Transportation Research Part A 80 (2015) 277–287

Contents lists available at ScienceDirect

Transportation Research Part A

journal homepage: www.elsevier .com/locate / t ra

The impact of transit station areas on the travel behaviorsof workers in Denver, Colorado

http://dx.doi.org/10.1016/j.tra.2015.08.0040965-8564/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 303 871 2654.E-mail addresses: gkwoka2@gmail.com (G.J. Kwoka), eric.boschmann@du.edu (E.E. Boschmann), andrew.goetz@du.edu (A.R. Goetz).

Gregory J. Kwoka, E. Eric Boschmann ⇑, Andrew R. GoetzDepartment of Geography and the Environment, University of Denver, 2050 E. Iliff Ave., Denver, CO 80208, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 July 2014Received in revised form 28 July 2015Accepted 10 August 2015Available online 2 September 2015

Keywords:Light railTransitTravel behaviorsWorkplaceSustainable transportationDenver

Transit development is one planning strategy that seeks to partially overcome limitationsof low-density single use car oriented development styles. While many studies focus onhow residential proximity to transit influences the travel behaviors of individuals, theeffect of workplace proximity to transit is less understood. This paper asks, does workingnear a light rail transit station influence the travel behaviors of workers differently thanworkers living near a station? We begin by examining workers’ commute mode based ontheir residential and workplace proximity to transit station areas. Next, we analyze theways in which personal travel behaviors differ between those who drive to work and thosewho do not. The data came from a 2009 travel behavior survey in the Denver, Coloradometropolitan area, which contains 8000 households, 16,000 individuals, and nearly80,000 trips. We measure sustainable travel behaviors as reduced mileage, reducednumber of trips, and increased use of non-car transportation. The results of this studyindicate that living near a transit station area by itself does not increase the likelihood ofusing non-car modes for work commutes. But if the destination (work) is near a transitstation area, persons are less likely to drive a car to work. People who both live and workin a transit station area are less likely to use a car and more likely to take non-car modes forboth work and non-work (personal) trips. Especially for persons who work near a transitstation area, the measures of personal trips and distances show a higher level of mobilityfor non-car commuters than car commuters – that is, more trips and more distant trips. Theuse of non-car modes for personal trips is most likely to occur by non-car commuters,regardless of their transit station area relationship.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction and background

When considering cities, sustainability, and climate change, urban transportation and the use of private automobilesplays a fundamental role in terms of energy consumption and greenhouse gas emissions. While technological advancementssuch as hybrid cars offer solutions of reducing pollution and energy consumption, a more fundamental shift may comethrough behavioral changes, including a reduction in the need to travel (particularly by car), increased use of public transport(as well as walking and cycling), and a reduction in travel distances (Cervero and Murakami, 2010; Banister, 2011; Chatman,2013). Given the urban forms commonly characterized by low density, sprawling, automobile-oriented development, many

278 G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287

U.S. cities face challenges in shifting travel behaviors due to the lack of infrastructure that facilitates nearby and non-cartravel.

Transit development is one planning strategy that seeks to partially overcome these limitations of low-density single usecar oriented development styles. For example, transit oriented developments (TODs) are high density development projectswith mixed land uses (including housing, retail, services, and jobs), access to transit (bus or rail), a high level of walkability,and design principles that emphasize pedestrians over automobiles (Calthorpe, 1993; Ratner and Goetz, 2013). Theirobjective is to create an urban space that increases overall livability and quality of life. Indeed, TOD has gained supportby planners and policymakers for its potential to address many issues associated with unsustainable transportation systems.Specifically, the level of interest in transit oriented developments in the U.S. has been spurred by traffic congestion growth,distaste for suburban-style development, new desires for urban walkable lifestyles, costs of private transportation, andincreases in integrating environmental and sustainability consciousness with lifestyle choices (Ratner and Goetz, 2013).Demographic projections further suggest public demand for such developments will become greater in the future (Myersand Gearin, 2001).

A key concern of urban and transportation planning research is how urban form, including new transit corridors, transitstation areas, or TODs, influences travel behaviors. While the study of the relationships between built environment and tra-vel behaviors is not new (Ewing and Cervero, 2001, 2010; Guo and Chen, 2007; Boarnet, 2011), it has taken on greateremphasis recently within the context of sustainability and sustainable urbanism (Guo and Chen, 2007; Cervero, 2009;Chatman, 2013). The extensive body of literature examines whether different arrangements of the built environment (e.g.high or low density, balance of jobs and housing, land uses) affect travel behaviors of people, particularly looking for reduc-tions in automobile usage and associated environmental and social impacts. Most commonly, changes in travel behaviors aremeasured using a variety of metrics, including trip frequency, trip length, or mode choice. For example, does residence in onetype of built environment foster changes and reductions in these measures? However, self-selection bias suggests peoples’residential choice is influenced by their preferences in travel and commuting (Guo and Chen, 2007; Olaru et al., 2011), thus itbecomes difficult to disentangle the effects of the built environment on travel.

