CHAPTER 1: Introduction
1.1 BACKGROUND
Geographic Information Systems (GIS) has evolved as a powerful analytical tool in the last two decades. Due to the slashing down of prices of GIS software, this technology is now easily available to any organization and individual. It is now commonplace for Business, Government and Academics to use GIS for many diverse applications. Some of the applications of GIS are, P Environmental application, Transport application, Municipal application, Engineering application, Forest development and Management application, Military application, Police application, Marketing Management application and Academic Research application. Still various efforts are taken by GIS scientists to find out the application of this potent technology in several areas.
GIS technology is finding many applications within the police agencies across the developed and developing countries. In United Kingdom, United States and other developed countries, GIS is used for policing in a successful manner as it has the capability in managing large amounts of data and performing various analyses on spatial data. The use of GIS to help understand and fight crime is a relatively recent phenomenon and is rapidly growing. This is in part due to the increased availability of data and the increased promotion of crime mapping research and development ( 2000). A good example is the Mapping & Analysis for Public Safety (MAPS) programme, formerly called the Crime Mapping and Research Center (CMRC), which was established in 1997 by the US National Institute of Justice. Its goal is to “promote research, evaluation, development, and dissemination of GIS technology for criminal justice research and practice” (, 2002). The programme provides many services including “grant funding, annual conferences, information on training centers, publications, and research” (, 2002). More and more police departments use GIS to varying extents to assist them in their daily tasks. GIS has a lot to offer law enforcement agencies. Although it will never replace the expertise of police officers, detectives, or investigators, it has significant and largely untapped potential to complement that expertise through its mapping and analytical capabilities. GIS is a very powerful tool that can much more efficiently perform tasks that were traditionally performed by police officers themselves. It can also make possible certain analyses that would not otherwise be feasible to perform.
Since the researcher is a police officer, he is most interested in how GIS can help identify patterns of crime, which can in turn be used to mobilize resources to specific geographic areas to prevent crime from reoccurring. Prevention is a critical part of policing as it results in a more proactive rather than reactive form of policing ( 1992). Unfortunately, resources available for prevention efforts, like most other resources, are often quite limited. It is therefore imperative to know how and where to use these resources most effectively. GIS is a tool that can help recognize areas or neighbourhoods that could benefit from increased prevention efforts, by identifying hot spots of criminal activity for example. Resources can be effectively deployed where they are most likely to yield results, rather than dispersing them over areas of lesser need.
Finding patterns in crime, which can then be used to predict and therefore prevent future incidents of crime, is highly effective in enhancing the ability of the police and crime analysts in simple mapping of the crime to, the apprehension of serial criminal by the usage of advanced criminal geographic targeting models and the association of demographic and socio-economic characteristics which can be correlated with the locations of crimes provide a clearer picture of crime. The feature of GIS that fascinates the police is its ability to swiftly and dynamically represent the crime from a variety of spatial perspectives. This technology supports a broad variety of problem solving and spatial decision-making applications in crime and crime locations. The ability of GIS to query over political boundaries improves utilization of databases and is far more efficient than paper document-based searches or straight database searches. Briefly, GIS lets agencies utilize more information more intelligently (, 1999).
CHAPTER 2: Literature review: Theoretical framework
2.1 THE EVOLUTION OF ENVIRONMENTAL CRIMINOLOGY
Crime has traditionally been associated with the disciplines of criminology, sociology and psychology. More recently the importance of geography and environmental studies in the study of crime has been and continues to be widely recognized ( 1978; ). The following section is meant to provide a general background of the evolution of criminology so that the link with geography and environmental studies can be made clearer.
Criminology was born at the end of the eighteenth century in Europe through the development of the Classical School legal reforms which “codified and systematized criminal law, mandated the creation of police and prisons, and introduced utility theory” a method of assessing criminal justice strategies ( & 1981: ). This assessment or utility testing of criminal justice strategies required statistical information, which opened the doors for the first wave of environmental criminology in the 1800s. This first wave’s findings led to the development of the School of Positivism which eventually replaced the Classical School ( & 1981: ).
Positivist criminology was committed to the gathering of facts about crime and to explaining criminal behaviour through biology and ‘s theories on evolution, leading to the notion of the born criminal ( 1998: ). This differed quite significantly from classical criminology, which revolved around that idea that people have free will and that offenders go through a complex but rational process of decision-making when committing crimes, driven by the maximization of pleasure and minimization of pain ( 1998: ). Ideas from both of these schools of thought have been incorporated in theories still used today.
It is when studying these statistics from the first wave of environmental criminology, in the 1830s, 1840s and 1850s, that the problem of spatial dependence was discovered ( & 1981: ). This realization that crime had a spatial component brought geography into the picture. Studies of spatial variation then started moving from high levels of aggregation, such as the county, province and country level, to lower more local levels of aggregation such as towns. Significant pattern differences were found between various types of crimes (violent versus property crimes). It was found that crime rates varied across space. that the various levels of aggregation showed different patterns, that spatial patterns through time could be used as a prediction tool, that crime data could be compared to other types of data, and that crime was in fact correlated with other problems such as poverty and high population density ( & 1981: ). These latter correlations with population characteristics established the link between crime and environmental studies. The first wave of environmental criminology sparked the association of crime to geography and environmental studies, but much of the work was descriptive and lacked a theoretical framework.
The second wave of environmental criminology was developed in America between 1900 and 1970. This is when many crime theories flourished. It was the Chicago School of Sociology that “developed America’s first large-scale theoretical and research approach to the study of the nature of crime” ( 1978: ). The Chicago School of Sociology’s social ecology concept was the underlying theoretical system to which much of this period was committed. The use of the biological theory of ecology in a human or social context is credited to , back in the early 1920s ( 1978: ).
