Abstract
The spatial dimension of the journey-to-work has important implications for land use and development policymaking and has been widely studied. One thrust of this research is concerned with the disaggregation of workers into subgroups for understanding disparities in commute. Most of these studies, however, were limited to the disaggregation by single socioeconomic class. Hence, this research aims to examine commuting disparities across commuter subgroups stratified by two socioeconomic variables—income and race—using a visual analytics approach. By applying the doubly constrained spatial interaction model to the 2014 Longitudinal Employer-Household Dynamics data, this research first synthesizes commuting flows for Downtown Houston workers across income-race subgroups at the tract level in Harris County, Texas, USA. It then uses bivariate choropleth mapping to visualize the spatial distributions of major Downtown Houston commuter neighborhoods by income-race classes, and significant commuting disparities are identified across income-race subgroups. The results highlight the importance of considering income and race simultaneously for commuting research. The visualization could help policymakers clearly identify the unequal commute across worker subgroups and inform policymaking.
Commuting disparities across worker subgroups by single socioeconomic class, such as occupation, gender, or income, are well understood (Hu et al., 2017; Hu and Wang, 2018; Kwan and Kotsev, 2015; O'Kelly and Lee, 2005). However, to what extent such observed commuting differences may change if another worker subgroup is accounted for remains less studied, especially at finer spatial scales. For instance, would the further consideration of race change the observed income disparities in commute? This research aims to explore the interaction between two socioeconomic variables—income and race—in understanding commuting disparities using a visual analytics approach. As no publicly-available census data provide commuting flows by more than one socioeconomic variable at a fine spatial scale (e.g. census tract level) in the U.S., this research applies the doubly constrained spatial interaction model (O'Kelly and Lee, 2005) to the 2014 Longitudinal Employer-Household Dynamics (LEHD) Origin-Destination Employment Statistics (LODES) data for synthesizing and visualizing commuting patterns across income-race worker subgroups at the tract level in Harris County, Texas.
This research focuses on six selected income-race classes including low- and high-income subgroups for Whites, Blacks, and Asians. 1 For better visualization, only commutes ending in Downtown Houston, the largest job center in the study area, are illustrated. Figure 1 illustrates a bivariate choropleth map in which each tract is colored by the largest income-race worker subgroup in terms of commutes to downtown from that tract. In the color matrix, the top row represents low-income workers and the bottom row high-income counterparts. The left, middle, and right columns demonstrate White, Black, and Asian commuters, respectively. Being the most significant job center, the commuter shed of Downtown Houston includes workers, both low- and high-income, from across the county. Clearly, low-income downtown commuter neighborhoods are sparse and mostly confined within the Beltway 8 loop, whereas high-income commuter neighborhoods are abundant and spread all over the county.

Spatial distributions of the largest income-race worker subgroup commuting to Downtown Houston.
As the county and the Greater Houston region have grown in racial diversity, it would be intriguing to also examine how race would affect the observed income differences in commutes. Overall, substantial commuting differences are observed across these income-race categories. Specifically, both low- and high-income Black downtown commuters co-locate in space, concentrated mostly within the Beltway 8 loop (e.g. southern areas of the county). Quite differently, low-income White downtown commuters are sparsely scattered within the Beltway 8 loop, while their high-income counterparts are all over the county, especially in the suburbs in the outskirts, but not the southern areas. Most interestingly, high-income Asian downtown workers concentrate in only a few western Houston suburbs, such as Sugar Land, and the Bellaire Chinatown, while there exist no neighborhoods where low-income Asian workers are the majority commuters to downtown jobs. Their absence in the map may suggest the presence of a spatial mismatch problem for low-income Asian downtown commuters.
This research provides several novel findings and policy implications. First, commuting spatial patterns measured along a single sociodemographic dimension may change remarkably when another variable is considered. This is obvious in the present study when race is considered to further stratify income subgroups of workers. Second, it draws attention to the low-income Asian downtown commuters that appear to be the most underrepresented (of the three racial categories) in this area's labor market. Policymakers should consider the income-race disparities in commuting patterns when making transportation and land-use policies. The methods, including the trip synthesis model and bivariate choropleth mapping, can benefit a variety of studies examining other socioeconomic classes, such as race-gender, and/or travel behaviors like mode choice and trip-chaining (Hu and Li, 2021; Niedzielski et al., 2020).
A potential issue that may bias the measurement of income differences in commutes deserves some discussion. In the LODES data, low- and high-income workers are defined as workers earning less than $1250 or more than $3333 per month, respectively. The definitions, however, are likely not representative of both income groups in today's workforce. Therefore, researchers and policymakers should use caution when using the LODES data to study income differences in commuting patterns.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
