polices, or operational procedures (Boba 2000). Different from intelligence analysis, crime analysis aims at identifying patterns and trends of a crime while the former aims at examining the association and identification of criminals with any criminal activity.

2.3.1

Analysis Process

The book Exploring Crime Analysis published by the International Association of Crime Analysts (2005) presents a more precise definition for a crime analyst in conducting crime analysis which "focuses on the study of crime incidents, the identification of patterns, trends, and problems; and the dissemination of information to develop tactics and strategies to solve patterns, trends, and 20

problems.” In the same book, crime pattern can be defined as when two or more incidents are related by a common casual factor, usually to do with an offender. Trend represents long-term increases, decreases or changes in crime. The concepts of pattern and trend provide an overarching framework to identify relationships of crimes.

By and large, there are five major types of crime analysis conducted on a regular basis by law enforcement agencies: tactical crime analysis, strategic crime analysis, administrative crime analysis, criminal investigative analysis, and police operations analysis.

2.3.1.1

Tactical crime analysis

Tactical analysis provides information to assist operations personnel (patrol investigative officers) in the identification of specific and immediate crime problems and the arrest of criminal offenders. Analysis data is used to promote response to field situations (Gottlieb 1994). The goal of tactical analysis is to: (1) identify emerging crime patterns as soon as possible; (2) complete comprehensive analysis of any patterns; (3) notify the agency of the pattern’ existence; and (4) work with the agency to develop the best strategies to address the pattern (Bruce 2004). In comparison with all the related crime incidents, it is always possible to identify some commonalities among them. The most significant element in tactical crime analysis is to identify the pattern and that can be achieved in a number of ways, from tabulation comparison to statistical analysis and from simple pin map to automated GIS mapping. This timely qualitative analysis enables the frontline officers to make good use of the available resources for interdicting recent criminal and potential 21

criminal activity and leading to the apprehension of offenders. After all, tactical crime analysis relies totally on the methods however refined and scientific, flexible and intuitive that the analyst can utilise.

GIS has an important role to play in the compilation of tactical crime analysis mapping – a common process of using GIS in combination with crime analysis techniques to focus on the spatial context of criminal and law enforcement activity (Boba 2001).

2.3.1.2

Strategic crime analysis

Strategic analysis is concerned with long-range problems and projections of long-term increases or decreases in crime (crime trends). Strategic analysis also includes the preparation of crime statistical summaries, which are generally referred to as exception reports (Gottlieb 1994). Since there are always changes in the operations of criminal activity, such as spatial and temporal modification, targeting properties and modus operandi. From time to time, resources allocated for tackling will vary according to the crime situation. Exception reports are an effective means to deliver the information for better communication. Strategic crime analysis incorporates two primary functions: (1) to assist in the identification and analysis of long-term problems; and (2) to conduct studies to investigate or evaluate responses and procedures (Boba 2001). After a long-term study, trends can be mapped and tested by repeated hypothesis and subsequently provide a findings of the correlation of a specific crime. From the perspective of the management, these crime projections are useful in the anticipation of crime trends and future challenges and the assessment can assist them to have an insight to initiate strategies, priorities, 22

resource deployment and organisational and planning needs (Baker 2005). Crime trend is the focal point of compiling a sound strategic crime analysis.

2.3.1.3

Administrative crime analysis

This is different from the previous types of analysis in that administrative crime analysis helps facilitate strategic goals (Baker 2005) and focuses on the provision of economic, geographic, or social information to administrators (Gottlieb 1994). It refers more to presentation of findings rather than to statistical analysis or research (Boba 2001) and it is a broad category including an eclectic selection of administrative and statistical reports, research and other projects not focused on the immediate or long-term reduction or elimination of a pattern or trend (Bruce 2004). Administrative crime analysis supports law enforcement agencies to initiate special research projects, feasibility studies and questionnaire surveys which ultimately provide additional information to the administration for better crime prevention responses. Its subsidiary outcome may improve and promote public relationships by making available a picture of the overall crime situation for publication to the public.

2.3.1.4

Criminal investigative analysis

Criminal investigative analysis is the study of criminal personality behaviour, especially in violent crimes. It involves profiling of offenders, victims and geographical features with a view to linking and solving current serial criminal activity. This is a very specific type of crime analysis that is primarily carried out by law enforcement agencies (Boba 2001) as it requires professional skills and a high 23

degree of expertise.

2.3.1.5

Police operations analysis

Police operations analysis depicts the review of the organisation and operations of a police department. Generally speaking, it involves the measurement of effective allocation of police resources in deterrence of crime and disorder (Bruce 2004).

2.3.2

Crime Analysis Model

From the perspective of information requirements, the crime analysis model introduced by Boba (2001) explains the relations between the aforementioned first four types of crime analysis and the levels of aggregation. In short, types with low levels of aggregation aim at individual cases and use qualitative data and analysis techniques and those with high levels of aggregation aim at a limited scope of larger amounts of data and information. Tactical crime analysis relies heavily on latest and immediate crime information with the aim of putting together a timely and qualitative analysis of crime patterns. Strategic crime analysis depends greatly on massive quantified crime data for statistical operations. Administrative crime analysis when compared with the former two analyses depends on additional information, perhaps the inclusion of economic, geographic, or social information, for the compilation of an overall crime summary. Criminal investigative analysis, however, deals primarily with profiling information.

24

Figure 2.3: Crime Analysis Model Source: Boba 2001

From the perspective of crime analysis outcome, these five categories of analysis produce different outcomes. The ultimate aim of tactical crime analysis is to deter reoccurrence of a crime with an intention to apprehend the offender. Strategic crime analysis serves to inform the concerned stakeholders of the prevailing crime situation. Administrative crime analysis fosters closer public relationship and garners community support. Police operations analysis helps with the preparation of annual budgeting and resources allocation. The ultimate expectation of criminal investigative analysis is no more than to bring the serial offender to justice.

2.4

Problem Solving Philosophy in Crime Analysis

As GIS is capable of handling large volume of spatial information and performing high-speed analysis. The use of GIS in policing has two broad applications in crime analysis: tactical crime analysis and strategic crime analysis 25

(Canter 2001). In tactical crime analysis, GIS can be used to map recent criminal incidents and related information for the purpose of identifying a crime problem for interdiction or prevention. Visualisation of GIS allows effective display of incident locations over time or case attributes suggesting commonalities in target, offender, or victim. Together with the integration of relevant information, GIS can improve an analyst’s ability to associate crime with other factors. As such, the use of GIS in tactical crime analysis boosts the accuracy of a crime assessment. In strategic crime analysis, since most information has a geographic component, GIS can review these data and information collected for qualitative or quantitative analysis and determine whether a crime problem was displaced, reduced or unchanged within a target area. GIS can, therefore, be deployed to assist in formulating crime prevention programmes, planning for resources allocation, and even improving problem solving capability.

2.4.1

Problem-Oriented Policing (POP)

Reinforced by crime analysis, conventional policing, or traditional incident-driven policing, relies heavily on patrolling, rapid response, and follow-up investigations (Eck and Spelman 1987). After reviewing considerable research, they find that conventional policing effectiveness depends largely on how effective are the tactics employed. For example, focused patrolling on hot spots can have a large impact only for a short period of time and these patrols cannot be maintained for long. Similarly, rapid response to crime reports may not always be applicable and the delay in reporting gives the offender time to escape. In addition, follow-up investigation may not always lead to detection of crime.

26

Figure 2.4: Incident-driven Policing Source: Eck and Spelman 1987

The deficiency in conventional policing encouraged Goldstein to explore a new way of policing. In 1979, he developed the concept of Problem-Oriented Policing (POP). In his book, Problem-Oriented Policing, Goldstein (1990) states that “in a narrow sense, [Problem-Oriented Policing] focuses directly on the substance of policing – on the problems that constitute the business of the police and on how they handle them. This focus establishes a better balance between the reactive and prospective aspects of policing…. In the broadest context, Problem-Oriented Policing is a comprehensive plan for improving policing in which the high priority attached to addressing substantive problems shapes the police agency, influencing all changes in personnel, organisation, and procedures.” He further states that “the Problem-Oriented approach calls developing – preferably within the police agency – the skills, procedures, and research techniques to analyse problems and evaluate police effectiveness as an integral continuing part of management.” (Goldstein 1990). 27

This new approach to policing has changed the usual response in crime control and crime prevention and has shifted the traditional problem-solving concept in crime analysis too.

The concept of POP is quite simple and it can be summarised in four steps as follows: (1) scan data to identify patterns in the incidents being routinely handled; (2) conduct an in-depth analysis of cause from the patterns identified; (3) employ appropriate tactics to intervene and prevent any future reoccurrence; and (4) assess the impact of the interventions and if they have not worked, repeat the process all over again. This approach involves a well-defined problem, a root cause analysis, a tailor-made strategy to prevent future crime from developing. The strategy so employed relies less on arresting offenders and more on developing long-term methods to deter crime by preventing potential offenders, protecting likely victims and reducing hazards of a potential crime location. In essence, POP transforms the concept of policing from a reactive response to a proactive response.

28

Figure 2.5: Problem-oriented Policing Source: Eck and Spelman 1987

The philosophy of POP has seen rapidly adopted by many law enforcement agencies worldwide and a tremendous number of successful case studies have been recorded, such as those reported in the book Problem-Oriented Policing: Success in the Field, Volume 1 to 3 published by Police Executive Research Forum (Shelley and Grant (Ed) 1998, Brito and Allen 1999 (Ed) 1999, Brito and Gratto (Ed) 2000). There have openly recognised the effectiveness of the problem solving approach, and some of the essential ingredients of which are also incorporated into community policing.

2.4.2

SARA Model

To cope with the introduction of POP and its implementation, a common and 29

widely accepted problem solving technique in crime analysis has also been developed – SARA Model, a simplified process of POP. SARA is the acronym formulated by Eck and Spelman to refer to the four phases of problem solving – Scanning, Analysis, Response and Assessment (Center for Problem-Oriented Policing, http://www.popcenter.org/about-SARA.htm)

Problem solving is crucial to crime analysis because underlying factors leading to crime and disorder problems for which effective responses can be developed and through which assessment can be conducted may affect the relevance and success of the responses (Boba 2003).

The SARA model employed in crime analysis proposes that the first step is a Scanning phase to identify a problem, such as a cluster, related or reoccurring crime incident, and to accord priority to the crime problem for future examination. Following that, the second step is an Analysis phase to conduct a comprehensive review, both qualitative and quantitative, of the crime problem from various sources of information to determine the root cause of it. After that, the next step is a Response phase to formulate a tailored set of intervention and enforcement actions to be implemented and if necessary modify the implementation. Finally, the last step is an Assessment phase to evaluate the effectiveness of the response phase, both before and after the responses have been implemented (Boba 2001, Braga 2002, Vann and Garson 2003).

Overall, the Scanning phase is crucial in determining the direction of the effort in that the problem should be clearly defined and validated. Thereafter, the Analysis phase should focus on the particular elements of crime, such as offender, 30

victim and venue of location. During the Response phase, the emphasis is placed on the actions to be taken by key personnel and time scales. When it comes to the Assessment phase, the whole operation should be reviewed and assessed to the degree of effectiveness and if it was not, why particular actions were not successful and what other alternatives might be able to achieve the same goal (Patrick 2002).

Figure 2.6: SARA Model Source: Clarke and Eck 2003

Still, the SARA model can be misleading in suggesting that these four phases should follow one another in a strictly linear manner (Clarke and Eck 2003). In the problem solving process, it is always possible to receive and collate new information, especially during scanning and analysis phases, and the process can be revisited at any phase. SARA model is a repetitive process in thinking and it allows the analyst to refocus on the problem and continue to examine the best possible solution to the problem. 31

Crime analysis and crime mapping can be interacting with each other in all phases of the problem solving process. It is imperative to assess the problem accurately during the scanning, analysis and the assessment phases (Boba 2001). In the scanning phase, crime maps can show the analyst where the problems are located and that can help test the hypothesis about the problem in the subsequent analysis phase. Analyst can also monitor the situation in the response phase by advising the effective allocation of resources of that operation requires, based on the time span of occurrence and the locations where offences are mainly occurring. After the enforcement action, it is very likely to see changes in the maps, be it a spatial or temporal change or both. After all, it is vital for crime analysts to have a comprehensive understanding of the model in support of the mapping process while it is equally important for members of a law enforcement agency to understand the role of crime analysis in problem solving even though it always relies very much on their experiences and understandings of a problem.

2.5

CompStat

The most effective crime control techniques and later with the aid of GIS crime mapping witnessed in the success of the CompStat process pioneered by the former commissioner William Bratton of the New York Police Department (NYPD) and his management team in 1994 (Shane 2004). CompStat is short for “computer comparison statistics” or “computerised statistics”. The implementation of CompStat has remarkably caused 65.99% reduction in the total number of reported crimes for the seven major crime categories (murder, robbery, first degree rape, felony assault, burglary, larceny and grand larceny) in New York from 430,460 in 1993 to 146,397 32

in

2003

(CompStat

Report

2004

http//www.nyc.gov/html/nypd/pdf/chfdept/cscity.pdf).

Vol.11 The

No.50

ultimate

at

goal

of

CompStat is to reduce crime and to improve quality of live.

Major Category

1993

1997

% chg vs. 1993

2003

% chg vs. 1993

Murder

1,927

767

-60.2

598

-68.9

Rape

3,225

2,783

-13.7

1,875

-41.8

Robbery

85,892

44,335

-48.3

25,919

-69.8

Fel. Assault

41,121

30,259

-26.4

18,774

-54.3

Burglary

100,936

54,866

-45.6

29,215

-71.0

Gr. Larceny

85,737

55,686

-35.0

46,877

-45.3

G.L.A.

