A case for race and space in auto ownership modeling: A Los Angeles County study.

Auto ownership behavior is driven by complex relationships that can vary dramatically across different traveler groups and communities. Differences in auto ownership among racial groups have been of particular interest, given ongoing efforts to advance equity in transportation outcomes. There are a number of studies documenting racial disparities in auto ownership associated with racial and ethnic residential clustering, termed "automobile mismatch." Yet, these differences in auto ownership behavior by race and residential location are virtually never considered in models of travel behavior, despite calls for the consideration for race in transportation planning and decision making. This study aims to bridge the gap between understandings of the connections between race and space and transportation outcomes, using Los Angeles County as a case study. A series of auto ownership model specifications are used to investigate statistical connections between the racial and ethnic categories of residents, and neighborhoods, revealing systematic variations across racial and spatial dimensions. The composite model, which includes racial and spatial indicators, outperforms the base model, suggesting that the inclusion of race and space explains significantly more information on variations in auto ownership and provides a superior fit to the data. Our findings also suggest that the exclusion of racial and spatial indicators may lead to overestimation of certain effects, and may completely misrepresent the importance of certain household, individual-level, and built environment effects in explaining auto ownership preferences. Given the increasing attention to equity and representation in transportation outcomes, models that exclude considerations for race and space may be poorly positioned to support meaningful transportation equity analyses. • Current travel models of auto ownership lack consideration for race and space in shaping auto ownership outcomes. • We systematically compare auto ownership model specifications, that include racial and spatial indicatiors. • The results show that models with such indicators offer better statistical fit and explain more about auto ownership. • Excluding these indicators may bias certain effects and misrepresent important household, individual, and built environment effects. • Interactions between race and space also matter and can siginificantly shape auto ownership outcomes. [ABSTRACT FROM AUTHOR]