Bicycle-metro integration in a growing city: The determinants of cycling as a transfer mode in metro station areas in Beijing

Transportation Research Part A 99 (2017) 46–60

Contents lists available at ScienceDirect

Transportation Research Part A journal homepage: www.elsevier.com/locate/tra

Bicycle-metro integration in a growing city: The determinants of cycling as a transfer mode in metro station areas in Beijing Pengjun Zhao ⇑, Shengxiao Li College of Urban and Environmental Sciences, Peking University, China

a r t i c l e

i n f o

Article history: Received 6 May 2016 Received in revised form 22 December 2016 Accepted 6 March 2017

Keywords: Cycling Bicycle-metro integration Bicycle-and-Ride Bicycle-sharing program Beijing

a b s t r a c t Bicycle-transit integration, in which cycling is used to as a transfer mode to/from transit station is widely believed to be one very important way of promoting a transit city and achieving efficient and sustainable urban transport systems. However, the empirical evidence for the determinants of people’s choices to transfer by bicycle as a travel mode remain largely unstudied. This paper investigates this issue, using Beijing and its metro system as a case study. Using a multilevel logistic model, we found that travel distance is the most important influence on rates of cycling for transfer trips between metro stations and home or workplace. There were also socioeconomic influences, with young people being less likely to cycle and more likely to use buses. Middle- and high-income earners were more likely to drive than cycle, while low-income earners were more likely to take the bus. Personal attitudes are also influential—those who prefer cheap travel were more likely to cycle. Above results suggest that the increasing city size and urban expansion are great challenges to cycling systems in growing cities. The presence of bicycle-sharing programs, mixed land use, and green parks in metro station areas were associated with greater rates of cycling transfer. In order to promote Bicycle-and-Ride schemes in metro station areas, education initiatives designed to influence behavior should be integrated with investment in bicycle infrastructure. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Cycling is considered to be a sustainable, economic and convenient mode of transport because of its low cost and moderate degrees of travel speed and flexibility (Akar and Clifton, 2009). It also contributes to people’s physical health and decreases their likelihood of being overweight (Lawlor et al., 2003). Compared with motorized transportation modes, cycling generates no air pollution and consumes no fossil fuels. Moreover, cycling is considered to be the most equitable form of transportation (Pucher and Buehler, 2008) because it is affordable for most low-income earners. In some developing countries, cycling is now regarded as a travel mode used mainly by people with low incomes (Nkurunziza et al., 2012; de Dios Ortuzar et al., 2000). The renaissance of cycling is regarded as an effective way to alleviate transportation problems caused by car dependence, such as air pollution and traffic congestion (Moudon et al., 2005). There are many policies designed to encourage cycling, such as bicycle-sharing programs (for a review, see DeMaio, 2009), the improvement or provision of bicycle lanes (e.g. Pucher and Buehler, 2008) and the improvement or creation of ⇑ Corresponding author at: Room 3267, Yifu 2 Building, College of Urban and Environmental Sciences, Peking University, 5 Yiheyuan Road, Beijing 100871, China. E-mail addresses: [email protected] (P. Zhao), [email protected] (S. Li). http://dx.doi.org/10.1016/j.tra.2017.03.003 0965-8564/Ó 2017 Elsevier Ltd. All rights reserved.

