Abstract:This study analyzes the 2009 National Household Travel Survey (NHTS) and attitudinal data pooled across four California studies, to investigate the impact of household demographics, individual attitudes, and residential location on vehicle ownership and usage decisions in California. We classify each household as lower-than-expected (LTE), about-as-expected (AAE), or higher-than-expected (HTE) vehicle owning, based on the comparison of the actual vehicle ownership level with the expected value computed from a model that predicts vehicle ownership based on household size and composition. Households that do not own any vehicles are classified as zero-vehicle-owning (ZVO). We are especially interested in exploring the reasons for which a household would own fewer-than-expected or no vehicles, and have low vehicle miles traveled (VMT). We first estimate a set of models to explore the impacts of the most natural constraints that could explain low- or zero-vehicle-ownership status, and lower VMT, namely income and driving limitations. We then focus on the reasons for the voluntary choice of low/zero vehicle ownership through controlling for the impacts of personal attitudes.  Finally, we analyze the role of residential location and land use traits in affecting household vehicle ownership and VMT.  We find that, consistent with expectations, lower-income households and those containing someone with driving limitations are more likely than others to own zero or fewer-than-expected vehicles and travel fewer miles. Among the segment of the population with higher income and no driving limitations, households that own zero or fewer-than-expected vehicles, and have lower VMT, tend to be more diverse, have fewer children, and live in rental units in very high density neighborhoods. The inclusion of attitudinal variables improves the ability to explain household vehicle ownership by a limited, but not trivial, amount. Individuals with more pro-environmental attitudes and who like transit, biking and walking are more likely to live in zero-vehicle-owning households. Conversely, those who like driving and living in spacious homes with large yards are more likely to be in higher-than-expected vehicle-owning households.  With respect to land use characteristics, both local density and regional status (the latter being a three-part classification based on metro-area size and the presence/absence of rail) yield strong associations. Households in higher-density neighborhoods are more likely to own zero or fewer-than-expected vehicles, and have lower VMT. But even lower-density living is associated with lower VMT if located in larger metropolitan areas (especially those with rail). Similarly, residential locations in smaller regions are found to have lower VMT if residential neighborhoods are denser. Overall, density has non-linear effects on travel behavior: a given increase in density is associated with larger reductions in households’ VMT in lower-density neighborhoods than in higher-density ones, and this difference (between lower- and higher-density neighborhoods) is larger in smaller regions. Specifically among higher-density neighborhoods, however, the strongest relationships between density and VMT are found in large regions with rail: a given density increase is associated with larger reductions in VMT for households living in rail-served regions, all else equal.  The study provides useful insights for promoting the adoption of more sustainable travel behavior. In particular, it improves the understanding of what policy levers could lead to the adoption of environmentally-beneficial behaviors and help meet the required reductions in VMT. It also highlights the importance of collecting attitudinal data to improve the ability to explain vehicle ownership and use, especially decisions (such as voluntary “down-shifting” of vehicle ownership and VMT) that cannot be explained by traditional socioeconomic and demographic variables.