111,622

51,312

-54.0

23,139

-79.2

TOTAL

430,460

240,008

-44.24

146,397

-65.99

Table 2.1: Historical Statistics of CompStat between 1993 and 2003 Source: NYPD at http//www.nyc.gov/html/nypd/pdf/chfdept/cscity.pdf (accessed on 2005.01.15)

The CompStat process is a hybrid management style that combines the best and most effective elements of various organisational models and policing philosophies (Henry 2003). Not only does it retain the best practices of traditional policing, it also incorporates the philosophy of Problem-oriented policing (POP), including the SARA model, to formulate crime control strategy. One of the most important theory resulting from this process is the “Broken Windows” theory which was introduced by George Kelling and James Wilson in 1982, which state that if minor offences are left unchecked they will lead to more serious crime (Kelling 1995). The dominant role of police has long focused on serious crime, such as murder and robbery etc., and rarely geared up to focus on minor crime, such as petty cash theft. Broken Windows theory raises the level of awareness of the police and the public that if minor problems are not taken care of they will develop into serious

33

problems. Kelling and Sousa (2001) later revealed in their research on the impact of Broken Window policing to violent crime in New York that the adoption of Broken Window policing had successfully prevented 60,000 violent crimes between 1989 and 1998.

Another significant policy encompassed in the CompStat process is the “zero-tolerance” policy which connotes a complete lack of responsiveness on the part of police officers in the manner they enforce the law (Henry 2003). The promotion of zero-tolerance policy transforms law enforcement agencies into focused, efficient and accountable organisations by motivating the officers who are guided by a focused mission, disciplined and stimulated by a rigorous concern of direct and personal accountability.

The advocacy of Broken Windows theory and zero-tolerance policy as well as other management philosophies meant that new police cultures then evolved to a four-crime reduction principle model, namely the CompStat model. These changes in the management and practices of law enforcement agencies have transformed the traditional police culture in devising crime reduction strategy.

The first step in the CompStat model is the collection of accurate and timely intelligence so that police have to respond to crime effectively and immediately by having accurate knowledge of particular types of crime occurrence. The effectiveness of police response to crime will increase proportionately as the accuracy of criminal intelligence increases. The second step is to employ effective tactics designed to reduce the number of crimes. The tactics should be comprehensive, flexible and adaptable to the shifting crime trends to avoid any 34

possible displacement of crime. The third step requires rapid deployment of personnel and resources. Immediate task forces or “split-forces” should be formed and tasked to tackle and even eradicate the specific crimes, which are accorded with high priority or assigned as a primary responsibility. The fourth step is to adopt an ongoing process of relentless follow-up and assessment to ensure the desired goals are actually achieved. The evaluation process allows the agency to assess the feasibility of a particular response and to incorporate the knowledge acquired for future tactical options (Shane 2004).

One of the vital facilities in CompStat, especially for the purpose of data analysis and data presentation, is the weekly CompStat Report which meticulously captures all the details of crime statistics, ranging from summaries of weekly crime complaints, arrest and summons activity, crime patterns to specific times and locations at which the crimes and enforcement actions took place. These reports will be presented by the personnel from each of the 76 precincts and other departments in the weekly Crime Control Strategy Meeting where the responsible precinct commander will be challenged on the skills and effectiveness of their staff in the capability of crime reduction and police performance. Crime maps obviously have a significant role to play in this regard.

35

Figure 2.7: Principles of CompStat Model Source: Reconstructed based on Shane 2004

In another research conducted by Willis, Mastrofski and Weisburd (2003), they commented that the characteristics of CompStat can be generalised using six core elements: (1) mission clarification; (2) internal accountability; (3) devolution of decision making authority; (4) organisational flexibility; (5) data-driven analysis of problems and assessment of department’s problem solving; and (6) innovative problem solving tactics. CompStat has not been so successfully implemented in the three study cities, Lowell, Minneapolis and Newark, as its pioneer in the New York City. One striking effect on key strategic decision-making process was noticeably placed on new information technology and the emphasis of the implementation of CompStat process. The data-driven analysis process in CompStat relies heavily on the accuracy and timeliness of intelligence of crime information and this has caused the middle managers to be highly sensitive to the need of what detail of information they were looking for. Without the support of GIS crime mapping techniques, 36

particularly the visual display of aggregate data explaining the relationships among criminal offences in time and space, immediate spatial and temporal analysis would not be possible.

2.6

Literatures on Campus Crime

Campus crime appears to be a sensitive topic to discuss. Any search for basic information, even locally, has always been fruitless, and no mention of any form of utilisation of GIS in analysing crime in tertiary institutions can be found. Perhaps, any disclosure of crime situations may cause an adverse effect on the reputation of an institution. This research has made repeated attempts to locate any updated articles and reports from the Western countries through internet and publication, and through email subscription from the Crime Mapping Association and the Law Enforcement Analysts Network in the United States but almost all has been in vain.

In spite of this, some respondents replied and gave several useful websites on campus crime. The website of the Office of Post-secondary Education (http://ope.ed.gov/security/index.asp) is supplemented by the United States Department of Education offering a wealth of campus crime information of over six thousand tertiary institutions the United States. The search engine allows the user to make one enquiry at a time and yet it does not yield very useful results. The enactment of the Higher Education Act of 1965 (amended in 1998) requires tertiary institutions to report criminal offences to the Department of Education and the Department of Education has an obligation to inform the students and their parents of the safety of the environment in which they are studying (Campus Crime and Security

at

Post-Secondary 37

Education

Institutions,

http://ope.ed.gov/security/index.asp). No similar law seems to be enacted in Hong Kong to keep the members of the public abreast of the campus safety policies.

An email reply from Sean Bair, ([email protected]) the programme manager of the Crime Mapping Analysis Program (CMAP) of the National Law Enforcement and Corrections Technology Centre under the administration of University of Denver affirmed that they have just begun to get involved in analysing crime activity on campus and agreed that they have not come across too much literature on the captioned research topic. He further suggested discrete analysis would suit the purpose for a confined campus.

2.7

Summary

The development of crime mapping over the past few decades can be generalised by four contributing factors: (1) the technological evolution from manual mapping to automated GIS mapping has improved the mapping process greatly in quality and in quantity; (2) the conceptual transformation of crime prevention from criminal behaviour based to crime and place theory assists law enforcement in formulating effective strategies focusing more on the geographical effects; (3) the philosophical advance from traditional policing to POP by adopting the SARA problem solving model forces law enforcement to take a proactive approach to tackling crime; and (4) the expansion of scope of crime analysis from investigative-centred to a multifaceted dimension allows the management a better understanding of the impact of crime to resources allocation and may even strengthen public confidence in fighting crimes. CompStat model creates a management structure that helps law enforcement agencies to control crime and 38

disorder effectively in the communities. CompStat is an information-driven process relying on accurate and timely intelligence, partly based on the effective performance of GIS crime mapping to identify and plot the occurrences of crime. The visual display of aggregate data enables crime analysts to explain the relationships among criminal offences in time and space makes immediate spatial and temporal analysis possible. The world of policing therefore is able to benefit from these theoretical development and refinement.

In the absence of any useful literature on the subject of campus crime, this research will evaluate two dimensions of the campus crime situation from the available data. From the analytical perspective, this research will utilise crime mapping techniques to identify crime hotspots and their spatiotemporal correlations in the campus environment. From the management perspective, it will examine the current organisation structure and public safety procedures of the campus security and derive an effective crime reduction strategy. The methodology used will be discussed in the following chapter.

39

CHAPTER THREE METHODOLGY

3.1

Introduction

From the previous chapter, it seems no one would argue that the distribution of crime incidents is spatially random. Pioneer environmental criminologists have concluded that the types of built environments have a direct link with many types of crime.

The University of Hong Kong was selected as the study area. Section 3.2 will give a brief introduction of the study area and the security system of the campus as well as the source of research data and the tools to be used in this study. All these serve to provide some basic information of the study area.

Section 3.3 will explain the design schema of this research. Since the study is intended to assist the management and frontline

personnel to formulate effective

crime prevention measures and patrol strategies in the campus, a practical approach was adopted for this research. The methodological framework will gear up the operation of data management and data analysis processes. Section 3.1.1 will explain the functions of data management how data can fully be optimised for analysis purpose. Section 3.3.2 will outline the design of analytical flow and determine the appropriate techniques in visual identification of crime hot spots.

3.2

The Research Data 40

3.2.1

Study Area – The University of Hong Kong

Due to the sparse distribution of buildings in the campus of the University of Hong Kong, this research sub-divides the campus into three study areas, namely the Main Campus, Western Region and South Region and these are the terms later referred to in this research. The first study area is the Main Campus which is situated at the north side between Pokfulam Road and Bonham Road in the Mid-levels. It covers most of the academic buildings, administration departments and student facilities in the whole university. The second study area is the Western Region where a dense convergence of student dormitories can be found along the west bank of Pokfulam Road. The third study area is the Southern Region in Sassoon Road where the medical school and staff residences are located. The size of the study area covers about 39.5 hectares (Sustainability Report 2004 available at http://www.hku.edu.hk). The map of the campus of the University of Hong Kong can be found at Figure 3.1.

Special off-campus facilities, such as the Swire Institute of Marine Science in Stanley and Kadoorie Agricultural Research Centre in Taipo, however, will not be included in this research simply because there was no residences and no crime reported and they are distanced from the study area. These facilities occupy an area of

about

9.8

hectares

(Sustainability

http://www.hku.edu.hk).

41

Report

2004

available

at

Figure 3.1: Sub-division of the study area of HKU

The ownership of the campus of the University of Hong Kong is under the Authority of the University of Hong Kong (Sustainability Report 2004). Members of the public may have a wrong perception that they have access rights to the open campus. It is actually a private property and therefore police do not patrol the campus on a regular basis. Indeed, there are many blocks of buildings in the campus restricted to authorised access only and controlled by campus security guards.

42

The total number of regular students of the University of Hong Kong between the year of 2002 and 2003 was 19,000, including 11,700 undergraduate students and 7,300 postgraduate students of which more than 1,000 were international students (University at a Glance available at http://www.hku.edu.hk). The number of regular students between 2003 and 2004 was reported as 19,562 (Review 2004 available at http://www.hku.edu.hk). This figure, however, does not take into account the staff population which was reported as 6,724 in December 2003, including 4,462 regular staff and 2,262 temporary staff (Quick Statistics available at http://www.hku.edu.hk). In standard calculations in campus crime statistics in American universities staff have never been counted. Neither does it take into account of the headcount enrolment on School of Professional and Continuing Education (SPACE) programme and the out-reach programme which amounted to 105,427 and 730 respectively in the year 2003 and 2004 (Quick Statistics available at http://www.hku.edu.hk) as these programmes are operating outside the university campus or outside of Hong Kong.

The geographic location of the campus of the University of Hong Kong is very unique. The Main Campus and the Western Region are enclosed by medium density and middle to high-income residential blocks while the Southern Region is quite remote from the urban area and surrounded by low density and high-income residential blocks as at Figures 3.2 and 3.3. There is neither any entertainment centre, such as cinema, nor any liquor premises, such as bar, closely found in these three areas. Residents nearby have seldom entered the campus probably because of the topography and because academic institutions rarely attract any unnecessary visitors, and therefore the access to the campus is almost confined to the students and staff only. 43

Figure 3.2: Population Density around HKU Campus Source: Lands Dept, 2001 B1000, Digital Topographic Map

Figure 3.3: Income Groups around HKU Campus Source: Lands Dept, 2001 B1000, Digital Topographic Map

44

That is to say, the campus area has rarely been influenced by any potential crime prone factors, such as high population density, high poverty level and high accessibility in an area. This corresponds to two observations made by crime analysts that low-socioeconomic-status areas have more crime and high-status areas have low crime; and high pedestrian flow has accelerated higher rate of crime (Kamber, Mollenkopf, and Ross 2000).

Unlike many other crime analyses conducted by police that frequently focus their targeting efforts on mixed socio-economic areas within their jurisdiction. Campus crime mapping, however, is a discrete analysis in that the targeting area is an isolated area where activities are confined to academic affairs and have little interaction with socio-economic events.

3.2.2

Campus Security

An interview was conducted with Mr. Teddy Wong, manager of the Security Section of the University of Hong Kong on 26th Jan 2004. Responsibility of campus security falls on the shoulder of the security manager under the management arm of the Estates Office. He directly supervises the security of the Main Campus with a strength of ninety five guards and co-ordinates security issues with security management of the Western region and Southern Region. The staff turnover rate is low and about one third of his personnel are pension staff. Most of his staff are experienced and trained guards but none of them are capable of performing crime analysis. Surprisingly, upon the interview with the security manager, it was revealed that the primary duty of campus security guards is to maintain a good degree of public relationship with the staff and the students. Their secondary duty is to 45

respond to incident calls including reporting building defects, and the tertiary duty is to patrol the campus. The Main Campus is divided into nine patrol zones and there is no static post so that they have to patrol their zone and record their attendance with an electronic device, very much like an electronic visiting book, at all the check points. Unlike most of the security guards employed in the tertiary institutions, they work on a three shifts system instead of a two shifts system and therefore their turnover rate is very low.

The situation outside the Main Campus is slightly different. Security of the Western Region is carried out by dormitory attendants whose primary responsibility is for caretaking duties. Security of the Southern Region is outsourced to three private security companies, which were awarded by the lowest bids, for building management purpose, such as the staff quarters and medical blocks. This explains the diversified management style of security within the campus. Each of the attendants and guards is responsible for the security of their own buildings and the rotation of staff of the private security companies is quite frequent. The map at Figure 3.4 shows the sub-division of the campus security.

Reports of crime can be made directly to the Security Section and a security officer will scrutinise each incident. Cases requiring police attention will be referred to the police immediately for action. Minor cases, such as loss or theft, will be recorded for documentation purpose and it is up to the victim to make a report to the police personally. There is no designated computer system to store the crime incident records, instead they were kept in a word processor, Microsoft Word, for easy retrieval and report writing. On a monthly basis, the Security Section emails to all staff and students a crime report keeping them abreast of the current crime 46

situation of the campus and an extract is at Figure 3.5. Around the middle of each year, the Security Section stages a one-day security campaign advising participants of the importance of protection of personal belongings.

Figure 3.4: Sub-division of the campus security of HKU

Since the campus falls within the boundary of the Western Police District, the Security Section maintains a close liaison with the police for exchanging crime information. Police undertake a proactive role in crime prevention initiatives by 47

providing District Intelligence Section (DIS) to monitor the crime situation of the campus and Regional Crime Prevention Office (RCPO) to hold crime prevention seminars to the fresher at the beginning of each academic year.

Figure 3.5: Sample of the monthly crime report of HKU

3.2.3

Research Tools

Software for mapping crime has made great strides in crime analysis. There are many GIS products specialised in crime analysis available on today’s market and some are even free software, such as the CrimeStat III developed by Ned Levine. CrimeStat is a spatial programme for the analysis of crime incident locations with six common tools, such as journey-to-crime estimate, distance analysis and hot spot analysis etc., made available from National Archive of Criminal Justice Data at http://www.icpsr.umich.edu/CRIMESTAT. 48

This

application

promotes

crime

mapping practice to individual researchers and smaller scale law enforcement agencies.

One of the most common GIS programmes used is ArcGIS, an advanced version of ArcView developed by Environmental Systems Research Institute (ESRI). This is the software to be used in this research and more importantly it is available at the University of Hong Kong. This software application allows crime analysts to capture and create an integrated picture of information in the form of interactive maps and reports and in a Graphical User Interface (GUI) environment. Different from ArcGIS, CrimeStat is standalone software for spatial analysis and its output can be visualised using GIS software such as ArcGIS.

Along with the software package, a course workbook on crime mapping and analysis programme, “An Introduction to Crime Analysis using ArcGIS”, can be obtained from the National Law Enforcement and Corrections Technology Centre (available at http://www.nlectc.org) as a reference book for conducting the analysis in this research.

3.3

Methodological Framework

Establishing a systematic framework for this research is the foundation to produce an accurate and meaningful picture of the crime situation of the campus at Figure 3.6. On this basis, there are two structures to be constructed and they are the data management and spatial analysis. Proper data management helps eliminate the chances of errors and appropriate spatial analysis helps increase the validity of the data presented. 49

Crime data provided by the Security Section may not be always useful for crime analysis or even further for spatial analysis function, particularly when most of the data was collected based on their subjective purposes and their own data standards. The first step is therefore to evaluate the usefulness of data and examine the data in order to fit in for further spatial analysis purpose.

Data integrity and data structure are crucial in assessing the crime analysis and the maps produced. Data integrity affects the quality of data, such as consistency and accuracy, in the outcome of an analysis. The design of data structure determines the entities of data to be collected. Most of the crime data are not always spatially referenced and neither are they properly captured as it depends on the needs of individual agency, so they need to be processed and transformed. Geocoding or address matching is a prerequisite in plotting the crime incidents on a map. Data classification will assist in re-arranging the data by putting it into the designed arrays for precise measurement.

Spatial analysis brings a closer focus to explore the cause of a crime at a specific crime spot and leads to a holistic crime picture. Spatial distribution of crime clearly demonstrates that certain areas attract more crime than the others as mentioned in Chapter 2. This is the hot spot identification revealing the co-relationship between a specific geographic location and other potential variables, such as land uses and population characteristics (Anselin et. la. 2000). This research will make use of single layer hot-spot operation to identify the concentration of crimes in the campus and examine the spatial and temporal changes of these crimes to determine an appropriate strategy to prevent and tackle the crime problems. 50

The adoption of the SARA model in crime analysis has always helped uncover the underlying causes of crime problems. Strategic crime analysis and tactical crime analysis will be useful for devising effective strategy and implementing specific enforcement efforts to reduce and prevent crimes. Details of this analysis hierarchy will be discussed in later sections.

51

Figure 3.6: Methodological framework adopted for this research

3.3.1

Data Management

3.3.3.1

Data collection 52

Collecting data from the police on campus crime is nearly impossible due to Personal Data (Privacy) Ordinance in which it stipulates that data collected should not be used other than for its originated purpose, except that it is used for the prevention and detection of crime specified in Data Protection Principle 3. Nonetheless, the Security Section provided some useful crime statistics of the campus between the years of 2002 and 2004. This is the database used for the construction of this research and Table 3.1 shows part of the tabular data given.

When crime complaints are made to police, detailed information can be obtained by questioning the victims and witnesses, interrogating suspects, and collecting evidence from the scene by exercising their power under the laws for crime detection and crime prevention. Valuable information, such as victim and suspect description and property loss description etc. are crucial for immediate police action and investigation. However, due to concerns over confidentiality, Estates Office does not enjoy this privilege for data collection and data collected are very much depersonalised. In most of the situations, only limited information or even aggregated data can be obtained, such as type of crime, time or location of a crime etc. This limits the variety of analysis techniques to be conducted for crime analysis. One of the alternative techniques which can be used is the hot spots analysis.

Offence

Date/time

Location

Victim

Property

Summary of

Police

Facts

Involvement

A personal bag containing cash 800, mobile phone and IDs

VTM went to the toilet with the bag. She forgot to fetch back the bag after the toilet. When she returned to the toilet, the bag

of offence 1

Lost / Stolen

BET 2015 – 2130 on 02-JAN-2 002

3/F female toilet, Main Library

A female student

53

Remarks

2

Car found damaged

BET 25-DEC-2 001 and 06-JAN-2 002

Space No. 89, Tam Tower, Sha Wan 25

A resident of Tam Tower

A motor cycle

3

Criminal Damage

1550 hrs on 21-JAN-2 002

Inside HK Bank, RunRun Shaw Bldg

HK Bank

The leaflet holders

4

Lost / Stolen

1435 hrs on 29-JAN-2 002

4/F Main Library

A female Student

A handbag with cash, mobile phone and ID

5

Lost/Stol en

BET 1530 and 1730 on 29-JAN-2 002

1/F Old Library

A female student

A cloth bag with cash and and IDs

went missing VTM parked his m/c at AA for a week. When he returned to the bay, he found the m/c body scratched and suspected that his m/c had been tampered with A male Chinese student whilst doing transaction inside the bank suddenly turned violent and threw all the leaflet holders on the ground. The student was identified to have mental problem. VTM left the handbag at the reading desk and went to toilet. On her return, the handbag was found missing VTM studied at Library at the material time. During the period, she had left her desk for an hour. And she subsequently found the bag missing

VTM would report to police direct

Table 3.1: Extract of raw crime data of HKU between 2002 and 2004 Source: Estates Office (Security Section) of the University of Hong Kong in January 2005.

3.3.3.2

Data set in this research

The crime statistics extracted and shown in Table 3.1 are the records stored in a table form of Microsoft Word format along with some information about the crime incident. The format of tabular crime data is still commonly adopted by the Hong Kong Police for their daily crime analysis due to the fact most of the crime information was not captured electronically nor from any non-spatial databases, except the Incident Mapping System (IMS) which is the only spatial database system capable of providing a visual display of geographic location of police incidents. 54

To an extent, the crime statistics kept by the Estate Office consist of some types of geographic variable, such as address. Yet, many of these crime locations reported were not precisely maintained in the tabular records. For example, the location of a crime occurred in the library only provides the setting of crime but it does not yield an exact location where crime took place. Data, which carries no inherent geographic information, renders little value in spatial analysis. With the aim of producing an effective measurement of crime situation, quantitative geographic data is required by means of geocoding or address matching to map the location from these textual geographic variables or non-spatial attributes to a geographic location recognised in a digital map.

The incompleteness of temporal records in the data set poses a serious problem in assessing the accuracy of temporal findings. Of the 225 crime records, only 164 records contain information of time in hours when the offences were committed, which represents about 73% of the total crime data. That means the accuracy of temporal analysis of this study comes is approximately 73% accuracy. Table 3.2 presents the percentage of the known and unknown hours of the temporal analysis in this research. There was no reason given as to absence of this valuable information. This valuable temporal information was concatenated in one data field called date / time which squeeze the date and time information in one single column

Known / Unknown

2002

2003

2004

Total

66

42

56

164

Known %

84.62%

65.63%

67.47%

72.89%

Unknown

12

22

27

61

15.38%

34.38%

32.53%

27.11%

Known

Unknown %

55

Total

78

64

83

225

Table 3.2: Statistics of crime data with reported and non-reported time between 2002 and 2004.

Apart from the lack of spatial and temporal attributes in the data set for sustaining possible crime mapping performance, there are other data which require further verification. For instance, a lengthy narrative illustrating the criminal act of a suspect in a single data field - the summary of facts field makes them difficult to differentiate from each other for data matching and data comparison. Obviously, the data structure needs to be revised to cater for more precise measurement.

In addition to that, other deficiencies were also detected from the data set where arrays of information were maintained inconsistently and inaccurately. For instance, floor numbers may be missing and building names are not standardised. Main Library sometimes refers to the New Wing of Main Library but it happens to be the Old Wing occasionally. These data, which cannot be matched against the campus map and the too simplified victim information in some records, create a great deal of problem in data integrity.

3.3.3.3

Data classification

The original data structure and the deficiencies spotted in the data set prohibit further automation of crime analysis. It was at this stage the rectification process started to re-examine the value of each entity and to collate them into expanded common attributes, which were ignored in the original design.

In fact, each type of crime carries some significant attributes to establish the 56

hypothesis of a modus operandi (M.O.). For example, robbery analysis aims at identifying the properties robbed and the method employed to succeed; burglary analysis also aims at not only identifying the stolen properties but also on the method the burglar entered the premises; Analysing theft of conveyance without authority is aimed at identifying the types of stolen vehicle, the method employed to prise open the vehicle and the way to dispose of it. Furthermore, police is able to construct offenders profiling by means of utilising their criminal records system, criminal intelligence system and MO system to compile a comprehensive targeting report. Yet, all these go beyond the scope of this study.

Having said that, there are some common elements determined from all crimes; such as geographic factors, time factors, victim descriptors, property loss descriptors, physical evidence descriptors, specific MO factors, suspect descriptors, and suspect vehicle descriptors (Gottlieb, Arenberg and Singh 1998). They are the basic data fields considered to supplement the original data set. The date and time file expanded to include start date, end date, weekday (Monday etc.), start time and end time. Some other files were also created to include type of stolen properties and MO descriptors derived from the original data fields of property and the summary of brief facts.

In some of the lost or stolen incidents, especially those which were reported in 2002, it was up to the victim’s preference to decide whether or not he or she wished to report it to the police, and hence data reliability becomes doubtful. Along the same lines, most of these cases, if they were carefully examined, should be classified as theft since none of the stolen properties were recovered at a later time. It is hard for anyone to imagine a loss of notebook computer in the library be 57

classified as lost or stolen. At least, there is a prima facie case to establish that offender has no intention to return the properties to the owner. It is for this reason that all the lost or stolen cases were reclassified to theft cases. Nonetheless, the situation improved in 2003 when the Estates Office changed the handling procedure and maintained a closer liaison with the police in crime reporting. This shift of administrating crime complaints indicates an apparent strategic change in handling crime reports by the management probably due to the upsurge of theft offences. Classified theft offence will be investigated by the Criminal Investigation Department (CID) of the police, while making a lost or stolen report will be processed merely for documentation purpose.

There are other deficiencies found in the data classification stage. For example, there are non-crime incidents recorded in the original crime data classified as crime incidents. Report of peeping tom in itself is not a crime and it is only a police report of suspicious behaviour due to probably, one’s sexual instinct. Yet, if one is found performing a suspicious act in a place where he could not give a reasonable explanation to his existence, he may be arrested for a loitering offence according to the Crimes Ordinance. Another example is incidents of trespassing which should not be counted as a crime, except in the case of trespassing with armed weapons or explosives stipulated in the statutory law of Hong Kong. Trespassing is merely a civic action against violation of intruding one’s property under the regulations or by-laws. However, it might give rise to the suspicions of other potential criminal activities, such as loitering. Therefore, peeping tom and trespassing are also included in the data set of this study to reveal fully the overall crime situation of the campus.

58

3.3.3.4

Temporal data conversion

Monitoring temporal change of crime would always yield a good prediction of time of offences because of the offenders’ temporal behaviour but it depends largely on the data accuracy and data availability. In order to able to allow the existing data to be measurable and comparable with other temporal variables, particularly the hours variable, all the available data was transformed in a spread sheet of an Excel format by means of manual input and converted into an expanded data table, including fields of date, day in week, length of crime, automatically.

3.3.3.5

Geocoding

Geocoding or address matching is the process of transforming tabular and geographic data, some non-spatial attributes or addresses in this study, to georeferencing spatial data. The process involves matching the addresses against a geographically referenced base map constructed from address data in the Centamap (http://www.centamap.com/). Since nearly all the crime incidents, except one street crime snatching, took place inside buildings and the geographic variables kept in the records did not indicate the exact whereabouts the crime occurred, the centre point of a building block inevitably becomes the georeference point or the geocode of a crime location. Regarding the street crimes, the geocodes were manually converted direct from the referenced base map. This is the analysis unit to be used in the research.

The geocodes collected were stored with the crime data in a tabular Excel format and subsequently plotted in ArcGIS against the topology of the base map 59

obtained from the Lands Department. All the crime locations were charted accordingly.

3.3.3.6

Data re-classification

Data classification is deemed necessary to uphold the data integrity requirements and to explore the extent of parameters and the value of crime information for future data collection. Finally, data cleaning was initiated to eliminate the manual input errors after which data classification was completed. Figure 3.7 depicts the process of data classification.

Figure 3.7: Re-classification process of crime event data

60

3.3.4

Data Analysis

3.3.4.1

Visual inspection

Boba (2000) stated clearly that there are six general areas to be included in making a crime analysis and mapping needs assessment: (1) current state of affairs; (2) strategic goals; (3) data sources; (4) technology; (5) current crime analysis and mapping products and their uses; and (6) crime analysis and mapping needs. In the area of crime analysis and mapping needs, she argues that information presented in a map are subjective information but more important is how this information is perceived by the audience.

In data analysis, as well as in crime analysis, one needs to apply visual thinking to generate ideas and make a hypothesis of a problem (Harries 1999). This visual identification allows map users to see some relationships between different phenomena. Map production, therefore, depends not only on the availability of data and the objective goal but more importantly the expectations of the readers. Since, campus administration and front line personnel are the targeted readers of crime maps in this research, crime maps to be produced have to gauge their level of thinking.

