Introduction
Methodological literature review
Research methodology
Study area
Datasets and selection of variables
Dimension | Indicator | Description | Rationale | Status | Data source(s) |
---|---|---|---|---|---|
Socio-economic (N = 15) | V1: Population density | Dividing the total number of people by the total land area (km2) | Population density can be associated with high rates of violent crime in urban areas [40] | 5 | 1 |
V2: Average household income | Average after-tax income of households ($) | Low income and income poverty can play an important role in the occurrence of violent behaviour and crime [37] | 3 | 1 | |
V3: Unemployment rate | Unemployed population/total population in the labour force aged 15 years and over ×100 | There is an association between unemployment rates and the occurrence of violent behaviour, such as homicide [36] | 4 | 1 | |
V4: Rate of adults lacking tertiary education | Population lacking tertiary education/total population aged 15 years and over ×100 | Lack of tertiary education can associate with many crimes, including violent ones and homicide [31] | 2 | 1 | |
V5: Visible minority rate | Total visible minority population/total population×100 | 3 | 1 | ||
V6: Sex ratio | Total number of males/total number of females×100 | Evolutionary behavioural models suggest that when the sex ratio is high (more available men than women), violence against women is more likely to occur [81] | 4 | 1 | |
V7: Residential instability | This measure refers to area-level concentrations of people who experience high rates of family or housing instability, weighted average residential instability score - higher values mean more instability | Social disorganization theorists argue that residential instability can associate with the local violence crime rate by disrupting residential networks that are protective factors against crime [18] | 2 | 2 | |
V8: Material deprivation | Material deprivation is closely connected to poverty and it refers to inability for individuals and communities to access and attain basic material needs. The indicators included in this dimension measure quality of housing, educational attainment and family structure characteristics [82]. Weighted average residential instability score – higher values mean more instability | Some studies have shown that homicide rates were higher in urban areas with higher material deprivation [83] | 5 | 2 | |
V9: Ethnic concentration | Proportion of the population who self-identify as a visible minority, weighted average material deprivation score – higher values mean more deprivation | Some studies revealed that ethnic concentration exhibits a significantly positive but spatially different association with violent crime rates [30] | 2 | 2 | |
V10: Dependency ratio | Dependency ratio (total population 0-14 and 65+ / total population 15 to 64), weighted average dependency score – higher values mean more dependency | 1 | 2 | ||
V11: Mobility status | Mobility status 5 years ago – 25% sample data= total movers/total population × 100 | 1 | 1 | ||
V12: Youth rate | Youth 15-34 years old/total population×100 | 3 | 1 | ||
V13; Rate of rented homes | Total number of renter households/total number of private households×100 | The highest crime rates are in neighbour-hoods where a significant portion of all homes are rented [32] | 3 | 1 | |
V14: Rate of homes needing major repairs | The number of private households whose dwellings are in need of major repairs/total number of private households×100 | 3 | 1 | ||
V15: Unsuitable house rate | Total number of private households who are living in unsuitable accommodations /Total number of private households×100 | Poor housing condition is a potential risk factor for crimes and may be associated with areas with higher crime rates [89] | 3 | 2 | |
Built-environment (N = 10) | V16: Property units | The total number of property units/total land area (km2) | 3 | 3 | |
V17: Commercial establishments | The total number of commercial places/total land area (km2) | The rate of violent crimes, especially property theft, is higher in commercial spaces than in other spaces and may associate with homicide [41] | 5 | 3 | |
V18: Sport places | The total number of sport places/total land area (km2) | 1 | 2 | ||
V19: Places of interest | The total number of places of interest/total land area (km2) | 3 | 3 | ||
V20: Intersections | Dividing the total number of road intersections by total land area (km2) | Intersections provide opportunities for death by shooting, intentional car crashes or during escapes from crime scenes [44] | 3 | 3 | |
V21: Public secondary schools | The total number of public secondary school locations/total land area (km2) | 3 | 3 | ||
V22: Large buildings | The total number of buildings that includes >5 independent homes/total land area (km2) | 5 | 3 | ||
V23: Parking lots | The total number of parking lots/total land area (km2) | Some studies have shown that the incidence of violent crimes, such as homicide, is higher in certain places such as parking lots [96] | 2 | 3 | |
V24: Subway stations | The total number of subway stations/total land area (km2) | 1 | 3 | ||
V25: Public parks | The total number of municipality public parks/total land area (km2) | 1 | 3 |
Data analysis
Linear and geographically weighted regression
Results
Temporal clusters
Spatial and spatio-temporal clusters
Pearson’s correlation, ER and OLS model
GWR model results
Variable | Bandwidth | Mean | STD | Minimum | Median | Maximum |
---|---|---|---|---|---|---|
Intercept | 123 | -0.027 | 0.060 | -0.146 | -0.046 | 0.075 |
Population density | 123 | -0.267 | 0.011 | -0.300 | -0.265 | -0.245 |
Material deprivation | 123 | 0.424 | 0.083 | 0.311 | 0.437 | 0.555 |
Commercial establishments | 123 | 0.350 | 0.062 | 0.251 | 0.333 | 0.526 |
Large buildings | 123 | 0.401 | 0.042 | 0.288 | 0.404 | 0.480 |
Diagnostic name | Value | Value | |
---|---|---|---|
Residual sum of squares | 64.627 | AICc | 309.528 |
Effective number of parameters (trace (S)) | 8.459 | BIC | 335.827 |
Degree of freedom (n – trace (S)) | 131.541 | R2 | 0.538 |
Sigma estimate | 0.701 | Adj. R2 | 0.508 |
Log-likelihood | -144.541 | Adj. alpha (95%) | 0.030 |
Degree of Dependency (DoD) | 0.894 | Adj. critical t value (95%) | 2.199 |
AIC | 308.000 | - |
MGWR model results
Variable | Bandwidth | Mean | STD | Min | Median | Max | Monte Carlo test |
---|---|---|---|---|---|---|---|
Intercept | 139 | 0.006 | 0.012 | -0.030 | 0.007 | 0.030 | 0.840 |
Population density | 139 | -0.254 | 0.005 | -0.272 | -0.252 | -0.249 | 0.905 |
Material deprivation | 123 | 0.415 | 0.079 | 0.301 | 0.427 | 0.522 | 0.143 |
Commercial establishments | 70 | 0.375 | 0.247 | 0.068 | 0.301 | 0.036 | 0.002 |
Large buildings | 139 | 0.430 | 0.009 | 0.409 | 0.432 | 0.447 | 0.905 |
Diagnostic name | Value | Value | |
---|---|---|---|
Residual sum of squares | 61.667 | AICc | 305.235 |
Effective number of parameters (trace (S)) | 9.432 | BIC | 334.066 |
Degree of freedom (n – trace (S)) | 130.568 | R2 | 0.560 |
Sigma estimate | 0.687 | Adj. R2 | 0.527 |
Log-likelihood | -141.259 | - | - |
Degree of Dependency (DoD) | 0.872 | - | - |
AIC | 303.380 | - | - |
Model | AIC | AICc | R2 | Adj. R2 | Increased Adj. R2(%) |
---|---|---|---|---|---|
OLS | 1027.85 | 1028.95 | 0.526 | 0.505 | - |
GWR | 309.528 | 335.827 | 0.538 | 0.508 | 0.003= 0.6% |
MGWR | 303.380 | 305.235 | 0.560 | 0.527 | 0.022=4.35%, 0.019=3.74% |