Professional Documents
Culture Documents
Abstract: The purpose of this study was to test whether a compact neighborhood design is associated with trip-chaining behavior. Trip
chaining is regarded as a growing phenomenon in travel and activity behavior because people seek to minimize the travel time and cost
required to accomplish their daily activities based on the available time budget and other needs. In this study, trip-chaining patterns were
examined over a survey day, giving insight into the association between land use and the planning of the trip and its distance, as well as the
preferred mode of transportation. A tour consists of a combination of individual trips, including all of the stops that are made along the way. A
series of multivariate models was used separately for different types of tours segmented into simple (one destination) and complex (more than
one destination), and into work and nonwork tours. The results confirmed the idea that trip chaining diminishes the likelihood of using
nonmotorized modes of transportation. In addition, local land use and walkability have a limited effect on work-related tours, but these
factors are significantly related to nonwork tours. A resident who lives in a more walkable neighborhood is likely to take simple, albeit
more frequent nonwork tours and conduct their nonwork activities on foot or by public transport, which reduces their use of vehicles during
the day. These findings lead to the conclusion that there is more opportunity to use urban design policies to influence nonwork tours than work
tours. DOI: 10.1061/(ASCE)UP.1943-5444.0000312. © 2015 American Society of Civil Engineers.
Author keywords: Trip chaining; Neighborhood walkability; Travel behavior; Tour-based approach; New urban design.
subset of broader research planned for the future, and the scope of ever, if they are accustomed to shopping in their neighborhoods as a
the research could be expanded to include other metropolitan single trip that is separate from commuting, there is a high likeli-
areas. However, the case study of Los Angeles County will aid hood that they will choose a nonmotorized mode for shopping
in evaluating the effects of land-use and transportation strategies (Shiftan 2008).
in a sprawling, auto-oriented, urban spatial structure. Previous research that used tour-based modeling (also known as
trip chains–based modeling) considered chained (linked) travel in
which two or more activities were connected. The results of this
Literature Review research indicated that the complexity of travel is an impediment
to changing the transport mode and that chained trips are more
Urban planners and transportation professionals have studied the likely to be taken by private vehicles (de Nazelle et al. 2010).
relationships between land use and travel behavior. Many recent Similarly, Krygsman et al. (2007) identified a relationship between
studies of travel behavior have provoked methodological debates. transport mode and chaining behavior, but there was no evidence of
The most prominent methodological debate relates to the causality this relationship when location factors of intermediate activities
between the built environment and travel behavior. Bagley and were considered within a tour.
Mokhtarian (2002) posited that personal characteristics and The empirical results are mixed in terms of the effects of land
propensities are critical factors that influence travel behavior. That use on trip-chaining behavior. Analyzing the relationships between
is, individual demographic status (e.g., race, ethnicity), socioeco- the type of tour and neighborhood access, Krizek (2003a) found
nomic status (e.g., age, gender, income), and lifestyle are related to that higher levels of accessibility were associated with a larger
one’s propensity and needs for activity. The results of these studies number of tours, but the tours were less complex. Maat and
implied that the association between the built environment and Timmermans (2006) studied the complexity and frequency of tours
travel behavior was insufficient to establish causality. The role and found that higher densities resulted in a greater number of tours
of attitudinal propensity and self-selection is essential to the and more chained trips.
analysis of the effect of residential location on travel demand. There Although the recent literature has reported notable develop-
have been several attempts to control for self-selection bias, such as ments in theoretical conceptualizations and analytical methodolo-
studies using longitudinal data (e.g., Handy et al. 2006) and gies, questions remain regarding how the local environment
studies using structural equation models and instrumental variables influences an individual’s planning of a trip and its complexity.
(e.g., Bagley and Mokhtarian 2002; Bento et al. 2005). Investigating these factors by using the land-use attributes of the
Recently, some publications in the literature have stated that a rich origins and destinations in the trip would be an important addition
set of sociodemographic and socioeconomic variables might to the current research, which was focused on the residential
partially alleviate the self-selection effect (Bhat and Guo 2007), environment.
and, later, this study used multivariate analysis to account for
the effect of individual socioeconomic status and lifestyle.