In a criminology context, social ecology has two major components. The first is a theory of urban form, based on the idea that competition for the scarce spatial resources of a growing city leads to the development of radically structured cities containing well-defined areas into which different social groups will be sorted due to competition for space ( & 1981: ). The second component is a social psychology that predicts the nature and quality of social organization in different areas, and also the consequences of exposure to the varying states of that social organization ( & 1981: ). Both of these human ecology components are still present in theories used today.
and of the Institute of Juvenile Research in Chicago led some of the most important research of the time ( & 1981: ). They used the zonal development model of , the reputed Chicago School sociologist. The model suggested that a city develops outwards from a central business core in concentric circles that are dominated by certain socio-economic groups of the population due to competition for space. The model further hypothesized that crime rates would be highest in the centre of the city and decrease with distance away from it. Results of quantitative studies by and seemed to confirm their hypothesis ( & 1981: ).
Things have changed considerably in environmental criminology since the hiatus period of the 1960s. Many substantial contributions have been made to the field and many more are yet to come as the field is rapidly expanding. “Since the late 1970′s there has been a realization that there is a spatial aspect to crime the study of which can be the legitimate objective of geographers” ( 2001). Whether or not links between crime, geography and environmental studies were made consciously at the time is not as important as the fact that the connections definitely exist today, and that they are responsible for much advancement the field has seen in recent years.
2.2 THE GEOGRAPHY OF CRIME
Many studies recognize the importance of the context in which criminal victimization occurs, whether it is social, geographical or otherwise ( 1988; -2000). Since the realization that crime has a spatial component, geography has been as much a part of studying and understanding crime as have the more traditional disciplines. Social ecologists that studied crime were often criticized for their approach simply being associational descriptive analysis (1978: ). Geographers came to fill the gap by providing an extra dimension: that of spatial analysis. The role of geography in crime studies is now widely recognized.
“Victimization is not a phenomenon that is uniformly distributed; it occurs disproportionately in particular times and places; it occurs disproportionately by offenders with particular demographic characteristics; it occurs disproportionately under certain circumstances (e.g., according to whether or not the person is alone); it occurs disproportionately according to the prior relationship between the potential victim and the potential offender; and so forth” (. 1978: ).
“The objectives of the geography of crime are to describe and map the spatial distribution of crime in greater detail and meaning than has been done before. This field of research attempts to relate the spatial patterns of crime to environmental, social, historical, psychological (cognitive), and economic variables that may explain these patterns. It hopes to develop associational and predictive models that explain crime manifestation in regard to locale. Last but not least, it is hoped that its contribution to the analysis of the dynamics of crime manifestation will help those charged with the responsibility of crime control to assess better the effectiveness of programs they currently use” ( 1978: ).
2.3 THE THEORIES OF CRIMINAL BEHAVIOUR AND CRIME PREVENTION
A myriad of theories exist regarding criminal behaviour. However, none of them alone can be used as a strong underlying framework for this study because of the particularities of the data used and the types of analyses performed. Many of the existing theories however, contribute greatly to the understanding of deviant behaviour, repeat victims and vulnerability, so it is important to describe them briefly. This section outlines some of these theories and highlights how they contribute to the understanding of deviant behaviour, repeat victimization and personal crimes, specifically domestic abuse, as well as how they can be used to develop effective prevention programmes.
2.3.1 Routine activity theory
Routine Activity Theory (RAT), developed by criminologist in the late 1970s, is one of the most commonly cited theories in the crime literature. It suggests that for crime to occur three things must converge in time and space: a willing offender, a suitable target, and the lack of a capable guardian (average citizens rather than police or security officers) that could interfere in the commission of a crime against the target. Similarly, & (1981) identify four components or dimensions of crime: the law, the offender, the target and the place. Environmental criminology is the study of the fourth dimension: place. For a crime to occur, the other three dimensions must meet at a place in time and space: an offender must find a suitable target and break the law. It seems each of these two views lack one component that is present in the other, so the RAT components and the ‘ four dimensions of crime are combined here to make a comprehensive diagram of the necessary components of crime (see Figure 1).
These components provide guidance in understanding and studying deviant behaviour. Studies have often separated the study of offenders and victims. Traditionally, offenders received greater attention, but in the past decades the focus has shifted significantly to the victims. This study focuses primarily on the victim but could provide some information regarding the offenders since domestic abuse often occurs within the home, in relationships that often involve cohabitation between the offender and the victim. Therefore studying the victims could tell us something about the offenders as well. This study not only focuses on the target component (victim), but also focuses on the place (spatial) component as well as the law component (crimes studied are all violations of the Criminal Code of Canada).
Figure 1: The dimensions of crime
Source: & 1981 and in ‘s Theory
As stated by (1992), the routine activity approach can be used as a way to think about crime and how it can be prevented. It looks at the flow of offenders, targets and capable guardians in space. It offers a different way of preventing crime by working on the flows of the three elements necessary for crime to occur. The lack of a capable guardian, an offender and a target must converge in time and space in order for a crime to occur. If any of these elements is kept from converging with the other two, then the crime will be prevented.
suggests keeping the flow of likely offenders away from the flow of potential targets and keeping the flow of capable guardians close to potential targets. Micro-environmental design changes, awareness, and community involvement are all ways to affect these flows to increase prevention of crime ( 1992: ). Respective examples of these include having a receptionist at the front of an office suite to increase protection to internal offices by decreasing the flow of offenders in that area, making people aware that a particular street is prone to muggings after dark to discourage potential targets from taking that route, and keeping the community involved in neighbourhood watch programmes to maximize the presence of capable guardians.
Routine activities theory also identifies four elements that make a target suitable to an offender. These are represented by the acronym VIVA: value inertia, visibility, and access (, 2007). Value refers to the benefits that target can bring to the offender: money or the satisfaction of committing the came, for example. Inertia refers to the size or weight of an item. This is most relevant in robberies and thefts where small portable objects are most desired. Visibility is how obvious the target is to the offender, whether an object or person is seen as a potential target in the offender’s mind. Access is how easily the offender can get to the target.
