GIS and the Social Sciences

LEARNING OBJECTIVES

▪ Becoming familiar with relevant key types of crime data and relevant GIS data sources

▪ Using GIS for the allocation and management of policing resources

▪ Using GIS for mapping and analysis of crime patterns

▪ Using GIS to support crime prevention strategies

This chapter explores the potential of GIS to support the work of police and fire authorities around the world. Thus the chapter focuses largely on GIS and crime pattern analysis. The primary uses of GIS technology in law enforcement are in detection and the analysis of crime patterns and in the development and management of preventative measures. This chapter will first discuss the types of data that are potentially available for crime analysis (although this will vary by country) and the issues surrounding usage and sharing. We will then review a series of tools that are available for GIS mapping and the analysis of crime. Key questions we might typically want to address include: when and where should resources (e.g. officers and vehicles) be most effectively deployed? How can future crimes be prevented? Can we find patterns to help us apprehend repeat offenders? What are the displacement effects of crime prevention? These types of questions have an inherently spatial-temporal aspect and GIS is an obvious choice to help in law enforcement.

In many countries there are often two different types of crime data available for analysis with GIS. The first are data collected by police forces themselves. In many countries these data have to be made routinely available for inspection and analysis by law. Typical features of basic crime data recorded by police authorities include when and where a crime has been committed and various details of the nature of the offence. These (point) data are collected by the police or law enforcement agencies for various purposes including both strategic and operational planning.

The fundamental issue regarding crime data is that police data can only provide a limited picture, which does not implicitly indicate true patterns of crime. For instance, even if a crime takes place it is not guaranteed that the crime will actually be recorded by the police. In fact, Dodd et al. (2004) estimated that only 30% of all crime is actually represented by recorded crime data. This is, of course, an average and will vary hugely by offence type and by country. Burglaries, for instance, are more commonly reported for insurance purposes. However, unreported crime is expected to cluster in the same places as reported crime (Chainey and Ratcliffe, 2013), thereby effectively increasing the association between reported crime figures and actual amounts of crime (Malleson, 2010).

Figure 7.1 shows a screenshot of a publicly available GIS site created by the police in the UK. It shows all crimes in a single month in the city centre of Leeds, UK (mainly shoplifting and assaults in bars and night clubs). The user can plot any one of ten crime types and for different months of the year. (In the US the FBI collects crime data from each law enforcement agency across the US and then publishes the ‘uniform crime reporting statistics’ at various spatial scales.)

The data given by police forces often have limitations. As noted above, only recorded crime can be included. Second, some crime locations might be reported as an approximate location (by victims or eye witnesses) but they are always recorded as a precise point location. This might be more of an issue for certain crime types over others.

Due to the increased developments in mobile phone technology, the Police.UK website shown above is also available through a number of phone applications. While there are a number of different applications, the applications essentially allow users to look at the level and types of crime in a given area based on the phone’s internal GPS. The main applications include: Crime Spy UK, Crime Map UK, Crimes Near Me, UK Crime View and Crime Finder. The Crime Finder application is demonstrated in Figure 7.2 and even includes Google Street View capability for the visualisation of the street (see Malleson (2010) for more details).

As an alternative to using data provided by the police, it is also often possible to use survey data. In the UK, for example, it is possible to access the newly named Crime Survey (CS) for England and Wales, formerly known as the British Crime Survey (BCS), which measures the amount of crime in England and Wales. The original BCS started in 1982 and initially covered England, Wales and Scotland. In 2011/12 around 67,000 private households across England and Wales were invited to participate in the survey and about 75% of those responded (50,000). The main topics include information about the crimes people (aged 16 and over) have experienced in the last year, attitudes towards the police, attitudes about the criminal justice system and perceptions of crime and anti-social behaviour. From January 2009, 4,000 interviews have also been conducted each year with children 10–15 years old, although the resulting statistics remain experimental. The UK Home Office (the government department responsible for policing and crime reduction) believe that in some respects the CS provides a better reflection of the true level of crime than police statistics, since it includes crimes that have not been reported to, or recorded by, the police. The survey is also a better indicator of long-term trends because it is unaffected by changes in levels of reporting to the police or police recording practices (Malleson, 2010). In the US, the Bureau of Justice is responsible for a similar survey: the ‘National Crime Victimisation Survey’ which samples from 90,000 households across the US.