Beyond the residential selection debates, there has been some Methods
solid evidence of an association between land use and travel
behavior. Much of the high-quality research on travel behavior that
Research Design
uses individual travel data has found a relationship between land
use and travel after controlling for the self-selection effect Fig. 1 shows a conceptual model of the tour-based modeling that
(Brownstone and Golob 2009; Boarnet and Sarmiento 1998). Most was conducted. Tours represent the foundation of current activity-
of these studies have focused on trip-based modeling, and they have based transportation research. The literature most often defines a
provided a useful framework of a straightforward means of tour as a “home-to-home loop.” This study extended Bowman
predicting travel outcome by examining each trip in isolation. and Ben-Akiva’s (2001) study in which it was concluded that tour
Recent improvements in this line of research have included patterns are conditioned by the scheduling of the day’s activities.
tour-based modeling. Scheduling activities involves a set of choices, including the agenda
Tour-based modeling is a step toward an activity-based of activities, the priorities of the activities, arranging the activities in
approach. Advocates of the activity-based approach (e.g., Bhat a sequence, and choosing the locations of activities and the time
and Koppelman 1999; Kitamura 1988; Bowman and Ben-Akiva spent at those locations. Conversely, the choice of activity patterns
2001) have concluded that individuals tend to optimize their also is affected by the alternative tours that are available. The pro-
entire activity patterns rather than just maximizing the utility of pensity to chain multiple trips in a single tour may alter the indi-
separate trip choices (Maat et al. 2005). Most importantly, vidual activity patterns. In combination, these activities and tour
activity-based travel theory views most travel as an activity-driven patterns have direct influence on VMT and the choice of mode.
demand (Bowman and Ben-Akiva 2001). If trips are induced by This study was based on the assumption that personal and
activities, then the conventional, trip-based modeling that considers household characteristics influence activities and tour patterns.
trips discretely will provide, at best, only simple adjustments for The characteristics of all of the people involved, their households,
the behavioral analysis of travel (Kitamura 1988). Tour-based and their neighborhoods were merged into the data set of the tours.
modeling improves the model’s ability to capture tour-based travel Hagerstrand (1970) and Rosenbloom (1989) pointed out that there
patterns over the period may not have been large enough to have the day of the survey. The models were run separately for two
had significant influence on the travel behavior of regional different types of tours, i.e., work tours and nonwork tours. The
residents. equations for the regression model to test the significance of the
tour characteristics with respect to neighborhood walkability are
as follows:
Model Specification
The number of trips per tour ¼ fðNW; RA; HH; PSÞ ð4Þ
The hypothesis of this study was that neighborhoods’ walkabilities
and the characteristics of households are associated with several
assessments of personal mobility, i.e., planning and chaining of The number of tours per day ¼ fðNW; RA; HH; PSÞ ð5Þ
a trip, distance of the trip, and choice of the mode of transportation
for the trip. A series of regression models was developed to test The distance traveled per day ¼ fðNW; RA; HH; PSÞ ð6Þ
this hypothesis with regard to the characteristics of tours. The
complexity, frequency, and distance of travel were compared where NW = neighborhood walkability; RA = regional accessibil-
to household and geographical characteristics. Four outcome ity; and the control variables include HH, which denotes the
variables were estimated independently in separate regression household-level travel demand variables, including socioeconomic
models developed with identical structures to ensure consistency status and lifestyle, and PS denotes the person-level attributes, such
between the models. The models were consistent internally with as age, gender, race, work status, and the possession of a driver’s
regard to the definition and estimation of costs on which travel license.
choices are based. The models had a feedback mechanism among Finally, multinomial logit (MNL) models were used to measure
the tour complexity, tour distance, and mode split. the overall effect of the built environment on mode choice of tours.
Fig. 2. Study area and household location (Los Angeles County data from NHTS-CA 2009)
Results
Descriptive Statistics
In this study, work tours were distinguished from nonwork tours.
Fig. 3 illustrates the basic descriptive statistics that show the
usefulness of studying trip chaining. Almost 21% of all tours
involved commuting, and about 45% of commuting trips were
complex tours that linked two or more two intermediate stops.