To ease the understanding of map users, single layer hot spots identification is used in this study that is similar to a digital pin map displaying points of crime incident and forming an overall picture of spatial distribution of crime over a specific period of time. Graduated size point which serves to distinguish the locations with multiple occurrences will be applied and since most of the crime 61

locations are represented by building blocks it also overcomes the problem of overlapping.

3.3.4.2

Hot spots identification

In crime analysis, hot spots are always referred to as clusters of crimes. Even though there is no fixed definition for hot spot, a common interpretation recognised by most is that hot spot is a small place where crime occurred repeatedly, such that it becomes highly predictable over a period of one year at least (Sherman 1995). Some criminologists and analysts generally define hot spot as a somewhat larger area in size that even the extended surroundings of a building are included (Vann and Garson 2003). These two definitions will be employed in this research, especially the latter because of the limitation of geographic variables obtained from the original data set, where a large number of points are geocoded from the centre points of buildings.

There are other arguments in formulating the definition of a hot spot by focusing on the issues of public space and private property, nature of boundaries and some variants in calculation of a linear block (Buerger, Cohn and Petrosino 1995). These quickly become confusing as the location of a crime can be measured to the accuracy of which side of a street, including “street cartilage” – an American term to describe a large open area in front of the property lines where police can park their patrol car. Obviously, there are geographical differences in the design and layout of urban land use between the United States and Hong Kong and high precision of calculating hot spots is not required in this research. Besides, the characteristics of the campus of the University of Hong Kong can be generalised as a single 62

ownership campus and a discrete campus with less influence from its surroundings. Boundary problems and demarcation of public space and private properties do not exist in this research.

In order to determine whether there is a hot spot or no hot spot, the crime data has to satisfy that crimes occurred frequently in an area at least once during a one year period. Initial review of the data set apparently indicated that there are clusters of crime incidents more contagious than other areas in a long span of time. In short, hot spots are detected where the density of crimes is high. The crime data gathered for this study fulfils this basic requirement for crime analysis.

Eck, Gersh and Taylor (2000) suggest that if any one of the following hypotheses is established, mapping hot spots will be made possible: (1) a location where crimes are spilling into surrounding areas – central place hypothesis; (2) a location which appears to have a weak attraction to offenders but there are interesting crime target in the nearby places – side effect hypothesis; and (3) some locations (hot spots) are disproportionately vulnerable than other locations (non-hot spots) but not in crime-resistant areas (area effect). The first and last hypotheses seem to fit into the spatial distribution of crime in this study and these reasons suffice to sustain the adoption of hot spot analysis.

3.3.4.3

Hot spots analysis techniques

Identifying a crime hot spot has an element of subjectivity resulting from individual’s interpretation of a crime hot spot and the analysis techniques applied accordingly may produce different results (Harries 1999, Vann and Garson 2003). 63

In general, there are five hot spot analysis techniques to help visualise spatial distribution and they are: (1) visual interpretation which is the most basic method of presenting crime patterns similar to that of a digitised pin map but the major problems of it are stacking points and over-lapping points; (2) choropleth mapping which is very much like a thematic map in which areas are shaded according to the data values and the fallacy of it may be resulted from generalisation within a boundary; (3) grid cell analysis, which is based upon artificial boundaries, uses a uniform-sized grid cell to overlay an area of interest, yet error may occur when a point falls closely within a neighbouring grid forcing it to give an aggregated value in the neighbouring cell; (4) point pattern or cluster analysis involves identifying an arbitrary starting point or a “seed” and thereafter calculating the following seeds statistically in order to identify the clusters. A statistical algorithm is required in this employment; (5) spatial autocorrelation which is designed to establish spatial relationships among clusters of points based on spatial similarities and traits among the points of interest, yet there is a chance of generating a negative autocorrelation where no cluster may be located (Harries 1999, Vann and Garson 2003). Although there are a variety of techniques for detecting hot spots in crime data, there is no single approach superior to the others (Grubesic and Murray 2001). The argument of adopting the appropriate visualisation technique really depends on the targeted audience.

As mentioned above the data provided in this research has its own limitation and taking into consideration that the primary objective is to produce a single layer hot spot analysis and that the analysis is to be explicable to the management and frontline security personnel, visual interpretation technique is chosen to perform the 64

analysis because of its simplicity in visual presentation. The stacking point and over-lapping point problems resulting from this technique can be overcome by increasing the scale of a map. That is, the use of smaller scale map will prevent the occurrence of these problems.

3.4

Summary

The methodological framework of this research was discussed in full detail in this chapter. Though most of the theoretical frameworks are designed for a different setting they are purported to have a more universal applicability. One commonality they all share is that a well-structured data management and data analysis framework is essential.

The operation of data management process was explained by maintaining a high degree of data integrity and data quality control as well as data validity and data reliability. The importance of it cannot be emphasised any higher. The logical design of analytical workflow was also outlined by examining each and every consideration required for making data visualisation possible. These are determining factors in the success of the following spatial and temporal analysis.

The methodological framework selected in this study serves as the springboard to research into the crime situation within a University of Hong Kong sample. The hot spot identification and visual interpretation techniques employed proved to be able to present crime data accurately and meaningfully.

65

CHAPTER FOUR CAMPUS CRIME ANALYSIS

4.1

Introduction

From reading the literature review in chapter two, it seems no one would argue that the distribution of crime incidents is spatially random. Pioneer environmental criminologists have concluded that the types of built environments have a direct link with many types of crimes. These are the spatial crime patterns expecting that certain types of crimes will follow.

Section 4.2 will present the overall crime situation of the campus from the year 2002 to 2004. The spatial and temporal analysis in section 4.3 and section 4.4 will aim at examining the available data to uncover the spatiotemporal relationship between the crime problems. Serial crimes are detectable through a “signature”, that is unique behaviours on the part of the offender, and the commonalities they share, such as timing, location, victim similarities, and modus operandi will be discussed. Regarding the overall crime distributions, this study will focus on the severe hit areas and the prevailing crimes that may affect the law and order of the campus. Finally, section 4.5 will make a comparison between the crime situation of the campus and that of the universities and colleges in the United States.

4.2

Campus Crime - the University of Hong Kong

Overall, the total number of crimes which occurred on the campus of the 66

University of Hong Kong between the year 2002 and 2004 was not considered serious at all. The statistics at Table 4.1 shows that there were altogether 225 crime incidents reported within fourteen categories of crimes, and the total numbers of crime cases reported in these three years were 78, 64 and 83 respectively.

Criminal Offence

2002

%

2003

%

2004

%

Total

%

Burglary

0

0.00%

4

6.25%

4

4.82%

8

3.56%

Criminal Damage

2

2.56%

1

1.56%

1

1.20%

4

1.78%

Criminal Intimidation

0

0.00%

0

0.00%

1

1.20%

1

0.44%

Deception

0

0.00%

0

0.00%

1

1.20%

1

0.44%

Indecent Exposure

0

0.00%

0

0.00%

1

1.20%

1

0.44%

Loitering

0

0.00%

1

1.56%

1

1.20%

2

0.89%

Object Fell From Height

0

0.00%

0

0.00%

1

1.20%

1

0.44%

Peeping Tom #

0

0.00%

0

0.00%

7

8.43%

7

3.11%

Robbery

0

0.00%

1

1.56%

0

0.00%

1

0.44%

Snatching

0

0.00%

0

0.00%

1

1.20%

1

0.44%

TCWA

0

0.00%

2

3.13%

0

0.00%

2

0.89%

Theft

73

93.59%

55

85.94%

63

75.90%

191

84.89%

Theft From Vehicle

3

3.85%

0

0.00%

1

1.20%

4

1.78%

Trespassing #

0

0.00%

0

0.00%

1

1.20%

1

0.44%

Total

78

100.00%

64

100.00%

83

100.00%

225

100.00%

Table 4.1: Crime statistics of HKU between 2002 and 2004 Source: Estates Office (Security Section) of the HKU in January 2005. # Peeping tom and trespassing are not crimes under the statutory law.

Further observations from Table 4.1 indicate that more than 90% of the total number of crimes was property crime, including burglary, taking conveyance without authority (TCWA), theft and theft from vehicle. Of concern are the number of peeping tom activities reported, which rose from 0 in the years 2002 and 2003 to 7 cases in the year 2004, which must surely catch the attention of the management. The number of burglaries remained steady with 4 in each year from 2003 to 2004. When examining the bar chart at Figure 4.1 it can be seen that loitering and related 67

activity, such as peeping tom, and burglary figures have become alarming. For these reasons, this study will focus on the following crimes: theft, loitering and peeping tom and burglary, which pose a greater threat to the law and order in the campus environment. Targeting offenders of these prevailing crimes will become the primary objective in formulating an effective crime control and crime reduction initiative.

Statistics of Crime (2002 - 2004)

No. of Crimes

1000 100

2004 2003 2002

10

C Cr rim Bur im ina gla in l D ry al In ama tim g id e In de De atio ce ce n nt p O Ex tion bj po ec su tF el Lo re l F it ro eri m n Pe H g ep eig in h t g T Ro om b Sn ber at y ch in TC g Th W ef A tF ro Th m e V ft eh Tr icl ep e as se r

1

Offences Figure 4.1: Crime Statistics of HKU between 2002 and 2004 Source: Estates Office (Security Section) of the University of Hong Kong in January 2005.

Tables 4.2 and 4.3 also suggest that the overall crime situation and violent crime rate of HKU are quite low when compared with the figures for the whole of Hong Kong between the years of 2002 and 2004.

Year

2002

Hong Kong

Total no. of

Overall Crime HKU Student

Population

Crime

Rate *

6,787,000

75,877

111.80

68

Total no. of

Overall Crime

Population

Crime

Rate *

19,000

78

41.05

2003

6,803,000

88,377

129.91

19,562

64

32.72

2004

6,845,000

81,315

118.79

19,562

83

42.43

Table 4.2: Comparison of Overall Crime Rate between HK and HKU between 2002 and 2004 Sources of Information: Hong Kong Police (http://info.gov.hk/police) and Census and Statistics Department (http://info.gov.hk/censtatd) * represents crime rate per 10,000 population

Year

Hong Kong

Total no. of

Overall Crime HKU Student

Total no. of

Overall Crime

Population

Violent Crime

Rate *

Population

Crime

Rate *

2002

6,787,000

14,140

20.83

19,000

0

0

2003

6,803,000

14,542

21.38

19,562

5#

2.56

2004

6,845,000

13,890

20.29

19,562

4#

2.04

Table 4.3: Comparison of Overall Violent Crime Rate between HK and HKU between 2002 and 2004 Sources of Information: Hong Kong Police (http://info.gov.hk/police) and Census and Statistics Department (http://info.gov.hk/censtatd) * represents crime rate per 10,000 population # burglary and robbery are the only violent crimes took place in HKU

On average, it takes nearly five days for one crime to occur on the campus as opposed to two hundred and twenty four crimes reported in Hong Kong each day. It collaborates with the early presumption that the campus of the University of Hong Kong University is a peaceful and harmonious institution.

In the following spatial and temporal analysis, it will be examining the crime data from the approach of tactical crime analysis to identify crime patterns by means of visual identification as well as from the approach of strategic crime analysis to evaluate the effectiveness of crime intervention initiatives. Alongside this, it will be aided with elementary statistical analysis to confirm its validity.

4.3

Spatial Analysis 69

Spatial analysis is the study of location. Spatial distribution of crime events helps crime analysts to understand what has happened and what may happen in the future. Computerised mapping speeds up the process of spatial analysis in locating crime clusters of an area and identification of hot spots becomes visible.

Before going into the analysis stage, it is always good to review the spatial distribution of crimes from the statistical data. Table 4.4 portrays the geographical distribution of crimes in the campus. Apparently, the Main Campus was heavily attacked and suffered slightly over 80% of the total number of crimes from the year 2002 to 2004. Following that was the Southern Region with 14% and thereafter the Western Region with 5%. Interestingly, there was a drop of crime in the Main Campus in the year 2003 and no crime was reported in the year 2002. Cartographic distribution of crimes at Figure 4.2 also signifies the changes of crime locations between the year 2002 and 2004.

Study Area

2002

2003

2004

Total

Percentage

Main Campus

68

47

66

181

80.44%

Western Region

0

7

5

12

5.33%

Southern Region

10

10

12

32

14.22%

Total

78

64

83

225

100.00%

Table 4.4: Geographical distribution of crimes of HKU between the year 2002 and 2004

70

Year 2002

Year 2003

Year 2004

Figure 4.2: Spatial distribution of crimes of HKU between 2002 and 2004

The following crime statistics in Table 4.5 provide a better presentation of crime distribution in the Main Campus, Western Region, and Southern Region within the campus between the years 2002 and 2004. Most of targeted offences, theft, loitering and peeping tom and burglary, were committed in the Main Campus which is the most congregated region amongst the three study areas.

Criminal

2002

2003

2004

Total

Offences

M

W

S

M

W

S

M

W

S

Burglary

-

-

-

3

-

1

3

-

1

Criminal Damage

1

-

1

-

-

1

1

-

Criminal Intimidation

-

-

-

-

-

-

1

-

-

1

Deception

-

-

-

-

-

-

1

-

-

1

Indecent Exposure

-

-

-

-

-

-

1

-

-

1

Loitering

-

-

-

1

-

-

1

-

-

2

Object Fell From H

-

-

-

-

-

-

1

-

-

1

Peeping Tom

-

-

-

-

-

-

6

-

1

7

Robbery

-

-

-

1

-

-

-

-

-

1

Snatching

-

-

-

-

-

-

1

-

-

1

TCWA

-

-

-

-

-

1

-

-

-

1

Theft

66

-

7

42

7

6

48

5

10

191

Theft From Vehicle

1

-

2

-

-

1

1

-

71

8 4

5

Trespassing Total

-

-

-

-

-

-

1

-

-

1

68

0

10

47

7

10

66

5

12

225

Table 4.5: Crime statistics of Main Campus (M), Western Region (W), and Southern Region (S) of HKU between 2002 and 2004

One of the characteristics of crime in the University of Hong Kong is that over 90% of crime was property crime. Property crime usually has a direct relationship with the types of victims that the offenders normally profiled. The statistics of victimisation briefly gives a hint as to who was usually being targeted.