The downside to survey data is that they are often poorly referenced in spatial terms. For confidentiality reasons, individuals and their locations are rarely named. Thus it is impossible to plot crimes from the survey with any geographical accuracy – however, it is often possible to reweight such survey data to link the attributes of individuals to census data in order to add small-area geography (see the discussion of microsimulation in Chapter 6).

One of the first major uses of spatial analysis for crime fighting is likely to be to help allocate resources. Indeed, most public service organisations have funding models that are based on geographical variations in socio-economic data. In the UK, the police use such a formula to allocate £11 billion (2014 figure) between the different police forces shown in Figure 7.3.

The allocation formula includes the following census-based indicators: population total, population density, (youth) unemployment rates, bars per hectare (a proxy for where assaults are more likely to be committed), CACI’s geodemographic classification ‘Hard Pressed’ label (which picks up areas of low income), student housing rates, population inflow (to measure community cohesion) and single-parent households (Home Office, 2013). A GIS containing these layers of information would be a good starting point to be able to not only calculate the allocations on a regular basis, but also to be able to experiment with different weightings in order to see how resources could be allocated differently and perhaps more fairly (especially as the values of these variables change over time).

While we explore the use of GIS for crime pattern analysis it is useful to try to answer the questions why is geography important in crime analysis, or why does crime vary spatially? Figure 7.4 shows the spatial variations at a single point in time across London, UK. As the map shows, there are widespread variations across the city. Crime data can be mapped using raw numbers or normalised to produce rates. City centres, for example, can have large actual crime figures (as there are lots of people in town during the day and night) but low rates when expressed as a percentage of the population that resides there. Figure 7.4 shows there is, indeed, a high concentration of crime in the city centre but also in the east of London, with much lower rates to the south-west. These high crime areas tend to correspond with higher levels of poverty. Thus, the first explanation for spatial variations in crime rates is simply socio-economic. Areas with high levels of unemployment, drug use, poor educational attainment, etc. can become breeding grounds for crime.

Figure 7.5 shows the spatial variation in crime classified as homicides (murders) across Washington, DC. Again we can see a clustering effect with much higher rates in certain notorious inner-city or downtown districts, north and east of the US Capitol. You will be able to explore recorded crime rates in Washington, DC, further in the practical activity linked to this chapter.

There have been many studies exploring this correlation between crime and deprivation. A GIS application for exploratory analysis of crime patterns has been widely used by researchers at the University of Liverpool in the UK, who have concentrated particularly on the mapping of crime patterns in relation to socio-economic structure. An early application used ArcGIS to plot maps of crime incidence and examined spatial and temporal trends in these patterns (Hirschfield et al., 1995; Bowers, 1999; Hirschfield and Bowers, 2001). The researchers also used GIS to study the hypotheses that:

▪ crime can be associated with low socio-economic status and social heterogeneity.

The studies plotted the location of disorder incidents in relation to deprived areas, police beats, main roads, pubs and night clubs. The STAC software (Spatial and Temporal Analysis of Crime) was developed to identify clusters or hotspots of criminal activity, which can be displayed in relation to the distribution of affluent and disadvantaged areas from various geodemographic classifications. One of the primary discoveries was a link between crime such as burglaries and changes in household occupancy (as represented by changes in council tax registrations), which conforms to the second of the above hypotheses.

Marxist interpretations of crime claim that such links between crime and deprivation are inevitable given the capitalist system and its inherent ability to create class divisions around income (Peet, 1975; Lowman, 1986). Some argue that in more deprived areas, individuals (especially youths) are more likely to have had a more difficult if not violent upbringing with few good role models to draw on. If more individuals who are likely to commit crime live in certain areas then we have the first part of Cohen and Felson’s (1979) crime triangle – a motivated offender. Add a victim in the neighbourhood of the offender, and the absence of a capable guardian, and we complete the crime triangle shown in Figure 7.6.

Brantingham and Brantingham (1981) show how the geography of crime can be better understood by further considering the ‘spatial activity map’ of criminals. Criminals tend to operate close to home (but not too close), in and around major destinations they are likely to visit (friend’s house, shopping centre, even workplace). Thus individuals build up a spatial awareness map over time around key anchor points in their neighbourhoods. Figure 7.7 demonstrates this idea geographically.