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

47

a cycling-friendly environment (Cervero and Duncan, 2003). Additionally, policies designed to discourage car use, such as congestion charging and road pricing, could benefit the promotion of cycling (Buehler et al., 2016). Recently, attention has turned to the integration of bicycle and transit systems (e.g. Martens, 2004; Rietveld and Daniel, 2004; Wang and Liu, 2013; Singleton and Clifton, 2014). This integration aims to encourage transit passengers to use a bicycle as a transfer mode to and from transit stations. In such integrated systems, the bicycle is a desirable feeder mode of travel for trips to and from transit stations (Rietveld, 2000a). One reason is that cycling has a higher speed than walking and a more flexible service than public transport (Keijer and Rietveld, 2000). Another reason is that in most countries, cycling is free or much cheaper than buses for trips to transit stations. A third reason is that in large cities, for example, Paris, London, Beijing, and Amsterdam, many residents live in the suburbs. The ‘‘last mile” between home and train or metro stations is a major factor influencing residents’ usage of train or metro systems. ‘‘Bicycle + transit” provides a chance to promote transit ridership in large cities. Therefore, it is considered to be an effective way to promote both transit and cycling (Bachand-Marleau et al., 2011), and to reduce car use in transit station corridors (Martens, 2004). This has been promoted by policymakers and planners worldwide, for example, in northern European countries such as the Netherlands and Denmark, and in some developing countries, for instance, China, Colombia, and Brazil. ‘‘Bicycle + transit” is also encouraged in some car-dependent cities, such as in the USA, Australia, and New Zealand, although bicycle trips to/from stations still account for only a small proportion of all trips in these countries (Wang and Liu, 2013). Bicycle-transit integration has recently become an important research theme attracting greater research attention (La Paix and Geurs, 2015; Keijer and Rietveld, 2000). In the field of research on bicycle-transit integration, several research gaps need to be filled. Firstly, studies exploring use of bicycles as a transfer mode to underground train (metro) station areas in cities remain scarce. The existing literature mainly focuses on the use of cycling to and from regional railway or bus stations (e.g. La Paix and Geurs, 2015; Wang and Liu, 2013; Heinen and Bohte, 2014). There are differences in using bicycles as a transfer mode between metro station areas and railway/bus stations areas. In a city with metro lines, especially in large cities, bicycle-metro integration could be more important than bicycle-railway integration. The amount of cycling needed for metro trips could be much higher than for trips to railway stations in a city. One reason for this is that metro stations are more numerous than railway stations. Another reason is that urban metro services are usually intended for daily intra-city transport, such as commuting. Therefore, integration with cycling will be advantageous for most residents. However, railway stations mainly provide services for inter-city transport. Bicycle-bus integration is also an important way of using bicycles as a transfer mode in some countries, in particular, cities in Europe. However, in some dense Asian cities, ‘‘bus + bicycle” systems play a limited role in promoting cycling. One reason is that bus stops are densely distributed and most bus passengers integrate bus travel with walking, rather than cycling, as a transfer mode. Another reason is that bus stops are usually small spaces, and parking spaces for bicycles are very limited, particularly in central city areas. This makes cycling unattractive to many bus passengers. The other reason is that buses provide a lower proportion of trips for long distance travel than metro trains in large cities. Metro systems have much higher speeds and are more punctual than buses. Increasing road traffic congestion tends to make road buses less attractive in dense Asian cities, such as those in China. As a result, the rate of cycling to buses stations is much lower than to metro stations. In this sense, studies on bicycle-metro integration could be more important to policymaking and promoting cycling in large Asian cities such as Tokyo, Beijing and Seoul. Secondly, attitude variables, especially respondents’ attitudes towards bicycles, have rarely been taken into account in the existing literature. Thirdly, the impacts of public bicycle sharing programs in transit station areas have scarcely been examined. Public bicycle-sharing programs have been implemented in many countries. It has been claimed that bicycle sharing could significantly promote bicycle-transit integration (DeMaio, 2009; DeMaio and Gifford, 2004). However, empirical evidence for this claim remains lacking in the literature. Fourthly, the existing literature on bicycle-metro integration has mainly been conducted in Western countries, which are characterized by high car prevalence. However, case studies in growing cities in developing countries are absent. In developing countries, cycling used to be a major travel mode, including as a transfer mode to access transit services. However, cycling rates have been declining in many countries such as China (Zhao, 2014; Kenworthy and Hu, 2002). Exploring the determinants of bicycle-metro integration in growing cities in developing countries will add to the exiting literature, and aid bicycle renaissances in these countries. This paper attempts to fill the above research gaps by looking at the case of Beijing and exploring the determinants of metro passengers’ use of the bicycle as a transfer mode for their trips to/from metro stations. China used to be called the ‘kingdom of the bicycle’, as cycling accounted for the largest proportion of trips in Chinese cities, even in the 1980s. However, it has since lost this title due to rapid motorization, particularly in large cities (SCMP, 2015). In Beijing, the share of all cycling trips decreased from 62.7% in 1980 to 13.9% in 2012 (BTRC, 2014). Both the central and Beijing municipal governments have introduced many policies to encourage cycling, such as a bicycle-sharing program and controls on car use. At the same time, a huge amount of investment has been put into improving metro services. Since 2007, on average, 100 km of new metro lines have been built annually. The total length of metro lines in Beijing will reach 550 km by the end of 2015. Trips by metro have already reached 10.73 million per day on work days, with an additional 1.5 million on Fridays (Du, 2015). This provides a great opportunity to encourage bicycle-metro integration in Beijing. However, to date, little attention has been paid to bicycle-metro integration as part of government policy. One of the reasons for this is that empirical evidence for cycling as a transfer mode to metro station areas in Chinese cities remains scarce. While there are some studies on cycling, they are mainly focused on people who only use bicycles for commuting (Zhao, 2014). This paper aims to specifically examine the determinants of cycling as a transfer mode between metro stations and homes and workplaces in Beijing. It examines a passenger’s transfer mode either for the trip between home to metro station or the