2.3.2 Lifestyle theory
. (1978) developed what is now known as lifestyle theory, as a theory of personal victimization. The model is based on the idea that the likelihood of someone suffering personal victimization is highly linked to the person’s lifestyle. . explain that people adapt to role expectations, which are closely related to gender, age, marital status and culture, as well as structural constraints (economic, familial, educational, and legal) that are imposed on them. Individuals adapt to structural constraints and role expectations by developing attitudes and beliefs that become embedded in their daily activities resulting in regularities in behavioural patterns (. 1978: ). It is these daily routines that . refer to as lifestyle: vocational and leisure activities for example.
They argue that differences in lifestyle result in differences in “exposure to situations that have a high victimization risk” (vulnerability). For example, married women are more likely to spend more time at home taking care of the house and the children than a single male. Also, people with higher income have more flexibility to go out and enjoy more leisure activities outside the home, than those with a limited income. Lifestyle theory reinforces the idea that victimisation is not uniformly distributed. “Because different lifestyles imply different probabilities that individuals will be in particular places, at particular times, under particular circumstances, interacting with particular kinds of persons, lifestyle affects the probability of victimization” (. 1978: ). Closely related to lifestyle theory is the theory of demographic composition, which “emphasizes social class, ethnicity, and stage in the life cycle as the main determinants of crime” ( 2000: ).
2.3.3 Crime prevention
Throughout human history, there have been many theories attempting to explain why crimes occur, as some of them have been stated previously. This reflects the human desire to understand the forces behind the commission of crime. Based on implicit and explicit assumptions that the causation or causations of crime need to be understood and explained to deal with them, it was hoped that such theories would guide us regarding how the occurrence of criminal events can be best addressed (, 2002). One way to interpret the meaning of best is to reduce or prevent crime by attacking the causes or factors facilitating crime. Such concern for reducing crime has always been accompanied by concerns for public safety in human history (, 1997). Recently, such concern has been termed crime prevention which came to denote “a set of ideas for combating crime” (, 1997: ).
With the growth of the scientific study of crime, the criminal justice system has incorporated more prevention oriented responses to crime rather than employing simple responses such as retribution (, 1997). One of the crime prevention techniques, secondary crime prevention, focuses on individuals and places with a high probability of deviance (). In secondary crime prevention, the already existing factors fostering deviant behaviour are the focus of interest. Therefore, an accurate identification and prediction of future criminal offences, whether based on individuals or places, is pertinent in secondary crime prevention techniques (). Yet when identifying individuals at risk of offending and when predicting their future possible offending behaviours there should be concern for the level of accuracy and relevant moral issues. Many may feel uncomfortable with the idea of predicting certain individuals as criminals when those particular individuals have not committed any criminal offences yet; and when many of them may never commit a crime. And it may not be possible to predict with high accuracy whether certain individuals may or may not take up a certain course of action due to the complex nature of the human world. Unlike making predictions about individuals, making predictions on places does not usually bring out a moral dilemma or self-fulfilling prophecy. Fortunately, places and buildings do not have emotions. Therefore, when predicting future criminal offences, shifting the focus from individuals to places would resolve the moral issues related to identification and prediction of future offences based on individuals.
The higher concentration of crime amongst places than amongst repeat or habitual offenders is another reason that the prediction of future crime should be based on criminal places rather than criminal persons. For example, (1995) compares the result of his 1986 calls to the police in Minneapolis with the highly regarded research by on a 1945 Philadelphia cohort. found that 18 per cent of the individuals were responsible for over 50 per cent of the total arrests while found that three per cent of places produced 50 per cent of calls to the police.
Time and again, empirical research has echoed the same conclusion; there is a high concentration of victimization in a few places, and this clustering exists especially in crime hot spots ( & , 2001). Depending on a level of interest, the clustering of crime can be distinguished in terms of repeat victims or crime hot spots (, 1998). A crime hot spot is defined as a small geographical area where crime is highly predictable for at least one year ( & , 2001, , 2000).
In fact, the observation of clustering in criminal events and criminal residences is not new. For over a century, the close proximity of criminals’ residences and criminal occurrence has been observed and such clustering has remained relatively persistent till the present day ( & , 1995; & , 1991; & , 1991 & 1984). European and North American cities in the 19th century demonstrated that clusters of poor neighbourhoods and criminal opportunities were closely located to each other.
Considering the identifiable socio-demographic characteristics in age, sex, economic status, and ethnicity of offenders, it is not surprising to find that criminal residences are frequently clustered together in space ( & , 1991). In fact, both criminal behaviour and victimization “are strongly associated with lower socio-economic status” ( & , 1993: ) and there are large differences in likelihood of becoming a victim of crime (i.e. robbery) across different social groups ( & , 2001).
It has been in only the last two to three decades that the implications and importance of repeat victimization on small numbers of targets, causing substantial differences in crime rates, have been recognized in criminological theory and public policy research areas (, 2001; , 2001; & , 2001; & , 2001; & , 1999; , 1997; ., 1995; & , 1993). Even in crime hot spots, only about 20 per cent of people suffer personal crime, and about 30 per cent suffer property or vehicle crime ( & , 1999). In the commercial sector, the concentration of victimization is even higher than those of persons and households. Depending on the business type, three to eight per cent of premises suffer 59 to 63 per cent of crime. In Scotland, about 10 per cent of premises suffer 66 per cent of crime, and 5 per cent suffer 54 per cent of crime (., 1999). Thus, the high crime rates in crime hot spots are due to the higher repeat victimization rates and not due to the widespread victimizations in the area ( & , 2001; , 1998; , 1997; ., 1995). In other words, crime hot spots are areas where repeat victimization rates are the highest. This means that the prevalence the chance of becoming a victim is lower in crime hot spots than if crime were to be randomly distributed (., 2002).