While transit development may seek to increase transit usage as a planning objective, some research shows that transitstation areas attract residents for a variety of reasons and that close proximity to transit does not measure highly. For exam-ple, TODs tend to attract smaller households without children, who make residential choices based on housing quality andtype, cost, and neighborhood quality (Lund et al., 2004; Lund, 2006; Dill, 2008). Furthermore, this same research indicatesmost TOD residents are not transit dependent but in fact still own cars, and are not as motivated by access to transit as wouldbe expected (see also Chatman, 2013). Rather, TODs attract persons seeking shorter non-work trip destinations whether bycar or walking, not necessarily by transit. These findings support early skepticism that TODs and neo-traditional neighbor-hoods do in fact reduce travel distances or increase transit usage among residents (Gordon and Richardson, 1997). Anddespite the growth of residential TODs, a large citizenry maintains a preference for ‘ideal’ residential neighborhoods oflow density single family suburban living (Loukaitou-Sideris, 2010).

Despite the growth of residential lifestyles in New Urbanist style developments such as TODs, the arguments above sug-gest living preferences are not necessarily motivated by proximity to transit. To date, studies conducted on TODs’ ability toinfluence travel behavior have shownmixed results. According to a study of TOD impacts on transit ridership in California byLund et al. (2006), TODs do confer meaningful ridership benefits, but complementary policies and programs may be neces-sary before hoped-for ridership gains can be met. Cervero (2007) acknowledges that transit oriented development producesan appreciable ridership bonus in California, but it is largely due to residential self-selection and employer policies thatreduce free parking and automobile subsidies. Dill (2008) found that residents of surveyed TODs in Portland are not transitdependent, although they did commute by transit at a significantly higher rate than residents citywide. Cao and Schoner(2014) show that the Hiawatha light rail transit (LRT) line in Minneapolis promoted transit use of residents who had livedin the corridor before its opening, but that residents who moved to the corridor after its opening use transit with the samefrequency as new residents in comparable urban corridors without LRT. This literature suggests that TOD area residents gen-erally tend to use transit more than residents outside of TOD areas, but that this relationship is not always conclusive, andthe reasons for higher transit use include other factors such as supportive policies.

To increase ridership, one key objective in urban transportation planning is to make transit more accessible to residents.However, new research proposes that transit stations located near workplaces are more effective than stations near resi-dences (Tsai, 2009), thus arguing for a stronger emphasis upon the non-residential components of transit developments, par-ticularly retail and employment. This is particularly evident as most transit trips are work or school related (Kim et al., 2007).For example, in California TODs, a sizable 26.5% of work trips for TOD residents were made by bus or rail, whereas only 8.1%of home based non work trips were made by bus or rail (Gard, 2007). In Chicago, Lindsey et al. (2010) found that a largepercentage of trips that originate from households close to transit also terminate at work destinations close to transit.The success of transit in Stockholm, Sweden suggests that a key design principle is to distribute industry and offices roughlyin proportion to residential population, i.e. to achieve a jobs-to housing balance (Cervero, 1996). In the U.S., high densitydevelopment of both jobs and housing along rail lines, however, is important for influencing travel behaviors from theresidential side of transit development. This is particularly true along the Arlington (VA) Metrorail corridor, where residentsare twice as likely to commute by transit than residents outside the corridor, and every 100,000 square feet of added officeand retail floor space increases average daily boardings by nearly 50 (Cervero, 2009). For San Francisco’s BART transit

G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287 279

systems to remain competitive with the private automobile, Cervero and Landis (1997) argued that station areas need tocapture even larger shares of future employment growth in addition to housing.

Much of the previous research has emphasized the relationship of transit station areas and TODs on work-related trips.But given the nature of transit developments as encompassing sites to fulfill many needs of everyday life, and that mosturban travel is non-work related (Santos et al., 2011), we examine travel behaviors of personal (i.e. non-work) trips. If a majorsustainability planning goal in urban transportation is to reduce negative travel behaviors, is it more effective to providetransit access nearer to residential populations, or workplace centers? With this in mind, this research seeks to answerthe question: does working near a transit station area influence personal travel behaviors more than living near a transit sta-tion area? Specifically we examine how worker proximity to light rail transit stations impacts non-work related personaltrips in Denver, Colorado.

