The development of a social, demographic and land use profiler for areas of high crime.

Maps that show the patterns of crimes do not allow the investigator to analyse, in depth, the relationships between levels of crime and the social and physical environment. In order to address these shortcomings, a demographic, land use and crime risk profiling system has been developed and piloted on Merseyside. This paper sets out the need for a profiling system, discusses the data sets which are used and addresses a number of issues concerning profile production and design. The paper concludes with a series of examples showing different applications of the profiler and describes an agenda for future research.

This paper discusses the development of a system for producing demographic, land use and crime risk profiles for crime `hotspots' and user defined areas in Merseyside in north-west England. Merseyside is an appropriate venue for the development of such a tool because of the availability, in the county, of a wide range of spatially referenced data sets (e.g. residential and non-residential property databases, digital street networks, computerized crime records and command and control incidents) which can be cross-referenced to produce valuable insights into relationships between crime, land use and social conditions. The profiling tool was developed during the course of a two-year study into relationships between crime and disadvantage on Merseyside. More general results of this study are reported in Hirschfield et al. (1995).

The Evidence of Crime Patterning

Individual crimes and incidents are not unique random events but rather share a number of common characteristics or features. The latter might include a tendency for incidents to occur in the same areas (spatial clustering), to afflict the same households or individuals (repeat victimization), to occur at certain times of the day, or to feature specific modus operandi, affect particular types of property, or to be perpetrated against victims with similar demographic and social characteristics (e.g. unemployed single people).

Evidence of crime patterning in terms of the spatial distribution of offences, the spatial and social distribution of the victims of crime and the demographic characteristics of offenders appears in a number of studies (Sherman et al. 1989; Cohen and Felson 1979 Mayhew et al. 1993;. Trickett et al. 1992).

The very fact that such patterns are discernible raises questions about the processes which underpin their manifestation. Ecological regularities in the distribution of `direct contact predatory' crimes involving theft and/or damage to persons or property have been explained in terms of the convergence in space and time of three key elements; likely offenders, suitable targets and the absence of `capable guardians' against crime (Cohen and Felson 1979). The probability that all three of these preconditions are met is sensitive to the spatial and temporal organization of routine social activities (for example, those pursued in fulfilment of employment, education, leisure, child-rearing and social interaction). According to the `routine activities theory' structural changes in patterns of everyday life have facilitated the convergence of the various risk factors and as a result have been an important influence on the crime rate. These changes are characterized by a move away from leisure/social activities centred on the home and involving household members to non-household activities involving non-household members.

Other theories have been advanced which explain spatial crime patterns in terms of the intersection of criminal opportunities with offenders' motivation, mobility and perceptions of target areas (Brantingham and Brantingham 1991), social disorganization (Shaw and McKay 1969; Bursik 1988; Sampson and Groves 1989), social disadvantage (Schmid and Schmid 1972; Brown 1982; Corsi and Harvey 1975; Pyle 1974; Simcha-Fagan and Schwartz 1986; Smith and Jarjoura 1988) and spatial concentrations of poverty (Braithwaite 1979; Wilson 1987; Taylor and Covington 1988).

The spatial configuration of different types of neighbourhood (i.e. spatial relations) may also be important. The temptation for potential offenders to commit crime may be greater when deprived people are in daily contact with affluence. Thus crime risks may be raised where poor areas either border or are close to areas of affluence. This is an under-researched area particularly in the British context.

It is essential that spatial patterns of crime are not to be viewed in isolation to the functions of different areas. Wikstrom (1991), stresses how particular areas may be devoted to different types of land use (residential development, retailing, industry, leisure, open space) and how the activities and population profile of an area may vary according to the day or time of day (e.g. city centres on weekday mornings and Saturday nights).

The Need for a Multi-Factor Profiling System

Clearly, spatial variations in crime at the neighbourhood level are the result of a complex web of inter-relationships between demographic, social, cultural, lifestyle and land use characteristics. This diversity of factors and processes influencing the levels and distribution of criminal events is generally not conveyed in maps and tables describing high crime areas. Most commonly, the latter will indicate the approximate location of areas with an above-average incidence or prevalence of offences and occasionally may display these areas in relation to readily identifiable land use features (e.g. main roads, schools, housing estates). Other processes which may be operative in such areas, some of which may be highly criminogenic (e.g. a highly segregated poor population, low levels of guardianship, the bordering of poor and affluent neighbourhoods) are not represented usually because the data sources required for their identification are either unavailable or are not coded geographically.

Three main types of map can be identified which delineate areas of high crime. They include point distribution maps which reveal the actual locations of individual incidents (e.g. offence locations), conventional shaded maps which show variations in crime rates across territorial units (e.g. police beats, census tracts) and probability, surface or generalized maps which show boundaries or contours delineating areas of unusually high criminal activity. The latter include maps which depict standard deviational ellipses or crime hotspots, derived from point distribution maps.

Maps which show the distribution of high crime areas, irrespective of the cartographical method which has been adopted, often convey interesting information. For example, areas of high crime may be large, small, contiguous or fragmented, they may be located in city centres or suburban areas, they may coincide with disadvantaged neighbourhoods and they may overlap with each other (which might be a finding of significance if the areas at risk had been derived for different types of crime). However, the analysis of relationships between areas of high crime and the social and physical environment, and the ability to compare rank different areas (e.g. burglary hotspots) require more than simply a description of the size and spatial distribution of areas at risk or individual crime `hotspots'. Information is also needed about the types of social and physical environment which characterize areas of high crime.

A demographic, land use and crime risk profiling system (the `profiler') has been developed to fulfil these requirements. The profiler is a specially designed piece of software which produces aggregate statistics on land use, demography and levels of crime for any area (restricted currently to Merseyside) using census enumeration districts (EDs) as building blocks. EDs are the smallest areas for which census data are available and …