Exploring the Link between Crime and Socio-Economic Status in Ottawa and Saskatoon: A Small-Area Geographical Analysis

7. Study # 2 Findings: Neighbourhoods of Saskatoon

7. Study # 2 Findings: Neighbourhoods of Saskatoon

7.1 Descriptive Statistics

Table 7.1 (p.84) displays the descriptive statistics for the 31 variables used in the study. The five crime variables were calculated as a rate per 1,000 population in each of Saskatoon's 55 residential neighbourhoods, a standard procedure in crime analysis. The remaining 26 census and development/planning variables were calculated at the ratio scale with the exception of two, average value of dwelling ($DWELLING) and average selling price of a dwelling (AVG_SELL_PRICE) which were left at the interval scale. The table shows that several of the variables, including those related to crime, have high coefficients of variation (the standard deviation divided by the mean) indicating greater dispersion of individual values around the mean. In particular, violent offences and drug offences have values that fluctuate widely among the 55 neighbourhoods, suggesting significant geographic disparity within Saskatoon. As very few of the 31 variables have "approximate" normal distributions coupled with the fact that they originate from different sources (police data, census indicators, development/planning information) each of the 31 variables was transformed into a Z-score for every neighbourhood in the dataset (55).

7.2 - Principal Components Analysis

Table 7.2 shows that the PCA produced a 6-component solution, accounting for more than 84% of the total variance in the dataset, a large proportion indicating a significant degree of inter-correlation amongst the variables. Component 1 alone accounts for just under 47% of the total variance. An examination of the component loadings in Table 7.3 reveals that there is a positive association between crime (particularly violent and major property crimes) and socio-economic disadvantage in Saskatoon (Component 1). The characteristics of disadvantaged neighbourhoods are marked by high loadings on variables relating to mobility (SINGLE, MOVERS), ethnicity/family (ABORIGINAL, LONE_PARENT), education (NO_HS_DIP), income (GOVT_TRANSFER, LOW_INC_FAM), housing (DWEL_RENTED, MAJOR_REPAIRS) and labour force activity (UNEMP, YOUTH_UNEMP). A closer scrutiny of the table reveals that 3 variables on Component 1 have high loadings above 0.9 (ABORIGINAL, LONE_PARENT and LOW_INC_FAM) pointing to the presence of a particularly vulnerable group of people in Saskatoon with respect to both social deprivation and crime, namely Aboriginal single mothers living in low-income. Component 2 isolates the incidence of minor property and drugs offences in neighbourhoods with apartment high-rises and lower labour force participation. The remaining components identify several dimensions of socio-economic status and land-use characteristics in the city, including "park space" (Component 3), "housing value and education" (Component 4), and "density and old housing" (Component 5).

7.3 Multiple Regression Analysis

A step-wise multiple regression was performed on the Saskatoon neighbourhood dataset. Table 7.4 shows the results of the regression analysis between each of the 5 crime variables and a selection of neighbourhood indicators. When using SPSS for Windows for this procedure, tests were run to measure the level of multicollinearity among the independent variables. Each independent variable in the 5 regression models had tolerance values close to 0.1 and variance inflation factors (VIF) well below the rule-of-thumb value of 5 (Rogerson 2001, p.136). In fact, all were below 2 indicating low multicollinearity thus strengthening the predictive value of the regression models.

The table clearly demonstrates that there is a strong association between crime and neighbourhood characteristics in Saskatoon with each of the crime variables recording fairly high coefficients of multiple determination (R2) and adjusted R2. Regression model # 1 shows that four independent variables, 'youth not at school' (YOUTH_NO_SC), 'population that is single' (SINGLE), 'level of social assistance' (GOVT_TRANSFER) and 'apartment high-rises' (APT_HIGH_RISE), when taken together, account for nearly 70% (adjusted R2 = .666) of the variation of total crime in Saskatoon's 55 residential neighbourhoods. The variable GOVT_TRANSFER recorded the highest beta coefficient (.339) among the four independent variables indicating that the level of social assistance in a community (in the form of residents who require welfare and other benefits as a major source of their income) can be considered as a significant predictor of crime. As can be expected, the need for social assistance is clearly associated with socio-economic disadvantage and is related to factors such as low-income, unemployment, lower educational attainment, and rented housing.

