Diurnal movements and the ambient population: an application to municipal-level crime rate calculations.

• Author: Andresen, Martin A.
• Position: Canada

Introduction

One of the most fundamental components of criminology and of criminal justice is the assessment of risk: the risk of criminal victimization, the risk of recidivism, and the risk of repeat victimization, to name only a few. With regard to the general risk of criminal victimization, the crime rate is the most common measure employed. Crime rates are used to compare the risk of criminal victimization between neighbourhoods, municipalities, provinces/states, and nations. But how useful is this measure for making these comparisons?

The crime rate is calculated by multiplying a ratio by a scalar. The ratio has the number of criminal incidents in the numerator and the population at risk of criminal victimization in the denominator; these numbers are typically representative of a particular year. The calculation of the numerator receives a lot of attention; the numerator varies for each crime and is also scrutinized/criticized because of the dark figure of crime. The denominator, however, receives little, if any, attention. In fact, most often the denominator is the same for all crimes--the Bureau of Justice Statistics in the United States uses some alternative denominators in its calculations, usually differentiating among ethnic groups. This may lead to difficulty of interpretation because the population at risk of criminal victimization for one crime is different from the population at risk of criminal victimization for another. And, of course, there may be overlap between two (or more) populations at risk of criminal victimization.

In the 1960s, Sarah Boggs investigated the impact of alternative denominators on various crime rates: the number of automobiles present for automotive theft, the pairs of persons present for violent crime, commercial land use for business-related theft, the number of females for forcible rape, and the number of occupied housing units for burglary. In some cases, changing the denominator had little impact on the crime rate (residential day burglary, forcible rape, criminal homicide, and aggravated assault), but in other cases the impact was significant such that the alternative denominator completely changed the interpretation (automotive theft, grand larceny, nonresidential day and night burglary). This finding led Boggs (1965) to state that we must "construct crime occurrence rates on the basis of environmental opportunities specific to each crime" and that "the number of events, or the numerator, varies with the type of crime, [so] the denominator should likewise vary so that the whole number of exposures to the risk of that specific event is incorporated as the base" (Boggs 1965: 899-900). Indeed, as stated in Harries (1981), a crime rate has the potential to be a meaningful statistic only with reference to the appropriate denominator (see Cohen, Kaufman, and Gottfredson [1985] for an alternative view on alternative denominators).

Despite the importance of this issue, very little research has been undertaken in recent years. This small literature, however, has been instructive. O'Brien (1989) tests the Easterlin hypothesis using age-specific crime rates, concluding that larger population cohorts have greater propensities toward property crime because there are relatively fewer legitimate economic opportunities. Investigating the trend in Canadian homicide, Andresen, Jenion, and Jenion (2003) compare the conventional homicide rate with an alternative rate that uses the number of young (15-30 years) males as the denominator. The two trends are radically different. The conventional homicide rate increases up to the mid-1970s, falling thereafter, whereas the alternative homicide rate increases right up to the early 1990s, then starts to fall. Andresen (2006) shows that in the neighbourhood-level spatial crime-analysis literature, the standard ecological theories of crime (social disorganization theory and routine activity theory) fare better with an alternative denominator; and Andresen and Jenion (2008) show that for a number of different crime classifications, there is very little relationship between spatially referenced crime rates that use conventional denominators and those that use alternative denominators--these two studies employ the same data as are discussed below. In sum, recent research (especially on crime in Canada) shows that the alternative denominator issue is still important.

This paper contributes to this literature through the incorporation of a relatively new data set into municipal-level crime rate calculations. This relatively new data set is the ambient population, a measure of how many people are expected to be present in an area on any given day and at any given time. To date, the criminological applications using these data (Andresen 2006; 2007; Andresen and Brantingham 2007; Andresen and Jenion 2008) are all at the neighbourhood level. However, as shown below, the same processes that change ambient population counts at the neighbourhood level are also present at the municipal level. Thus, understanding the nature of population movements and the corresponding impact on the population at risk of criminal victimization is important.

Diurnal movements and the ambient...