In the case of nonwork tours, which accounted for about 79%
of all tours, 35% of the tours were complex tours that had multiple,
nonwork stops. The proportions of trip chaining were slightly
greater among commute tours than noncommute tours.
Low- and high-walkability neighborhoods were compared to
determine whether these differences led to different distributions
of tour patterns and travel modes. The neighborhoods were
categorized as two different types based on the distribution of
the walkability index scores. A low-walkability neighborhood
was identified as a neighborhood within the first, second, third,
Fig. 4. Share of transportation modes by different purposes of tours
and fourth deciles of the walkability index scores, while those
and neighborhood types (LW = low-walkability neighborhood; HW =
within the seventh, eighth, ninth, and tenth deciles were defined
high-walkability neighborhood)
as high-walkability neighborhoods. This approach was consistent
These results seem to be in line with the findings of previous stud- number of chains, while residential walkability had a negative ef-
ies, which show that higher levels of neighborhood access matters, fect on the complexity of tours. Intuitively, whether or not the land
in particular as they relate to formation of tours (Krizek 2003a). use around work locations offers more opportunities for nonwork
The models were tested using the z-scores of the walkability. activities is relevant when travelers chain their trips to draw advan-
The effect of walkability on tour formation was mixed. For work tages from the affluent environment of their places of employment.
tours, walkability of the workplace had a positive influence on the For nonwork tours, the results showed that people in higher levels
conducted locally in one’s own neighborhood. In order to further results showed that a good walkability of origins could potentially
explore whether most of these tours were pursued by nonmotorized reduce driving and encourage walking. People who live in a neigh-
means (i.e., walking and biking) or transit mode, mode-choice borhood that is more pedestrian and transit supportive conduct their
models were used in this study, and their results are provided in nonwork activities by walking or cycling. However, the evidence
the next section. showed that this was the case only when both the origin and the
destination neighborhoods were walkable. People who travel to
destinations that have a high level of walkability tend to rely on
Mode-Choice Model Results private vehicles for their nonwork activities if they live in a
Table 5 presents the results from the mode-choice models. The car low-walkability area. This finding likely reflected the fact that
is the base unit of comparison. The models here are comprehensive people are more likely to drive from outside of the immediate
in that they included variables of travel time differential between neighborhood to shop and spend leisure time at the more busi-
ness-concentrated (mostly walking-oriented) centers that are larger
the alternatives in order to reflect the popularity of the alternatives.
than the local resident-based market. A key contributor to transit
The results showed that a shorter travel time aboard transit relative
use was transit accessibility rather than walkability. That is, good
to the car increased the transit use, consistent with theory. Similarly,
regional transit accessibility appeared to be far more important than
a longer travel time via walking relative to the car lowered the prob-
neighborhood attributes for the use of public transportation for both
ability of walking versus driving, all other things being equal.
work and nonwork trips.
Several attribute factors of a tour also affected travelers’ choice
of mode for making the trips. People’s tour patterns were affected
by the time their trips began. Travelers were more likely to Discussion
commute via walking or cycling when their trips began during
the morning period of peak traffic. The results also confirmed To test the relationship between neighborhood walkability and
the hypothesis that trip chaining is a barrier to the propensity to travel behavior, tour-based models were developed in this study
use a nonautomobile mode. This means that a traveler is less likely to investigate the sequence and combinations of all trips and
chained nonwork tours. They conduct their nonwork activities Department of Transportation and the Southern California Associa-
on foot or by public transport, enabling reduced vehicle use during tion of Governments. The author thanks Genevieve Giuliano and
the day. This confirmed the concept that New Urban Designs with Marlon Boarnet for their constructive comments on an earlier
higher density, mixed land uses, grid-pattern street networks, and version of this paper and the anonymous referees for their careful
proximity to the locations of commercial activities will decrease the review and useful questions. This work was supported by the
number of miles that private vehicles travel. There was some evi- Research Fund of University of Ulsan.
dence of induced automobile travel for nonwork trips to destina-
tions in a compact location. This finding indicated that the
reduced generation of automobile trips to walkable destinations References
is not as robust as the evidence for trips from walkable residences.
