Methodology
The study area for all analyses was the City of Vancouver; suburbs were not included. All socio-economic data was acquired at the dissemination level from Statistics Canada as part of the National Household Survey of 2011. Skytrain data was obtained from the City of Vancouver. Centroids were derived from the dissemination areas which were then used to calculate the distance to the nearest Skytrain station. A total of 8 independent variables were used and inputted into an exploratory regression analysis, including the nearest distance measure. The details are listed below.
| Description | Variable | Mean | Standard Deviation | Minimum | Maximum |
| Dependent Variable | |||||
| Median Value of Dwellings (includes value of land and dwelling, in $) | VALUE | 847,815.16 | 554,832.35 | 31.12 | 4,004,171.00 |
| Independent Variables | |||||
| Average number of rooms per dwelling (includes kitchens, bedrooms, and finished rooms in basement and attic) | ROOMS | 5.39 | 1.77 | 1.50 | 11.10 |
| Median total income, 15 years and over, in 2010, in $ | INCOME | 28,661.62 | 10,900.83 | 0.00 | 74,546.00 |
| East or west side (dummy variable using Main Street; East = 1) | EAST | 0.49 | 0.50 | 0.00 | 1.00 |
| Proportion major repairs needed on dwelling | REPAIRS | 0.06 | 0.08 | 0.00 | 0.54 |
| Distance to nearest Skytrain station (metres) | NEAR_DIST | 1,902.39 | 1,597.72 | 31.12 | 9,655.32 |
| Proportion constructed from 2001-2011 | NEWCONST | 0.08 | 0.15 | 0.00 | 0.99 |
| Proportion part of a condominium development | CONDO | 0.21 | 0.29 | 0.00 | 1.06 |
| Proportion of pop. with postsecondary certificate, diploma, or degree | EDUC | 0.62 | 0.15 | 0.00 | 0.98 |
For the exploratory regression analysis, I decided to set a maximum of five explanatory variables; I realized if I were to include any more variables, the results would not necessarily get better. Here are the results with the associated R2 and AICc scores:
| Top results, 1 variable | ||||||||
| AdjR2 | AICc | VIF | Model | |||||
| 0.62 | 28469.07 | 1.00 | +ROOMS*** | |||||
| 0.16 | 29270.16 | 1.00 | +NEAR_DIST*** | |||||
| 0.13 | 29298.74 | 1.00 | -CONDO*** | |||||
| Top results, 2 variables | ||||||||
| AdjR2 | AICc | VIF | Model | |||||
| 0.72 | 28150.46 | 1.00 | -EAST*** +ROOMS*** | |||||
| 0.68 | 28303.83 | 1.02 | +EDUC*** +ROOMS*** | |||||
| 0.65 | 28400.96 | 1.11 | +NEAR_DIST*** +ROOMS*** | |||||
| Top results, 3 variables | ||||||||
| AdjR2 | AICc | VIF | Model | |||||
| 0.73 | 28131.27 | 1.19 | +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
| 0.73 | 28145.16 | 1.55 | -CONDO*** -EAST*** +ROOMS*** | |||||
| 0.72 | 28148.40 | 1.86 | +EDUC** -EAST*** +ROOMS*** | |||||
| Top results, 4 variables | ||||||||
| AdjR2 | AICc | VIF | Model | |||||
| 0.73 | 28127.01 | 1.55 | -CONDO*** +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
| 0.73 | 28130.73 | 1.19 | -NEWCONST* +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
| 0.73 | 28130.87 | 1.89 | +EDUC +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
| Top results, 5 variables | ||||||||
| AdjR2 | AICc | VIF | Model | |||||
| 0.73 | 28124.54 | 1.97 | +EDUC** -CONDO*** +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
| 0.73 | 28126.50 | 1.65 | -CONDO*** +NEAR_DIST*** -EAST*** +INCOME +ROOMS*** | |||||
| 0.73 | 28128.69 | 1.91 | -CONDO** -NEWCONST +NEAR_DIST*** -EAST*** +ROOMS*** | |||||
I then proceeded to use the best model as determined by exploratory regression (5 variables) which included EDUC, CONDO, NEAR_DIST, EAST, and ROOMS. For the results of the actual regression, continue onto the Results & Analysis section.