Regression model # 2 points very clearly and directly to a troublesome association between violent crime and Aboriginal people in Saskatoon. The two independent variables explain more than 70% (adjusted R2 = .715) of the variation in violent crime in the city's neighbourhoods with ABORIGINAL standing out as the key predictor recording a beta coefficient of .651. As seen in Table 4.5 above, assaults accounted for about 75% of violent crimes in Saskatoon in 2003. These results suggest that it is primarily Aboriginal people who are at risk of being victims of physical violence.

Regression model # 3 reveals that the most significant predictors of major property crime in Saskatoon are average home selling price (AVG_SELL_PRICE); people who have recently moved (MOVERS) and the presence of older homes (OLD_HOUSE). In other words, older neighbourhoods with low real-estate values appear to be vulnerable to major property offences particularly residential break and enter and motor vehicle theft. Regression model # 4, which measures minor property crime, has a slightly smaller adjusted R2 value of .596 and identifies apartment high-rises (APT_HIGH_RISE), youth who are not at school (YOUTH_NO_SC) and dwellings that are rented (DWEL_RENTED) as the most important predictors. Offences related to minor theft (under $5000) and mischief are especially prevalent in neighbourhoods with these characteristics. Finally, drug offences in Saskatoon (Regression model # 5) appear to be influenced by the presence of apartment high-rises (APT_HIGH_RISE), low-income families (LOW_INC_FAM) and youth who are not at school (YOUTH_NO_SC). It should be noted, however, that drug offences accounted for less than 2% of all incidents in Saskatoon in 2003 (Table 4.5).

7.4 Mapping and GIS Analysis: Identifying Spatial Patterns of Crime and Neighbourhood Characteristics

ArcGIS was used to create a series of maps displaying the geographic distribution of criminal offences and selected neighbourhood characteristics in the City of Saskatoon. A classification scheme was devised to rank neighbourhoods according to their crime rates per 1,000 population:

  1. Lowest
  2. Low
  3. Moderate
  4. High
  5. Highest

For example, as can be seen in Figure 7.2, the classification for 'Total Offences' is as follows with the numbers in parentheses denoting the crime rate:

  1. Lowest (0-35)
  2. Low (35-100);
  3. Moderate (100-200)
  4. High (200-300)
  5. Highest (300-600)

For the purposes of this study, High Crime Areas (HCAs) are defined as any neighbourhood that is placed in the 'High' or 'Highest' categories. This classification scheme was adjusted to correspond with the crime rates for each of the four remaining offence types.

The map in Figure 7.2 shows that there is a noticeable presence of 'Moderate' and HCAs in the south and central sections of Saskatoon, particularly on the west side of the South Saskatchewan River. In fact, all of the HCAs are located there and most are clustered in and around the inner city including the 4 with the 'Highest' crime rates in 2003 – the Central Business District (546), Pleasant Hill (450), Kelsey-Woodlawn (374) and Riversdale (347). These 4 neighbourhoods are adjacent or very close to several other HCAs including Caswell Hill (296), King George (243), Westmount (234), and Mayfair (209). The two remaining HCAs are Confederation Park (227) located on the western edge of Saskatoon and the Airport Business Area (281) situated in the northwestern section of the city. While there is clearly a presence of elevated crime rates in the western part of the city, Figure 7.2 also reveals an interesting pattern of neighbourhoods with 'Moderate' crime rates adjacent to 8th Street, a major commercial thoroughfare on the east side of the South Saskatchewan River, pointing to a possible geographical association between crime and mobility/accessibility. These neighbourhoods include Nutana (150), Buena Vista (120), Varsity View (121), Grosvenor Park (147), Greystone Heights (104) and Brevoort Park (162).

