Principles of Geographical Offender Profiling

David Canter, Donna Youngs

• Publisher: Routledge
• Year: 2017
• Print ISBN:
  9780754625490
• eBook ISBN:
  9781351908986
• Topic: Social Sciences
• Subtopic: Criminology
• Index: Social Sciences

Geographical Offender Profiling (GOP) is the term that has emerged for the examination of where offences take place and the use of that examination to formulate views on the nature of the offender and where s/he might be based. As such, it has become the cornerstone of 'offender profiling'. By its nature, GOP bridges psychology, geography, criminology and forensic science and is of academic interest to all those disciplines as well as practical significance to police investigators. This book brings together a cross-section of the major papers published in the field that lay out the concepts and foundations of this area - including some widely quoted but difficult to obtain 'classic' papers - with an introduction that puts the papers into an overall context and a concluding extensive bibliography of the publications relevant to this rapidly growing area.

Chapter 1

Geographical Offender Profiling: Origins and Principles

David Canter and Donna Youngs

The Emergence of Geographical Offender Profiling

In the mid-1980s scientists assisting police investigations [Kind 1999, Canter 1994, Canter 2004, Rossmo 1995]¹ realised that they could estimate where an offender was likely to be based from an analysis of the geographical locations of that offender’s crimes. This process of indicating the possible area in which police should search for an offender eventually became known as ‘geographical profiling’ [Canter 2004], or in North America – where they worry less about syntactical niceties – ‘geographic profiling’ [Rossmo 1995].

The procedures for making such inferences grow out of studies of criminal spatial behaviour that can be traced back over a century and a half. It owes most, however, to the discovery first reported by Canter and Gregory [1994] for serial rapists, that an offender’s home or base was more often than not likely to be within an area circumscribed by his crimes. This discovery was first put to practical use in the case of the Railway Rapist in London [Canter 2004]. In that case the location of the offender’s residence was inferred from the geographical distribution of his offences. At that time a great deal of publicity was being given to the contributions which some FBI agents were claiming for how they solved crimes by drawing on a process that they called ‘offender profiling’. It was therefore a small step to see the analogy to FBI ‘profiling’, which made inferences about the characteristics of the offender from crime scene behaviours. When deductions about where an offender may be found were derived from the distribution of offences on a map it seemed appropriate to call this process ‘geographical profiling’.

There is now a recognisable field of geographical offender profiling that sits within the general discipline of Investigative Psychology and goes far beyond the ‘wheredunit’ search for an unknown offender’s likely residential location. This field has, from its inception, and even before it was labelled, consisted of a number of different aspects:

a) Studies of criminal spatial behaviour – basic research

b) Development of decision support tools that incorporate research findings

c) Explorations of how such tools may help police investigations and the effectiveness of these applications.

It thus straddles the divide that so often exists between, on the one hand, those purely academic studies that attempt to contribute to our knowledge about crime and criminals (as well as human behaviour more generally), and on the other, directly operational contributions to criminal investigations. An involvement at the borders between these academic and operational domains is also emerging out of the work on geographical profiling. Nowadays this is often characterised as ‘strategic’, policy development, in which a better understanding of criminals’ activities informs how law enforcement agencies approach their overall tasks.

Although the direct study of the spatial distribution of individual offender’s crimes only became a focus for research in the mid-1980s, some of the principles on which geographical profiling were based had been established over a century earlier. This had grown out of early criminological studies of the aggregate geographical distribution of crimes and criminals, most notably in the work of Guerry (1833). These studies revealed that offenders tended to live in distinct areas of a city and did not travel very far to commit their crimes.

The most graphic early description of this process was in the writings of Mayhew (1861) in which he described the areas in which there was a high concentration of criminals as ‘Rookeries’. These criminal areas were located close to places that offered attractive opportunities for crime. For example, the St Giles rookery was situated between the City of London boundary and Oxford Street that was then, as now, one of the major shopping streets in the capital. From the Rookery, less than half a mile away, thieves could commit crimes in Oxford Street and never be too far from the sanctuary of their home.

Such findings were rediscovered over and over again in the following 150 years, being given considerable impetus by US sociologists (e.g. White 1932), and were reflected in the day-to-day experience of local police officers who knew the villains who lived and committed crimes on their beat. So, although the emergence of geographical profiling started with very serious crime series, notably serial rapists and serial murderers [Canter and Larkin 1993], its origins are in studies of general, volume crime. There therefore is no a priori need to assume that the findings and approach will only be of relevance to serious crimes or series of offences. The great majority of criminal investigations can be enhanced by considering the geographical aspects of criminality.