Crime is not only spatially concentrated but also temporally concentrated as well depending on seasons, weekdays, and times of day. From Report data analysis, & (2001) found that robberies are less likely to occur in the spring. The majority of street muggings happen during daytime hours, and the most popular days for robbery are Thursdays and Fridays (, 2002; ., 1987).
Backed by the empirical research findings in the 1980s that crime is disproportionately concentrated in a few locations, interest in repeat victimization and hot spots in policing and controlling crime has experienced a renewed growth ( & , 2001). This has also brought a shift of primary focus in crime prevention from larger units such as community and neighbourhood to smaller units such as specific places, that is, crime hot spots ( & , 2003; ., 2002;, 2001; & , 2001; , 1997; ., 1995; & , 1995).
In thinking about prevention and detection of crime, there are at least three reasons why accurate and prompt identification of repeat victims and crime hot spots should be performed, and that information be utilized in crime control strategies and policy. Firstly, repeat victims are clustered in places, and this clustering is relatively stable over time (, 1997). If a spatial concentration of crime is only temporary or sporadic, or if an area appears to be a high crime area due to the random nature of crime since crime is not usually committed with fixed schedules, then location-based strategies are likely to fail in curbing crime rates (., 2000; , 1995). Fortunately for criminologists and those who are interested in preventing crime based on spatial patterns, analyses show that concentrations of crime remain fairly stable over time. Thus, for successful crime prevention, identifying hot spots and the factors promoting such persistent differences in crime rates among locations will be important.
Secondly, research has shown that revictimization tends to occur quickly; from 40 per cent to 80 per cent taking place within one month to eight weeks following the previous victimization ( & , 2001; ., 1999). Research in various countries has repeatedly shown that as the number of prior victimizations increases, the probability of further victimization of both personal and property crimes increases as well. ( & , 2003; , 2001; & , 1999; , 1998; , 1997; ., 1995). Therefore, an incident of victimization has predictive power regarding both where and when future crimes will likely occur at the level of individuals and areas (., 2002; & , 2001; , 1998). This means that crime prevention and control resources can be both spatially and temporally deployed depending on future victimization risks ( & , 2003; , 2001; & , 1999; , 1998; ., 1997; ., 1995).
Thirdly, since most offenders commit offences closer to their homes, identifying repeat victims will automatically bring the offenders to the attention of the police (, 2001; & , 2001). Research has shown time after time that only a small proportion of individuals continue to offend, becoming habitual offenders ( & , 1993). These habitual offenders are responsible for a disproportionate number of the most serious crimes (, 2002). (1997) estimates that about 10 per cent of offenders are responsible for more than 50 per cent of crimes. Self-report on robbery by inmates in California, Michigan, and Texas shows that while 50 per cent of inmates committed less than four robberies per year, about 10 per cent of inmates reported committing more than 70 robberies per year ( & , 1987). It is also hypothesized that the high crime area over an extended time period may correspond to the small percentage of habitual offenders committing a large number of criminal offences (., 2000).
Recently, an effort to find a link between habitual offenders and repeat victimizations has begun ( & , 2003; ., 1995). If most prolific offenders are largely responsible for serious crime and repeat victimization, arresting the prolific offenders will have more impact in decreasing crime rates. Clearly, it is only logical to apply more resources where many crimes occur rather than on places where fewer crimes occur ( , 2001).
CHAPTER 3: Literature review: Crime and Analytical framework
3.1 A REVIEW OF LEEDS CRIME RATE
The population of Leeds, as measured in the 2001 Census, was 715,402 of which 48% were male and 52% were female. The population of Leeds accounted for one third of the residents of the county of West Yorkshire. In 2003/04, 39.1% of all crime committed in West Yorkshire took place in Leeds with the rate of 177.95 crimes per 1,000 population. This was higher than the average crime rate in West Yorkshire, which was at 155.8 per 1,000 populations and significantly in excess of the national picture at 114.04 per 1,000 populations (Table 5.1). However, the number of crimes and the crime rate in Leeds dropped in 2003/04. There were 127,304 offences recorded in Leeds in 2003/04, which represents a 1.5% decrease in crimes over the previous year. However there has been a 20.32% increase overall since 2000/01.
Table 1: The rate of all crime per 1,000 populations in Leeds compared to West Yorkshire and England and Wales
Year
Leeds
West Yorkshire
England &Wales
2003/2004
177 95
155 80
114 04
2002/2003
180.67
154.45
113.36
2001/2002
168.61
143.73
106.20
2000/2001
147.89
124.29
102.28
Source:
3.2 CRIME ANALYSIS AND BENEFITS OF UTILIZING GEOGRAPHIC INFORMATION SYSTEMS
As known that 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 the operational and administrative personnel in planning the deployment of resources for the prevention and suppression of criminal activities, aiding the investigative process, and increasing apprehensions and the clearance of cases (, 2005). It supports a number of department functions including patrol deployment, special operations, and tactical units, investigations, planning and research, crime prevention, and administrative services.
A Geographic Information System (GIS), accordingto , 2003), is “an interactive mapping system that permits information layering to produce detailed descriptions of conditions and analyses of relationships among variables. A GIS is based on drawing different spatial distributions of data and overlaying them on one another to find interrelated points. Conditions, or filters, can be used in a GIS to refine searches at any level an analyst chooses”. It can be used for analyzing physical space, assigning perspective, and producing visual images of different styles of data in map layouts. Data displayed in the form of a map facilitates understanding of the significance of where, when, and by whom crimes are committed (, 2003). It also has the capability to change, visualize, query, and analyze tabular and geographic (spatial) data.
In the 1990s, there were improvements and growth achieved in spatial pattern analyses utilizing Geographic Information Systems (GIS) that allow flexible mapping and spatial analyses with various analytical tools ( ., 2000; , 1999). With the improvements, GIS and their techniques became critical in measuring and representing the geographic properties and spatial relationships of data (., 2000). Considering the many possible benefits of utilizing GIS, it is not surprising to see an increased appreciation of the value and potential utility of GIS in spatial data analyses, and many government agencies, especially the police in Britain and the United States, investing in GIS for more effective policing ( & , 2003; , 1999).