2. Study design

The purpose of this study is to examine differences in travel behaviors between workers who live near transit stationareas versus workers who work near them. We hypothesize that positive (more sustainable) changes in travel behaviors(as measured by trip generation, mode choice, and trip distances) are influenced more by persons who work near a transitstation area than live near one. This has particular implications for urban transportation planning policy and practice.

The analysis is conducted in the Denver, Colorado metropolitan area, which had a 2010 population of approximately2.5 million persons. A decidedly automobile-oriented city, in the 1990s the Regional Transportation District (RTD) began aprogressive movement toward building a region-wide transit system, and in 2004 a voter-approved program (dubbed‘‘FasTracks”) called for extensive regional transit and land use development funded through sales tax increases. This planincludes future expansion of rail systems to northern communities and Denver International Airport, as well as bus rapidtransit lines, and the redevelopment of Union Station into a multimodal transit hub (Jonas et al., 2014). Our analysis focusesupon residences and workplaces along the existing Central, Southeast, and Southwest light rail lines (Fig. 1), which contain35 miles of track and 34 stations. Since the West corridor light rail line only began service in April 2013 – four years after thetravel diary data were collected – it was not included in this analysis.

At the same time, Denver’s transit oriented development has also been aggressive, as noted by the establishment of full-time staffed TOD programs at the City and County of Denver, Denver Regional Council of Governments (DRCOG) – themetropolitan planning organization, and RTD. Together, stated indicator goals were developed for TOD measuring successin Denver, including location efficiency, a rich mix of choices, value capture, place making and portal or entry point. In addi-tion to over 18,000 new housing units, over 9.3 million square feet of new and existing space for office, retail, government,

Fig. 1. Denver light rail system map, 2012 (Image source: www.rtd-denver.com).

280 G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287

cultural, medical and education are located within a half-mile of the existing light rail stations (Ratner and Goetz, 2013).Many of the light rail stations exhibit the primary characteristics of TODs: high density pedestrian oriented spaces with lightrail and bus line access, parking, bicycle infrastructure, and walkable access to retail, jobs, and housing developments. Whilethere is increasing TOD activity at each station, at the time the data was collected not all of the light rail stations exhibitedthe full suite of TOD features. Thus, the analysis considers the travel behavior effects of proximity to all light rail transitstation areas.

The data for this analysis come from the Front Range Travel Counts survey provided by DRCOG. This travel diary surveyincludes 6966 households and 16,210 persons, with 80,090 out-of-home trips. The data were collected from households on asingle weekday between October 2009 and June 2010. In the dataset 8011 individuals were identified as workers employedout of home. As a whole, these Denver metropolitan area workers are highly reliant upon the private automobile for both jobcommuting (83.9% car use) and personal non-work (83.7% car use) travel (Table 1). Since our focus is upon differencesbetween workers who live or work near transit stations, the study universe is reduced to 3439 workers. Some results of theirtravel behaviors will be compared with all area workers in the dataset (represented in Table 1).

To address our questions of travel behavior differences, the data were disaggregated by proximity to transit stations, aswell as modes of travel for work commute and personal trips. Preparation of the data for analysis was therefore conducted inthree phases (Fig. 2). First, all workplace and household locations of workers in the survey dataset were spatially referencedusing ArcGIS (Fig. 3). Proximity to a transit station area is defined at three different distance thresholds: a .5-mile straight-line buffer that is a standard TOD boundary from the literature (Guerra et al., 2012), a 1-mile buffer widely used in local poli-cies and reports (see Ratner and Goetz, 2013; DRCOG, 2010), and a 15-min walkshed that provides a more realistic scenarioof access to a station. With this approach, if a worker either lives and/or works near a transit station area at these threethresholds, we define this as the worker–transit relationship in all subsequent analyses. It can be assumed that by using apopulation whose homes or workplaces are within a relatively similar proximity to transit, each worker will have equalaccess to the services provided by the light rail. While Boarnet (2011) suggests using an experimental and control groupapproach for studying the effects of transit upon land use and travel behaviors between those living within and outsideof transit areas, our design excludes all workers who do not have a transit relationship as defined above. Second, each workerin our study sample is then identified by their ‘‘typical mode to work” as defined in the Front Range Travel Counts survey.These are identified as ‘car’ commuters (primarily using a private vehicle as driver or passenger), or ‘non-car’ commuters(walking, biking, light rail, or bus). The purpose here is to separate the effects of commute mode upon personal travelbehaviors. Finally, travel behavior measures for all personal (non-work) trips were extracted from the data, including numberof trips (generation), distance of trips (in miles), and mode of trips.

Table 1Worker travel characteristics, entire Denver metropolitan area survey respondents.