48

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

trip between workplace and metro station. The metro stations, which were selected in the study, could be either a metro station where a passenger embarked or disembarked. The factors which could affect cycling as a transfer mode are investigated, including passenger socioeconomic characteristics, personal preferences for travel mode, the built environment in metro stations, and availability of public transportation, facilities and services. This study has important policy implications. The share of total trips by cycling decreased from 61% in 1985 to 16.7% in 2014. Of all transfer modes between metro stations and homes or workplaces, cycling still occupies only a small proportion. It was only 7^ according to the survey data used in this study (Table 1). There is a great need to promote cycling as a feeder mode for trips between metro stations and homes or workplaces. As mentioned in the Introduction, the encouragement of bicycle-transit integration could be helpful, not only to promote cycling, but also to increase transit ridership. In China’s large growing cities such as Beijing, trip distances have been increasing due to the rapid growth of the city’s area. Trip distance is one of the most important factors influencing cycling rates. This point was supported by the findings of this study (Table 4). It means cycling as a transfer mode may continue to decline if policy interventions are absent. The results of the analysis in this paper could be valuable for policy making with regards to promoting cycling in China’s large cities, and in other cities around the world. The paper is organized into the following sections: Section 2 provides a literature review on the determinants of bicycle travel and bicycle-transit integration; Section 3 offers an introduction to the survey, data and models used in the paper; Section 4 presents the results of our analysis of the data; Section 5 presents a discussion of the findings and policy implications; while Section 6 presents our conclusions. 2. Literature review There are usually two types of bicycle use, depending on whether it is combined with other transport modes. One type is as a transfer mode to access other travel modes, such as train and car. The other is as a direct transport mode without combination with other travel modes. When bicycles are used as a transfer mode, cycling trips have some special characteristics. For instance, travel distances may be shorter, and cycling may be more significantly affected by the availability of reliable transit services and bicycle parking facilities at transit stations. There is a third type of bicycle use where transit passengers are allowed to take their bicycles on the public transport system, usually on trains. This occurs in some countries such as the UK and the Netherlands. However, in China, bicycles are not allowed to be taken on trains or metros. Therefore, this case is not addressed in this paper. Below, we review previous studies which have examined the determinants of bicycle use as a transfer mode to/from transit stations. Some of the literature explores the determinants of cycling as a transfer mode to metro stations. However, this literature is scarce. Most of the literature is about the determinants of general bicycle use. The socio-ecological model is widely used to explain the factors influencing active travel (e.g. Green et al., 1996; Stokols et al., 1996; Sallis et al., 2006). According to the socio-ecological model, cycling is affected by both personal socioeconomic factors and structural context. The built environment, transportation infrastructure, and services are important factors of the structural context. The natural environment, social contexts, and the policies designed to encourage cycling also belong to the structural context (for a more detailed review, see Heinen et al., 2010; Zhao, 2014). In this study, the theory of the socioecological model is applied. The factors of personal socioeconomic attributes, the built environment, and other factors are discussed. The literature review section is organized as follows: firstly, the effects of individual and socio-economic attributes are reviewed; secondly, the impacts of the built environment, transport infrastructure and services are discussed; and thirdly, the effects of other factors are reviewed. 2.1. Individual and socio-economic attributes Many studies have already found that cycling is significantly correlated with the socioeconomic features of the individual and their family. Men (Garrard et al., 2008; Stinson and Bhat, 2004) and young people (Stinson and Bhat, 2004) are more likely to cycle than others. Women are less likely to cycle, partly due to personal safety concerns (Dill and McNeil, 2013). However, in some cases, women cycle more for commuting trips than men, such as in Belgium (Witlox and Tindemans, 2004). When it comes to ethnicity, some studies have found that ethnic minority groups are less likely to cycle than others (Rietveld and Daniel, 2004; Parkin et al., 2008; Moudon et al., 2005). Household income and car ownership play different roles in cycling in different contexts. Generally, in countries where cycling dominates transport culture, such as the Netherlands, Germany, and Denmark, people who use the bicycle as a transfer mode to and from transit stations exhibit mixed social statuses (Krygsman et al., 2004; Pucher and Buehler, 2008). In developing countries, a higher household income is associated with a lower probability of cycling (Guo et al., 2015), such as in Chile (de Dios Ortuzar et al., 2000) and Tanzania (Nkurunziza et al., 2012). In these countries, cycling is often regarded as the main travel mode for low-income earners because they simply cannot afford a car. However, in some developed countries, high-income earners have a higher likelihood of cycling than low-income earners. This was found, for example, in the USA (Zahran et al., 2008) and in the UK (Parkin et al., 2008). This may not be the case when it specifically comes to bicycletransit integration. For example, Wang and Liu (2013) found that in the USA, trips to transit stations by bicycle are mostly undertaken by those who have a low level of education or come from a low-income group. In relation to the effects of car

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

49

ownership, some scholars report that a high level of car ownership is related to greater rates of cycling (Dill and Carr, 2003; Buehler and Pucher, 2012; Parkin et al., 2008) because of the intrinsic relationship between income and car ownership. However, a higher level of car ownership may be related to lower rates of cycling, such as in the USA where car travel competes with cycling, even for short trips (Wang and Liu, 2013; Pucher et al., 1999). People’s attitudes towards cycling may affect their choice to use it to transfer to transit station areas. Unsurprisingly, cyclists have a more positive attitude towards cycling than non-cyclists (Gatersleben and Uzzell, 2007; Gatersleben and Appleton, 2007). People who believe cycling to be a time-saving, comfortable and convenient travel mode have a greater preference for cycling (Heinen et al., 2011). More positive perceptions of the built environment of railway station areas, bicycle parking sites, and railway services by railway passengers could all contribute to higher rates of bicycle-railway integration (La Paix and Geurs, 2015, 2016). Those who have more positive attitude towards both bicycle and public transport are more likely to become bicycle-transit commuters (Heinen and Bohte, 2014). People with a high level of environmental awareness cycle more than those who don’t (Dill and Voros, 2007). Perceived traffic safety (Akar and Clifton, 2009; Dill and McNeil, 2013; Parkin et al., 2008; Advani and Tiwari, 2006) and perceived social safety (Parra et al., 2011) are also important considerations of whether to cycle to and from metro stations. In many cases, cycling may be the result of ‘self-selection’ (Xing et al., 2010), where people who choose to live in communities favorable to cycling are more likely to cycle than others.