GIS offers at least four benefits over other traditional methods such as listing addresses of criminal events. First, once the spatial and temporal aspects of crime are mapped out, GIS can be employed as an improved method in identifying repeat victims, crime hot spots, and other spatial patterns that might be inconceivable or confusing from a mere list of crimes by bringing diverse pieces of information together ( & , 2002; & , 2001; , 1999; ., 1990). With hundreds or even thousands of street names in any given city, discerning patterns from a list of addresses may be confusing if not impossible (, 1999). Utilizing GIS, crime patterns such as hot or cold spots can be indicated by clusters or lack of clusters of point data, density maps, and various statistical tests ( & , 2003).
Second, once crime is mapped out, the user has flexibility in selecting what kind of crime will be displayed on a map in what manner incorporating various analytical tools (, 1999; ., 1990). Additional information about crime can be easily added to crime data analyses in GIS. For example, not only the times, locations, and types of offences but also demographics of the areas and offenders can be incorporated in data analyses. With prompt and accurate crime pattern analyses, crime forecasting and prevention can be achieved as well; police administrators and officials can take one step closer to a proactive management and deployment of resources.
Third, the relationship of geography such as a street network and crime on a map can be readily identified using maps rather than from the names of locations (., 1990). For example, any possible impacts of bus routes, rapid transit stations, main traffic arteries and other establishments on crime are much easier to see on a map.
Fourthly, GIS allows flexible spatial aggregation at various levels such as census geographies such as Districts, Wards and Output Areas. This can lead to facilitating place-based spatial patterns analyses of crime ( ., 2000). Once crime is mapped out, using fixed boundaries, the crime counts can be standardized with denominators such as population or dwelling units. Such aggregation and standardization would allow comparison with neighbouring areas and would be useful in trend analysis as well ( & , 2003).
The disproportionate clustering of crime on a few victims within a few geographically small hot spots means that a substantial reduction in crime rates can be achieved by preventing future repetition of crime in those small areas ( & , 2002; & , 2001). By simply focusing on the unique characteristics of crime hot spots, it has been proposed that perhaps 50 per cent of calls to the police from the most dangerous locations can be reduced (, 1995).
According to & (2001; ) in England and Wales, “repeat victimization has entered the mainstream of crime prevention discourse.” Combined with a desire to make policing and crime control more efficient, the importance of preventing repeat victimization has at last been recognized ( & , 2001). Crime reduction research and programs that aimed at thwarting repeat victimizations not only have shown a positive impact in decreasing all types of crime but also have been shown to give reassurance to victims that police are determined to stop further victimization (, 2001). Strategies that concentrate on repeat victims fuse crime prevention, detection, and victim support together, thereby improving the quality of life in the area in general ( & , 2003; & , 1999; , 1998). Therefore, it is of policy interest, in curbing crime, to examine and identify factors that will aid in predicting repeat victimization (., 2002).
The present research is structured in a way to demonstrate how GIS is able to discover crime patterns, recognize extraordinary concentrations of crimes or hot spots that show clusters of criminal events, or characterize criminal behaviour in the different neighborhoods of concern. This will be performed by visualizing and analyzing the spatial and temporal patterns of crime in Leeds City.
3.3 SPATIAL AND TEMPROAL CONCENTRATION OF CRIME
As stated previously, many theories have been presented in attempts to define and explain criminal activity. Some of these have focused on individual criminals while others examine the aggregate crime within an area. Although the public perception may be that crime is randomly distributed in space, extensive evidence now exists that it is not.
, & (1989) detailed the many early problems associated with the analysis and identification of high crime areas or hot spots. An inability to pinpoint the location and/or the time of some crimes added error into the analysis. Also, until recently, reliable, complete data was not generally available.
As ever smaller police departments make the transition to computerized systems, both the quantity and quality of crime data available for research has seen tremendous improvement. However, the overwhelming quantity of data and the difficulty associated with processing such multidimensional data presents new difficulties (, & , 2003). Exponential growth in the power of computers and the development of sophisticated GIS systems has advanced this research tremendously and it is now possible to process large amounts of spatial data and determine trends where they may not have been previously visible.
Several techniques have been used to analyze crime data. One limitation that has arisen when techniques are applied to administrative areal units such as census tracts or police beats is called the Modifiable Areal Unit Problem (MAUP). This problem occurs when different results are produced depending on the selection of areas within which the data is aggregated ( & , 2003). This issue is of particular importance in the study of crime. Since crime does not necessarily occur in any particular relation to the land area or the residential population, analyzing a simple spatial concentration of crime can be more valuable (, 2004). Studies have produced several methods of spatially identifying hot spots. Among these methods are the Spatial and Temporal Analysis of Crime (STAC), which produces ellipses which cluster crime points (, 1995), and methods using kernel density estimation (, 1998).
, , & (2001) note that the most common feature of all categories of crime reported was the spatial concentration into a relatively small number of hot spots allowing the anticipation of high crime in these areas in the future. It was also noted that these concentrations peaked at fairly specific times on a regular basis.
According to & (2003), the prevalent forecasting method is currently primarily spatial in nature, with the assumption that clusters of crime will persist in the near future. Researchers have noted that a system capable of utilizing existing information in real-time and able to predict where and when high rates of crime are expected would be of great value for police resource allocation (, , & , 2003). Also, a greater understanding of the patterns of crime can result in a better grasp on the causes, and lead to improvements in urban planning and design to reduce susceptibility to crime (., 2001).
One problem encountered when attempting to analyze the temporal qualities of the crime data involves the lack of detail in many police crime databases. In crimes such as burglary or theft, it can be impossible to pinpoint the actual time of occurrence and the possible window of opportunity may cover many hours. The Aoristic Temporal Analysis
Method, a temporal weighting method, was developed to attempt to deal with this problem (, 2002). (2004) also proposed several general categories of temporal clustering based on prevalent patterns.