Total workers in survey n = 8011

Typical commute modeCar 83.9%Non-car 16.1%

Personal trips modeCar 83.75%Non-car 16.25%Average number of daily trips 4.51

Average trip distanceCar 4.51 milesNon-car 2.39 miles

Fig. 2. Analysis design: workers’ personal travel behaviors and commute mode by transit station area relationship.

Fig. 3. Map of survey respondent locations (red dots = workplaces, green dots = residences) and transit station area thresholds (walkshed and .5 mile and1 mile buffers). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287 281

3. Analysis and results

For analytical simplicity the broader research question, ‘does working near a transit station area influence the travelbehaviors of workers differently than workers living near one?’, is broken down into four sub-questions. (1) Is there a differ-

ence in the commute mode of workers either working near a transit station area, living near a transit station area, or living

and working near a transit station area? (2) Is there a significant difference in the number of non-work trips made between

those who drive to work and those who do not? (3) Is there a significant difference in the mode of non-work trips made

between those who drive to work and those who do not? (4) Is there a significant difference in the average mileage of

non-work trips between those who drive to work and those who do not? Each of these analyses is presented here.Since we are examining differences between groups (worker relationship with transit station areas), our analysis utilizes

statistical significance tests. While much of the built environment and travel behavior relationship literature utilizes regres-sion models (Ewing and Cervero, 2001, 2010), we deviate from this tradition, as we are not examining determinant factors ofobserved travel behaviors. Differences are measured between car commuters and non-car commuters in terms of the num-ber, distance and mode of personal trips made. In an instance where several trips were chained together, the chain was splitinto separate legs and recorded as multiple trips. Independent sample t- and z-tests were used for both the mean tripnumber and mean trip distance tests, while chi-square testing was used for trip mode. As in the case of the commute modeanalysis, personal trip mode groups were divided in the same manner as commute mode: ‘car’ or ‘non-car’. Outlier sampleswho recorded total trips or trip distances greater than ±3 standard deviations from overall means were removed from testing;trips identified as ‘‘walking transfers” were also eliminated. The results of these tests are compared between worker–transitrelationships within similar distance thresholds. Each section below corresponds to one of the four questions above.

3.1. Commute mode

The first analysis examines the ‘typical commute’ modes reported by each survey respondent. Table 2 illustrates howcommute mode varies significantly by worker–transit relationship type (household, workplace or both). Workers who liveand work, or work near light rail transit, at the 1-mile threshold, are more likely to use non-car modes (34.8% and 26.1%,respectively) for commuting than workers who only live near light rail (11.0%), highlighting the importance of transit accessat the workplace. When compared to the entire travel survey sample, workers livingwithin 1 mile of a transit station area are

Table 2Worker commute mode, by transit relationship and threshold.

Threshold Commute mode Worker–transit relationship

Workplace andHousehold

Household Workplace

n % n % n %

1 Mile Non-car 216 34.8 61 11.0 591 26.1Car 404 65.2 496 89.0 1671 73.9

Total 620 557 2262

.5 Mile Non-car 60 50.0 39 18.0 464 31.3Car 60 50.0 178 82.0 987 68.9

Total 120 217 1451

Walkshed Non-car 29 61.7 29 25.7 352 36.8Car 18 38.3 84 74.3 605 63.2

Total 47 113 957

Notes: Study universe (n = 3439) contains all workers with transit relationship: workplace and/or household near a transit station area. Walkshed is basedon 15-min walking distance.

282 G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287

also more likely (z = 3.227; p = 0.001) to use a car for work commuting (Table 2) than all survey respondents (Table 1) outsideof the transit station area zones in this analysis. While this particular result is surprising, other persons with close access tolight rail transit (here, workers who live and work, or only work within a mile of a station) are far more likely to use non-carmodes for commuting than the entire survey population. Typical commute mode changes further as the distance thresholddecreases (reading Table 2 vertically). While car commuting remains dominant for respondents either living or working intransit station areas, the share of non-car work commuting increases for all worker–transit relationships at the smallerthresholds. For those both living and working in transit station areas, the non-car mode share at the smaller thresholds isthe same or higher than for the car. The findings of particular note here are that: (1) the workplace–transit relationship ismore effective at non-car commuting than is the residential–transit relationship, and (2) those who both live and work ina transit station area are most likely to use non-car modes in their commutes. In the subsequent analyses we focus upondifferences in personal travel behaviors, and given that commutemode influences personal trips, we use the typical commutemodes defined here as one distinguishing parameter.