2.2. Built environment, transport infrastructure and service The term bikeability is used to indicate whether the features of the built environment are perceived as a help or hindrance to cycling (Wahlgren and Schantz, 2012). It is also related to cyclists’ perceptions of their route’s environment, and their well-being. The built environment and transportation system affect the bikeability of a specific area (Winters et al., 2013; Wahlgren and Schantz, 2012). Low population density is related to low cycling rates (Rietveld and Daniel, 2004; Dill and Voros, 2007). However, over-density might reduce cycling because of crowding and high traffic volumes. Therefore, cycling is most common in medium-density areas (Rietveld and Daniel, 2004). Mixed land use is seen as an important indicator of environments that encourage cycling (Cervero and Duncan, 2003). For instance, residential developments mixed with a high level of other daily living facilities (e.g. convenience stores, fast food restaurants, hospitals) could increase cycling rates (Moudon et al., 2005). When it comes to cycling as a transfer mode to and from transit stations, destination accessibility and the built environment in the transit station areas seem to play important roles. Previous studies have found that a transit passenger’s maximum tolerance of the cycling distance between home and a transit station ranges from 1.2 to 3.7 km (Taylor and Mahmassani, 1996; Rastogi and Krishna Rao, 2003; Krygsman et al., 2004). Therefore, a high level of walking connectivity, and walking-friendly built environments near transit station areas, are vital to bicycle-transit passengers (La Paix and Geurs, 2015, 2016). Bicycle infrastructure directly affects cycling rates. Generally, a greater number of bicycle lanes contributes to a higher probability of cycling (Akar and Clifton, 2009; Moudon et al., 2005; Krizek et al., 2007; Dill and Voros, 2007; Krizek and Roland, 2005). When the distance between the trip origin and a bicycle trail is over 2.5 km, the probability of cycling decreases dramatically (Krizek et al., 2007). In particular, cyclists with low incomes or over 30 years old seem to be more sensitive to bicycle network connectivity (Guo et al., 2015). Moreover, different types of bicycle trails, lanes and paths may have different effects on cycling (for a review, see Buehler and Pucher, 2012). On-road bicycle lanes with bicycle parking are more attractive to cyclists than off-road bicycle trails (Tilahun et al., 2007). It should be noted that some researchers have found that the supply of bicycle lanes has a limited or even no effect on cycling rates (e.g. Dill and Voros, 2007; Parkin et al., 2008). Another type of bicycle infrastructure concerns the availability and design of bicycle parking sites (Martens, 2004). Exclusive and safe parking sites designed for bicycles in transit station areas contribute to a higher rate of bicycle-transit trips (La Paix and Geurs, 2015). However, unsafe parking near transit stations is an important obstacle to bicycle-transit integration (Advani and Tiwari, 2006). Other facilities, for instance, better lighting along bicycle lanes, may also contribute to higher rates of cycling (Akar and Clifton, 2009). Provision of transit services which are able to load bicycles on board could encourage the use of cycling as a transfer mode (Krizek and Stonebraker, 2011). It has been widely claimed that the availability of public bicycle-sharing systems could encourage cycling, in particular, transfer trips to transit station areas (DeMaio, 2009; Liu et al., 2012; DeMaio and Gifford, 2004). However, empirical evidence for this claim remains scarce. People’s choices to use bicycles as a transfer mode are also affected by the relative services and costs of alternative transport modes such as cars and buses. Passengers are willing to endure greater cycling distances if they travel to a station which is serviced by faster public transport, for example, in the Netherlands (Martens, 2004). One of the likely reasons for this may be that faster transit brings more time savings to these cyclists. Recent studies in the Netherlands also confirm the essential role of highly efficient railway services in encouraging bike-and-ride integration (La Paix and Geurs, 2015, 2016). A high level of feeder bus services between the origin and the transit station encourages people to use buses as a transfer mode rather than bicycles (Rietveld and Daniel, 2004; Lee et al., 2011). Low cost transit could decrease cycling rates (Rietveld, 2000b; Heinen et al., 2013). Similarly, free and low-price car parking, and free car use provided by workplaces during the daytime discourages people from cycling (Heinen et al., 2013; Martens, 2004).

50

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

2.3. Other factors Other variables influencing cycling include the natural environment, social context, and transport policies. Although the literature mainly focuses on cycling as a direct travel mode, it may also work as a transfer mode to and from metro stations. Cycling is more sensitive to climate, weather and topography than motorized modes of travel such as cars, metro trains or buses. Cold and rainy weather (Bergström and Magnusson, 2003; Parkin et al., 2008), and hilly terrain (Vandenbulcke et al., 2011; Rietveld and Daniel, 2004) discourage cycling, although few studies claim that these factors are not significantly related to the possibility of choosing cycling as the travel mode (Buehler and Pucher, 2012). A cycling-friendly social atmosphere is very important in promoting cycling. Enthusiastic bicycle advocates in the neighborhood (Heinen and Handy, 2012), family (Sigurdardottir et al., 2013) and the workplace (Stinson and Bhat, 2004; Handy and Xing, 2011) incite more people to cycle. In contrast, cycling is less popular among people in places where it is considered abnormal and unrespectable (Pucher et al., 1999; Gatersleben and Appleton, 2007). 3. Research design 3.1. City context Beijing is the capital of China and one of the biggest cities in the country. It had a population of 21 million and a land area of 16,410 km2 as of 2014 (BMBS, 2015). In Beijing, cycling used to be the dominant travel mode, contributing to China being known as the ‘kingdom of the bicycle’ (SCMP, 2015). In 1986, bicycles accounted for more than 60% of all daily trips (Fan et al., 2013). However, bicycle usage has been decreasing in Beijing since the 1990s (Fig. 1) with current trips by bicycle accounting for less than 20% of all trips. This trend seems likely to continue in the near future. In the meantime, car ownership by Beijing households has been soaring. In 2015, for every 100 households in Beijing, 63 had private cars. This number is double the average level of car ownership in China (SCMP, 2015). Such rapid growth in car ownership is mainly attributed to increasing household income and urban sprawl (Zhao et al., 2011). One of the results of motorization is severe traffic congestion in the city. The mean traffic speed on sixty of Beijing’s main roads during peak hours is only about 12 km/h. Another result is that most exclusive bicycle lanes have been redesigned to give space to cars. In particular, in the center of Beijing, many exclusive bicycle lanes are illegally occupied by cars due to the inadequacy of parking facilities. The cycling system has broken down and the availability of bicycle lanes and facilities has been shrinking (Li, 2010). Moreover, driving is considered to be a symbol of high social status, while cycling is now seen as a travel mode for lowincome earners (SCMP, 2015). Like other large cities in China, Beijing has become a car-dominated city. The promotion of cycling has become an imperative for China’s governments. Beijing’s municipal government made an attempt to promote cycling in the lead up to the Beijing Olympic Games (PGBM, 2009) by building more exclusive bicycle lanes and bicycle parking sites. In particular, 1000 bicycle share sites were built in metro station areas and more than 50,000 bicycles were supplied for public use. Apart from these huge investments in facilities, new policies, including plans, regulations and rules, were introduced to encourage cycling, such as the Bicycle Transportation Plan 2009 and the Green Beijing Action Plan 2010–2012. According to these new policies, existing regulations discouraging bicycle travel were to be revised or abolished, and bicycle transportation planning was to be officially included in Beijing’s future transportation planning (PGBM, 2009, 2010). Currently, the Beijing Municipal Commission of Transport plans to build more exclusive bicycle lanes, not only near metro station areas but also in neighborhoods within 3–5 km of metro stations (Huang, 2014). A recent practical endeavor of the Beijing government mainly included extensions of the bicycle share system and attempts to create more bicycle parking sites near metro stations (Beijing Morning, 2015).

