A spatial statistical appraoch to assessing socio-economic equity in transit-based accessibility to urban services in Franklin County, OH

Abstract:

Analyzing spatio-temporal accessibility to urban services (jobs, healthcare facilities, groceries, etc.) via public transit has become increasingly popular for evaluating socio-economic equity. Various studies on accessibility have shown that different socio-economic groups experience varying levels of access to urban services. However, most studies attempt to evaluate equity by overlaying socio-economic variables on transit accessibility and generalizing their influences on a global scale. In many US cities, people with similar socio-economic characteristics tend to cluster in specific areas. Therefore, relying on generalized correlations of these variables may fail to capture local variations. To address this gap, we conducted a GIS-based spatio-temporal analysis of accessibility to urban grocery stores and supermarkets in Franklin County, OH, combining walking and transit travel using General Transit Feed Specification (GTFS) data. We incorporated key demographic (race), economic (poverty rate, vehicle ownership, home ownership), and spatial (distance from downtown) data for each Census Block Group to generate both linear (OLS) and spatial (SLM, SEM, and GWR) regression models. The results indicate that spatial models better fit the dataset than the linear model by accounting for spatial dependencies. Moreover, spatial model diagnostics reveal that GWR fits better than both SLM and SEM. Although the influence of variables varies across the study area, the global coefficients from the GWR model indicate that both poverty rate and distance from downtown negatively influence transit accessibility to grocery stores. In contrast, non-white population, renter households, and households without a vehicle positively influence accessibility, with the latter having the highest impact.

Keywords: Transit Accessibility, Spatial Statistics, Equity, Geographically Weighted Regression (GWR)

Authors: