Thematic issues

To complement the crime diagnostics, a number of additional issues were addressed in exploratory analyses. In some cases, these analyses highlighted the importance of using a variety of approaches to prevent crime, while in others cases, they illustrated the limitations of the available data.

Spatial variations by time of crime: robbery in Winnipeg

While criminal incidents can take place at any time of day in a city, research has shown that the different types of crime do tend to occur at specific times (Assunção, Beato and Silva 2002). The Winnipeg analysis supported this finding in the Canadian context.

Chart 2 shows the distribution, by time of day, of all robbery offences reported in Winnipeg in 2001. The data indicate that fewer robberies took place in the morning and more occurred in the evening before midnight.

Chart 2
Robbery incidents by time of day, Winnipeg, 2001

Maps 1 to 3 show the distribution of robbery hot spots at three times of the day, that is, morning (between 7 a.m. and 9 a.m.), evening (8 p.m. to 10 p.m.) and night (1 a.m. to 3 a.m.). In 2001, 4% of all robberies reported occurred between 7 a.m. and 9 a.m., 14% between 8 p.m. and 10 p.m. and 12% between 1 a.m. and 3 a.m.

Spatial distribution varied depending on the time of day, both in terms of the number of criminal offences and the areas where crime density was highest (hot spots), indicated by the darkest red.

Map 1
Kernel density distribution of morning robbery incidents, 7:00-9:00 AM, Winnipeg, 2001

Map 2
Kernel density distribution of evening robbery incidents, 8:00-10:00 PM, Winnipeg, 2001

Map 3
Kernel density distribution of night time robbery incidents, 1:00-3:00 AM, Winnipeg, 2001

Travel-to-offence patterns of persons charged: analysis of distances travelled in Montréal

According to the opportunity theory of criminal behaviour, the distribution of crime is determined by the convergence of three elements. The opportunity for crime exists when an offender, a target and the absence of a guardian converge in time and space (Felson and Poulsen 2003).

Using data provided by the Service de police de la Ville de Montréal on the location of criminal incidents and the place of residence of persons charged, it was possible to calculate the distance travelled by persons charged.

Overall, persons charged in violent incidents travelled less (0.9 km) than those charged in property incidents (4 km) (Table 3). Other research papers have also shown that persons accused of violent offences travel shorter distances than those accused of property crimes (LeBeau 1987; Turner 1969).

Table 3
Median distances travelled by charged persons by type of offence, Montréal, 2001

The median distance travelled also varied according to the closeness of the relationship between the person charged and the victim. Those who knew their victim travelled little, while those who did not covered the greatest distances and converged toward the city centre (Table 4).

Table 4
Median distance travelled by persons charged, by relationship with the victim

This study also found that distances travelled varied according to the age of the persons charged (Chart 3). The youngest travelled the most in violent incidents and the least in incidents involving property. Many foreign studies have produced similar results (Groff and McEwen 2005; Wiles and Costello 2000; Chapin and Brail 1969; Harries 1999). In the case of violent offences, the distance travelled was greatest among adolescents (those between 12 and 17 years of age) and diminished with increasing age.

Chart 3
Median distance travelled by persons charged, by age, Montréal, 2001

This variation is due to the fact that young persons aged 12 to 17 were more likely to target acquaintances (51% of their victims) and strangers (40%) than persons with whom they had any other type of relationship. Starting at age 25, persons charged were consistently more likely to target their spouse (between 26% and 29%), followed by acquaintances (between 24% and 34%) and ex-spouses (between 12% and 16%). Charged persons aged 18 to 24 were the most likely to target strangers (43%). In comparison, the distance travelled to commit property offences was shortest for male and female youth, and peaked between 18 and 34 years of age, after which it stabilized. This pattern may be related to access to various modes of transportation.

Descriptive analyses of the median distance travelled by persons charged led to the finding that distances travelled vary by type of offence, age of the persons charged and their relationship with the victim (Savoie 2006). Results from the Montréal study support British research findings indicating that most offender movements are relatively short, and that travel associated with crime is driven by opportunities presenting themselves during daily activities and routine travel rather than plans to offend (Felson and Clark 1998; Wiles and Costello 2000). Charged persons and their targets vary according to the initial reason for travelling—or not travelling in the case of spousal violence. In this regard, trips initiated for work, school or recreation offer specific opportunities for crime (Felson and Clark 1998). The longest median travel distances, which were recorded for auto theft incidents, may be related to a more organized criminal effort.

Spatial distribution of youth crime: Montréal as an example

The characteristics of distances travelled by charged youth aged 12 to 17 were different from those of adults, as was the spatial distribution of crimes involving at least one youth. Indeed, when youth crime alone was considered, it was found to be distributed in many locations spread across the entire island (Map 4).