70% 60%

62.7%

50% 40%

38.5% 30.3%

30% 20% 10%

23.0% 20.3%

18.1% 16.4% 15.1% 13.9%

0%

Fig. 1. Bicycle trips as a proportion of all travel modes, Beijing, 1986–2011. Source: The authors, edited from BTRC (2014).

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

51

Additionally, an official survey in Beijing showed that due to the problems of motorized transportation, 49.8% of respondents stated that they preferred the metro for daily travel (Beijing Daily, 2014). The awareness of bicycle revitalization in Beijing, and the large amount of public metro passengers, provide Beijing with a great opportunity to promote bicycle-metro integration. 3.2. Survey, data and variables This paper aimed to explore the determinants of bicycle use by commuters as a transfer mode to Beijing metro stations. Transfer mode refers to transit passengers’ travel mode between a transit station and their home or workplace. It includes walking, cycling, buses, cars or other modes. The data analyzed in this paper comes from a series of surveys conducted by the Centre for Urban and Transport Planning Research at Peking University between April and August in 2015. The study used a cross-sectional two-stage sampling technique to select respondents. At the first stage, 36 metro stations were randomly selected. The selected stations and the built environment at the stations were representative in Beijing in terms of their locations and transit-oriented development situations. In the second stage, metro passengers who either boarded or disembarked at the above stations between 5 and 7 PM were randomly chosen to participate in surveys. During this stage, some investigators travelled with passengers for several stops while conducting an interview. About 40 passengers were randomly chosen at each metro station. A total of 800 passengers agreed to conduct the survey. The response rate was 56%. The respondents were asked to report their transfer modes between the metro station and their home or workplace. They were also asked about travel distances, individual socioeconomic characteristics, housing ownership, travel preferences, and so on. After a check for data quality and missing data, 739 samples were retained for use in the study. Seven metro stations were not included into the analysis because of small numbers of respondents. The metro stations used were distributed across the whole city (Fig. 2). Eleven were located within the 4th Ring Road of Beijing (defined as the urban area in this paper). The other stations were located outside the 4th Ring Road (defined as the suburban area). Table 1 shows the measurements and descriptive data obtained during the survey for individuals’ travel and socioeconomic attributes. It indicates that 7.04% of the respondents choose cycling as their transfer mode to metro stations. This share rate is much higher than that of North American cities (Wang and Liu, 2013; Kim et al., 2007) or India (Advani and Tiwari, 2006). Table 1 also shows that the respondents were mostly young people (74.15% under the age of 30) and low-income earners (80.65% of the households earned less than RMB 8000). These values are consistent with the

Fig. 2. Distribution of sampled metro stations.

52

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

Table 1 Measurements and descriptive statistics for individuals’ travel and socioeconomic features (n = 739). Variable

Measurement and value

Category

Number of observations/ mean

Percentage (%)/SE

Transfer mode to metro station

Travel mode of the access/egress trip between the respondent’s home/workplace to the selected metro station

Walk

542

73.34

Bicycle Bus Car

52 129 16 2.08

7.04 17.46 2.17 2.95

Over 30 Under 30 Female Male <5000

191 548 347 392 315

25.85 74.15 46.96 53.04 42.63

5000–8000 >8000 Yes No Yes

281 143 234 505 191

38.02 19.35 31.66 68.34 25.85

No Yes

548 153

74.15 20.70

No Yes

586 80

79.30 10.83

No

659

89.17

Trip distance

Age

Continuous variable, the product of reported travel time between the respondent’s home/workplace and the selected metro station and average speed according to BTRC (2014): walking-5 km/h, bicycle-10 km/h, bus-20 km/h, car-40 km/h =1 if the age of the respondent is over 30 years old, other = 0

Gender

Dummy variable, =1 if the respondent is male, =0 if the respondent is female

Household income (RMB/monthly)

Monthly household income of the respondent

Car ownership

=1 if the respondent has private cars, other = 0

Preference for economic travel mode

=1 if the repondent’s answer to the question ‘‘Do you agree that you prefer cycling to other modes because it saves money?” is yes, other = 0

Environmental awareness

=1 if the respondent’s answer to the question ‘‘Do you agree that protecting the environment is one important reason for you to choose a travel mode?” is yes, other = 0 =1 if the respondent’s answer to the question ‘‘Do you like driving?” is yes, other = 0

Preference for driving

city-level passenger information reported by the municipal government transport agency. In this sense, the sample is representative of Beijing metro passengers. Table 2 below presents a comparison of the average travel distances and times of cycling and other travel modes. The mean travel distance for bicycle transfer trips to metro stations was 3.22 km, which was about half of the mean transfer distances of cars and buses. Post hoc tests showed that the mean differences between the travel times of cycling and that of other modes was statistically significant at the P < 0.01 level. On average, cyclists spent an additional 2.72 min transferring to/from metro stations compared with passengers who walked. In contrast, they took 5.24 min less than bus travelers. Descriptions of the built environment and transportation facilities are shown in Table 3. The transport facilities and services within a radius of 400 m (about 0.25 miles) were measured. The transportation variables included bus services, bicycle share stations, and parking sites. The data for land use and public facilities were obtained from the Beijing Municipal Institute of City Planning and Design. The data for bus services and parking were from the Beijing Transport Institute. Population Table 2 One-way ANOVA for travel distances and durations of different transfer modes to stations (post hoc method: Tamhane).

