picture gallery:
A typical "Vancouver Specials", a single detached house able to house 3 families.
general results
We begin this research with a hypothesis that a strong increase in multi-family houses in Vancouver will indirectly lead to an increase or change in crime pattern in Vancouver due to the increase in rental properties. We establish our hypothesis based on physical observations within Vancouver. Over the past years, we notice an increase in cars parked on the residential streets as well as increase in houses with multiple units for potential tenants. Based on these observations, we set our research goal and ask ourselves if there is a positive correlation between the multi-family houses and crimes. We carry out various steps to display the changes in crime pattern and demographic pattern in Vancouver. Overall, auto theft shows huge increase in east side Vancouver while Break & Enter reflects a relatively small decrease compared to other neighbourhoods in the west. Changes in multi-family houses and population both indicate huge increases in the east side of Vancouver. To further support our finding, we decide to perform a geographically weighted regression (GWR) within the ArcToolbox to obtain some statistical values and to get a better view on the correlation based on our results.
All in all, we produce 6 different maps from both the crime data and the census data. The maps include:
1. Change in Auto Theft Crime between 2002 - 2006
2. Change in Break & Enter between 2002 - 2006
3. Distribution and Change in Multi-Family Houses in Vancouver (2001-2006)
4. Distribution and Change in Rental Properties in Vancouver (2001-2006)
5. Distribution and Change in Population in Vancouver (2001-2006)
6. Crime Rate based on the number of Private Dwellings in Vancouver (2006)
change in crime from 2002-2006
Map showing the change in Auto Theft in Vancouver . click to enlarge
The above two maps are produced using the crime data that are obtained from the Vancouver Police Department. The datasets are entered into an excel spreadsheet to calculate the percentage change of crime in the various neighbourhoods in Vancouver. We focus on both the "Residential Break and Enter" as well as the "Auto Theft" data to compute both maps. On the top map, we are able to effectively present the areas that show increases in auto theft. From our results, we note that within the 22 neighbourhoods, 11 of them indicate an increase in auto theft while the rest indicate decrease. The majority neighbourhoods in the "West" show a decrease in auto theft except for Point Grey and Dunbar Southland. We suspect this is due to the large amount of commercial districts within the neighbourhoods. Further observation on the maps shows that many neighbourhoods in the east side of Vancouver (such as Killarney and Sunset) have a strong increase in the number of auto theft.
On the bottom map, we are able to clearly identify areas that have large increases in break and enter crimes from year 2002 to 2006. Namely, Marpole is the only neighbourhood that has experienced a strong increase in Break and Enter beside Downtown Eastside. However, one thing of note is the small decrease in break and enter crimes within the east side of Vancouver. Meanwhile, the neighbourhood "Killarney" is especially significant as it only shows a small decrease (less than 5%) in break and enter crimes within the 4 years while some of the neighbourhoods display a decrease of over 30% during the same period. All in all, we find an apparent trend in the map where neighbourhoods in the northwest side of Vancouver have experienced a large decrease in break and enter crimes when comparing to the east.
distribution and change in multi-family houses
The above map is created as it effectively displays the distribution of multi-family houses as well as the changes in multi-family houses across Vancouver. The "hot spot" of multi-family houses are concentrated in Victoria-Fraserview, a very large neighbourhood within the east side of Vancouver, with Killarney, Sunset, and Renfrew-Collingwood being the following neighbourhoods that contain the most multi-family houses in Vancouver. In addition, the results from the map show the neighbourhoods that reflect increases in multi-family houses. Namely, the two neighbourhoods in the southeast side of Vancouver show the most increases in multi-family houses (up to a maximum of over 300 houses within the 4 years period). These two neighbourhoods are very significant because the majority of the neighbourhoods in Vancouver have a decrease or a small increase in the number of multi-family houses within the same period.
distribution and change in rental properties
In addition to looking at the changes in multi-family houses in Vancouver, we want to look at the change in the distribution of rental properties within Vancouver. Except for the hot spot in Downtown Vancouver, the result from the above map shows a correlation between the distribution of multi-family houses and crime. Neighbourhoods in the southeast side have experienced the strongest increase in the number of rental properties. This can then relates back to the increase in multi-family houses from the previous map where the increase in multi-family houses leads to generating more rental properties for potential tenants.
