For my Professional Report (PR), I am developing an evaluation framework to facilitate the incorporation of equity into transit project performance metrics. The evaluation framework links project questions and goals with relevant indicators and data sources. The indicators are grouped into broad themes, such as the Transportation System, Transportation Outcomes, Transportation and Land Use, and the Planning Process.
In addition to the evaluation framework, I will build web-based tools, such as Python functions and web maps, to calculate and visualize metrics relatively quickly. The tools are intended to enable more meaningful community participation as web-based metrics are more interactive and engaging than static reports.
Today’s maps fall under the Transportation Outcomes theme. The first map looks at the relationship between the location of bicycle and pedestrian collisions and the percentage of households that do not own a car in Alameda and Contra Costa Counties. My hypothesis is that census tracts where the majority of households do not own a car will have more collisions that involve bicyclists and pedestrians because more people are bicycling, walking, and walking to/from transit, which means there are more opportunities for collisions to occur.
To create this map, I acquired a shapefile of census tracts for Alameda and Contra Costa Counties from the TIGER/Line Database provided by the U.S. Census Bureau. I downloaded a table of Household Vehicle Ownership for Alameda and Contra Costa Counties from the American Community Survey (ACS) provided by the U.S. Census Bureau. I joined the ACS table to the census tracts shapefile to visualize the percent of households that do not own a vehicle. Lastly, I downloaded collision data – filtered to bicyclist and pedestrian-involved collisions – from UC Berkeley’s Transportation Injury Mapping System (TIMS) Database; the TIMS data geocodes the California Highway Patrol’s Statewide Integrated Traffic Records System (SWITRS).
In the map below, the darker shaded census tracts have a higher percentage of households that do not own a vehicle. In Contra Costa County, most of the collisions are clustered around highways/main roads. This clustering intuitively makes sense as vehicles typically travel at higher speeds on large roads. In Alameda County, the collisions are also clustered around main roads and have a higher incidence of collisions in densely populated areas, such as Downtown Oakland and Downtown Berkeley. This clustering also intuitively makes sense. Overall, there are more collisions involving bicyclists and pedestrians in census tracts with higher rates of zero vehicle ownership. The high rates of zero car ownership, however, are related to other factors, such as the density of land use or presence of high-quality transit. Therefore, the relationship between car ownership and rates of collisions involving bicyclists and pedestrians is interesting but isn’t directly causal.
I also mapped collisions involving bicyclists and pedestrians with the Metropolitan Transportation Commission (MTC)-defined Communities of Concern; MTC is the Metropolitan Planning Organization (MPO) for the San Francisco Bay Area. The MTC conducted an equity analysis to identify Communities of Concern and inform equity measures for Plan Bay Area, the MTC’s Regional Transportation Plan. MTC defines Communities of Concern as a “diverse cross-section of populations and communities that could be considered disadvantaged or vulnerable in terms of both current conditions and potential impacts of future growth.”
Prior to 2013, MTC defined Communities of Concern as census tracts where 70% or more of residents were a racial minority or 30% or more of households were low-income (earning less than 200% of the FPL). In 2013, the MTC revised how Communities of Concern were identified for Plan Bay Area 2040: census tracts where (1) 70% or more of residents are a racial minority and 30% or more of households are low-income or (2) census tracts where 30% or more of households are low-income and are characterized by three or more of six factors, which are described in Table 1 below.
Table 1: Communities of Concern Factors
| Indicator | Description |
| Limited English Proficiency | 20% or more of census tract residents have limited English proficiency |
| Zero-Vehicle Households | 10% or more of census tract households do not own a vehicle |
| Seniors | 10% or more of census tract residents are 75 years or older |
| Disability | 25% or more of census tract residents have a disability |
| Single Parent Family | 20% or more of census tract households are single parent families |
| Cost-Burdened Renter | 15% or more of census tract residents are cost-burdened renters |
The map below shows the locations of Communities of Concern throughout the Bay Area, but collisions involving bicyclists and pedestrians are only mapped for Alameda and Contra Costa Counties (for now). I created the map using the same collision and census tract data from the first map and used a shapefile of Communities of Concern. I hypothesized that Communities of Concern would have higher rates of collisions involving bicyclists and pedestrians because there are likely more people biking, walking, and walking to/from transit (zero-vehicle households), people traversing streets slowly (seniors and people with disabilities), and denser land use (renters in apartments). In Alameda County, collisions primarily occured in Communities of Concern while that is not the case in Contra Costa County. Mapping collisions involving bicyclists and pedestrians throughout the Bay Area may provide more insights, but based on Alameda and Contra Costa Counties, the factors that contribute to the rates of collisions involving bicyclists and pedestrians vary based on context (e.g. dense parts of Alameda County with multi-modal mode share versus car-oriented parts of Contra Costa County).
Notes:
- While designing these maps, I chose blue census tracts with red dots to make the color scheme color-blind friendly. I also tried to size the dots so viewers could easily identify clustering; this involved making the dots large enough that they contrasted with the background census tracts’ colors but also small enough to avoid significant overlap among dots.
- I used Table B25044 to calculate the percentage of households that do not own a vehicle for each census tract.
- All datasets cover the 2012-2016 timeframe to match with the census tract geometries.