Research Lead: Riley O'Brien
UC Campus(es): UCLA
Problem Statement: This graduate student thesis project analyzes socioeconomic and built environment predictors of pedestrian- and bicycle-involved crashes in the six-county jurisdiction of the Southern California Association of Governments (SCAG). As the nation’s largest metropolitan planning organization, with a per-capita pedestrian fatality rate higher than the state and national averages, SCAG has a unique responsibility to improve active transportation conditions. SCAG plays a direct role in allocating a large portion of statewide Active Transportation Program (ATP) funding, which has the potential to prevent injuries and deaths, reduce automobile use, and promote environmental justice more broadly. By providing clear evidence that high-poverty communities experience a disproportionate share of crashes, this report demonstrates the importance of the ATP and provides justification for its expansion.
Project Description: The student researcher developed six linear regression models to identify predictors of pedestrian- and bicycle-involved crashes at three geographic scales. The student considered 14 possible predictors, including built environment factors such as schools and commercial land use, transportation variables such as transit stops and vehicle miles traveled, density variables such as the number of people and jobs, and socioeconomic variables such as poverty rate and Hispanic/Latino population share. Using geographic information systems (GIS) software, the researcher aggregated pedestrian- and bicycle-involved crashes and predictor variables to all census tracts in the SCAG region, as well as 1⁄4 mile buffers around the region’s Metrolink commuter rail stations and Los Angeles County’s Metro rail and busway stations. Using statistical software, the researcher conducted each regression to determine the relationship between each crash type and each predictor variable while controlling for the remaining predictor variables. The regression results suggest that more vulnerable communities have less safe conditions for walking and biking, especially at the census tract level. The tract-level models account for 57% of the variation in pedestrian crashes and 4 49% of the variation in bicycle crashes, with all predictor variables statistically significant at a 95% confidence level. A higher poverty rate and Hispanic/Latino population share predict more bicycle and pedestrian crashes per tract, and a higher Black/African-American population share also predicts more pedestrian crashes per tract. The number of major transit stops per tract is the top predictor of pedestrian crashes and third-strongest predictor of bicycle crashes. These trends are less consistent in the station-level regression results, yet poverty is still one of the strongest predictors of both crash types near Metrolink stations and bicycle crashes near Metro stations. At the Metro station level, vehicle- miles traveled is also one of the strongest predictors of both crash types.