In the case of theft of notebook computers, nearly half of them belonged to academic departments and the thefts took place in administration blocks, offices and classrooms. Thirty seven notebook computers belonged to the students of which only a small portion of victims was female. Perhaps, the staff were more cautious about their personal notebook computer and their victimisation rate was the lowest. The situation seems slightly different in the case of theft of wallet and cash, where some 60% of the victims were female students and nearly 83% of the victims were students. In the case of burglary, 5 out of 8 cases were involved in a loss of 28 computers and another 3 cases in a loss of 5 projectors, and 7 out of 8 victims were academic departments. In the case of loitering and peeping tom, all victims were noticeably female students. Tables 4.6 and 4.7 present the victimisation rates of the former two situations.

Office / Gender of Victim

No. of

Percentage of

No. of stolen

Percentage of stolen

of Theft

Victim

Victim

computer

computer

Office

12

21.82%

41

48.81%

Female student

5

9.09%

5

5.95%

Male student

11

20.00%

11

13.10%

72

Student (gender unknown)

21

38.18%

21

25.00%

Female staff

1

1.82%

1

1.19%

Male staff

3

5.45%

3

3.57%

Staff (gender unknown)

2

3.64%

2

2.38%

Total

55

100.00%

84

100.00%

Table 4.6: Victimisation rate of theft of notebook computer between 2002 and 2004

Victim of Theft of Wallet & Cash

Total

Percentage

Female student

53

61.63%

Male student

4

4.65%

Student (gender unknown)

14

16.28%

Female staff

5

5.81%

Male staff

0

0.00%

Staff (gender unknown)

10

11.63%

Total

86

100.00%

Table 4.7: Victimisation rate of theft of wallet and cash between 2002 and 2004

4.3.1

Spatial Analysis - Theft

The statistics at Table 4.8 show the locations which suffered severely from theft between the year 2002 and 2004. The library area was severely hit by thieves with 60 theft cases and ranks on the top of the list. Following that the Knowles Building was attacked 24 times while K.K. Leung Building ranks the third with 9 reports. The statistics also table the top ten hit locations of theft.

Location of Theft

No. of Crime

No. of Theft

Percentage

Main Library

36

36

100.00%

Knowles Building

24

17

70.83%

Old Library

17

17

100.00%

K.K. Leung Building

9

7

77.78%

Composite Building

7

3

42.86%

Haking Wong Building

7

6

85.71%

73

Library Extension

7

5

71.43%

Swire Hall

7

4

57.14%

Faculty of Medicine Building

6

6

100.00%

Chow Yei Ching Building

6

6

100.00%

Simon K.Y. Lee Hall

6

6

100.00%

132

113

85.61%

Total

Table 4.8: Prominent locations of theft between 2002 and 2004

Theft is a prevailing crime and a property crime. Understanding the properties of the offenders targeted helps unearth the relationship of its spatial and temporal characteristics. The most popular items being stolen were: (1) notebook computer because of its mobility (83 were stolen); (2) wallet and cash which involved 46 cases because of easy disposal; (3) projectors because of its high re-sell value (33 were stolen which added up to nearly a million Hong Kong dollars); (4) mobile phone because of its trendy attraction (15 were stolen); (5) camera because of its compactness (13 were stolen of which 11 were digital cameras).

Hence, studying statistics of stolen properties helps to identify the black spots. The prominent black spots of theft of computers at Table 4.9 show that the Knowles Building and Chong Yuet Ming Complex were primarily targeted.

Location of Theft of Computer

No. of Crime

No. of Stolen Computer

Knowles Building

11

12

Chong Yuet Ming Physics Building

4

4

Chong Yuet Ming Amenities Centre

2

2

Chong Yuet Ming Chemistry Building

3

3

Main Library and Library Extension

4

4

St. John's College

4

4

James Lee Hall

3

3

Table 4.9 Prominent locations of theft of notebook computer between 2002 and 2004

74

The prominent black spots of theft of wallet and cash at Table 4.10 indicate that in the library area, the classrooms of the Faculty of Medicine Building and the dormitory at the Starr Hall a total of 53 students were victimised.

Location of Theft of Wallet & Cash

No. of Crime

Victim Gender

New Library

28

25 female students and 3 male students

Old Library

15

all female students

Library Extension

2

all female students

Faculty of Medicine Building

4

all female students

Starr Hall

4

all student, gender unknown

Table 4.10 Prominent locations of theft of wallet and cash between 2002 and 2004

In the total of 191 cases of theft, 88 cases, representing 46% of the total thefts, were related to theft of wallet and cash. 45 of which took place in the library area, especially in the year of 2002, and a few at the Faculty of Medicine Building and the Starr Hall, classroom and dormitory were also targeted.

Statistics at Table 4.11 indicate that the well-known black spots of theft of projectors were at the Knowles Building and the Ming Wah Complex and a total of 11 projectors were stolen. Between the year 2002 and 2004, there were 33 projectors stolen. 5 cases reported in the year 2002 with a loss of 6 projectors, 13 cases in the year after with 20 projectors and finally 5 cases in 2002 with 7 projectors being stolen. The upsurge of this trend, particularly in the year 2003, was a result of the legalising soccer gambling when the demand became high.

75

Location of Theft of Projector

No. of Crime

No. of Stolen Projector

Knowles Building

5

7

Ming Wah Complex

3

4

Total

8

11

Table 4.11 Prominent locations of theft of projector between 2002 and 2004

When compared with the statistics at Table 4.12 which shows computer projection facilities in the buildings of the Main Campus, it tends to suggest the block with more classrooms attracts more thefts (theft of computers and projectors in particular) from the blocks with less classrooms. However, the Main Building, which is one of the oldest buildings in the Main Campus and ranks the third in the number of most classrooms blocks, was rarely attacked with only 3 theft cases reported and they all happened in the year of 2004. When examining the security devices installed in the building, such as surveillance camera and electronic door locks, they are comparatively weaker than other targeted blocks. One of the premises drawn from this observation is that the top 4 buildings with most classrooms, except the Main Building, are situated in the centre of the Main Campus or in the radius of communal area, such as cafeteria and student associations. The centrality of its geographic location with accesses to almost all the blocks in the campus allows students to muster and socialise, and it becomes a common assembly point. This provides potential offender an excellent setting to conceal himself or even make good of his escape in case something went wrong. The bus stops and mini-bus stops along Boham Road which are within close proximity of the Knowles Building and K.K. Leung Building offer good escape route to culprits. In contrast, the Main Building does not provide offenders with many of these advantages.

76

Building

No. Classroom /

Seating Capacity

Theatre

No. of Computer Projection

Meng Wah Complex

16 classrooms

1,529

16

K.K. Leung Building

12 classrooms

755

12

Main Building

11 classrooms

953

11

Library Extension

9 classrooms

1,227

9

Knowles Building

5 classrooms

793

5

Chong Yuet Ching Building

4 classrooms

720

4

James Lee Building

4 classrooms

269

4

Run Run Shaw Building

4 classrooms

186

4

Chow Yei Ching Building

3 theatres

395

3

T.T Tsui Building.

3 classrooms

330

3

Eliot Hall

2 classrooms

82

2

1 theatre

298

1

Rayson Huang Theatre Shaw Bldg

Table 4.12: Classroom facilities in the Main Campus buildings Source: HKU Examination Unit (http://www.hku.hk/exam)

In the year 2002, over 93% of the total crimes were theft of which 66 cases occurred in the Main Campus and 7 in the Southern Region. Of the 66 cases of theft which cropped up in the Main Campus, the commonly attacked areas were the Main Library and the Old Library with 47 reports. Of the 7 theft reports made in the Southern Region, four took place in the Faculty of Medicine Building and three were in the nearby buildings. Figure 4.3 shows that the hot spots of theft in the Main Campus are centred around the library blocks. Seemingly, the library is one of the most popular spots with the university for staff and students, and targeting victims for petty cash and personal belongings at the library always yields good rewards for the criminals.

77

Figure 4.3: Hot spots of theft in the Main Campus in 2002

In the year 2003, theft dropped to 55 cases from 77 cases in the preceding year. The Main Campus was still the targeted area with 42 thefts reported. Meanwhile, the Western Region was also attacked with 7 cases and Southern Region remained very much similar to last year with 6 cases. Surprisingly, only one report of theft took place in the Main Library and another two in the Library Extension. This was as a result of a thief being arrested red-handed in the Main Library in October 2002 as well as the strengthening of consciousness of crime prevention and the tactical option employed by the Security Section. There were signs and notices affixed to the studying desks warning the library users to be cautious about their belongings. At the same time, plainclothes guards were deployed to identify any suspicious or potential offenders mingling in the vicinity. Figure 4.4 portrays the hot spots of theft in the Main Campus and the Western Region. 78

Equally astonishingly, 13 cases of theft were recorded in the Knowles Building and 5 cases in the Chong Yuet Ming Physics and Chemistry Building blocks. It seems the trend of theft has shifted away from the library area eastward and westward due to displacement factor and a small number of thefts started to spread around from the centre of the library to all the classroom blocks.

Dormitory theft became one of the characteristics of crime in the year 2003. All 7 reports of theft which occurred in the Western Region were located in the dormitories with 5 reports at the St. John’s College. 6 reports of theft took place in the Southern Region where 3 were found in the Wei Lun Hall and one in the Lee Hysan Hall. In the case of Wei Lun Hall, 2 thefts were reported in the neighbouring flats.

Figure 4.4: Hot spots of theft in the Main Campus and the Western Region in 2003

79

In the year 2004, theft experienced a slightly upward trend and increased from 55 in the previous year to 63. 48 reports of theft were made in the Main Campus, 5 in the Western Region and 10 in the Southern Region. Figure 4.5 shows the hot spots of theft of theft in the Main Campus and the Western Region.

The spatial distribution of theft of the year 2004 has changed in the Main Campus. It has expanded from congregating around the Knowles Building and the Chong Yuet Ming Physics and Chemistry Building blocks northward to the Swire Hall and westward to the Simon K.Y. Lee Hall. A two-fold increase of theft with 7 reports was made in the Main Library and the Library Extension but there was a reduction by nearly a half with 6 reports in the Knowles Building. Dormitories were also heavily attacked with 4 cases of theft occurring in the Swire Hall where one of the thefts took place in the amenities centre and 4 in the Simon K.Y. Lee Hall.

A similar situation happened in the Western Region where 4 out of 5 theft reports were made in the Starr Hall shifted northward from the St. John’s College, which was the targeted dormitory block in the previous year. The situation in the Southern Region remained not much different from last year except that 4 out of 10 theft reports were made in the R.C. Lee Hall moved southward from the Wei Lun Hall, which was the badly hit dormitory block in the preceding year.

80

Figure 4.5: Hot spots of theft in the Main Campus and the Western Region in 2004

4.3.2

Spatial Analysis – Loitering and Peeping Tom

Of deep concerned was that number of loitering and related peeping tom and indecent exposure complaints rose to 10 cases between the latter part of the year of 2003 and the end of the year 2004 which appeared outrageous. There was no history of complaints of this nature made to the Security Section before that and the upsurge of this activity requires further attention.

Spatially, both loitering offences, the 7 reports of peeping tom and one report of indecent exposure in the female toilets all occurred in the Main Campus except for one in the Southern Region. One loitering and one peeping tom case were located in the Library Extension, one peeping tom and one indecent exposure took

81

place in the same disabled toilet in the Fong Shu Chuen Amenities Building, two peeping tom cases were reported in the same toilet complex in the K.K. Leung Building, one loitering case on the upper floor and one peeping tom case on the lower floor were located in the Knowles Building, one peeping tom case occurred in the Hui Oi Chow Building; and last peeping tom case occurred in the dormitory of the Madam S.H. Ho Residence in the Southern Region.

In connection with this peculiar activity, a case of theft was reported in July 2004 where some woman’s clothing were stolen from a laundry room at the R.C. Lee Hall in the small hours. Yet, there is no evidence to link this case with peeping tom activity.

In one of the loitering cases a suspect was found acting suspiciously in a female toilet by a female student and later charged with the offence but he was acquitted in court. In the rest of the cases, no one was arrested. There were three cases reported in connection with weird behaviour in the use of camera. Security Section has been actively investigating this unusual behaviour and has advised the management to heighten the cubicle partition in order to avoid any re-occurrence of “sneak a look” activity in the female toilet around the black spots as a preventive measure.

All the spatial evidence suggests that the female toilets in the communal blocks and classroom blocks, especially at the lower floors where 8 out of 10 cases were reported, were seriously targeted. It is easy to understand that the lower floors always foster a better chance of escape upon discovery. As such, it is strongly believed that the suspect has a good local knowledge of the area and probably even 82

resides not far away from the scenes as a hideaway. Figure 4.6 plots out the hot spots of this activity.

Figure 4.6: Hot spots of loitering and suspected peeping tom activities in the Main Campus in 2004

4.3.3

Spatial Analysis – Burglary

Between the year 2003 and 2004 there were 8 cases of burglary reported and 4 in each year of which 3 were in the Main Campus and one in the Southern region. The spatial pattern appeared to be the same in these two study areas. In general, two locations were burgled twice where the Run Run Shaw Building and the New Clinical Building was hit once in each year and another six cases were single incident.

Figure 4.7 shows an even amount of spatial distribution of burglary in the 83

Main Campus. Due to small number of burglary reports and light density of its distribution, no hot spot can be identified. Nonetheless, one significant observation can be generalised that the burglars tended to target the administration blocks or office blocks.

Figure 4.7: Spatial distribution of burglary in the Main Campus between 2003 and 2004

4.4

Temporal Analysis

Temporal analysis is the study of pattern changes in time. Temporal distribution of crime events helps crime analysts to understand what has happened and what will happen in the future. Measuring time can be done in many ways, such as annually, seasonally, monthly, weekly, daily and even hourly. When considering the measurement interval, one has take into account the characteristics of an area, clusters and serial occurrence of crime events. The University of Hong Kong is a tertiary institution where activities of staff and students are worked on regular 84

monthly, weekly and daily schedules. This is the framework of measurement interval to be applied in temporal analysis.