A crime data analysis model utilizing information on spatial location as well as time of day would have the advantage of being able to pinpoint more accurately when additional policing may be needed in an area. Also, if certain areas are noted as problematic primarily during holidays or specific weekdays, this information can then be used to make better predictions and to improve the allocation of resources. When utilized with real-time data, such a model could possibly help to identify crime trends as they develop in unexpected areas. Research continues on the integration of temporal components with spatial data.
(2001) notes that extending the relational model to include data in the temporal dimension can be effective when the temporal data is linear. However, the common use of temporal “snapshots” leads to inefficiency and inflexibility. Future work will continue to develop the theoretical framework and techniques to deal with both linear and nonlinear temporal data. . (2003) have demonstrated the use of artificial neural networks to facilitate predictive modeling. Any effective approach to modeling the temporal dimension must incorporate both absolute and relative views of space-time to adequately represent temporal dynamics in geographic information systems (,
1994).
3.4 SOCIO-ECONOMIC AND DEMOGRAPHIC CHARACTERESTICS AND CRIME
Of the theories examined in the literature review, the study of socio-economic and demographic characteristics and crime has the longest history. This practice was started back in the 19th century with governmental administrators examining variations of crime and various socio-economic and demographic characteristics within cities. It was believed that differences in these conditions could explain crime.
Much of the research on social dimensions and crime has been based on social ecology. With regard to crime, it is believed that there is a spatial distribution of crime and offenders, and this distribution is not uniform. Due to spatial concentrations of crime in some areas and the absence of crime in others, it is believed that the characteristics of the general demographic, social and economic dimensions are very important in the distribution of crime and offenders (and , 1971).
The review of literature suggests that the social disorganization theory is the best framework to use for examining the social dimensions related to crime. This section will discuss and review studies that have used some variables to measure the relationships of the socio-economic and demographic characteristics to crime.
The economic status of neighborhoods influences the level of social disorganization. (1980) found that robbery varied based on variations in the economic status of suburban areas. found that areas with low socioeconomic status and high social problems contained robbery rates significantly higher than the mean robbery rate. Robbery was significantly lower in areas classified as upper and upper-middle class and in rural areas. Much of the research looking at the urban underclass has found that the greater the levels of economic disadvantage, the higher amount of social problems (, 1987).
Numerous studies have established a link between crime and disadvantage. It has been found that there is a correlation between poverty and crime (, 1979; , 1980). The strength of this correlation is dependent on the type of crime. It has been shown that there is a stronger relationship with violent crime and disadvantage. (1987) noted that the highest rates of violent crime occur in the neighborhoods of Chicago where the underclass resides and where public housing developments have a concentration of both underclass and violent crime. The following provides a brief review of studies that have examined the correlation between disadvantage and crime.
Socio-economic inequality may have a stronger effect on crime rates than poverty alone. and (1982) examined the effects of socioeconomic inequality, racial composition, and poverty on violent crime. They hypothesized that variation in crime rates for urban violence resulted from variation in inequality of socioeconomic characteristics. They found that socioeconomic inequality correlated with rates of violent crime. When inequality was controlled for, racial composition had a much smaller effect on violent crime rate and poverty did not have an effect.
Disadvantage may have a stronger effect on crime in urban areas. (1986) looked at crime victim survey data to examine the differences between urban, suburban, and rural victimization, and the structural determinate each place had on victimization rates. He found that urbanization and housing density had positive effects on victimization and that poverty increases the risk in urban areas alone. He notes that studies have shown that urban areas are more heterogeneous with regard to social class as opposed to suburban and rural areas, which are more homogeneous. Sampson’s study found that “regardless of age, racial composition, poverty, and density, residents of central cities had a risk of theft and violent victimization considerably higher than their suburban and rural counterparts.”
Some studies have shown that poverty has the same influence on crime regardless of the population composition. and (1996) examined the hypotheses put forth by (1987) that asserted that extremely disadvantaged neighborhoods had high rates of crime and structural disadvantage influences both black and white neighborhoods equally. They noted that most studies looking at disadvantage did not look to see if crime rates are `unusually high’ in areas with high disadvantage. They proposed that neighborhoods with extremely high levels of disadvantage will have considerably higher rates of crime, and that this should apply to both black and white neighborhoods. The study looked at Columbus, Ohio and found that communities with extreme disadvantage had higher levels of crime rates then those with moderate to low disadvantage. With regard to structural differences between black and white neighborhoods with extreme disadvantage, white neighborhoods had higher than normal rates of crime similar to black neighborhoods. The study also found that property crime was lower and violent crime was higher in black disadvantaged neighborhoods compared to white counterparts. and noted limitations with the study that included the exclusion of public housing, level of middle class families, proximity to other disadvantaged areas, as well as crime reporting and police deployment differences.
Disadvantaged areas may also produce more victims. and (1999) explored the relationship between crime and the distribution of disadvantaged, middle income and affluent residential neighborhoods in North-West England. They found that demand for police services was higher in and around disadvantaged neighborhoods. They also examined the victims and found that victims living in disadvantaged areas are more likely to be a victim closer to home, and many times in their own neighborhood, than victims who live in affluent areas. and (1997) also noted that many robbers preferred to select victims from poor areas because of the greater likelihood of the victim carrying cash.
The last variable included in the social dimension is unemployment. There are two ways to look at this variable, by level of unemployment and by the types of positions held in the neighborhood. These variables have been affected by advancements in technology, which has led to a displacement of jobs and requires increases in skills in many sectors (, 1987). In addition, foreign competition with sectors of industry and manufacturing has led to decline in employment (., 1987), especially in the northern and northeastern states, where the study areas for this research are located. (1987) noted that changes in the economy, such as shifting from production of goods to service producing, moving industries from city centers, partition of high and low wages, weakness in the economy, and technical innovations, all have a strong effect on the disadvantaged areas and can lead to long-term unemployment. (1987) also noted that age is a factor in crime, and teenagers who are not in school and are jobless affect social organization.