3.2. Number of personal trips

Trip generation indicates the total number of trips made by an individual. Here we extract the number of non-work per-sonal trips taken by each worker to determine if their proximity to a transit station area is influential. In all instances non-carcommuters make more mean number of personal trips than do car commuters; however, only the shaded rows are statis-tically significantly different from each other (Table 3). This suggests a kind of elevated level of mobility occurring due tocommute mode, but this is not an unusual finding for travel behaviors in TODs (Lund et al., 2004). Persons who work neara transit station area appear to have some of the highest average personal trips. For instance, at the 1-mile threshold personswho work near a transit station area make more personal trips than persons who live near one. This is true both for thosewho commute by car (t = 9.607; p < 0.0000) as well as non-car commuters (t = 2.8071; p < 0.01). This too is not unexpected,and illuminates the effect of workplaces near transit stations as dense areas of urban development that provide many oppor-tunities for individuals to conduct personal activity trips.

3.3. Personal trip mode

Next we analyzed the effects of transit station area proximity on transport mode of personal trips. Here again the analysisbecomes complex as we make distinctions between persons and their commuting mode. Generally, car commuters use a pri-vate automobile for around 90% of their personal trips while non-car commuters use a car for half of their personal travels.Within each worker–transit relationship, and at each threshold distance, there is a statistically significant difference in themode of personal trips between non-car commuters and car commuters (Table 4). For instance, persons who work within1 mile of a transit station area made 10,674 personal trips. Those workers who used non-car commute modes also usednon-car modes for 55.2% of their personal trips, while car commuters only made 5.7% of their personal trips usingnon-car modes. These relationships are consistent throughout the subgroupings in Table 4, and highlight the expecteddependence upon car travel from work commutes to personal travel. At the same time, we see that 50–60% of all personaltrips made by non-car commuters in all worker–transit relationships are done using non-car modes. What we do not knowfrom this dataset is the decision making process and the reasons for mode choices in personal trips.

Table 3Personal trip generation, by transit relationship and threshold.

Notes: ‘n’ denotes total number of personal trips. Shaded areas are significantly different trip means between non-car and car commuters, 2-tailed t-test,p < 0.000.

Table 4Personal trip travel mode, by transit relationship and threshold.

Notes: Shading indicates differences between commute mode that are statistically significant, 2-tailed Pearson Chi-square, p < 0.000.

G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287 283

3.4. Personal trip distance

Finally, we examined how transit station areas affect personal trip distances. An initial look at Table 5 suggests that per-sonal trips are longest (in miles) for people who work near a transit station area. There is no statistical difference in personaltrip distances between persons who both live and work in a transit station area and those who live in one. There is, however,a statistical difference (for all thresholds and modes, except one) between persons living in a transit station area and those

Table 5Personal trip distances, by worker–transit relationship, threshold and trip mode.

Notes: Statistical tests compared differences in mean miles between comparable threshold and trip mode. Differences between Workplace/Household withHousehold are not significant. Shading indicates significant differences between Household andWorkplace transit relationships (2-tailed t-test, p < 0.0000).

284 G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287

who work in one. Whether by car or by non-car modes, persons who work in a transit station area make longer personal tripsthan do persons who live in one. In some instances the difference is nearly 4 miles. For example, the persons in the walkshedthreshold using non-car modes made trips on average of 2.63 miles (living near transit) versus 6.94 miles (working neartransit). In a further disaggregation of the data, there are significant differences in personal trip distances based upon thetypical commute mode (Table 6). What is clear is that non-car commuters are more willing to make non-car personal tripsthat are further than those who typically drive to work. For example, among persons who work within 1 mile of a transitstation area, car commuters made 426 personal trips by non-car mode, and non-car commuters made 1766. These personaltrips by non-car commuters were also nearly 3 miles further, on average, than for car commuters. This certainly highlightsdiffering travel behavior tendencies of persons who are comfortable making a variety of trips via non-car modes. But expla-nations remain elusive here, calling for further studies.

4. Discussion and conclusion

Several key findings of this analysis are summarized as follows. Living near a transit station area by itself does notincrease the likelihood of using non-car modes for work commutes. But as would be expected, if the destination (work) isnear a transit station, persons are less likely to drive a car to work. Especially for persons who work near a transit stationarea, the measures of personal trips and distances show a higher level of mobility for non-car commuters than carcommuters – that is, more trips and more distant trips. The use of non-car modes for personal trips is most likely to occurby non-car commuters, regardless of their transit relationship.

In terms of transit relationships influencing commute modes, our findings here are consistent with research that suggestsincreased proximities to transit stations leads to decreased levels of car-based commuting (Dill, 2008). Here, persons withboth households and workplaces near transit station areas are most likely to use non-car modes, and persons with only theirhousehold near transit station areas are least likely to use non-car commute modes. This might suggest that living near lightrail transit does less to influence alternative commuting modes than does working near places well serviced by transit. How-ever, it should be noted that the marginal increase in non-car commuters is greater for those thresholds nearest the transitstation areas. This supports the implication that increasing household proximity (particularly within walking distance) totransit stations areas can increase the use of alternative travel modes (Cervero, 2001; Lee and Senior, 2013).

While it is intuitive that increased workplace proximity to transit opportunities will lead to higher levels of non-car usagefor job commuting, a broader implication is upon how personal trips are subsequently influenced (Cairns et al., 2010). Othersfound that personal trips were less frequent for TOD users than commute trips (Lund et al., 2004; Dill, 2008). While this maybe true at an aggregate scale, these studies did not analyze disaggregate personal travel based on job commute mode, as wasdone here. In all situations, workers who commute to work by non-car modes showed higher levels of trip generation fortheir personal travels. This indicates that non-car commuters need more trips by walking, biking, or transit to accomplishsimilar errands otherwise executed by the automobile.

Table 6Personal trip mean distance (miles), by transit relationship, threshold, and modes.

Notes: ‘n’ denotes total number of personal trips. Shading indicates statistically significant differences in personal trip miles, between commute mode, bypersonal trip mode (2-tailed t-test, p < 0.001).

G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287 285

Similarly, when analyzing the results of the trip distance tests, there are cases where non-car commuters made longerpersonal trips than car commuters, and others where car commuters made longer personal trips than non-car commuters.However, there are no instances where non-car commuters travel farther distances by the private automobile than car com-muters for personal trips. Again, these may be situations where non-car commuters are more reliant upon transit, walking,or biking modes for their commute, and thus are using them more extensively for personal travel as well. Another explana-tion is that voluntary non-car commuters develop a greater sense of comfort and confidence with these modes and perhaps ahigher level of assurance to use them for successfully accomplishing other tasks beyond commuting.

However, in the cases of workers with workplaces located within a whole and half mile of transit station areas, car com-muters make longer personal trips with the private automobile than do non-car commuters. In these instances, transit istheoretically capable of serving these workers’ commuting needs, though they choose to use the private automobile toaccomplish their work trips instead. This may suggest that workers in these groups have no interest at all in transit, and thismindset could also be applied to travel in their personal lives as well. Thus, if workers in these groups do in fact need to usetransit for their personal travel, they may likely only use it as minimally as possible. This theory could also be used to explainwhy non-car commuters take significantly longer personal trips with non-car modes than do car commuters. Personal per-ception may also play a role in explaining these results. To many car commuters who travel long distances multiple timeseach week to work, the relative distances required to complete personal errands may seem short in comparison. Conversely,non-car commuters may feel that exactly the same distances required to complete personal errands are too long, since when

286 G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287

compared with their own commute, the personal trip may be perceived as requiring too much excessive driving. As an exam-ple, a person who drives forty minutes to work each day may perceive a twenty minute drive to a store as being short,whereas a person who rides their bike to work each day may perceive that same twenty minute drive to the store as beingtoo long.

For personal trips, non-car commuters are much less likely to use a private automobile for personal travel than car com-muters. Non-car commuters made less than 50% of personal trips by car, while car commuters made upwards of 90% by car.This is an expected finding, with many plausible explanations: the commute is often the foundation around which all othertravel is scheduled (Redmond and Mokhtarian, 2001); non-car commuters may be more receptive to accomplishing dailytasks without the private automobile, especially if they are working and/or living in a higher density, mixed use transit area;and non-car commuters’ personal choices may be motivated by mindfulness of sustainability and the impacts of theirtransportation.

The primary motivation of this study was upon how different transit station area relationships affect travel behaviors ofworkers. In most cases, statistically significant differences in personal travel behaviors between non-car commuters and carcommuters occur in groups where workers’ workplaces are near transit station areas. An implication of this might be thatlocating workplaces nearer to transit or providing transit services to employment centers could be a more effective wayof encouraging greater use of non-car modes beyond the work commute. Since significant differences were not identifiedbetween most of the commute groups with households near transit station areas, this suggests that workplaces locatednearer to transit may be more effective at promoting the use of more non-car modes than locating households nearer to tran-sit. This idea is shared by studies of TODs’ impacts on travel behaviors (Tsai, 2009; Dill, 2008; Cervero, 2009; Lee and Senior,2013). For planning policy purposes, these findings might enhance the argument to reduce zoning constraints and allow formore retail and services developments in transit station areas or TODs that are otherwise zoned strictly for office or indus-trial space. However, we are not implying that the clustering of households near transit is unimportant. After all, it was thesubgroups of workers both working and living near transit station areas who in fact used non-car modes most frequently forcommuting and personal travel. As a result, it should be emphasized that clustering both residential areas and employmentcenters around transit stations is perhaps the best planning strategy (Cervero, 1996).

Finally, this research also revealed mixed and unexpected findings with regards to sustainable travel behaviors. As notedearlier, much of the climate change and sustainability dialogue for urban travel argues for reduced auto usage, reduced dis-tances travelled between places, and reducing the need to travel (Cervero and Murakami, 2010; Banister, 2011; Chatman,2013). The findings do support the conclusion that workplace transit station areas are far superior to residential transit sta-tion areas in influencing persons toward non-car modes for personal and work trips. Yet unexpectedly, persons who usemore non-car modes are making more trips and more distant trips, and this finding is true across the transit relationships.This confounds some traditional measures of sustainable travel behaviors, and suggests two important implications. First,the ‘build it and they will use transit’ expectation of residential-led transit development may be misguided (see Chatman,2013), and planning strategies should emphasize the importance of transit, walking, and biking access to clusters of highdensity workplaces. Second, it may be worth considering that higher measures of trip generation and trip distance arenot unsustainable behaviors if they are conducted by non-car modes of travel.

There are specific limitations to this particular study. First, as noted earlier, our research design does not use anexperimental control group approach to compare travel behavior differences of transit station areas relative to the largermetropolitan population, beyond simple aggregate comparisons (Boarnet, 2011). We therefore cannot generalize how transitstation areas might make improvements across the entirety of the working population. Also, it should be noted that the datacollected in this travel diary occurred during the Great Recession of 2008–2009. Undoubtedly this economic collapse createdmany changes in personal lifestyles due to work layoffs, higher fuel prices, and housing foreclosures. Yet any potentialimpacts are not explicitly addressed in our study.

Furthermore, some transit station areas included in this study were farther along in their TOD evolutionary scale thanothers. While it is reasonable to expect stations with more advanced TOD characteristics like those in the ‘‘downtown” areasto have greater impacts than those in more rudimentary stages, our analysis did not consider these spaces separatelysince our focus was to explore the overall effect of commute mode on non-work travel. Future analyses could study thecontribution of each specific light rail station in greater depth. Similarly, our study does not attempt to correlate the amplesocio-economic and built environment factors that contribute to the observed travel behaviors. As transit stations in Denverfully evolve into transit oriented developments, future studies might make comparisons to these findings and offer insight onwhether travel behaviors differ any further in more complete TODs. Future studies could more explicitly look at howpersonal and work travel behaviors within these different transit relationships are impacted by a variety of factors suchas weather, income, weight, parking pricing, disabled persons, gas prices, and households with school-age children. Alongthese same lines, the travel diary dataset does not allow us to understand the decision-making process of why persons maketheir travel and residential choices. This also includes the inability to disentangle potential self-selection biases of personswith personal economic-environmental motivations to live/work near transit station areas, use non-car modes, orpreferences toward daily interactions with mixed-use dense urban environments. Future studies should build upon existingqualitative and survey research to better understand personal level motivations.

To conclude, this research finds that workplace proximity to transit station areas has greater influences upon the com-mute and personal trip mode of workers than residential proximity to transit station areas, although both working and livingnear transit station areas results in the highest levels of non-car commuting and personal (nonwork) travel. At the same

G.J. Kwoka et al. / Transportation Research Part A 80 (2015) 277–287 287

time, given the findings that non-car commuters have longer and more personal trips (albeit in a less auto-dependent fash-ion) we argue for nuanced discussions on the measures of sustainable transportation indicators. Given the singular contexthere, more research is needed to corroborate these findings in other cities with growing transit developments.

References

Banister, D., 2011. Cities, mobility and climate change. J. Transp. Geogr. 19, 1538–1546.Boarnet, M.G., 2011. A broader context for land use and travel behavior, and a research agenda. J. Am. Plann. Assoc. 77, 197–213.Cairns, S., Newson, C., Davis, A., 2010. Understanding successful workplace travel initiatives in the UK. Transport. Res. Part A 44, 473–494.Calthorpe, P., 1993. The Next American Metropolis: Ecology, Community, and the American Dream. Princeton Architectural Press, New York, NY.Cao, X., Schoner, J., 2014. The influence of light rail transit on transit use: an exploration of station area residents along the Hiawatha line in Minneapolis.