As stated in Section 4.2 for examining crime distribution this study only aims at studying temporal changes of the three specific crimes affecting the law and order of the campus which are theft, loitering and peeping tom, and burglary. If other crimes are prevailing, the application of temporal analysis will be equally applicable.

4.4.1 Temporal Identification – Changes in Months

The comparative frequency of crime per month at Figure 4.8 demonstrates that the number of crime events in the year 2002 were highest in the months of February, May, September and October with an average of 9 or more crimes. Summer vacation seems to provide a favourable buffer period when crime dropped dramatically.

The overall crime situation in the year 2003 started to decline as the total number of crimes decreased 19% from 78 in the year 2002 down to 64, and the peak months were shifted to March, May, July, September and December with a month interval in between. The upsurge of nine cases of crime in July seems unusual where 2 burglaries and 3 thefts were purposely targeted at the administration and office blocks. One of the possible explanations for this escalation of crime is believed to be that a good number of offices were either left unoccupied or closed during summer holidays turning them into a more vulnerable situation thus attracting the attention of opportunity offenders. 85

In the year 2004, the trend of crime was changed totally in the second half of the year that the average crime rate rose to slightly more than 8 cases each month. It may be conclusive to say that the rise was resulted from the emergence of all crimes, particularly peeping tom, indecent exposure, snatching and other minor offences. March was the worst month in which 10 cases of theft and 2 cases of burglary took place.

No. of Crime

Frequency of Crime by Month (2002 - 2004) 16 14 12 10 8 6 4 2 0

2002 2003 2004

Jan Feb Mar Apri May Jun Jul Aug Sept Oct Nov Dec Figure 4.8: Comparison of frequency of crime by month between 2002 and 2004

The following accumulative bar chart at Figure 4.9 depicts the monthly distribution of crimes in the campus over the duration of the research period. It can be concluded that February and March, and September and October are the peak periods of criminal activities. It is highly suggested that these high crime seasons always come around after the winter and summer holidays.

86

Frequency of Crime by Month (2002 - 2004) 35

No. of Crime

30 25 2004 2003 2002

20 15 10 5 0 Jan Feb Mar Apri May Jun Jul Aug Sept Oct Nov Dec

Figure 4.9: Accumulative frequency of crime by month between 2002 and 2004

4.4.2

Temporal Identification - Changes in Weekdays

Figure 4.10 displays the changes of temporal pattern of campus crime in weekdays and shows that Monday, Tuesday, Thursday and Friday are more vulnerable than the other days in the rest of the week. This is probably having a direct link with the school days as many students have no classes in the low risk days. This is quite true until the year 2004 when almost every day was targeted and there is no apparent trend identified.

87

No. of Crime

Frequency of Crime by Weekday (2002 - 2004) 18 16 14 12 10 8 6 4 2 0

2002 2003 2004

Mon

Tue

Wed

Thur

Fri

Sat

Sun

Figure 4.10: Frequency of crime by weekday between 2002 and 2004

4.4.3

Temporal Identification - Changes in Hours

The deficiency of some missing temporal records found in the data classification stage certainly has an adverse effect in presenting the findings of temporal analysis.

Table 4.13 shows that the extent of missing temporal data of hours is quite scattered around in the three study areas. 42 crime events in the Main Campus where time was not recorded represents about 23% of the total number of crime data in this study. It becomes worse in the Western Region and Southern Region where one third and nearly half of the respective records were absent. Therefore, the temporal study will focus on the Main Campus. In order to not to cause any confusion with the missing hours records in this section, those records containing hours information will be referred as known records whereas those which do not will be the unknown records.

88

Main Campus

2002

2003

2004

Total

Unknown

6

15

21

42

Known

62

32

45

139

Total

68

47

66

181

Unknown %

8.82%

31.91%

31.82%

23.20%

Western Region

2002

2003

2004

Total

Unknown

0

3

1

4

Known

0

4

4

8

Total

0

7

5

12

Unknown %

0.00%

42.86%

20.00%

33.33%

Southern Region

2002

2003

2004

Total

Unknown

6

4

5

15

Known

4

6

7

17

Total

10

10

12

32

60.00%

40.00%

41.67%

46.88%

Unknown %

Table 4.13: Statistics of known and unknown crime records between 2002 and 2004

The incompleteness of temporal data records in this research has been explained in Chapter three and how it affects the measurement on the frequency of crime committed in a 24-hour period. As mentioned previously, the accuracy of hours assessment in this study is about 73%.

Table 4.14 outlines the temporal distribution of the known records of crimes. This study will assess the temporal distribution of crime in three sessions. The tuition hours of the University of Hong Kong start from eight o’clock (0800hrs) in the morning to six o’clock (1800hrs) in the evening. The time before and after the school hours will be sub-divided into two time segments using the mid-night line (2400hrs). Between six o’clock (1800hrs) in the evening and before mid-night is the 89

evening session; and from mid-night to eight o’clock (0800hrs) in the morning is the morning session. The delineation of three separate periods will explain the temporal distribution of crimes as some crimes occur at a specific time.

Time

2002

2003

2004

Total

Percentage

0:00

0

0

1

1

0.61%

1:00

1

0

0

1

0.61%

2:00

0

2

0

2

1.22%

3:00

0

0

1

1

0.61%

4:00

0

0

1

1

0.61%

5:00

0

0

2

2

1.22%

6:00

0

0

1

1

0.61%

7:00

0

0

1

1

0.61%

8:00

0

3

1

4

2.44%

9:00

3

0

1

4

2.44%

10:00

5

3

2

10

6.10%

11:00

3

1

1

5

3.05%

12:00

6

2

2

10

6.10%

13:00

5

5

4

14

8.54%

14:00

8

2

7

17

10.37%

15:00

8

4

3

15

9.15%

16:00

5

3

7

15

9.15%

17:00

7

6

2

15

9.15%

18:00

8

4

5

17

10.37%

19:00

2

5

4

11

6.71%

20:00

3

0

4

7

4.27%

21:00

2

0

3

5

3.05%

22:00

0

0

0

0

0.00%

23:00

0

2

3

5

3.05%

Known

66

42

56

164

100.00%

Unknown

12

22

27

61

---

Total

78

64

83

225

---

Table 4.14: Statistics of known crime records by hours between 2002 and 2004

Again from Table 4.14, it can be seen that between the year 2002 and 2004, 90

the occurrence of crime in the morning session took up slightly over 6% of the total of 104 crime cases. Nearly 77% of the total known cases occurred in the school session while another 27% took place in the evening session. Figure 4.8 shows the variation of time of the campus crime.

Frequency of Crime by Hours (2002 - 2004)

No. of Crime

10

2002 2003 2004

8 6 4

Known cases: 164

2

0: 00 2: 00 4: 00 6: 00 8: 00 10 :0 0 12 :0 0 14 :0 0 16 :0 0 18 :0 0 20 :0 0 22 :0 0

0

Time Interval

Unknown cases:61 12 in 2002 22 in 2003 27 in 2004

Figure 4.11: Frequency of crime by hours between 2002 and 2004.

In the following cartographic presentations, the temporal changes of crime between the year 2002 and 2004 are shown to be quite significant. Figure 4.12 and Figure 4.13 indicate s that in the year 2002 the temporal pattern is centred on the library area, especially during the class session. Figure 4.14 and Figure 4.15 reveal the situation in the year 2003 was changed that the pattern extended eastward and westward again from the library area with a much lower crime rate and that is similar to its spatial distribution. Figure 4.16 and Figure 4.17 shows that in the year 2004 the pattern shift to northeast during the class session while the pattern in the evening session remained much the same as that of the year 2003.

91

Figure 4.12: Temporal distribution of crime in the Main Campus between 0800 and 1800hrs in 2002

Figure 4.13: Temporal distribution of crime in the Main Campus between 1800 and 2400hrs in 2002

92

Figure 4.14: Temporal distribution of crime in the Main Campus between 0800 and 1800hrs in 2003

Figure 4.15: Temporal distribution of crime in the Main Campus between 1800 and 2400hrs in 2003

93

Figure 4.16: Temporal distribution of crime in the Main Campus between 0800 and 1800hrs in 2004

Figure 4.17: Temporal distribution of crime in the Main Campus between 1800 and 2400hrs in 2004

94

4.4.4

Temporal Analysis - Theft

Table 4.15 presents the temporal distribution of theft between the year 2002 and 2004. The upsurge of theft cases in the year 2002 was an exceptional phenomenon. An observation of this trend between the year 2003 and 2004 indicates that on average it is most active in February and March during the first half of the year and same is true in September, November and December in the second half. The reason is believed to be connected with the opening of school after the winter and summer holidays when the students returned to school and the opportunity for being targeted becomes higher. Yet, there is no pattern of weekday identified in this analysis.

Year

2400-0800 hrs

0800-1800 hrs

1800-2400 hrs

Total

2002

1

52

11

64

2003

1

27

7

35

2004

5

20

14

39

Total

7

99

32

138

Percentage

5.07%

71.74%

23.19%

100.00%

Table 4.15: Temporal distribution of theft between 2002 and 2004

Figure 4.15, however, suggests that theft varies in different time segments. It seldom occurred in the morning session which only takes a small portion of 5% of all the known records. In the year 2004 there were 5 cases reported in which 4 took place in the dormitory where some petty cash, clothing and notebook computers were stolen, and one in the computer laboratory.

Crime become active during the class session. In the year 2002, 24 cases of

95

theft took place in the library area falling into this time segment. The year after, there were 3 each found in the Chong Yuet Ming Chemistry Building, Knowles Building and the library area. In these 9 cases of theft, 5 took place in the office where notebook computers were targeted and one in the registry where 3 projectors worth HK$130,000 were found missing. Probably due to the frequent flow of staff and student in these areas over this period of time, offices and classroom became targeted. By the year 2004, no significant temporal pattern of theft of more than two occurrences could be detected.

Stealing activity does not diminish in the evening course hours. The only temporal pattern of theft was reported at the Knowles Building where 3 offices and one workshop were attacked with 3 stolen notebook computers and some cash taken, while the remaining one was an attempted theft of a notebook computer. In two of these cases, a male was found acting suspiciously in the building and another male was found to be at his age of early fifty’s. Again, it is believed that many staff and students remained in their offices and classrooms after normal working hours or studying hours, and their alertness was lowered so their negligence increased.

96

Temporal Pattern of Theft (2002 - 2004) 60 No. of Crime

50 40

2002 2003 2004

30 20 10 0 2400-0800 hrs

0800-1800 hrs

1800-2400 hrs

Time Interval Figure 4.18: Temporal variation of theft between 2002 and 2004

4.4.5

Temporal Analysis – Loitering and Peeping Tom

As mentioned before, loitering, peeping tom and indecent exposure did not occur prior to the end of the year 2003. 4 reports of peeping tom were made between May and July, and 2 in October in the year 2004. 3 occurrences were found on Wednesdays.

Of the nine known cases as shown at Table 4.16, one report of peeping tom was made at eleven o’clock (1100hrs) in the morning and 4 between half-past one (1330hrs) and four o’clock (1600hrs) in the afternoon. During the class session, female toilets probably have fewer visitors and therefore give the potential offenders an opportunity. The sole indecent exposure also falls into this time frame. Another two reports of peeing tom and one loitering took place between seven o’clock (1900hrs) and eleven o’clock (2300hrs) in the evening and there was not conclusive explanation to this phenomenon as the data sample is rather small.

97

Loitering

2400-0800

0800-1800

1800-2400

Total

2002

0

0

0

0

2003

0

0

0

0

2004

0

6

3

9

Total

0

6

3

9

Percentage

0.00%

66.67%

33.33%

100.00%

Table 4.16: Temporal distribution of loitering, peeping tom and indecent exposure between 2002 and 2004

4.4.6

Temporal Analysis – Burglary

Estimating the actual time of burglary is no easy task except with the assistance of technology. As a result, therefore, the temporal distribution at Table 4.17 only shows the starting time when the victim had last seen the stolen properties. Burglary reports are often received in the morning and it is usually a few hours before the victim discovered the loss.

Burglary

2400-0800

0800-1800

1800-2400

Total

2002

0

0

0

0

2003

0

2

2

4

2004

1

2

0

3

Total

1

4

2

7

Percentage

14.29%

57.14%

28.57%

100.00%

Table 4.17: Temporal distribution of burglary between 2002 and 2004

There was no report of burglary made in the year 2002 but 4 were made in each of the following two years. Of the 7 known records, there was one which was responded to immediately after the offenders triggered the burglar alarm and another one reacted to within 2 hours, while in the rest of the 5 cases the average time span

98

for making a report was more than 15 hours. This explains the complexity in measuring the exact time a burglary was committed. One significant observation is revealed that in the 4 known cases of burglary in the year 2003, 2 were made in March and July and in both cases they were just one day apart. That is not to suggest they were connected but the frequency of burglary in these two months was high. Further examination of the 7 burglary records shows that 6 burglaries took place between Tuesday and Thursday and burglars tend to aim at the office blocks and commit their act in the late evening or in the small hours.

In 4 burglary cases offenders smashed the window panel in order to gain the entry and stole the notebook computer therein, and in two of these cases 12 computers were stolen each time. In another case the offender prised opened the door to the theatre and two projectors worth HK$280,000 were stolen. Theft of a single notebook computer might be committed by a student who could hide it away easily and swiftly but the appropriation of 12 notebook computers and 2 theatre projectors might not. It is highly suspected that these serious burglaries could only be committed by professionals on a steal-to-order motive. They also required a vehicle to move away these heavy load of stolen properties.

Security Section also noticed the seriousness of theft of notebook computers and projectors and started to launch a series of crime prevention campaigns to control the situation by the end of the year 2003. They carried out risk assessment to the buildings at high risk of theft and burglary and advised the administration personnel to install closed circuit television (CCTV) in the area likely to be exposed to theft and burglary. They also revised their deployment strategy to task the patrols in the high-risk area and tighten up the exit control of the campus. More and more 99

CCTV and electronic door locks were installed in the classrooms equipped with expensive equipment. The crime statistics at Table 4.18 seem promising that in terms of crime cases, theft of projector was down significantly from 13 cases in the year 2003 to 5 in the following year and there was no report made between March and October in that year. Theft of notebook computer also witnessed a slight drop from 26 in the year 2003 to 20 in the year after but the actual number of loss of notebook computers went up due to the two incidents where 12 computers were stolen each time in July and September respectively.