There are underlying dimensions that also affect employment for those in disadvantaged areas. . (2000) found that, in 1990, the occupational segregation and inequality between whites and blacks was significant, and that the rate varied throughout the cities. They also established that blacks were employed in significantly lower status and lower income jobs. The study also revealed that inequality rose with a greater proportion of blacks in the population. They found that the larger the population, the greater the number of occupations that employed blacks. However, these were lower status and lower skilled positions.
The (1980) looked at the relationship of crime with employment and different occupation levels. They felt that different `economic specializations’ and `occupational distributions’ would create different opportunities and motivations in different cities. They found that the robbery rate rose with the population size, and that the rate related to black unemployment. Another study, a factor analysis conducted by (1974), found that unarmed robbery was correlated with the percentage of unemployed men, and the percentage of women employed in blue collar positions.
The overall employment structure of the community may affect other crimes that in turn affect the crime rate. In ‘ (2000) study of active robbers of drug dealers in St. Louis, he noted that, as a rustbelt city, urban decay has entrenched itself in the inner city and has allowed drug markets to thrive. This in turn produced much opportunity for robbers looking for victims who have cash and are unlikely to report the robbery to the police.
CHAPTER 4: Data and Methods
4.1 SOURCE AND DESCRIPTION OF DATA
To achieve the objectives of this study, three types of data were used. The first one is the data related to geography of Leeds. The second is all crime datasets. The third type is census data. Next, their sources will be defined and the three data sets will be described as well.
4.1.1 Spatial (Geographical) Data
All maps used in this study are based on data provided by the United Kingdom Boundary Outline and Reference Database for Education and Research Study (UKBORDERS) via Edinburgh University Data Library (EDINA) with the support of the Economic and Social Research Council (ESRC) and the Joint Information Systems Committee (JISC) and boundary material which is copyright of the Crown, Post Office and the EDLINE consortium.
UKBORDERS provides digitised boundary datasets of the UK, available in many Geographic Information System (GIS) formats (MapInfo MIF/MID, ArcView Shape, Arc/Info Export and several others), for teachers and researchers in the UK Higher and Further Education community to download and use in their work. The main group of boundaries available corresponds to the various levels of 2001 and 1991 Census geography which are designed to be used for spatial visualisation and analysis of Census statistics (, 2007). However, for seek of the purpose of this study, the census boundaries have been chosen, Wards boundaries and Output Areas boundaries of Leeds.
4.1.2 Crime Data
Data of Leeds crime have been collected by Leeds Police Service and provided by Dr and Dr of the University of Leeds. Detailed information on crimes committed, their geo-referencing, victims and offenders is available. Shut information on crime as reported to the police is the principal database enabling us to study the geography and determinants of crime in Leeds. Four fiscal years of recorded crime datasets were obtained Leeds Police Service. Each period of the following represents one fiscal year:
April 2000- March 2001
April 2001- March 2002
April 2002- March 2003
April 2003- March 2004
Crimes have been classified by Leeds Police Service into 14 classes. As can be seen from figure 2, each column represents one type of crime and the number of cases occurred of this class can be recognized as well. Each record identified a single report of criminal activity that was filed. The data fields used to identify the hot spots spatially were the starting date of the event, geo-reference of the location of the event (Easting and Northing) and the offence type, which identified the type of crime that was committed.
Figure 2: the graph shows 14 types of crime classified by Leeds PoliceService and shows the counts of each type.
Problems related to the quality of recorded crime data, from a geographical perspective are widely known al., 2004). For example, some data have no geo-referencing and others contain geo-referencing errors. Therefore, it is important to examine the quality of crime data before using them ( and, 2001). However, these datasets were cleaned to produce geographically more reliable data. This was done by Leeds Police Service. This has improved the spatial accuracy and reliability of the crime data quite considerably. Table 2 shows the variables in the recorded crime datasets.
Table 2: The description of variables used in recorded crime datasets
Variables
Description
CRIMENUMBE
Crime number
DATEENTERE
Date entered
HOCLASS
Sub-group of offence type
OFFENCE
Offence types
STATUS
Status (detected/undetected)
CRIMETYPE
Crime type
DATEFROM
Date from
TIMEFROMH
Time from (hour)
TIMEFROMM
Time from (minute)
DATETO
Date to
TIMETOH
Time to (hour)
TIMETOM
Time to (minute)
DIVISION
Police division
BEAT
Police beat
FEATURE
Detailed on where the crime occurs (ex. Roadside, garage etc.)
HOUSENUMBE
House number
STREETNAME
Street name
AREA
Area
TOWNCITY
Town city
POSTCODE
Postcode
OSREFERENC
OS reference
EASTING
Easting
NORTHING
Northing
IMPROVED
Improved
VICTIMAGE
Victim age
VICTIMGEND
Victim gender
VICTIMETHN
Victim ethnicity
NOMINALNUM
Nominal number
OFFENDERAG
Offender age
OFFENDERGE
Offender gender
OFFENDERET
Offender ethnicity
OFFVICRELA
Offender and victim relationship
POLICESTN
Crime committed in the police station
Source:
4.1.3 2001 Census Date
The 2001 Census statistics used in this dissertation are Crown Copyright and are produced by the Office for National Statistics (ONS). The statistics are licensed for academic use by the ESRC/JISC Census Programme, which funded access to the data for researchers in UK, free at the point of use. The ESRC/JISC Census Programme funds the Data Support Units which provide access to UK Census Data. The 2001 Census Area Statistics are provided by the Census Dissemination Unit (CDU) through the Manchester Information and Associated Services (MIMAS) of Manchester Computing, University of Manchester through an interface called CASWEB.