Indeed much general, non-criminal human activity is open to and informed by a spatial analysis. If routine activity holds sway then it can be assumed that criminal journeys will not be that different from other forms of legal journey. Certainly some of the patterns such as decay functions, are similar between legal and illegal activity. There is therefore a great deal to be learnt from studies of the spatial patterns of legal activities. This will enrich our models and theories of criminal behaviours in many ways.

One particularly powerful illustration of how studies of legal human spatial activities can enrich our conceptualisations of criminal geography are the very detailed studies of street traders reviewed by Bromley (1980). He explores the distinction between mobile and fixed traders and shows that quite different logical cycles can be modelled on the basis of different assumptions. One of the trader cycles he illustrates is remarkably similar to the ‘marauding’ pattern found for serial killers by Lundrigan and Canter [2001] and other types of criminals. So, the existence of other market cycles raises the possibility that we should be looking for them within criminal geographies. Many other issues raised by Bromley open up possible parallels with criminal activities. To take just one example, he draws attention to differences in part-time traders and those for whom it is a full time job. There are clear analogies here with casual offenders, who are often younger, and ‘career’ criminals whose life is devoted to crime. Are there similar important differences in how these different types of offender make use of their surroundings?

Some of the patterns of traders’ spatial mobility can be seen as a natural result of the economy of opportunities. Traders wish to maximise their access to markets whilst minimising the effort involved. When the process has such a strong relationship between effort expenditure and benefits it is to be expected that there will be parallels in the animal kingdom. Mitzutani and Jewell [1998] show such parallels in the movement of leopards and their ranges of activities, as well as the location of their kills (Mitzutani 1993). These studies are conducted by monitoring electronic tags on the animals. Developments in criminal justice may make closely similar studies possible with criminals in the future.

These animal studies do show that some, at least, of the processes we have been summarising have a fundamental logic to them and are thus a consequence of natural activities. We should therefore not be surprised that more than a century and a half ago models of human spatial activity were proposed to help explain noncriminal phenomena. In particular the cause of an outbreak of cholera in London was explained by locating the patients on a map and relating that to the available water sources on well pumps [Snow 1855]. The clear correlation between the density of cholera cases and the location of particular wells was used to support the view that cholera was water borne. The assumption here was that people would use their nearest well, just as we assume offenders will not travel far to commit their crimes. The well plays a role analogous to the criminal’s base and the cholera cases are distributed like crimes around that location. Snow’s study is often heralded as the origin of modern epidemiology. It serves to show us that the patterns of criminal behaviour we are finding are not unique to criminals and, equally, that there is much that can be learnt about criminals’ spatial activity from considering similarities in other domains of activity. These illustrations of approaches to spatial thinking drawn from other contexts then are important introductions to criminal geography.

Inferring a Criminal’s Characteristics from Crime Scene Information

There are two distinct qualities of criminal spatial behaviour that make the inference of an offender’s home location possible from examination of where the crimes are committed. One, already mentioned, is the common finding that the assumed journey from a base to a crime location, rather than being a lengthy one to an alternative city or locality, is often very short. This was well illustrated in a detailed study of young criminals in Philadelphia, drawn from a 10% sample of police records for 1960 by Turner [1969]. He reports that three quarters of offences took place within one mile of the juvenile’s place of residence. Although Turner does not seem to be aware of the practical implications of his results, his study is the starting point for a more systematic and scientific contribution to criminal investigations. Such findings are the first step towards combining informal experience and academic scholarship into investigative support tools.

Using Turner’s results to aid investigations is an example of a general process that can be summarised as using criminal psychology and criminology in the ‘reverse’ direction from that in which they are usually considered. Typically social scientists have information about people and are trying to relate that to their actions. The detective has the opposite task. He has information about the actions of an offender and he wants to infer from that crucial facts about the offender and his location. This is the process that Canter and Youngs (2003) describe as the A → C equation. A represents the Actions in a crime, more generally what is known to detectives when the crime is being investigated and before an offender is apprehended, which includes where and when the crime happened and to what or whom. The C describes the characteristics of the offender that the police can use to find him/her; most crucially, of course, where to look for the offender. The → is the science of investigative psychology. This arrow is a shorthand for the processes of inference that allow investigators to draw conclusions about offenders from information about the crime. In other words, from studies of criminals that are set up with the specific objective of finding patterns that could be applied to later investigations through the inference processes that are developed. The specificity of the information on the location of crimes and the residence of criminals make the study of criminals’ geographical patterns especially feasible and instructive.

