Crime specialization in rural British Columbia, Canada.

AuthorCarleton, Rebecca


Rural crime has received little attention. A common concern with the resulting urban focus is that there has been a tendency to romanticize the rural landscape as idyllic and position it as simply the opposite of the urban "underbelly." This assumption may be based on the understanding that rural areas are simply less dense urban areas, with fewer people spread over a larger geographical space. However, it is not certain whether this assumption is correct. Preliminary evidence suggests that there are differences between rural and urban crime patterns in British Columbia, Canada (Brantingham and Brantingham 1998), but more information is necessary to be able to assert this claim with confidence. Historically, it has been supposed that rural areas specialize in violent offending, whereas more urban areas show high concentrations of property crime (see, e.g., Guerry 1833; Quetelet 1835). The present article attempts to better understand rural patterns in crime across the province of British Columbia and asks whether rural areas can be considered particularly violent when compared to their urban counterparts.

Based on existing Canadian research, it would seem that crime increases moving east to west across the country (Hartnagel 1997). As the most western of Canadian provinces, British Columbia might be considered "most criminal" when compared to the other Canadian provinces (Andresen 2009). Given regional disparities in crime in British Columbia and questions about disparities for specific crime types (Brantingham and Brantingham 1998), it is necessary to evaluate these differences using an alternative measure to crime rates: the location quotient (LQC). Traditional crime rates are useful but insufficient for the present study since (1) British Columbia can be characterized as predominantly rural and crime rates are highly sensitive to changes in smaller populations, (2) there is a need to use population as an indicator to characterize an area as rural or urban, and (3) a crime rate does not allow for the assessment of a crime specialization. While crime rates are arguably useful in understanding crime patterns, an LQC helps supplement the understanding by providing additional information.

Numbering about 4.25 million residents, with a density of 4.4 persons per [km.sup.2] (Statistics Canada 2003), the British Columbian population is distributed across a largely rural landscape. The most western of the Canadian provinces, in 2011, British Columbia evidenced its lowest crime rate in over 30 years (Ministry of Public Safety 2012), which, as in the rest of Canada, was largely driven by decreases in property offences. Specifically, substantial property-crime declines included theft (-4%), mischief (-11 %), motor vehicle theft (-14%), and breaking and entering (-13%) (Ministry of Public Safety 2012). However, during the same period, the British Columbian crime-severity-index ranking remained about 27% higher than the Canadian national average (Ministry of Public Safety 2012). Further, the number of violent offences increased slightly from the previous year and remained higher than the national average (Ministry of Public Safety 2012). Thus, while British Columbia as a whole is reporting the lowest crime rates since the 1970s, it is acknowledged that the reduction is crime specific. Stated differently, while some types of offending are on the decline, others show an increase.

In terms of measuring crime, a location quotient allows for an analysis of crime that is not trapped by population counts and provides insight into the proportion of a specific type of crime in relation to all crime in a particular geographical area. Andresen (2007) suggests that traditional crime rates create issues inasmuch as different controls result in different spatial distributions of crime for the same area. Common in economic geography (see, e.g., Isard, Azis, Drennan, Miller, Saltzman, and Thorbecke 1998; Miller, Gibson, and Wright 1991) and in health research (see, e.g., Beyene and Moineddin 2005), location quotients have become an increasingly accepted measure within spatial criminology, as well (see, e.g., Andresen 2006; Breetzke and Cohn 2013; Groff and McCord 2012). While location quotients are useful for smaller populations, they also allow for the assessment of a particular type of offence in relation to all other offences within an area (Andresen 2007). Accordingly, location quotients are preferable for purposes of understanding rural-crime-specific patterns and identifying the crime specialization of these types of areas.

Location quotients in relation to crime

As a measure of "actual crime," crime counts represent a count of the total number of crime x in some geographical location or over some period. Thus, while useful in identifying hot spots for crime occurrence, crime counts are hampered for purposes of comparison across geographical areas if there is reason to suspect that some confounding variable, such as population size, might affect the total amount of crime X. In response to the problem of crime counts, crime rates present an alternative that normalizes the total amount of crime to present a measure of "the risk of crimes occurring to particular types of people in particular locations or at particular times" (Brantingham and Brantingham 1998: 266), in which the number of crimes acts as a numerator while the population at risk functions as a denominator. In the case of crime committed against individuals, the denominator would generally represent a measure of the number of people in a particular area (Brantingham and Brantingham 1998).

A traditional crime rate is of limited use for estimating crime in smaller population centres, since a small change in the total number of crimes has a substantial impact when the denominator (population count) is very small, and crime, as a result, can be overestimated. While this mathematical reality is not a problem outright, it is problematic when comparing across a range of geographical areas with a wide range of population counts. For example, within British Columbia, population counts for police jurisdictions range from the smallest jurisdiction (Atlin, British Columbia) that had 227 people to the largest jurisdiction (Vancouver Municipality), which had 601,159 people as of 2006 (mean = 27,036.5). Thus, a single event in Atlin would have a much larger effect on the crime rate than a single event in Vancouver.

Supplementing crime rates with location quotients for understanding geographical distributions of crime should be useful when it is necessary to use a population measure as an independent variable for the purposes of explaining a dependent variable. "In such instances," note Brantingham and Brantingham (1998: 271-72), "the overall strength of the model may be the result of the same numbers being used as the denominators on both sides of the equation." Such an approach would be problematic in the case where urbanization, or population mobility, was being used in a model to explain a geographical pattern of crime, but it is particularly problematic for a study that attempts to explain rural patterns of crime. The classification rural, or urban for that matter, is, by definition, a function of the population size in a given area and accordingly requires an outcome variable that is not related to population (which a crime rate is).

Finally, there is sufficient evidence to think that specific types of crime have different aetiologies in the British Columbian context (see, e.g., Andresen 2007; Frank, Andresen, and Felson 2012). There is also reason to believe that particular types of crime are over-represented when compared to all other crime types within the same area in British Columbia (Brantingham and Brantingham 1998). While a crime rate allows for an assessment of the risk of becoming a victim of a specific type of crime, it is less useful for examining whether an area could be considered, for example, disproportionately violent. Thus, traditional crime rates can be used in conjunction with location quotients if the objective is determining whether certain areas of British Columbia are disproportionately criminogenic and, if so, whether there are patterns that depend on the specific geographical characteristics of a jurisdiction.

An LQC addresses the problems associated with selecting a denominator for making total crimes comparable, and it eliminates disparities in crime rates across a range of differently sized population areas. Further, an LQC provides a measure of crime specialization or of the concentration of a specific type of offence when compared to all other types of offences in the same area. As noted by Brantingham and Brantingham (1998: 268-70), location quotients are really a "dimensionless measure" of the preference or choice of crime type in a smaller unit in comparison to a larger unit, which can have meaning in crime analysis "particularly when the analysis is related back to the relative mix of other socioeconomic conditions."

Despite the advantages of using a location quotient in conjunction with crime rates, and although LQCs have attracted increasing attention (see, e.g., Andresen 2006; Breetzke and Cohn 2013; Groff and McCord 2012), the practice remains relatively less common within criminology than the reliance on crime rates alone. As a "measure developed in regional planning and economics to try to address questions about the relative importance of local economies, local areas are placed within some wider context for analysis" (Brantingham and Brantingham 1995: 136). (1) When they have been used, LQCs have been applied to research involving the crime-generating / crime-attracting characteristics of neighbourhoods (Caplan, Kennedy, and Petrossian 2011) or the crime-generating role of neighbourhood parks (Groff and McCord 2012). In other work, location quotients have been used to "determine if drug arrests were clustered...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT