Review of the literature
As I discussed in Chapter 3, travel behavior research is driven by a variety of objectives, has taken place across a range of contexts, and has produced widely inconsistent results. In this chapter, I focus on my analysis on observed mobility patterns of lower-wealth, transit-captive (car-less) households. However, as with vehicle ownership, empirical research on the
travel/environment relationship in developing contexts is limited. Most of the existing research focuses on aggregate travel outcomes and, like with the vehicle ownership literature, tends to rely on one or a few measures of built form. For example, Lin & Shin (2008) found that subway ridership in Taipei, Taiwan was greater at station areas with greater floor-area-ratios, but was not related to density or land use diversity and was negatively related to connectivity. In contrast, Sung & Oh (2011) found positive associations between subway ridership and density, land use diversity, and connectivity in Seoul. Estupiñán & Rodríguez (2008) used factor analysis to evaluate associations between more finely grained measures of urban form and BRT boardings in Bogotá. Factors representing the presence of walking supports and deterrents to car use
(incorporating measures of density and land use diversity) were positively related to boardings, while a factor representing street network connectivity was not significant. Using similar urban form data but focusing on nonmotorized trip generation, Rodríguez et al (2009) found that land use diversity and factors representing connectivity and pedestrian friendliness were positively, but weakly, associated with pedestrian activity.
At the disaggregate level, Zegras (2004) found that individuals in Santiago, Chile made more walking trips for non-work purposes when they lived in neighborhoods with a greater
proportion of land area dedicated to commercial and service uses, and fewer non-work walking trips in neighborhoods with large vacant areas. In Chennai, India, Srinivasan & Rogers (2005) examined relationships among regional accessibility and travel patterns, finding that residents of a downtown neighborhood made more total trips, more nonmotorized trips, and fewer transit trips than residents of a peripheral neighborhood. Zacharias (2005) and Pan et al. (2009) both found greater rates of walking among Shanghai travelers surveyed at major destinations when the travelers lived in a more ‘walkable’ neighborhood, compared to travelers from three less walkable neighborhoods. Zacharias, (2005) also measured block-level attributes within the
neighborhoods: road density, road length, transit density (number of bus stops divided by land area), and block depth (most closely related to street network design, block depth was a measure of the mean street network distance between randomly selected points in a block and the
perimeter of the block). Block depth was negatively correlated with the likelihood of having used nonmotorized modes; none of the other measures was significant.
In the U.S., studies of relationships between travel behavior – particularly transit and nonmotorized travel – and the built environment tend to use a wider range of and more consistent measures of the environment; findings have been somewhat more consistent as well (Ewing and Cervero, 2010). Individual and household level studies have found positive associations between density and walking trip frequency (Targa and Clifton, 2005; Joh et al., 2008), transit trip
frequency (Chatman, 2009), the odds of walking for non-work travel (Greenwald, 2006; Frank et al., 2008), and the odds of walking and/or using transit for work trips (Zhang, 2004; Frank et al., 2008).
Land use diversity is consistently positively related to walking trip frequency (Targa and Clifton, 2005; Cao et al., 2009), the odds of using transit and nonmotorized modes for non-work trips (Cervero and Duncan, 2003; Rajamani et al., 2003; Zhang, 2004; Greenwald, 2006; Frank et al., 2008) and for work trips (Frank et al., 2008). Connectivity has been positively associated with walking trip frequency (Targa and Clifton, 2005; Chatman, 2009) and the likelihood of walking
and using transit for work and non-work travel (Greenwald, 2006; Frank et al., 2008); one study (Rajamani et al., 2003), however, found a negative relationship between connectivity and walking for non-work trips. Increasing transit access (measured in terms of station proximity and station density) is associated with increased walking trip frequencies (Targa and Clifton, 2005) and increased likelihood of using transit for non-work travel (Greenwald, 2006).
Measuring mobility
The findings to date shed important light on the ways the built environment can influence actual travel patterns. However, especially in a low-wealth, resource-constrained population, the existing literature does not examine the degree to which the travel needs of car-less households are being met by the built environments in which they live (McCray and Brais, 2007; Bocarejo and Oviedo, 2012; Jaramillo et al., 2012). What does an individual’s decision to walk to work, for example, tell us about the extent to which that individual’s neighborhood provides for non-car mobility? If that commute takes over an hour by foot, perhaps it tells us that, until the individual buys a car, walking is not a choice, but a last resort. Some measures work in different directions depending on the outcome of interest or on the population in question. For example, more distance traveled by nonmotorized modes may simultaneously indicate better health outcomes (because of increased physical activity) and greater risk of injury or illness (because of increased exposure to vehicular traffic and traffic-related pollution). More travel, measured in terms of travel time or distance, may suggest poor accessibility to destinations, while less travel may mean trips were shortened or forgone due to limited mobility.17 One of the main challenges of this analysis is to develop readily available measures of travel that are clear and consistent indicators of mobility. Based on the literature and theoretical framework described above, I propose three
17 Recent focus group work in Bogotá confirms the existence of a gap between households’ actual travel patterns and desired travel patterns, at least with respect to time and monetary costs of commuting: on average, residents of the city’s most impoverished neighborhoods spend 40 percent more time and 38 percent more money on traveling to and from work than they would like to spend (Bocarejo and Oviedo, 2012).
potential indicators of mobility: tour frequency, non-car travel purpose diversity, and vehicle independence.
Tour frequency
A tour is a set of trips that begin at a particular location, include one or more stops to participate in some sort of activity, and end back at the same location (Ortúzar and Willumsen, 2011). Tour frequency is the number of tours completed by household members during the survey day. While trip frequency has received more attention in the literature, tours more accurately reflect travelers’ decision-making processes than trips (Krizek, 2003b and Frank et al., 2008), and have been shown to offer better model fit than trips in travel modeling research (Chao and Ducca, 2012). Theoretically, tour frequency should be positively related to the ease of travel. Citing a threshold hypothesis put forth by Adler and Ben-Akiva (1979), Krizek (2003b) writes: “…unfulfilled household activities accumulate until some critical threshold is reached. At this threshold, a tour is scheduled to complete some or all of the activities.” He goes on to explain that as the ease of travel increases through increased density, diversity, and connectivity, the critical threshold is reached sooner, at lower levels of accumulation of travel need. Thus, in
neighborhoods with greater accessibility, we would expect to see more (and, by corollary, simpler) tours.
Despite the theoretical connection, only a handful of empirical studies have explicitly examined relationships between tour frequency and the built environment, with mixed results. In the Puget Sound region, Krizek (2003a) found that tour frequency was positively related to neighborhood accessibility, a measure incorporating population density, land use diversity, and street connectivity. Likewise, Frank et al. (2008) found positive associations between tour frequency and land use diversity and connectivity. However, Limanond and Niemeier (2004) failed to find significant connections between a range of built environment measures and frequency of shopping tours. Using 2001 NHTS data, Noland and Thomas (2007) found no
associations between tour frequency and density, but did find that tour complexity, which is commonly inversely related to tour frequency, decreased with increasing densities.
Purpose diversity
Non-car purpose diversity refers to the number of travel purposes/activity needs household members fulfill without relying on a car or taxi. Travel purposes are often classified into three categories: subsistence (i.e., work and school), maintenance (e.g., non-discretionary shopping, errands), and leisure/discretionary (e.g., recreation, socializing) (Cervero and Kockelman, 1997; Krizek, 2003b). Travel purpose has often been examined in travel behavior literature in combination with other measures such as trip generation (the number of trips or tours generated for a specific purpose, e.g., the number of shopping trips made by a household) or mode choice (e.g., the odds a commuter uses a particular travel mode for the journey to work). But to understand the factors driving households to purchase a vehicle, I argue that a more informative measure may be the diversity of travel purposes car-less households can fulfill without using a car or taxi (in other words, without temporarily altering their car-availability status). For example, if a household is able to meet its subsistence travel needs and its
maintenance travel needs without using a car or taxi, it may be experiencing lower pressures to motorize than a household that is only using non-car modes to engage in subsistence travel. Assume that, controlling for household attributes such as composition, structure, and financial resources, different households have similar demand to fulfill each of these travel purposes. A household whose members fulfill all three travel purposes via non-car modes will have (or will perceive) more mobility (and feel less pressure to enhance their mobility through motorization) than a household whose members only fulfill one or two travel purposes via non-car modes.
Vehicle independence
Finally, vehicle independence refers households’ ability to complete their travel without using a car (borrowed, rented, or taxi). Whether a car-less household completes its daily travel
without using a car or taxi may be an indication of the vehicle ownership pressures facing that household. Using a car to fulfill any household travel need represents a change in the household’s vehicle availability status, even if just temporarily. Such temporary changes may eventually lead to a permanent change (i.e., acquiring a private vehicle). Thus, I examine whether or not a household depends on a personal vehicle (perhaps either borrowed or rented) or taxi for any of their travel on the survey day.