As Canter and Youngs (2007) note, Geographical Profiling is thus a special and very important example of how the A → C equation can be solved for the crucial context in which the A components are essentially where crimes occur and the C component is the location of the base of the offender. Of course, matters such as the nature of the location or locations of crimes and related issues such as the behaviour of the offender and the time and target of the offences need to be considered. Geographical profiling is not simply the consideration of the meaning of a dot on a map. The locations have to be understood in the context of as many other aspects of the crimes as can be harnessed to the inference process.

The discovery and rediscovery over the years, that many offenders often do not travel far to commit their crimes could not form the basis of a decision support tool for investigators until it was realised that this general trend could be applied to individual offenders. This is a small and logical step, but had not been immediately obvious to detectives. So, for example, if 75% of offenders are found to travel less than 2 kilometres to commit their crimes, then it must be the case that 3 out of 4 actual offenders will travel less than 2 kilometres. This is not a perfect prediction but it gives an investigator some framework within which to evaluate possible suspects. For instance, all else being equal, the suspects who live less than 2 kilometres from the crime scene should be given higher priority in the follow up investigation. Geographical profiling, then, generally moves the broad criminological consideration of aggregate crime processes on to the application to individual offenders. This has interesting, and important parallels to other questions in human spatial activity, as discussed by Bromley [1980] when attempting to model the mobility of street traders.

Kind [1987] was probably the first person to recognise the investigative potential of geographical analysis of crime locations. In 1980 he produced a report on the locations of the crimes attributed to ‘The Yorkshire Ripper’ who, over a five year period starting in July 1975, in the North of England, had murdered 13 women. As an ex-RAF navigator (Kind 1999) saw the task of locating the offender as a navigational task, likening the offence locations to fuel depots, with the challenge of discovering the optimal location of an aircraft’s base by calculating the ‘centre of gravity’ of the crime locations. This proved reasonably accurate in determining where the convicted offender, Sutcliffe, was eventually found to live. Sutcliffe was arrested before the report could be acted upon, so Kind’s discovery lay dormant until Canter rediscovered it independently five years later (see Canter 2005).

Kind showed that the short average distance travelled by many offenders is only one aspect of criminal spatial behaviour that makes geographical profiling possible. He introduced the second important aspect of criminal spatial behaviour, i.e. the possibility that many offenders have a base within the area circumscribed by their crimes. This has been shown to be a quite distinct finding from the shortness of the distance between home and crime [Canter et al. 2000]. Indeed, the finding that criminals tend to operate over a particular size of area, having a particular ‘criminal range’ [Canter and Gregory 1994] helped to broaden the whole perspective beyond just those offenders who travel small distances. It opened the way to considering the patterns of offending in relation to the home rather than simply focusing on offenders whose base is close to the crime. These spatial patterns of offending, notably the significance of the home, drawing attention to the environmental psychological and perceptual processes that had to be involved in criminal spatial behaviour, and the characteristic range of offending, led to the recognition that under some circumstances the criminal’s base could be inferred with a useful degree of accuracy from relatively simple geometrical calculations. This encouraged the development of software systems that allowed precise predictions to be made, and tested, applying various geometrical models to the distributions of crimes [Rossmo 1995].

These early studies also demonstrated that there was enough replicable structure in criminal spatial behaviour to make this a fruitful area of research (see Canter and Youngs 2007e for a review). Furthermore, they showed that offence location patterns were relative, and needed to be adjusted, to the scale over which the offender operated. Very similar patterns were found for offenders who travelled large distances to those that travelled small distances (Canter 2005). This meant that geographical profiling could be much more than merely drawing attention to the many offenders having bases near their crimes. Rather broader models of offending could be derived that explored the geographical patterns independently of the range of any given offender.

The generality of these results suggested that they reflect general mental schemata (Bartlett 1930) of how offenders conceptualise their environment. These internal representations are often referred to as ‘mental maps’ (cf. Canter 1977). They have been explored by environmental psychologists such as Ladd (1970) using the device of getting people to draw a sketch map of an area. As Down and Stea (1977) and many others have shown, building on the seminal work by Lynch (1960), these mental representations do differ from person to person, relating closely to their actual experience of particular environments. The implication of this is that experience shapes understanding of places which in turn influences where a person does what. The findings open the way to exploring how the spatial distribution of offences may reveal fundamental cognitive processes that criminals share with law abiding citizens. The direct exploration of offenders’ ‘mental maps’ [Canter and Hodge 2000] as a way of predicting and understanding the spatial distribution of their activities thus became a further area for research. By this ...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Acknowledgements
• Series Preface – David Canter
• Preface
• 1 Geographical Offender Profiling: Origins and Principles
• Part 1 Spatial Patterns in Behaviour
• Part 2 Offenders’ Geography
• Part 3 Key Concepts in Geographical Offender Profiling
• Bibliography of Geographical Profiling
• Index