Travel distance

Travel time

Note: * p < 0.1, ** p < 0.5. *** p < 0.01.

Mode

Average distance (km)

S.E.

Mean difference with cycling

S.E.

Sig

Cycling Walking Car Bus F Sig.(P<)

3.22 0.85 6.29 6.29 285.80

1.47 0.42 3.47 4.49


2.37 3.07 3.07

0.21 0.89 0.45

***

5.89 5.10 10.60 13.68


2.72 3.49 5.24

0.85 2.77 1.46

***

Cycling Walking Car Bus F Sig.(P<)

*** ***

***

12.88 10.16 16.38 18.12 41.39 ***

***

53

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60 Table 3 Descriptive statistics of the built environment and transport service variables in metro station areas (n = 29). Variable

Mean

SE

Min

Max

Number of households Number of malls Number of parks Length of regional roads (m) Length of local roads (m) Length of exclusive bicycle lanes(m) Number of public bicycles Number of bus lines Number of car parking sites

26,271 4.61 1.65 15,875 34,375 1663 42.9 19.97 9.29

3086.13 0.93 0.44 1862.98 4.44.34 483.49 11.58 3.77 1.92

3647 0 0 0 5796 0 0 0 0

67,012 22 12 38,623 86,421 12,727 274 94 38

information was obtained from local authorities. The number of public bicycles near metro stations was measured on the basis of the Public Bicycle Map of Beijing (BMCT, 2015). The other variables were measured within a radius of 1.5 km from the metro stations. These variables refer to an area that is primarily focused on metro-oriented land development. In the North American and European contexts, the spatial range of a transit-oriented development area is usually defined by a walking distance from home to the transit station, and is in the range of 400–800 m (0.25–0.5 miles) for most transferring travelers (O’Sullivan and Morrall, 1996). In contrast, the spatial range of metro station areas in this study was larger than those of previous studies. One of the major reasons for this is that according to the Beijing municipal government’s TOD plans and designs, most metro station areas are defined by a circle of 1.5 km radius from a metro station. Another reason is that most pedestrians in Beijing are willing to walk for 1.5 km to get to a metro station (Huang et al., 2009). The number of households was measured as the number of households within 1.5 km of each metro station. Neighborhood location information was retrieved from BaiduMap (http://map.baidu.com), and the number of residents living in a neighborhood was obtained from the largest house broker website in China, Sounfun (http://china.soufun.com/). The number of malls was counted to indicate the degree of mixed land use. Parks refer to public parks in the city, not including small recreation sites such as neighborhood gardens. Road network data was calculated using the OpenStreetMap data retrieved from Long and Liu (2013). Exclusive bicycle lanes refer to lanes which are especially designated for cycling. These bicycle lanes are often curbed or fenced to exclude motorized vehicles. 3.3. Methodology This study attempts to investigate the determinants of cycling as a transfer mode in metro station areas in Beijing. A number of variables are estimates, including individuals’ trip features (mode, distance and duration), individuals’ socioeconomic attributes, public transport services, and public facilities in metro stations, etc. These variables may affect personal travel mode choices at different statistical levels (Cerin, 2011). Therefore, a multilevel logistic model was used in this study. Its dependent variable is the transfer mode for an individual’s trip between a metro station and a home or workplace, which may be walking, bicycle, car or bus. The independent variables in Level 1 are attributes of the metro stations, including transport types, facilities, and services. The independent variables in Level 2 are individual socioeconomic attributes. The intercept of the model was set to be random and the coefficients were set to be fixed. The data analysis was conducted with a multinomial logistic model using HLM for Windows. All independent variables were entered simultaneously in the model. The details of the model are presented as follows: P Level 1: Prob [Mode = k] = uk， 4k¼1 ¼ 1, and Log[uk/ubicycle] = b0(k) + f(individual) Level 2: b0(k) = c0(k) + f(station) + lk where Prob [Mode = k] denotes the probability of mode k satisfying the function Log[uk/ubicycle] = b0(k) + f(individual), which is a function of the individual and socio-economic attributes. The intercept for Level 1, b0(k), is random for every travel mode, and is a function of the metro station variables (built environment, transport infrastructure and service variables). 4. Results of the analysis Table 4 shows the results of the multilevel logistic regression. The travel mode of cycling was used as the reference group. The distance to the metro station was the most significant variable. As the distance to the metro station increases, people are less likely to choose to walk, while being more likely to choose cycling. The odds ratio (OR) of 0.028 indicates that when the distance to the metro station increases by one unit, the odds of walking (the probability of walking divided by the probability of cycling) will decrease by 0.028 times the original value. Compared with cycling, increasing distance to a metro station results in an increase in the likelihood of driving or taking the bus (OR = 3.834 for driving, OR = 2.713 for bus trips). The results suggest that cycling may be a popular transfer mode if people live or work a moderate distance from the metro station.