Property Crime

2002

2003

2004

Theft of projector

5

13

5

No. of stolen projector

6

20

17

Theft of notebook computer

16

26

20

No. of stolen computer

17

24

43

Table 4.18: Statistics of theft of notebook computer and projector between 2002 and 2004

4.5

Campus Crime outside Hong Kong

Subject to the Student Right-to-Know and Campus Security Act 1990 and upon request from the Department of Education, an institute, the National Center for Education Statistics (NCES) was appointed to conduct surveys annually on campus crime and security at tertiary institutions. The Act also demands tertiary institutions publish and distribute a security report containing information about campus policies and crime situation for all students and staff. The purpose of conducting the survey and compiling security report is intended to encourage institutions to put more emphasis on promoting campus safety and launching crime prevention campaign (National

Center

for

100

Education

Statistics.

http://nces.gov/surveys/peqis/publications/97402/3.asp).

The

website

of

Crime

On

Campus

(Crime

On

Campus,

Inc.,

http://securityoncampus.org/schools/cleryact/index.html) provides promising result on campus crime statistics of five hundred tertiary institutions in the year 2003. Of the 500 institutions, sixteen with an enrolment between 18,000 and 21,000 students were selected to compare with the University of Hong Kong. Table 4.19 shows the sharp contrast in crime situation between the United States institutions and the University of Hong Kong. One may argue the difference in social cultures, education systems and security measures may vary the result. Still, it demonstrates that the University of Hong Kong is comparatively safer than those in the United States.

University / College

Enrolment

Mur

Rape

Rob

Assault

Bur

Theft M Theft Arson

University of Delaware

20949

0

4

8

14

46

395

7

2

New York State

20855

0

2

5

5

43

638

14

*

20675

0

1

3

3

38

229

17

0

20373

0

4

1

4

58

358

4

3

University of Memphis

20332

0

2

4

0

50

225

34

0

University of Toledo

20313

0

1

2

2

61

315

12

3

Middle Tennessee State

20073

0

0

5

6

31

197

5

2

20007

0

0

0

5

30

342

40

1

19888

0

1

0

0

8

187

26

0

19762

0

4

0

4

4

133

1

0

University, Stony Brook California State University, Los Angeles University of California Santa Barbara

University California State University, Fresno Fresno Community College Grand Valley State University

101

Northern Arizona

19728

0

7

1

18

61

416

7

7

19642

0

0

0

5

6

263

3

0

East Carolina University

19412

0

2

9

2

22

290

4

0

University of Alabama,

19130

0

4

1

6

14

400

9

*

19041

0

1

3

0

40

252

30

0

Hong Kong University

19000

0

0

1

0

4

55

2

0

Bowling Green State

18739

0

2

1

0

8

255

3

2

University Columbus State Community University

Tuscaloosa California State Polytec University, Pomona

University

Table 4.19: Comparison of campus crime statistics between HKU and US Universities in 2003 Source of Statistics: Security on Campus. www.securityoncampus.org/crimestats/ucr03.pdf Mur – Murder & Non-negligent Manslaughter; Rape – Forcible; Rob – Robbery; Assault – Aggravated Assault; Bur – Burglary; Theft – Larceny / Theft; M theft – Motor Vehicle Theft Arson - Arson

* - FBI did not receive 12 months of arson data from either the agency or the state.

4.6

SUMMARY

This chapter has demonstrated the capability of data visualisation in spatial and temporal analysis for crime mapping. In comparison with the statistical analysis operated side by side in the study, hot spot area crime analysis can identify the hotspots of the prevailing crimes in the campus visually with the use of graduated size symbol, and presents far better result than the textual description of crime distribution. It further substantiates the feasibility of applying GIS computerised mapping techniques in the campus environment.

102

CHAPTER FIVE CONCLUSION

5.1

Introduction

The data visualisation method of crime mapping and its practicality in spatial and temporal crime analysis were discussed in the previous chapter. This chapter will discuss some of the limitations encountered in this research and summarise the major findings from the current analysis. To echo to the last focus stated in chapter one, it will also make some recommendations to improve the current crime control and crime prevention strategy and finally some suggestions for the future research.

5.2

Limitations of the Study

The first and perhaps most obvious limitation of the study lies in the data availability. Like most of the crime studies, comprehensive crime information from the police is anticipated to foster an accurate assessment on the development of crime. Due to personal confidentiality and data privacy control, some vital and useful information were not recorded or made available from the Security Section. Realistically, information on victims and potential offenders will assist in constructing profiling analysis and some interesting findings may be stemmed out from this analysis.

Another limitation of the current study is the data structure of the raw crime data. Not only was it presented in a textual descriptive format, it was not structured 103

nor suitably categorised for any possible comparative analysis. Not to mention the manual input errors, missing temporal data records and low data standardisation found in the data set. It took quite a long while to overcome these problems by verifying and rectifying each and every single crime incident record before the data could be graphically presented.

A final key limitation in this research lies in the small data sample. There were only 225 crime incidents reported between the year 2002 and 2004, including a few non-crime incidents explained in Chapter three. With an average of only 75 crime cases a year and given the disparity of crime distribution in some areas, the outcome of this analysis tends towards generalisation and therefore the inferences could only be tentatively drawn.

5.3

Analysis Findings

The findings of this research are summarised in the context of the effectiveness of spatial and temporal analysis in crime mapping and the prediction of future crime trend using counter-factual forecast method.

5.3.1

Crime in General

The application of GIS in mapping campus crime has been proved to be effective in plotting all the crime black spots on a two-dimension platform for spatial analysis. Some hot spots of prevailing crimes were more identifiable than the others because the occurrence of the former were more frequent in volume and more concentrated in an area than that of the latter. This reveals one fact that in the setting 104

of the University of Hong Kong due to low rate of some rare crimes, such as snatching and robbery, they do not produce any hot spot. Analysis into these rarely occurred offences require further justification by means of other analysis techniques, such as temporal and modus operandi analysis used in this study.

When comparing the crime rate between the year 2002 and 2004 with the annual average of 75, the crime rate of the year 2004 climbed to the highest point at 83, a recorded high in a three-year period. An alarming signal is the rising number of different crime types, which went from 3 in the year 2002 to 6 in the following year and finally reached to 14 in the year 2004, which poses a challenge to the security of the campus. The emergence of other crimes into the campus, even if the rate of which was low, if they are not properly controlled or enforced the opportunity for crimes will be promoted, such as snatching and robbery. The deployment of frequent patrols to the high-risk areas will initiate a displacement effect, which, in turns, might provide an opportunity to potential offenders to commit crime in the less-attention or low-risk areas.

5.3.2

Prevailing Crimes

The apprehension of an active thief in the year 2002 has neutralised theft activities in the library area, but theft is still prevailing throughout the whole campus and continues to spread out from the conventional targets, such as classrooms and office areas, to the dormitories. More importantly, theft of petty cash, notebook computers and some trendy appliances, such as mobile phones and digital cameras, are higher. Spatial analysis of theft is very effective in locating all the black spots as well as monitoring the annual changes of it. Since there was no one arrested or 105

confronted and victim kept on neglecting on their personal belongings, the identified target locations would still be vulnerable to opportunity thieves. If the economy of Hong Kong in the coming year stays very much like the year 2004, the current crime situation will not seem to be changed very much.

The disquieting burglary also remains problematic. There was no arrest made nor any stolen properties recovered in the past providing few clues to the analysts for further study. Spatial analysis in this research does not yield any useful information in identifying any hot spot due to small data samples and its sparse spatial distribution. Yet, temporal analysis and modus operandi (MO) analysis provide an important lead that most of these “bespoken” burglars target on the high value items, such as projectors and notebook computers, or perhaps some other expensive equipment. As long as these properties are well protected and kept confidential, it will render less chances to the burglars. The step up of preventive measures against burglary, such as installation of CCTV and strengthening of door locks, in the high-risk blocks recently will serve its purpose. However, spatial analysis reveals one significance phenomenon that they never take place twice in the same building block at least this is the situation so far. If the hypothesis is correct, the attention should be drawn to other blocks in the campus.

Of concern are the increasing numbers of the activities of peeping tom and indecent exposure in the year 2004. Not only have they caused anxiety to the females, the worst scenario is that they would turn into some serious sexual offences, such as indecent assault and rape, if “Broken Windows” theory is correct. Spatial analysis has precisely identified a hot spot of 4 building blocks in the Main Campus where the suspects usually frequent. Temporal analysis has also traced the time span 106

when the suspects become more active. This spatiotemporal analysis narrows down the parameter of the suspects’ peculiar behaviour in time and space and serves to provide the Security Section sufficient information and justification as well to allocate resources to keep the areas under close surveillance. That is to say, spatiotemporal analysis can be a handy tool for problem-solving and decision-making process without the assistance of modus operandi analysis.

5.3.3

Forecast of Prevailing Crimes

Based on hypothesis made in Section 5.3.2, the forecast of two prevailing crimes, theft and peeping tom activities would continue in the three study areas as shown in Figure 5.1 and 5.2. Theft will stay prevailing in the communal areas, such as library, classrooms, amenities hall and dormitories. Peeping tom activities will shift to the building blocks east and west from the Knowles Building, especially at the lower floor female toilets, which have not been renovated with heightened partition.

107

Figure 5.1: Forecast of crimes in the Main Campus and the Western Region

Figure 5.2: Forecast of crimes in the Southern Region

108

Figure 5.3 exhibits the crime trend between the year 2002 and 2004. Seemingly, it has been varied in its patterns from a seasonal variation pattern in the year 2002 to a cyclical fluctuation pattern in the year after, and finally to an irregular fluctuation pattern which was cyclical high in the first half and became steady in the second half. The changes of these patterns tend to suggest that the pattern in the coming year will be on a more stable development with a small level of fluctuation if there is no change in intervention or in other variables. In other words, the number of crime will be evenly spread out over the said period.

16 14 12 10 8 6 4 2 0

pt O ct N ov D ec

Se

ug

l

A

Ju

A pr i M ay Ju n

ar M

Fe

Ja

b

2002 2003 2004

n

No. of Crime

Statistics of Campus Crime (2002 - 2004)

Figure 5.3: Statistics of campus crime between 2002 and 2004

In a nutshell, GIS computerised mapping is found to be a feasible solution in mapping crime hot spots but not considered to be fully successful in the campus environment. In the case of burglary and rarely occurred offences, no hot spot was produced because of the thin distribution of burglary cases and low crime rate of the minor offences. In the case of theft, hot spots were identified spatially but not quite apparent temporally. In the case of peeping tom activities, hot spots were shown in 109

both spatial analysis and temporal analysis.

Hence, spatial analysis should not be operated independently and has to be supplemented by temporal analysis in order to reveal the spatiotemporal relationship of crimes. The four scenarios discussed in sections 5.3.1 and 5.3.2 summarise the current situation in this study. To unearth the root cause of a complex crime problem, other analysis techniques, particularly the modus operandi (MO) analysis, are required to operate together.

5.4

Recommendations

Conventional policing, or traditional incident-drive policing, is a reactive approach aiming at resolving each and individual crime incidents (Clarke and Eck 2003). This is the approach the Security Section adopting in the current situation. It is considered to be successful in reducing the crime rate in some of the hot spot areas but it does not control them from spreading out. Problem-oriented policing (POP) is a proactive approach aimed at assessing the underlying causes of problems behind a string of crime incidents (Goldstein 1990). This is a practical approach recommended to control the distribution of crimes and at the same time to eradiate the crime problems.

CompStat paradigm incorporates the best practices of traditional policing and the philosophy of Problem-oriented Policing (POP) in formulating crime control strategy. The four principles of CompStat are equally applicable in targeting crimes in a campus environment for better security protection, though it was originally designed for law enforcement environment and for high crime rate communities. 110

Though the subjects of protection between police and private security are similar, police aim at protecting the safety of the general public while private security aims at protecting corporate or individual assets. There are some commonalities in the nature of duties of the two professional which are overlapping, such as the methods and mandates of policing, and the organisational and operational structure (Henry 2003). The following recommendations in this study are made to improve the current situation.

The cultural change from conventional policing to Problem-oriented policing (POP) requires a clear and well-defined mission statement and scope of vision. It will provide all the personnel a better understanding of the direction an organisation is heading and a straightforward guideline to perform their designated task. It is, therefore, recommended that the Security Section under the Estates Office take on an active approach of Problem-oriented policing in controlling and reducing crime. Once the philosophy is instilled and anchored in their minds, personal accountability and job satisfaction will be raised accordingly.

The structure of the management should be revised as well to cope with the introduction of new technology and techniques, and to suite the changes of the new culture, particularly one of the essential ingredients of CompStat is to stress on the fast flow of information. Thus, exchanging timely information from the crime database and the crime map system makes immediate assessment possible and can devise functional tactics readily for operation. CompStat meeting is another way to facilitate the flow of communication, it allows all the stakeholders to evaluate the effectiveness of enforcement action and share their experience on fighting crime.

111

The complexity of today’s society means that no one can fight against crime on his own. Partnership with other organisations always garners better support, gathers more resources and trades new skills from each other. Resources in the campus are huge. The Geography Department can provide the Security Section a GIS platform for crime mapping analysis. The Sociology Department can conduct survey on the awareness of crime prevention from the students and provide Security Section detailed analysis to devise effective crime control and crime reduction strategies. Psychologists and criminologists may also be helpful in profiling specific offenders, such as those involved in peeping tom activities, so that frontline personnel can look for some characteristics of the potential offender. Liaison with the police surely is important, as they can provide professional advice on crime reduction strategy and detail analysis for targeting specific offenders.

Estates Office should take up the leading role in this regard to co-ordinate with the aforementioned departments and faculties for administration, research and strategic analysis functions. Annual assessment on campus crime should be put forward to the authority for endorsement of crime prevention policy as well as resources allocation for the coming year. Security Section should liaise with the police for intelligence purposes and the administration unit of all departments for routine security checks and risk evaluation functions. On the ground level, the section should conduct regular tactical analysis based on the outcome of spatial and temporal distribution of crime and other forms of intelligence, and task frontline security teams to mount targeting operation and tactical patrol duties. Besides, the section should always evaluate the effectiveness of crime maps and the result of tactical operations, and from time to time to fine-tune the operational plan.

112

In order to enhance the professional capability of the staff, training is definitely a must for all to acquire the required knowledge and skills for a specific function. Some indoor officers should be trained to perform crime analysis together with some basic understanding of GIS and crime mapping. Their analysis will be vital for the supervisors and the management to allocate resources to handle the problems at the right time and at the right place. Patrol guards should have a good grasp of local knowledge and always be briefed of the development of the prevailing crime for tactical deployment.

Education to the students is equally important in crime reduction. Crime prevention seminars should continue and be held more frequently to raise their awareness. Posting precaution notices and distributing anti-crime leaflets will be more effective than disseminating email crime information to individual students each month. Perhaps, an eye-catching presentation of crime information could be shown on the front page of the web before they logged on the university computer system or received their emails.

Crime analysis relies greatly on the clarity of data set and structure of the database. Data collected from the current crime incident reports needs to be expanded and cover more information for analysis, such as standardising keyword for modus operandi (MO) search which is handy for an advanced and multifaceted crime analysis. Also, special event data, such as dates of important school events, opening days and public holidays, are indicators to link some crime incidents with festive events for more precise temporal analysis. Other tabular data should be considered for future inclusion, such as victim’s particulars for victimisation or repeat victimisation analysis, yet issues on personal privacy should not be 113

overlooked.

5.5.

Suggestions for Future Research

Since this research represents the first attempt in studying the spatial and temporal distribution of crime on the campus of the University of Hong Kong, there are two areas that would benefit from further research.

There is a need for further research into the crime distribution from a multifaceted perspective. Acquiring more information on the demography and land use of the campus, such as census distribution on the campus, functions and categories of buildings etc., would enable future researchers to construct a multilayer spatial analysis. It will present a clearer correlation between crime and place and certainly will verify and extend the results of this study.

Future research should, where possible, address the issue of crime prediction. A longitudinal research using statistical analysis techniques, such as time-span analysis, regression analysis and correlation analysis, would enable researchers to forecast the crime trend and crime pattern in a specific time span. It will show the changes of crime over time period and also show the effects of displacement after intervention. Again, it will help researchers to develop and test their hypothesis for initiating a better crime prevention strategy.

5.6

Conclusion

GIS computerised mapping technology cannot in itself stand against crime, 114

and neither can spatial analysis nor temporal analysis. It is the joint efforts of the management and the frontline personnel. Crime mapping is a good means of communication providing management and frontline staff visual information of crimes so that they can easily explore relationships between crime, time and place to identify hot spots for targeting.

In this research, crime mapping techniques have proven to be an effective tool in mapping crime hot spots, and its potential for further development should not be ignored. Integrating computerised mapping technology in law enforcement is the way forward for the future of police and private security. In order to assist the management in formulating effective crime control and crime reduction strategies, and the frontline personnel in deciding appropriate tactical options at ground level, a reform in the current management system and policing strategy is deemed necessary. This is not a technical question but rather a management decision.

115

References

Ainsworth, P. (2001) Offender Profiling and Crime Analysis. Portland: Willan Publishing. Anselin, L.; Cohen, J.; Cook, D.; Gorr, W.; and Tita, G. (2000) Spatial Analyses of Crime. In Duffee (ed.) Measurement and Analysis of Crime and Justice. Vol. 4, Washington, DC: National Institute of Justice: 213-262. Baker, T. (2005) Introductory Criminal Analysis: Crime Prevention And Intervention. New Jersey: Prentice Hall, Pearson Education, Inc. Buerger, M., Cohn, E., and Petrosino, A. (1995) Defining the “Hot Spots of Crime” Operationalizing Theoretical Concepts for Field Research. In Eck, J.; and Weisburd, D. (ed.), Crime And Place. Crime Prevention Studies No. 4. New York: Criminal Justice Press and Washington, D.C.: 237-257. Boba, R. (2001) Introductory Guide To Crime Analysis And Mapping. Washington, DC: Police Foundation. Boba, R. (2003) Problem Analysis in Policing. Washington, DC: Police Foundation. Bowers, K. and Hirschfield, A. (ed.) (2001) Mapping And Analysing Crime Data: Lessons From Research And Practice. Taylor and Francis. Braga, A. (2002) Problem-Oriented Policing And Crime Prevention. New York: Criminal Justice Press. Brantingham, P.J.; and Brantingham, P.L.(ed.) (1981) Environmental Criminology. Beverly Hills: Sage Publications. Brantingham, P.J.; and Brantingham, P.L. (1984) Patterns in Crime. New York: Macmillan. Brantingham, P.J.; and Brantingham, P.L. (1995) Location Quotients And Crime Hot Spots In The City. In Carolyn Rebecca Block, Margaret Dabdoub and Suzanne Fregly (ed.), Crime Analysis through Computer Mapping. Washington, DC: 116

Police Executive Research Forum. Brito, C.S.; and Allen,T. (ed.) (1999) Problem-Oriented Policing: Crime-Specific Problems, Critical Issues And Making POP Work. Volume 2, Washington, DC: Police Executive Research Forum. Brito, C.S.; and Eugenia, G.E. (ed.) (2000) Problem-Oriented Policing: Crime-Specific Problems, Critical Issues And Making POP Work. Volume 3, Washington, DC: Police Executive Research Forum. Bruce, C. (ed.) (2004) Fundamentals of Crime Analysis. In Exploring Crime Analysis. The International Association of Crime Analysis: 11-36. Canter, P. (2000) Using A Geographic Information System For Tactical Crime Analysis. In Goldsmith, V.; McGuire, P; Mollenkopt, J.; and Ross, T. (ed.), Analyzing Crime Patterns: Frontier Of Practice. California: Sage Publications, Inc.: 3-10. Canter, P. (2001) Using Geographic Information For Problem Solving Research. In Blair, S.; Rachel, B.; Fritz, N.; Helms, D.; and Hick, S. (ed.) (2002), Advanced Crime Mapping Topics. National Law Enforcement and Corrections Technology Center, Rocky Mountain Region, University of Denver: 89-93. Clarke, R.V., and Eck, J. (2003) Become a Problem Solving Crime Analyst In 55 Small Steps. University College London: Jill Dando Institute of Crime Science. Cohen, L.E.; and Felson. M. (1979) Social Change And Crime Rate Trends: A Routine Activity Approach. American Sociological Review 44: 588-605. Cornish, D.; and Clarke. R.V. (1986) The Reasoning Criminal: Rational Choice Perspectives on Offending. New York: Springer-Verlag. Eck, J. (1998) What Do Those Dots Mean? Mapping Theories With Data. In David Weisburd, D.; and McEwen, T. (ed.), Crime Mapping and Crime Prevention, Crime Prevention Studies No.8. New York: Criminal Justice Press: 379-406. Eck, J.; Gersh, J.S.; and Taylor, C. (2000) Finding Crime Hot Spots Through Repeat Address Mapping. In Goldsmith, V.; McGuire, P; Mollenkopt, J.; and Ross, T. 117

(ed.), Analyzing Crime Patterns: Frontier Of Practice. California: Sage Publications, Inc.: 49-63. Eck, J.; and Spelman, W (1987) Problem-Solving: Problem-Oriented Policing in Newport News. Washington, D.C. : National Institute of Justice. Eck, J.; and Weisburd, D. (1995) Crimes Places In Crime Theory. In Eck, J.; and Weisburd, D. (ed.), Crime And Place. Crime Prevention Studies No. 4. New York: Criminal Justice Press and Washington, D.C.: 1-34. Felson, M (1995) Those Who Discourage Crime. In Eck, J.; and Weisburd, D. (ed.), Crime And Place. Crime Prevention Studies No. 4. New York: Criminal Justice Press and Washington, D.C.: 53-66. Goldstein, H. (1990) Problem-Oriented Policing. New York: McGraw-Hill. Gottlieb, S.; Arenberg, S.; and Singh. R. (1994) Crime Analysis: From First Report To Final Arrest. California: Alpha Publishing. Griffin, J. (2001) International Crime Mapping: Caveats And Considerations. In Crime Mapping News, Volume 3, Issue 1, 2001. Washington, DC: Police Foundation: 1-3. Grinder, D. (2000) Implementing Problem-Oriented Policing: A View From The Front Lines. In Brito, C.; and Gratto, E (ed.), Problem-Oriented Policing: Crime-Specific Problems, Critical Issues And Making POP Work. Volume 3, Washington, DC: Police Executive Research Forum: 141-155. Grubesic, T.; and Murray, A. (2001) Detecting Hot Spots Using Cluster Analysis and GIS. Paper presented at the 5th Annual International Crime Mapping Research Conference in Dec 2001, Dallas, Texas. Available at www.ojp.usdoj.gov/cmrc/whatsnew/hotspot/intro.pdf (assessed Jan 2005) Harries, K. (1990) Geographic Factors in Policing. Washington, DC: Police Executive Research Forum. Harries, K. (1999) Mapping Crime: Principle And Practice. Washington, DC: Crime Mapping Research Center. 118

Henry, V. (2003) The CompStat Paradigm: Management accountability in Policing, Business and the Public Sector. New York: Looseleaf Law Publications, Inc. Kamber, T.; Mollenkopf, J.; and Ross, T (2000) Crime, Space, and Place An Analysis of Crime Patterns in Brooklyn. In Goldsmith, V.; McGuire, P; Mollenkopt, J.; and Ross, T. (ed.), Analyzing Crime Patterns: Frontier Of Practice. California: Sage Publications, Inc.: 107-119. Kelling, G. (1999) Broken Windows and Police Discretion. Washington, D.C.: National Institute of Justice, Research Report series (NCJ 178259). Kelling, G. and Sousa, W. Jr. (2001) Do Police Matter? An Analysis of the Impact of New York City’s Police Reforms. New York: Manhattan Institute for Policy Research, Center for Civil Innovation. La Vigne, N.; and Wartell, J. (ed.) (1998) Crime Mapping Case Studies, Volume 1. Washington, DC: Police Executive Research Forum. La Vigne, N.; and Wartell. J. (ed.) (2000) Crime Mapping Case Studies, Volume 2. Washington, DC: Police Executive Research Forum. La Vigne, N. (1999) Computerizing Mapping As A Tool For Problem-Oriented Policing. In Crime Mapping News, Volume 1, Issue 1, 1999. Washington, DC: Police Foundation: 1-4. Mamalian, C.; and La Vigne, N. (1999) The Use of Computerised Crime Mapping by Law Enforcement: Survey Results. Washington, DC: National Institute of Justice, Research Preview. O’Connor T.S.; and Grant, A.C. (ed.) (1998) Problem-Oriented Policing: Crime-Specific Problems, Critical Issues And Making POP Work. Volume 2, Washington, DC: Police Executive Research Forum. Patrick, M. (2002) Proving the SARA Model: A problem solving approach to street crime reduction in the London Borough of Lewisham. London: InfoTech Enterprises Europe.

119

Ratcliffe, J.H. (1999) Terrorist, Patrons And Champions: Implementing Crime Mapping Across the UK. Presented at 3rd Annual International Crime Mapping Research Conference, Orlando, FL. Rich, T. (1995) The Use of Computerised Mapping In Crime Control And Prevention Programs. Washington, DC: National Institute of Justice, Research in Action. Rich, T. (1999) Mapping the Path to Problem Solving. In National Institute of Justice Journal. October 1999. Washington, DC: National Institute of Justice. Shane, J. (2004) CompStat Process. In FBI Law Enforcement Bulletin April 2004 issue: 12-19. Shane, J. (2004) CompStat Implementation. In FBI Law Enforcement Bulletin June 2004 issue: 13-21. Sherman, L. (1995) Hot Spots of Crime and Criminal Careers of Places. In Eck, J.; and Weisburd, D. (ed.), Crime And Place. Crime Prevention Studies No. 4. New York: Criminal Justice Press and Washington, D.C.: 35-52. Spelman, W. (1988) Beyond Bean Counting. National Institute of Justice Grant. Washington, DC: Police Executive Research Forum. Travis, J. (1998) Preface. In La Vigne, N. and Wartell, J. (ed.) In Crime Mapping Case Studies, Volume 2. Washington, DC: Police Executive Research Forum. U.S, Department of Justice (1999) Mapping Out Crime: Providing 21st Century Tools for safety Communities. Available at: http://govinfo.library.unt.edu/npr/library/ papers/bkgrd/crimemap/071299.pdf (Accessed Jan 2005) Vann, I.; and Garson, D. (2003) Crime Mapping: New Tools For Law Enforcement. New York: Peter Lang Publishing, Inc. Vellani,

K.;

and

Nahoun,

J.

(2001)

Applied

Crime

Analysis.

Woburn:

Butterworth-Heinemann. Velasco, M.; and Boba, R. (2000) Manual of Crime Analysis Map Production. Washington, DC: Police Foundation. 120

Velasco, M.; and Boba, R. (2000) Tactical Crime Analysis And Geographic Information Systems: Concepts and Examples. In Crime Mapping News, Volume 2, Issue 2, 2000. Washington, DC: Police Foundation: 1-4. Wartell, J.; and McEwen, T. (2001) Privacy in the Information Age: A Guide for Sharing Crime Maps and Spatial Data. Washington. DC: U.S. Department of Justice Office of Justice Programs. Weisburd, D.; and McEwen, T. (ed.) (1998) Introduction: Crime Mapping And Crime Prevention. In Crime Mapping and Crime Prevention. Crime Prevention Studies No.8. New York: Criminal Justice Press: 1-23. Weisburd, D.; Greenspan, R.; and Mastrofski. S. (1998) COMPSTAT And Organisational Change: A National Assessment. Washington, DC: National Institute of Justice Grant No. 98-IJ-CX-0070. Willis, J.; Mastrofski, S.; and Weisburd, D. (2003) CompStat in Practice: An In-depth Analysi of Three Cities. Washington D.C.: Police Foundation

121