As hypothesised above by Hirschfield and Bowers (2001), we might also expect crime to be higher in areas where affluent areas border lower income areas. Figure 7.8 plots the locations of offenders committing crimes in Roundhay, Leeds, an affluent suburb of the city. Roundhay has above average crime rates compared to other high-income areas in the city. As Figure 7.8 shows, part of the explanation might lie in the fact that Roundhay is bordered to the east and south in particular by low-income areas such as Harehills (shown on the map), Burmantofts and Seacroft. Many other high-income areas are themselves surrounded by high-income or middle-income suburbs. This unique geography provides an opportunity for offenders to quickly escape back into territories they are familiar with in order to escape police attention.

Deprived areas are often associated with more nuisance crimes such as anti-social behaviour, petty theft and vandalism. Wilson and Kelling (1982) referred to this as the ‘broken window’ theory. This refers to the fact that evidence of minor petty crimes in the landscape tends to encourage others to do the same. Thus a number of police forces around the world have tried to crack down on minor crimes in the belief that this would inevitably prevent more serious crimes. GIS can again help target those locations where petty crimes prevail.

As Figure 7.9 shows, the night pattern of crime is greater, more widespread and generally more serious. Understanding such patterns has enormous implications for real-time policing, crime prevention and resource allocation.

An extension of crime mapping is the understanding of repeat victimisation. For example, the risk of burglary doubles after a victimisation, but risk does then rapidly decay with time. This is said to be related to offender confidence (she/he knows that the property is able to be burgled). Figure 7.10 is taken from the work of Hirschfield and Bowers (2001) in part of Liverpool, UK. Although a rather crude GIS map by today’s standards, plotting the location of the victims of repeat burglary does help to show an important additional feature – that properties most likely to be repeat victims are closer to the main road network (thus again allowing quicker getaways).

This knowledge has been put to good use in Manchester in the UK. Police in the Trafford district of the city worked with GIS analysts from The Bartlett Centre for Advanced Spatial Analysis in London to examine the geography of burglaries and repeat burglaries. They showed that burglars were more likely to return to previously successful ‘hit’ locations than to choose new locations. This again was for reasons of offender confidence and the fact that the properties of the victim’s neighbours are more likely to have similar designs (hence could be broken into in the same manner as the original property chosen on the street). This allowed Trafford Police to develop a new model for reducing burglary, which has become known in the UK as the Trafford Model. By targeting resources at areas recently burgled Greater Manchester Police were able to reduce burglaries in Trafford by 26% in six months (see Fielding and Jones (2012) for more details).

GIS is increasingly being used in conjunction with psychology techniques to analyse spatial crime patterns to, in turn, estimate where offenders are most likely to live. This can help narrow down the search for criminals in the detective exercise following a (major) crime. This is becoming increasingly popular in attempts to capture serial murderers/rapists, etc., but is also more common today for many other applications; e.g. prolific burglary. The idea of geographical profiling is to use GIS to create a ‘probability surface’ of most likely offender home location based on key facts available (Rossmo, 1999; Canter et al., 2000; Canter, 2010; Canter and Youngs, 2008).

The most famous criminal never to be caught in the UK is probably Jack the Ripper. He lived in Victorian London in the infamous Whitechapel area, an area in 19th-century London that was synonymous with poverty, crime and prostitution. Jack is believed to have killed five prostitutes in and around the area but was never caught. Using modern GIS and geographical profiling new evidence can be brought to bear on the case. His victims were known to live in lodgings close to the Ten Bells public house (see Figure 7.12) although they drank and picked up clients in all the local pubs. If a polygon is drawn around the location of his victims it is argued that he probably lived close to the centre of that polygon. If you then add the fact that the victim lived close to the Ten Bells pub it could be argued that he himself was likely to frequent that pub and live nearby. Expert Canadian criminal profiler Kim Rossmo undertook this analysis and pinpointed ‘Flower and Dean Street’ as the most likely location of Jack’s residence (see Figure 7.12). Given the fact the police had several possible suspects on their files from that street at that time, then it is possible they could have concentrated their enquiries on that much smaller group of possible suspects.

The same type of profiling has recently been undertaken to find a so-called copycat killer operating in Yorkshire, UK in the 1980s. Nicknamed the ‘Yorkshire Ripper’, Peter Sutcliffe was found and convicted for the murder of 13 victims. However, he gave the police the run around for many years. Again modern GIS and geographical profiling might have helped capture Sutcliffe earlier. Using key evidence that he seemed to know the areas well and victims were killed close to major roads (it turned out he was a truck driver), profiling has shown that the most likely location of Sutcliffe was around Heaton and Manningham – the location, as it turns out, where he did actually live.

In many countries crime rates are often produced as league tables – the top crime areas at the top going down to low crime areas at the bottom. That is all well and good but such a table is often very predictable – most deprived areas at the top of the list, leafy, green suburban areas at the bottom. GIS and spatial analysis can be used to help create alternative league tables that might be more interesting. Harper et al. (2001) report the results of this procedure for the whole of the UK in a joint project with the UK Home Office. The study created ‘families’ of police authority zones that had similar socio-economic profiles – i.e. one family might be the result of clustering areas of lower income, terraced housing, low car ownership, etc. Once the families were created, Harper et al. could compare crime within these families rather than simply between them (more typical of normal crime lists). Figure 7.13 shows the results for ‘Family 4’, largely inner-city areas of the UK’s major cities. It shows how violent crime varied across the family – the Home Office could then ask why violent crime was much lower in inner cities of Leeds and Sheffield compared to Manchester and Nottingham. Hence, good police practice could, in theory, be exchanged between good performing and poorer performing police forces.

Alongside the police, the fire service provides a major emergency service provision and, once again, GIS and spatial analysis can provide very useful data support roles. Again, an interesting question is why would the distribution of fire incidents have a spatial dimension? Of course there are obvious urban and rural differences. Bush fires in countries such as Australia and the US have been very much in the news recently especially as hot summers provide ideal conditions for such fires to spread rapidly. Chen et al. (2003) used GIS to map the distribution and impacts of bushfires.

But even within urban areas geography seems important. Most importantly, fires are more likely to start in households within lower income areas. This is for a number of reasons. First those households might have older electrical appliances, with older wiring. Second, lower income households are less likely to be able to afford smoke detectors. Third, such households lie in areas more prone to arson. The number of malicious calls and incidents of arson are more common than many people suspect.

Corcoran et al. (2007a, 2007b, 2011) have analysed the spatial variations in incident types for fires across South Wales in the UK. Figure 7.14 shows these patterns, with marked concentrations visible in the coastal belt (which links the main cities of Cardiff and Swansea) and in the valleys (the old coalmining areas which are now suffering from high levels of poverty). Thus, as with many types of crime, there is a striking correlation between areas of deprivation and high fire incidence, especially for arson and malicious fires.

This chapter has provided an introduction to the application of GIS in the crime and fire service sectors. We have also demonstrated that GIS can be applied to crime issues at a variety of levels from illustrative mapping to more complex spatial analysis. The fundamental processing options for GIS in crime analysis include:

▪ Basic spatial mapping of crime geography, which permits a visual impression of crime distribution, offence types and victim and offender locations.

▪ Pattern detection analysis of point level data, which offers a statistical interpretation of crime hotspots. This might confirm or denounce hypotheses of distribution suggested by simple map reviewing.

▪ Location analysis to study crime in relation to other static spatial features such as schools, pubs, cash machines, etc.

▪ Spatial aggregation of crime data for comparison to contextual information on demography, land use, socio-economic conditions, etc.

▪ Temporal series analysis to indicate trends in criminal activity and to monitor and manage crime policy.

▪ Application of these results and analyses to target crime prevention strategies and link efforts with other governmental organisations and authority departments.

The Trafford example above shows how crime can be reduced by adding a GIS and spatial analysis capability. There are a number of other illustrations of police forces around the world advocating the benefits of GIS, usually on their websites. The ESRI website (2007), for example, quotes Police Chief Crisp of Columbia Police Force in the US as saying ‘the crime rate for Columbia has fallen dramatically with the implementation of GIS mapping’ and ‘using GIS mapping helps us concentrate efforts to maximize resources in the most effective manner possible. It has helped produce the lowest crime rate that Columbia has seen within the past 15 years’.

The final words can be given to Joe Kezon, GIS manager for the Chicago Police Force in the 2000s (quoted in Chen, 2004). He commented that using GIS technologies allowed officers to make better-informed decisions about which areas of the city need additional police power: ‘The commander of the Deployment Operations Center saw the importance of having the ability to do some mapping and analysis that would allow them to make key judgments of where they should create police deployment areas.’ The net effect, according to Kezon, was an 18% drop in murders compared with the same period the year before. That alone must be a good reason to employ GIS!