54

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

Table 4 Multilevel logistic regression results (fixed effects, reference group: bicycle, n of Level 1 = 728, n of Level 2 = 728). Walking

Level 1 Individual and socioeconomic attributes Trip distance to the metro station Age: under 30 Gender: male Monthly income: between 5000 and 8000 RMB Monthly income > 8000 RMB Car ownership Attitudes Preference for economic travel mode Environmental awareness Preference for driving Level 2 The built environment No. of households No. of malls No. of parks Live in suburban area Transport infrastructure and services Length of regional roads Length of local roads Length of exclusive bicycle lanes No. of public bicycles No. of bus lines No. of packing sites Intercept

Car

Bus

Coeff.

SE

Odds Ratio

Coeff.

SE

Odds Ratio

Coeff.

SE

Odds Ratio


3.572*** 0.199 0.090 0.354

0.583 0.78 0.571 0.692

0.028 1.220 1.094 1.425

1.344*** 1.017 2.562** 12.186**

0.245 1.191 1.024 4.785

3.834 2.765 12.962 1.96*10E5

0.998*** 1.258*
0.499
0.789*

0.180 0.568 0.497 0.567

2.713 3.518 0.607 0.454


0.509
0.796*

0.797 0.602

0.601 0.451

13.252*** 6.673***

10.397 2.213

5.69*10E5 790.764


0.304*
0.530

0.813 0.544

0.738 0.589


1.357**
0.139
0.243

0.679 0.723 0.131

0.257 0.870 0.784

1.008
0.357 3.119***

1.060 1.267 1.582

2.740 0.700 22.624


0.345 0.818
0.114

0.622 0.685 0.920

0.708 2.266 0.892

0.111 3.099*
3.628
4.065

0.561 2.969 2.023 3.126

1.117 22.176 0.027 0.017


0.112 3.523*
10.979* 3.232*

0.422 2.822 1.571 2.320

0.894 33.886 0.000 25.330


0.158 1.548*
7.280** 3.749**

0.298 1.260 3.148 1.407

0.854 4.702 0.001 32.479

0.096 0.267
0.405 0.033 0.007

1.116 0.623 1.383 0.121 0.042

1.101 1.306 0.667 1.034 1.007

0.423 1.484 1.689 1.059 1.030

5.349

0.078 2.465 0.027 0.980 1.051 1.770 0.002

0.674 0.408 0.918 0.095 0.024

2.902

0.979 0.544 2.711 0.261 0.046 1.179 4.370


0.860* 0.395 0.524 0.057 0.030*

1.677


2.545** 0.902*
3.610
0.020* 0.050 0.571
12.411**


3.467*

1.928

0.031

Note: * p < 0.1. ** p < 0.05. *** p < 0.01.

The results above also suggest that the relationship between cycling choice and travel distance to metro stations is not linear. There could be a range of within which people will choose to cycle. If the travel distance is above this range, they tend to choose a motorized mode of transfer, and if the travel distance is below this range, they tend to choose to walk. The data suggest that this range is between approximately 1 and 5 km in Beijing (Fig. 3). Fig. 3 demonstrates that 40% of drivers and 50% of bus users travel less than 5 km for trips to metro stations. About 80% of cyclists have a travel distance between 1 and 4 km. The upper threshold for cycling distance in Beijing is much greater than that reported by previous studies in North

Fig. 3. Cumulative percentage of trip distances for different travel modes.