These findings suggest that future interventions that are de- ArcGIS [Computer software]. ESRI, Redlands, CA.
signed to reduce car travel should target the creation of residential Bagley, M. N., and Mokhtarian, P. L. (2002). “The impact of residential
environments that better accommodate the needs of nonwork neighborhood type on travel behavior: A structural equations modeling
approach.” Ann. Reg. Sci., 36(2), 279–297.
activities. These types of nonwork trips can be encouraged, for in-
Bento, A. M., Cropper, M. L., Mobarak, A. M., and Vinha, K. (2005). “The
stance, by providing a convenient pedestrian environment, along
effects of urban spatial structure on travel demand in the United States.”
with sites for recreational activities that are closer to residential Rev. Econ. Stat., 87(3), 466–478.
neighborhoods and by making the neighborhood environment safer Bhat, C. R., and Guo, J. Y. (2007). “A comprehensive analysis of built
and easier to navigate, especially for children to walk, ride a bike, environment characteristics on household residential choice and auto
or use the transit by protecting them from the danger of traffic and ownership levels.” Transp. Res. Part B: Method., 41(5), 506–526.
by enhancing public safety (Lee 2013). Bhat, C. R., and Koppelman, F. S. (1999). “Activity-based modeling of
In view of the evidence concerning the automobile’s dominance travel demand.” Handbook of transportation science, Springer,
of chained trips, there was limited a priori expectation regarding the New York, 35–61.
role of a nonautomobile mode in conducting several activities in a Boarnet, M., and Crane, R. (2001). Travel by design: The influence of
chained trip. This evidence does not suggest that it is infeasible for urban form on travel, Oxford University Press, New York.
travelers to use nonautomobile modes of travel for organizing Boarnet, M. G., and Sarmiento, S. (1998). “Can land-use policy really af-
fect travel behaviour? A study of the link between non-work travel and
multiple activities into daily trip chains. Given the fact that many
land-use characteristics.” Urban Stud., 35(7), 1155–1169.
automobile users link trips on their way to and from work, it is Bowman, J. L., and Ben-Akiva, M. E. (2001). “Activity-based disaggregate
expected that a local balance of jobs and housing will encourage travel demand model system with activity schedules.” Transp. Res. Part
people to commute by transit and bikes to some extent, because A: Policy Pract., 35(1), 1–28.
they will be able to run errands on the way to and from work. Brownstone, D., and Golob, T. F. (2009). “The impact of residential
The findings and limitations of this study point to the need for density on vehicle usage and energy consumption.” J. Urban Econ.,
additional analyses and research. This study relied on cross- 65(1), 91–98.
sectional data that limit causal relationships and may be affected Cervero, R., and Radisch, C. (1996). “Travel choices in pedestrian versus
by unexpected situational variables at the time of assessment. It automobile oriented neighborhoods.” Transp. Policy, 3(3), 127–141.
also is true that the absence of detailed street data, such as the exist- de Nazelle, A., Morton, B. J., Jerrett, M., and Crawford-Brown, D. (2010).
ence of sidewalks, was a limitation of this study. For example, it is “Short trips: An opportunity for reducing mobile-source emissions?”
Transp. Res. Part D: Transp. Environ., 15(8), 451–457.
possible that the status of the sidewalks could be correlated with
Frank, L., Bradley, M., Kavage, S., Chapman, J., and Lawton, T. K. (2008).
travelers’ propensity to walk to their destinations, because the con-
“Urban form, travel time, and cost relationships with tour complexity
nectivity and quality of sidewalks could determine the level of and mode choice.” Transportation, 35(1), 37–54.
safety of the pedestrian environment (Ha et al. 2011). Furthermore, Frank, L. D., et al. (2010). “The development of a walkability index:
pedestrians’ perceptions about walkability could vary according to Application to the neighborhood quality of life study.” Br. J. Sports
different sidewalk environments (Kim et al. 2013). Med., 44(13), 924–933.