This chapter is accompanied by Practical 9: Crime analysis. The practical gives you an opportunity to explore data related to actual recorded crimes in the US city of Washington, DC. We walk you through the process of obtaining, downloading and importing crime data from an external source. You practise handling this data in the GIS, including aggregation to larger spatial units and hot spot analysis to identify clusters of neighbourhoods with high crime rates.

Wortley, R., & Mazerolle, L. (2008) Environmental Criminology and Crime Analysis, Willan Publishing, London.

Bowers, K. (1999) Exploring links between crime and disadvantage in north-west England: an analysis using geographical information systems. International Journal of Geographical Information Science, 13(2), 159–184.

Brantingham, P. J., & Brantingham, P. L. (1981) Notes on the geometry of crime, in P. J. Brantingham & P. L. Brantingham (eds) Environmental Criminology. Prospect Heights, New York, 27–54.

Canter, D. (2010) Offender profiling, in J. M. Brown & E. A. Campbell (eds) The Cambridge Handbook of Forensic Psychology, Cambridge University Press, Cambridge, 236–241.

Canter, D., & Youngs, D. (eds) (2008) Applications of Geographical Offender Profiling, Ashgate Publishing, London.

Canter, D., Coffey, T., Huntley, M., & Missen, C. (2000) Predicting serial killers’ home base using a decision support system. Journal of Quantitative Criminology, 16(4), 457–478.

Chen, K., Blong, R., & Jacobson, C. (2003) Towards an integrated approach to natural hazards risk assessment using GIS: with reference to bushfires. Environmental Management, 31(4), 0546–0560.

Clarke, G. P., Vickers, D., et al. (2006) Evaluation of the Leeds Arson Task Force: Final report to West Yorkshire Fire Service (unpublished) (copies available from the first author).

Cohen, L., & Felson, M. (1979) Social change and crime rate trends: a routine activity approach. American Sociological Review, 44, 588–608.

Corcoran, J., Higgs, G., Brunsdon, C., Ware, A., & Norman, P. (2007a) The use of spatial analytical techniques to explore patterns of fire incidence: a South Wales case study. Computers, Environment and Urban Systems, 31(6), 623–647.

Corcoran, J., Higgs, G., Brunsdon, C., & Ware, A. (2007b) The use of comaps to explore the spatial and temporal dynamics of fire incidents: a case study in South Wales, United Kingdom. The Professional Geographer, 59(4), 521–536.

Corcoran, J., Higgs, G., & Higginson, A. (2011) Fire incidence in metropolitan areas: a comparative study of Brisbane (Australia) and Cardiff (United Kingdom). Applied Geography, 31(1), 65–75.

Dodd, T., Nicholas, S., Povey, D., & Walker, A. (2004) Crime in England and Wales 2003/2004. Home Office Statistical Bulletin. Home Office, London.

Eftelioglu, E., Shekhar, S., & Tang, X. (2016) Crime hotspot detection: a computational perspective, in O. E. Isafiade & A. B. Bagula (eds) Data Mining Trends and Applications in Criminal Science and Investigations. IGI Global, Hershey, PA, 82–111.

Fielding, M., & Jones, V. (2012) Disrupting the optimal forager: predictive risk mapping and domestic burglary reduction in Trafford, Greater Manchester. International Journal of Police Science & Management, 14(1), 30–41.

Hirschfield, A., & Bowers, K. (2001) Mapping and Analysing Crime Data: Lessons from Research and Practice, Taylor & Francis, London.

Hirschfield, A., Brown, P., & Todd, P. (1995) GIS and the analysis of spatially-referenced crime data: experiences in Merseyside, UK. International Journal of Geographical Information Systems, 9(2), 191–210.

Lowman, J. (1986) Conceptual issues in the geography of crime: toward a geography of social control. Annals of the Association of American Geographers, 76(1), 81–94.

Malleson, N. (2010) Agent-Based Modelling of Burglary. PhD thesis, School of Geography, University of Leeds.

Malleson, N., & Andresen, M. A. (2015) Spatio-temporal crime hotspots and the ambient population. Crime Science, 4 (10). DOI: https://doi.org/10.1186/s40163-015-0023-8.

Peet, R. (1975) The geography of crime: a political critique. The Professional Geographer, 27(3), 277–280.

Wilson, J. Q., & Kelling, G. L. (1982) Broken windows. Critical Issues in Policing: Contemporary Readings, 7, 395–407.