The census is a survey of the whole UK population. It has been carried out every ten years since 1801. The latest census was held on 29th April, 2001. The data in the census describes the characteristics of the population of the UK including demography, households, families, housing, ethnicity, birthplace, migration, illness, economic status, occupation, industry, workplace, transport mode to work, cars, and language (., 2002). The aggregate outputs are counts of people or households broken down by demographic and socio-economic characteristics. These are contained in a series of tables on a specific topic or area of interest. The 2001 Census aggregate statistics datasets include, Key Statistics, Standard Tables, Standard Table Theme Tables, Census Area Statistics, Census Area Statistics Theme Tables, Census Area Univariate Tables and Armed Forces Tables.
The main dataset used in this study is the Census Area Statistics (CAS), which is equivalent to the Small Area Statistics (SAS) of the 1971, 1981, and 1991 Censuses. It is available for geographical levels down to output area (OA), the smallest unit of the 2001 Census geography. Each output area contains approximately 290 persons or 125 households. This is different from the 1991 Census when the smallest areas were Enumeration Districts (EDs) and electoral wards with an average size of about 180 and 2,000 households respectively ( and , 2002).
4.2 DESCRIPTION OF METHODS USED
4.2.1 Preparation of Recorded Crime Datasets
This data set was provided on shapefile format and covered the period of four fiscal years, April 2000- March 2004. Because it has geo-references for all crime locations as stated in Chapter 1. To display such format, ArcMap has been used that is one part of ArcGIS package version 9.1. For this research, the data has been prepared. First of all, the original data set have been divided into four datasets which each set stands for all crimes that occurred during one fiscal year, using selection by attribute tool which is available in ArcMap. Subsequently, four types of crime were selected for analysis where they have been derived by using the same approach as well. The four crimes selected are Burglary Dwelling, Criminal Damage, Theft from Motor Vehicle and Theft of Motor Vehicle. The reason of this selection is that these crimes are related to human beings possessions and a huge amount of them was committed as shown above in figure 1.
4.2.2 Identification of Crime Concentrations
Crime analysts are making greater use of GIS to analyze and display geographic concentrations or hot spots of crime events. One of the techniques for performing this analysis is Kernel Density Estimation or Kernel Smoothing, a spatial statistical method that generates a map of density values from the point event data. A critical issue in the smoothing process is the selection of a bandwidth size or the radius of the circular window in which smoothing is performed. Most GIS and spatial statistical programs that perform kernel smoothing calculate the bandwidth based on the geographic extent of the point pattern. These estimates for bandwidth do not reflect the geographic distributions of the points within the study area, only their geographic extent. This can result in misleading density values and maps that are either too smooth or too spiky in appearance ( and , 1995).
Many of techniques used to detect clusters apply the base idea of calculating the number of cases within a circle and then testing the count that results for statistical significance using a certain statistical test. They offer differ in detail and in the reliance they place on the circular distribution (, 2006). As this technique was used to identify the crime clusters, it was as well used to analyze events within specific spatial hot spots to identify peak times of criminal activity. Identifying the temporal distribution of crimes required the determination of the time the event occurred. In some crimes, such as violent crime, this is fairly straightforward, since most victims know with certainty when it occurred. But in other crimes, usually those without witnesses, it may be only possible to place the event within a span of time which can cover several hours, a day, or even longer. The event start date and time and the event end date and time were used to determine the span of time involved in hours
As mentioned previously, the method which has been used for detecting crime clusters and hot spots in study area (Leeds). It has been selected due to its availability where it is one of spatial analysis tools existing in ArcMap. Also, it has selected as and (1996) state the kernel estimation approach was developed specifically to address the point analysis hotspot problem in epidemiology. The bandwidth has been set 150 which is almost similar to default value (144.44) since and (1995) rule of thumb says the ArcGIS default is useful when the study area is large.
4.2.3 Identification the relationship between crime and certain demographic and socio-economic characteristics
The ‘s correlation coefficient is used in this study to investigate the possible relationships between demographic and socio-economic characteristics and crime. The unit of analysis is the Output Area. The correlation coefficient always takes a value between -1 and 1 indicating the strength and direction of a linear relationship between two variables. A value of 1 or -1 indicates perfect correlation. A positive correlation indicates a positive relationship between the variables (increasing values in one variable correspond to increasing values in the other variable), while a negative correlation indicates a negative relationship between variables (increasing values in one variable correspond to decreasing values in the other variable). A correlation value close to 0 indicates no relationship between the variables: thus the higher the absolute correlation coefficient the better in terms of finding relationships. The Statistical Package for the Social Sciences (SPSS) was used for the correlation analysis.
The analysis started with ‘s correlation coefficient calculated for (overall) crime in Leeds for each crime type. Note that numbers of crimes used in this section are from the period 2001/02 which is the same period as the 2001 Census.
4.2.4 Identification the best multiple regression model
A multiple regression was also employed to explore the relationship between crime and its related determinants. It can be used to describe the relationship between multiple variables precisely by means of an equation that has predictive value. The multiple regression modelling in this study was carried out using ‘stepwise variable selection’ which is a method of choosing the best predictors of a particular dependent variable on the basis of statistical criteria. Fundamentally, the statistical procedure decides which independent variable is the best predictor, the second best predictor, etc. It is a combination of forward and backward procedures. When each variable is added, variables which are entered in earlier steps are rechecked to see if they are still significant. If not, they will be removed.
As mentioned previously, there is considerable evidence that demographic, socio-economic characteristics are related to crime. Therefore recorded crime incidents for the period of April 2001- March 2002 and demographic and socio-economic characteristics of the people in those areas derived from the 2001 Census were analysed. Note that the models presented in this section are for Burglary Dwelling, Criminal Damage, Theft from Motor Vehicle and Theft of Motor Vehicle. As with the correlation analysis in the previous section, SPSS was used as well for the statistical modelling.
Credit:ivythesis.typepad.com
0 comments:
Post a Comment