Figure 7.3 is a map illustrating the distribution of violent offences in Saskatoon and it is immediately apparent that these crimes are far more concentrated in the west side of the city with all HCAs and all but 3 'Moderate' crime areas located there. Similar to total offences, there is a noticeably tight clustering of HCAs in and around the inner city including the 3 neighbourhoods with the highest violent crime rates – Pleasant Hill (131), Riversdale (105) and the Central Business District (85). Adjacent to these are other violent HCAs including King George (43), Meadowgreen (35), Mont Royal (36), Westmount (46) Caswell Hill (41); further to the west, Confederation Suburban Centre (36) and Confederation Park (38) and just to the north, Kelsey-Woodlawn (58) and the Airport Business Area (54).

'Major Property' offences (Figure 7.4) display a similar pattern as 'Total Offences' with a cluster of HCAs visible in and around the core area and 'Moderate' and 'Low' crime areas evident in the southeast portion of Saskatoon. While 'Minor Property' offences (Figure 7.5) have a somewhat more dispersed pattern, all the HCAs are again located in the west side of the city. However, compared to 'Major Property' offences, a number of suburban residential neighbourhoods have slightly higher rates of minor property crimes including several in the southeast (Wildwood, Lakeview, Lakeridge and Briarwood), in the northeast (Sutherland, Forest Grove, Erindale, Arbor Creek, and Silverspring) and just to the northwest of the South Saskatchewan River (River Heights, Lawson Heights and Silverwood Heights). Finally, while 'Drugs' accounted for less than 2% of all offences in 2003, these crimes display a similarly dispersed pattern (Figure 7.6).

Figure 7.8 is a map showing the geographic distribution of low-income families in Saskatoon. Due to the serious socio-economic problems related to families living in low-income, including higher unemployment, lower educational attainment and greater dependency on social assistance, it is appropriate to use this census indicator as composite measure of disadvantage. For the objectives of this study, a 'disadvantaged' area is defined as any neighbourhood having more than 20% of its families living in low-income. The map indicates that a socio-economic divide does exist in Saskatoon with 13 of the 16 'disadvantaged' neighbourhoods located in a relatively tight cluster in the west side of the city, particularly in and surrounding the core area. Several of these neighbourhoods have strikingly high rates of low-income families including the Airport Business Area (62.5%), Confederation Suburban Centre (57%), Pleasant Hill (57%) and Riversdale (51%). Other 'disadvantaged' neighbourhoods on the west side of the river include King George (34%), Holiday Park (21%), Meadowgreen (35%), Westmount (37%), Caswell Hill (26%), Kelsey-Woodlawn (29%), Mayfair (26%), Massey Place (27%) and Confederation Park (21%). While the east side of the river is generally more prosperous, there are 3 suburban neighbourhoods that can be classified as 'disadvantaged'- Grosvenor Park (20%), Nutana Suburban Centre (21%) and Sutherland (21%). It is evident that there is a geographical association between disadvantaged neighbourhoods and crime in Saskatoon. An inspection of the maps in Figures 7.2 and 7.7 indicates that 9 of the 16 disadvantaged neighbourhoods are also High Crime Areas (HCAs) and conversely that all but one of the HCAs are disadvantaged (see table below).

Neighbourhood % Low-Income Families Total Crime Rate
Airport Business Area 62 282
Pleasant Hill 56 450
Riversdale 51 348
Westmount 37 235
King George 34 244
Kelsey-Woodlawn 29 374
Mayfair 26 209
Caswell Hill 26 296
Confederation Park 21 228

As stated, Aboriginal people in Western Canada, including those living in urban areas, have been found to experience greater levels of socio-economic disadvantage and to have greater contact with the justice system, particularly as victims but also as offenders (La Prairie 2002). However, it is important to point out that a study conducted by the Canadian Centre for Justice Statistics (2000) found that in Saskatchewan, crime rates on reserves were double those in urban and rural areas. According to the 2001 Census, Saskatoon has a sizeable Aboriginal population (20,275) comprising 9% of the city's total. Between 1996 and 2001, the Aboriginal population grew by 25.5% compared to 1% for non-Aboriginals. Overall, Saskatoon has a relatively strong economy and in 2001 recorded an unemployment rate of 6.7%. However, Aboriginal residents have fared poorly in the labour market with a jobless rate of 22%. Even more striking is unemployment among North American Indian males at 33%. Furthermore, the 2001 Census shows that the average income in Saskatoon is $28,045 while for Aboriginal peoples it is $17,667 and for North American Indians $14,513. These statistics point to a high level of socio-economic disadvantage among the city's Aboriginal residents.

Figure 7.8 is a map showing the distribution of Aboriginal residents in Saskatoon's neighbourhoods. While it is clearly visible that there are more Aboriginal people living on the west side of the river, it is important to point out that there is not a high level of segregation in the city. A reasonable measure of ethnic segregation is when more than 30% of a neighbourhood's population is made up of one particular group of people sharing an identifiable characteristic such as race or ancestry. As can be seen Figure 7.8, only 3 neighbourhoods have more than 30% Aboriginal residents - Pleasant Hill (48%), Riversdale (43.5%) and Confederation Suburban Centre (37%). In addition to these, just 6 other neighbourhoods have 20% or more Aboriginal residents – Meadowgreen (28%), Airport Business Area (27%), Westmount (23%), Caswell Hill (25.5%), Massey Place (21%) and Mayfair (20%). It is clear, then, that the majority of residents in all of Saskatoon's neighbourhoods are non-Aboriginal suggesting that issues of victimization, crime and socio-economic status impact a much wider segment of city's population. However, the regression analysis (Table 7.4) indicates that there is a strong relationship between Aboriginal people and violent crime in Saskatoon. An examination of the maps in Figures 7.3 and 7.8 reveals that 7 of the 9 neighbourhoods with Aboriginal populations greater than 20% are also violent High Crime Areas (see table below).

Neighbourhood % Aboriginal Violent Crime Rate
Pleasant Hill 48.4 131
Riversdale 43.5 105
Confederation SC 37.4 36
Meadowgreen 28.0 35.5
Airport Business Area 26.7 54
Westmount 22.8 46
Caswell Hill 21.5 41.5

The two exceptions are Mayfair, located just north of the city's core, which is a 'Moderate' violent crime area and Massey Place, just west of the core, which is a 'Low' violent crime area.

7.5 Spatial Autocorrelation

Table 7.5 shows that each of the 5 crime variables has a calculated Moran's I value greater than 0 and a significant Z value (at the 99% confidence interval) indicating positive spatial autocorrelation. The most spatially concentrated crimes in Saskatoon are major property offences (I = 0.188, Z = 7.90) followed by violent offences (I = 0.125, Z = 5.49) and total offences (an aggregate variable) (I = 0.118, Z = 5.20) meaning that neighbourhoods with high rates of crime tend to be located close to other neighbourhoods with high rates of crime. The least spatially concentrated or most dispersed crimes are minor property offences (I =0.065, Z=3.21) and drug offences (I=0.064, Z=3.15). The table also indicates that low-income families (I=0.112, Z=4.99) and Aboriginal residents (I=0.185, Z=7.78) are also geographically highly concentrated in Saskatoon compared to the distribution of the population as a whole (I=0.011, Z=1.13).

Moran's I was re-calculated for each variable using the 'adjustment for small distances' option in CrimeStat, in which the distance weights between two locations Wij can never be greater than 1 mile. This ensures that "I" won't be excessively large for neighbourhoods (as represented by points) that are close together or adjacent. As shown in Table 7.5, while the adjusted Moran's I values are smaller than the original values, the first three crime variables (total, violent and major property) are positively and significantly spatially autocorrelated as are low-income families and Aboriginal residents.