Transport. Res. Part A: Policy Pract. 59, 134–143.Cervero, R., 1996. Mixed land-uses and commuting: evidence from the American Housing Survey. Transport. Res. Part A 30, 361–377.Cervero, R., 2001. Walk-and-ride: factors influencing pedestrian access to transit. J. Public Transport. 3, 1–23.Cervero, R., 2007. Transit-oriented development’s ridership bonus: a product of self-selection and public policies. Environ. Plann. A 39 (9), 2068–2085.Cervero, R., 2009. Public transport and sustainable urbanism: global lessons. In: Curtis, C., Renne, J.L., Bertolini, L. (Eds.), Transit Oriented Development:

Making it Happen. Ashgate, Farnham, England, pp. 23–35.Cervero, R., Landis, J., 1997. Twenty years of the Bay Area Rapid Transit system: land use and development impacts. Transport. Res. Part A 31, 309–333.Cervero, R., Murakami, J., 2010. Effects of built environments on vehicle miles traveled: evidence from 370 US urbanized areas. Environ. Plann. A 42, 400–

418.Chatman, D.G., 2013. Does TOD need the T? On the importance of factors other than rail access. J. Am. Plann. Assoc. 79, 17–31.Dill, J., 2008. Transit use at transit oriented developments in Portland, Oregon area. Transport. Res. Rec. 2063, 159–167.DRCOG (Denver Regional Council of Governments), 2010. Who is TOD in Metro Denver? Benchmarking the Evolution of TOD in Metro Denver. DRCOG,

Denver, Colorado.Ewing, R., Cervero, R., 2001. Travel and the built environment: a synthesis. Transport. Res. Rec. 1780, 87–114.Ewing, R., Cervero, R., 2010. Travel and the built environment: a meta-analysis. J. Am. Plann. Assoc. 76, 265–294.Gard, J., 2007. Quantifying transit oriented development’s ability to change travel behavior. Inst. Transport. Eng. 77, 42–46.Gordon, P., Richardson, H.W., 1997. Are compact cities a desirable planning goal? J. Am. Plann. Assoc. 63, 95–106.Guerra, E., Cervero, R., Tischler, D., 2012. Half-mile circle: does it best represent transit station catchments? Transport. Res. Rec. 2276, 101–109.Guo, J.Y., Chen, C., 2007. The built environment and travel behavior: making the connection. Transportation 35, 529–533.Jonas, A.E.G., Goetz, A.R., Bhattacharjee, S., 2014. City-regionalism as a politics of collective provision: regional transport infrastructure in Denver, USA.

Urban Stud. 51, 2444–2465.Kim, S., Ulfarsson, G.F., Todd Hennessy, J., 2007. Analysis of light rail rider travel behavior: Impacts of individual, built environment, and crime

characteristics on transit access. Transport. Res. Part A: Policy Pract. 41, 511–522.Lee, S., Senior, M., 2013. Do light rail services discourage car ownership and use? Evidence from census data for four English cities. J. Transp. Geogr. 29, 11–

22.Lindsey, M., Schofer, J.L., Durango-Cohen, P., Gray, K.A., 2010. Relationship between proximity to transit and ridership for journey-to-work trips in Chicago.

Transport. Res. Part A 44, 697–709.Loukaitou-Sideris, A., 2010. A new-found popularity for transit-oriented developments? Lessons from Southern California. J. Urban Des. 15, 49–68.Lund, H., 2006. Reasons for living in a transit-oriented development, and associated transit use. J. Am. Plann. Assoc. 72, 357–366.Lund, H., Cervero, R., Willson, R.W., 2004. Travel Behavior of Transit-oriented Development in California. Final Report. California Department of

Transportation, Sacramento.Lund, H.M., Cervero, R., Willson, R.W., 2006. A re-evaluation of travel behavior in California’s TODs. J. Archit. Plann. Res. 23 (3), 247–263.Myers, D., Gearin, E., 2001. Current preferences and future demand for denser residential environments. Hous. Policy Debate 12, 633–659.Olaru, D., Smith, B., Taplin, J.H.E., 2011. Residential location and transit-oriented development in a new rail corridor. Transport. Res. Part A 45, 219–237.Ratner, K.A., Goetz, A.R., 2013. The reshaping of land use and urban form in Denver through transit-oriented development. Cities 30, 31–46.Redmond, L., Mokhtarian, P., 2001. The positive utility of the commute: modeling ideal commute time and relative desired commute amount.

Transportation 28, 179–205.Santos, A., McGuckin, N., Nakamoto, H.Y., Gray, D., Liss, S., 2011. Summary of Travel Trends: 2009 National Household Travel Survey. U.S. Department of

Transportation, Federal Highway Administration Report, No. FHWA-PL-ll-022.Tsai, Y., 2009. Impacts of self-selection and transit proximity on commute mode choice: evidence from Taipei rapid transit system. Ann. Reg. Sci. 43, 1073–

1094.