distribution and change in population
Furthermore, we have assessed the distribution of population in Vancouver between the years 2001 and 2006. Because population has a strong correlation to the occurrence of crimes, studying the distribution of population allows us to determine the probability on the occurrence of crimes. According to the above map, the change in population is shown in the choropleth map while the population counts are shown by proportional symbols. By using these methods, we are able to produce a map that shows the trend in settlement and movement as well as the direct count of population in each neighbourhood. Similarly, except for the hot spot in downtown, both the changes in population and population counts are at their greatest in the southeast of Vancouver. We assume that population is positively related to the increase in multi-family houses which are then led to the increase in available rental properties. All in all, a trend that shows population moving from the west to the east is presented and is assumed to be due to the increases in multi-family houses.
Crime rate based on the number of private dwellings in vancouver (2006)
After looking at all the distributions of crimes and various census variables, we decide to create a map that shows the linkage between crime rate and the number of houses in each neighbourhood. Data that we use for the above map are from 2006 Census. The data are used in conjunction with the break and enter crime data of 2006 to compute the crime rate per 1000 houses in Vancouver. Referring to the above map, the number of break and enters per 1000 houses ranges from 14 to over 47 in Vancouver. In other words, a probability of up to 47 houses from 1000 houses within a neighbourhood may experience break and enter within that year. Similarly, a trend from west to east is apparent in the above map where neighbourhoods in the west (such as West Point Grey and Arbutus-Ridge) have experienced the lowest range of crime rates, and neighbourhoods in the east (such as Renfrew-Collingwood and Sunset) have experienced some of the highest range of crime rates in our study. Even though Victoria-Fraserview and Killarney have maintained a relatively low crime rates as shown on the map, we believe the reason is that these neighbourhoods have experienced an increase in multi-family houses while neighbourhoods such as Sunset and Renfrew-Collingwood have a substantial amount of multi-family houses prior to year 2006.
Geographically weighted regression analysis
After performing visual comparison and correlation between the different demographic and housing trends with the crime counts data, we perform a regression analysis using the Geographically Weighted Regression (GWR) tool within ArcToolbox in GIS. Regression analyses are always effective in dealing with a measurable dependent variable and several other independent variables in each location (often known as explanatory variables). Within our research, explanatory variables are defined as population, percentage of multi-family houses, and percentage of rental properties while crime count is used as the dependent variable. By performing GWR using these variables, we are able to obtain results that show the predicted and observed values of crime counts that are used to compute their differences. Furthermore, GWR allows us to obtain the R-Squared value which tells us the percentage of dependent variable that can be explained by the explanatory variables. The various R-Squared values computed using GWR are being displayed in the following table.
In general, we are able to obtain a range of R-Squared value using different explanatory variables with auto theft or break and enter crime counts as the dependent variables. A range of .31 to .65 for R-Squared value is obtained for the GWR analysis. According to our results, many of the auto theft crimes (up to 59%) can be explained by one of our three explanatory variables, while a maximum of 55% break and enter crimes can be achieved using the same variables. Even though the R-Squared values are not significantly high, we suspect this is due to possible errors within the census data. On the other hand, we have computed graphs showing the correlation with percentage of rental properties and population count versus the number of crimes (such as auto theft and break & enter). These graphs are shown below.
Graphs showing correlation between % of rental properties and crimes
Referring to the graphs above, we find that there is a strong correlation between the percentage of rental properties and population with crimes. The graphs are able to show us both the increases in population as well as rental properties can lead to an increase in crime counts. These graphs are significant because the number of people that reside within each neighbourhood depends on the housing trend. In addition, Andresen has noted in his study that “a greater number of people implies a greater number of both potential offenders and potential victims” (Andresen, 2005:258). If the number of multi-family houses continues to rise in the future, the numbers of rental properties and population increase correspondingly.