Home

Results

Regression analyses were performed to study the strength of the relationships between the number of residential burglaries and the factors described in the Methods section. The regression graphs indicated that there was a cluster of four census tracts on the western coast (tracts 7600, 7700, 7901, and 7904) that consistently had much higher numbers of burglaries than would be expected from the best fit lines. With the exception of census tract 7700 in the housing unit analyses, their residual values were consistently greater than 2.5 standard deviations above the mean. The number of burglaries in these census tracts seem to be influenced by, or associated with factors other than those that we investigated.

click on each image to enlarge

Total Population map: the total population in each census tract was compared to the numbers of burglaries

Total population and Burglaries correlation

Total population and burglaries regression

A regression analysis of the census tract populations and the number of burglaries produced a best fit equation of

[number of burglaries] = 2.6379 + 0.0029[total population],
or (y= 0.0029x +2.6379).

This suggests a general trend where census tracts with larger populations have a higher prevalence of residential burglaries. The correlation coefficient R² was 0.155163 which is a relatively low value, reflecting the large number of residuals that do not closely follow the trend, especially the four census tracts for which there are unexpectedly many burglaries relative to their population.

Income map: median household income versus number of burglaries in each census tract

The number of residential burglaries was also compared to the median household income per census tracts. The regression produced a best fit line of

[number of burglaries] = 22.6125 - 0.0001[median household income],

or y = -0.0001x + 22.6125.

This shows a decreasing number of residential burglaries as the median household income per census tracts increases, such that higher income neighbourhoods appear to experience fewer burglaries. The R² was 0.08236, which is an even lower value than that found for the comparison of population and burglaries.

Owners versus Renters map: the percentage of each category out of total occupants per census tract are compared to number of burglaries.

Owners

Percentage of burglaries compared with Owners

Regressioin relationship map of Owners and Burlaries numbers

We performed a regression analysis of the number of residential burglaries in a census tract relative to the number of owner-occupied residences. The equation of the best fit line was:

[number of burglaries] = -0.0011[number of owner occupants] + 18.3475,

or y = -0.0011x + 18.3475.

The small negative slope value suggested that there was a decreasing number of burglaries as the number of owner occupants increases. However, the R² value we obtained was 0.016607, indicating that there was not a strong correlation between the two variables.

Renters

Regression Graph of Renters and Burglaries

Another regression analysis was performed on the number of residential burglaries relative to the number of renter occupied residences. The equation for the best fit line was:

[number of burglaries] = 0.005[number of renter occupants] + 5.5043, or y = 0.005x + 5.5043.

The slope of the equation was small but positive, suggesting that the number of burglaries increases with the number of renters in a census tract. The R² value was 0.373826, a substantially larger value than the R² found in the previous owner occupants analysis.

Housing Units: Single Unit versus Multiple Units (apartment, condos) per Census Tract Compared to number of burglaries

Single Unit

Correlation between Single unit and Burglaries

Regression graph One unit and Burglaries

This regression analysis was performed for the number of residential burglaries in relation to the number of single unit homes. The equation for the best fit line was:

[number of burglaries] = 0.0021[single unit homes] + 13.7976, or y = 0.0021x + 13.7976.

The slope of the equation was quite small, but positive, which suggests that the number of burglaries would increase when the number of single unit homes increases. However, the R² value was 0.00826, indicating a weak correlation between the two variables.

Multiple Units

Correlation between multiple units and burglaries

Regression graph two units and burglaries

The correlation of multiple-unit homes and residential burglaries produced an equation of

[number of burglaries] = 7.5863 + 0.0115[number of multiple-unit homes],
or y= 0.0115x + 7.5863.

The R² value was 0.42887, the largest value out of all of our analyses, suggesting that the prevalence of multiple-unit home such as apartments is the factor most closely associated with the number of residential burglaries.