This research indicated the need for additional scholarly atten- Ha, E., Joo, Y., and Jun, C. (2011). “An empirical study on sustainable
tion in terms of improved methods. First, in this paper, the need for walkability indices for transit-oriented development by using the ana-
incorporating the interrelationship among mode choices, trip chain- lytic network process approach.” Int. J. Urban Sci., 15(2), 137–146.
ing, and VMT was visualized. Therefore, complex relationships Hägerstraand, T. (1970). “What about people in regional science?” Reg.
among these travel variables should be explored by using some Sci., 24(1), 7–24.
type of discrete-continuous modeling approach in future studies. Handy, S. (1993). “Regional versus local accessibility: Implications for
nonwork travel.” Transp. Res. Rec., 1400, 58–66.
Second, since multiple tours can be taken in a day, a better under-
Handy, S., Cao, X., and Mokhtarian, P. L. (2006). “Self-selection in the
standing of connections between the tours is critical because people relationship between the built environment and walking: Empirical evi-
tend to organize their activities within a daily schedule (Rasouli and dence from northern California.” J. Am. Plann. Assoc., 72(1), 55–74.
Timmermans 2014). Thus, there may be substantial interactions Hansen, W. (1959). “How accessibility shapes land use.” J. Am. Inst.
across tour decisions when decisions are made about scheduling Plann., 25(2), 73–76.
in order to manage the time and space constraints associated with Hausman, J., and McFadden, D. (1984). “Specification tests for the multi-
making trips. A model that incorporates the duration of activities nomial logit model.” Econometrica: J. Econ. Soc., 52(5), 1219–1240.
Krygsman, S., Theo, A., and Timmermans, H. (2007). “Capturing tour Conf., Transportation Research Board, Washington, DC, 7–28.
mode and activity choice interdependencies: A co-evolutionary logit Saelens, B., Sallis, J., and Frank, L. (2003). “Environmental correlates of
modelling approach.” Transp. Res. Part A: Policy Pract., 41(10), walking and cycling: Findings from the transportation, urban design,
and planning literatures.” Ann. Behav. Med., 25(2), 80–91.
913–933.
Sallis, J. F., et al. (2009). “Neighborhood built environment and
Lee, J. (2013). “Perceived neighborhood environment and transit use in
income: Examining multiple health outcomes.” Soc. Sci. Med.,
low-income populations.” Transp. Res. Rec.: J. Transp. Res. Board,
68(7), 1285–1293.
2397(1), 125–134.
Shaheen, S. A., Bejamin-Chung, J., Allen, D., and Howe-Steiger, L.
Maat, K., and Timmermans, H. (2006). “Influence of land use on tour
(2009). “Achieving California’s land use and transportation greenhouse
complexity: A Dutch case.” Transp. Res. Rec.: J. Transp. Res. Board, gas emission targets under AB 32: An exploration of potential policy
1977(1), 234–241. processes and mechanisms.” Institute of Transportation Studies,
Maat, K., Van Wee, B., and Stead, D. (2005). “Land use and travel UC Davis.
behaviour: Expected effects from the perspective of utility theory Shiftan, Y. (2008). “The use of activity-based modeling to analyze the
and activity-based theories.” Environ. Plann. B: Plann. Des., 32(1), effect of land-use policies on travel behavior.” Ann. Reg. Sci., 42(1),
33–46. 79–97.
Næss, P. (2009). “Residential self-selection and appropriate control Sundquist, K., Eriksson, U., Kawakami, N., Skog, L., Ohlsson, H., and
variables in land use: Travel studies.” Transp. Rev., 29(3), 293–324. Arvidsson, D. (2011). “Neighborhood walkability, physical activity,
NHTS-CA (National Highway Traffic Safety Administration California). and walking behavior: The Swedish neighborhood and physical activity
(2009). 〈http://nhts.ornl.gov/〉. (SNAP) study.” Soc. Sci. Med., 72(8), 1266–1273.
O’Sullivan, S., and Morrall, J. (1996). “Walking distances to and from Ye, X., Pendyala, R. M., and Gottardi, G. (2007). “An exploration of the
light-rail transit stations.” Transp. Res. Rec.: J. Transp. Res. Board, relationship between mode choice and complexity of trip chaining
1538, 19–26. patterns.” Transp. Res. Part B: Method., 41(1), 96–113.