Map 4
Spatial distribution of youth crime on the Island of Montréal, 2001

Violent crime among youth aged 12 to 17 occurred in a large number of small kernels scattered across the entire island. Several of these kernels corresponded to the location of a secondary school or, in some cases, other public institutions, such as youth centres. In 2001, 27% of violent youth crimes occurred in a school.

Shoplifting was the most frequent crime committed by youth aged 12 to 17, with the density kernels of property crime corresponding to the major shopping malls. The city centre, the Carrefour Angrignon and the Fairview Pointe-Claire shopping mall had the highest property crime densities.

The results of the multivariate analysis indicated that the characteristics of Montréal neighbourhoods had a minor impact on youth crime rates. The presence of a secondary school, commercial zoning and education were three common factors that had a slight influence on the variation of both violent and property police-reported youth crime.

Since school is where most young people spend a good part of their day, they commit many of their crimes nearby. In addition, as Tremblay and Ouimet (2001) have pointed out, the risk of assault or theft increases in places with a high density of urban movement and interaction; two such places are schools and shopping centres. These locations, particularly retail stores, also present crime opportunities for youths, whose property offences are mostly shoplifting and mischief. These findings are consistent with observations by LaGrange (1999), who noted that the presence of a secondary school or a shopping centre was a key factor having an impact on the number of mischief offences. LaGrange pointed out that such places attract a large number of non-residents to a neighbourhood, which would reduce the effectiveness of community surveillance.

Several socio-economic factors, such as the proportions of people living in low-income households, of dwellings requiring major repairs, of members of visible minorities and of people without a secondary school diploma were moderatly associated with crime, while the proportion of recent immigrants had the opposite effect. The property crime rate was also slightly increased by residential mobility but decreased by the male-female ratio and the proportion of people with a university degree.

These findings are partly consistent with the hypothesis put forward by Sampson and Raudenbush (1999) and the observations concerning Montréal made by Savoie et al. (2006) that crime varies with social capital and collective efficacy. They are also in keeping with the findings of Jacob (2006), who noted that certain neighbourhood characteristics had little influence on youth crime, but that level of education, occupation and residential instability were major factors.

Although these results reflect those of other studies of youth crime (Jacob 2006; LaGrange 1999), they nevertheless show that neighbourhood characteristics alone are not sufficient to understand youth crime, and that a better approach would be one that integrates family- and individual-level data. Some recent studies (Dupéré et al. 2007; Hay et al. 2006; Simons et al. 2005) suggest that for young people, certain neighbourhood characteristics have an influence primarily through their interaction with family- or individual-level factors. In a survey of self-reported youth delinquency in Toronto, Savoie (2007) noted that some individual and family characteristics were significant risk factors for delinquency among young people. Collecting data on victimization and self‑reported delinquency at the neighbourhood level might be particularly useful for the analysis of youth crime.

Exploration of the relationship between neighbourhood crime rates and the Aboriginal population: Regina, Saskatoon and Winnipeg

Numerous studies have shown that the Aboriginal population in Canada is over‑represented among victims and offenders (La Prairie 2002; Brzozowski, et al. 2006; Richards 2001). While it should be understood that the spatial analyses discussed in this report cannot make direct connections between Aboriginal residents of neighbourhoods and Aboriginal offenders or victims, the results do suggest that is important to understand the urban context in which Aboriginal people live. In 2001, Regina, Saskatoon and Winnipeg reported the highest proportions in Canada of Aboriginal people in census metropolitan areas, respectively 8.3%, 9.1% and 8.4% of the population.

Wallace et al. (2006) pointed out the need to identify the characteristics associated with neighbourhoods where a high proportion of Aboriginal people lived. In Regina, the neighbourhoods with higher proportions of Aboriginal people also tended to have a high unemployment rate, housing in need of repair, and high proportions of low-income households, populations receiving government transfers, low education rates, multi-family dwellings, tenants and people who had recently moved. It was shown that many of these factors were related to higher neighbourhood crime rates.

In the case of Saskatoon, the research showed that the city’s neighbourhoods differed according to several characteristics, notably socio-economic disadvantage, aging dwellings, residential mobility, the age of the population and commercial activity, but not by Aboriginal status (Charron 2008). In Saskatoon, Aboriginal people tended to live in neighbourhoods characterized by aging dwellings, socio‑economic disadvantage and residential mobility—all factors strongly associated with crime (Charron 2008). No direct relationship was found between variations in crime measured at the neighbourhood level and the Aboriginal population.

Winnipeg offered the first opportunity to examine the nature and extent of the over‑representation of Aboriginal people among police‑reported offenders at the neighbourhood level. Based on the more detailed analysis by Fitzgerald and Carrington (forthcoming in 2008 in the Canadian Journal of Criminology), Aboriginal people are almost seven times more likely than non-Aboriginal people to be identified by police as offenders. In Winnipeg, as in Regina and Saskatoon, more Aboriginal people live in high‑crime neighbourhoods. The authors determined that, together, socio‑economic disadvantage and residential mobility explained 61% of the differences in the percentages of the Aboriginal population among Winnipeg neighbourhoods. They also determined that the higher probability of Aboriginal people being identified as offenders in police‑reported data was due to the location of the offence and, more specifically, to the living conditions in the neighbourhoods in which Aboriginal people were presumed to have committed crimes. This result confirms La Prairie’s (2002) hypothesis that the structural conditions of cities contribute to the over‑representation of Aboriginal people as offenders in the Criminal Justice System.

Fitzgerald and Carrington (forthcoming in 2008 in the Canadian Journal of Criminology) concluded that a significant portion of the high police‑reported crime rate among Aboriginal people can be explained by the structural characteristics of the neighbourhoods in which they tend to live.

Variations in crime and neighbourhood characteristics over time: exploration of possibilities in Regina

An initial exploratory analysis to measure variations in crime and neighbourhood characteristics over time was included in the Regina research. While it was found that crime rates are statistically associated with certain neighbourhood characteristics, the changes in these same characteristics occur over several years. For this reason, trends in crime at the neighbourhood level can only be noticeable at a certain level (volume) and after a period of time.

The availability of data represented a major limitation in this initial exploration of change over time. An examination of changes over time in neighbourhood crime must account for changes in the populations and characteristics of neighbourhoods. This initial research was based on geocoded crime data for 1999 and 2003, and since these years were not census years, census data on populations and neighbourhood characteristics were not available. However, Statistics Canada’s Small Area and Administrative Data Division was able to provide neighbourhood population figures along with data on income categories and median income, which it drew from the annual tax data record provided by the Canada Revenue Agency. Given that in Regina, household income was found to be a key predictor of the variation in violent and property crime rates in 2001, the availability of income data is an important characteristic to explore over time.

One change over time that is likely to influence crime levels in neighbourhoods is the variation in the number of residents. The total population for the entire territory of the City of Regina fell 1% between 1999 and 2003. Demographic changes varied from neighbourhood to neighbourhood, ranging from a decline of 10% in Glen Elm to an increase of 18% in McNab. In the case of violent crime rates, it was found that a majority of neighbourhoods with violent crime rates that were higher than the overall rate for Regina in 1999 saw a decrease in their populations between 1999 and 2003 (Map 5 and 6). In their study of violent crime in Chicago between 1970 and 1990, Morenoff and Sampson (1997) observed that neighbourhoods with high homicide rates tended to experience a drop in population as residents sought to move to safer neighbourhoods.

Map 5
Kernel density distribution of violent crime incidents, Regina, 1999

Map 6
Kernel density distribution of violent crime incidents, Regina, 2003

While the multivariate analysis of the crime rates in Regina and of the neighbourhood characteristics for 2001 showed that income was a determining factor in the variations of the extent of crime among neighbourhoods, a second type of change that must be taken into account is access to economic resources. Although these same income indicators (median household income and proportion of low-income private households) were not available for 1999 and 2001, some data on personal income were.

To examine the relationship between income and neighbourhood crime rates over time, neighbourhoods were divided into two groups for each year. The first group consisted of the 25% of neighbourhoods with the highest proportion of persons reporting an income of $50,000 or more, referred to as “high‑income neighbourhoods.” The second group consisted of the other 75% of neighbourhoods, referred to as “lower‑income neighbourhoods.” Without taking other factors into account, significant differences were found between the high‑income neighbourhoods and the lower‑income neighbourhoods.

In general, high-income neighbourhoods had significantly lower property crime rates than lower-income neighbourhoods. This remained true over time. Differences between violent crime rates in high- and lower-income neighbourhoods did not remain significant over time.

The analytical work by Wallace et al. (2006) in Regina highlights the importance of having access to longer time series of criminal incidents in neighbourhoods and to multiple data sources. Furthermore, an examination of the distribution of the frequency of neighbourhood crime (hot spots) for 2001 and 2004 in all other cities studied indicated that the distribution of both property and violent crimes remained relatively stable during that period. Based on the results of the multivariate analysis conducted as part of this analytical series, it would seem essential to explore the change in socio‑economic and functional characteristics of neighbourhoods over time in order to understand the spatial distribution of crime.