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

55

American cities (Taylor and Mahmassani, 1996; Rastogi and Krishna Rao, 2003; Krygsman et al., 2004). This could be related to the greater distances of residential and employment locations from metro stations in Beijing. The results in Table 4 show that the respondents living/working near suburban metro stations are more likely to choose cars or buses than bicycles as their transfer mode, compared with those who live in the city center. There might be two reasons for this. One is that the distances to metro stations in suburban areas are, on average, greater than in the city center. Another reason may be that there are fewer cycling facilities and services in the suburbs than in the city center. The number of shopping malls is negatively related to the likelihood of an individual choosing to cycle to a metro station. In metro station areas with more shopping malls, a person seems to be more likely to choose modes other than cycling to transfer to/from metro stations. One of the major reasons for this might be that the large size of shopping malls reduces land use heterogeneity compared with smaller-scale land use in a given geographical space, thus discouraging cycling. Another reason could be that the metro station areas with a large number of shopping malls in the suburbs usually have plenty of free parking, which encourages car use. In relation to walking, a high concentration of shopping malls results in a busy environment and a high level of vibrancy in metro station areas, and thus enhances the attractiveness of walking. The number of public parks is positively related to cycling rates to/from stations. As the number of parks increases, the probability of driving and taking the bus clearly decreases. One possible reason is that the appealing scenery of parks makes cycling more attractive. Another possible reason is that cyclists may cycle through parks to avoid traffic, potential injuries, and waiting for traffic lights. This result suggests that creating more parks in metro station areas encourages more people to cycle to and from metro stations. Table 4 reveals that transportation infrastructure and services in station areas are also important determinants of people’s choice to transfer by bicycle. The length of local roads is positively related to the probability of driving rather than cycling (OR = 2.465). This is because a greater number of local roads in metro station areas favor motorized travel such as driving and taking the bus. In relation to public bicycle services, the number of public bicycles in metro station areas was significantly related to metro passengers’ choice to use cycling as a transfer mode. If there were more public bicycles supplied in station areas, metro passengers had a higher probability of choosing a cycling transfer (OR = 0.980). This result suggests that a public bicycle program in the metro station areas could play a positive role in encouraging cycling and thus discouraging car use as a transfer mode, although the effect is small. This result contributes to the existing literature by providing the new evidence of the positive effect of public bicycle programs on bicycle-metro integration. For bus services in metro station areas, the results presented in Table 4 show that the presence of more bus lines increases the probability of choosing to take the bus rather than cycling (OR = 1.030). This suggests that there is an obvious substitution effect between cycling and buses, and raises a dilemma for policymakers: an improvement in feeder bus services in metro station areas may reduce the rate of bicycle transfers. Admittedly, a positive effect of bus lines could be the relatively low costs of bus services in Beijing. The regression results in Table 4 show that residents’ socioeconomic features were related to cycling transfer rates. Compared with people over the age of 30, those under 30 were more likely to transfer by bus than cycling (OR = 3.518). Two reasons may contribute to this result. One is that younger generations in China rely on motorized transport more than older generations. Another reason is that younger generations may live further from metro stations due to housing costs. Placing a higher value on time because of peer pressure, they may prefer buses to bicycles for transfers. Men in particular seem to be more likely than women to drive to metro stations rather than cycle (OR = 1.021). This gender gap may be a result of differences in preferences for short-distance driving. The effect of demographic attributes such as gender and age on bicycle-metro integration should be paid more attention to in the pursuit of tailored policies for bicycle-metro integration in Beijing. This is because significant transportation inequity issues exist in terms of gender and age (Litman, 2002). The results also reveal that middle- and high-income earners (above RMB 5000 monthly) are more likely to choose bicycle than bus transfers. Apparently, they are also more likely to transfer by car. Passengers owning cars are more likely to use car transfers (OR = 2.213) and are less likely to cycle (OR = 0.451). The above analysis demonstrates that Beijing’s bicycle-metro users have a mixed socioeconomic status. The attitude variables in Table 4 present more interesting results. If a metro passenger has a preference for cheap travel, they are more likely to transfer by bicycle (OR = 0.257), after controlling for the socioeconomic attributes. This suggests that cycling is considered to be a cheap travel mode to some extent, even compared with walking. This is the image of cycling in many developing countries other than China (de Dios Ortuzar et al., 2000; Nkurunziza et al., 2012). Cycling is seen as an outdated travel mode used by low-income earners to save on travel costs. Table 4 also shows that respondents prefer to transfer by car instead of cycling (exp(b) = 22.624). This means the people who like driving are actually more likely to transfer by car than by bicycle. It is widely believed that people who have a preference for driving actually drive more (Handy et al., 2005). This is an example of travel mode self-selection. However, when it comes to transfer trips, evidence is scarce. The effect of a personal preference for driving on actual travel behavior could differ for transfer trips and direct trips, because transfer trips are usually shorter. The results of above analysis provide new evidence that the self-selection phenomenon still exists for transfer trips. Another interesting finding is that environmental awareness was not significantly related to cycling rates in Beijing. One of the reasons for this may be that environmental awareness is not relevant to bicycle-metro integration, compared with cycling as a direct travel mode. Another reason is that environmental awareness in relation to travel behavior is low in Beijing, as confirmed by empirical studies (e.g. Liu, 2011).

56

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

Fig. 4. Illegally parked cars in exclusive bicycle lanes in Beijing. Source: the authors.

The results in Table 4 show that the length of exclusive bicycle lanes was not related to a residents’ probability of cycling. There could be many reasons for this unexpected result. One possible reason is that these so-called exclusive bicycle lanes are, in most cases, not exclusively used by bicycles but are actually used by drivers to park their cars (Beijing News, 2013, Fig. 4). This situation is more severe near metro stations. Although the state and the municipal governments have launched regulations prohibiting parking in bicycle lanes, illegal parking still occurs. One of the major reasons for this is that local authorities (district or sub-district (jiedao) governments) don’t have enough resources to police the regulations. The other reason is that Beijing’s parking supply is severely inadequate. The number of private cars increased from 2.3 million in 2008 to 5.6 million in 2015. However, over the same period, the number of parking spaces only increased from 1.07 million to 2.8 million. Parking has become a vital transportation problem in Beijing, particularly in the old city center. The local residents make complaints to the authorities, who often have to give tacit consent to the illegal behavior of parking in cycle lanes. As a result, it is difficult to create a safe, comfortable cycling environment for bicycle-metro integration, and the potential advantages of exclusive bicycle lanes for cyclists may not eventuate. The provision of exclusive bicycle lanes thus has a lesser effect on cycling behavior.

5. Discussions This research explored the determinants of people’s choices in using cycling as a transfer mode to/from transit stations. Here, we also make policy suggestions for the promotion of bicycle-metro integration in Beijing. Based on our findings, several key points will be discussed. Firstly, a cycling-friendly built environment in metro station areas could play a positive role in encouraging transit passengers to choose cycling as a transfer mode. This finding echoes previous studies reporting that the positive effect of better built environments in railway station areas is beneficial to bicycle-metro integration (La Paix and Geurs, 2015). Transitoriented development policies designed to enhance mixed land use could result in shorter trips between metro stations and residential/employment locations, which would facilitate bicycle-metro integration. Having moderate or small distances between metro stations and homes/workplaces is a major factor promoting cycling transfers. According to the data used in this study, 80% of cycling trips between metro stations and homes or workplaces in Beijing were between 1 and 4 km (Fig. 3). For trips of less than 1 km, walking is the dominant transfer mode. There are good opportunities for promoting bicycle-metro integration in Beijing. More green parks may also increase the probability of commuters using cycling as a transfer mode. To our knowledge, this paper is the first attempt at revealing the importance of green space design in promoting bicycle-metro integration. This result may also be insightful for promoting bicycle-metro integration across the world. At present, land development in metro station areas is usually characterized by a high building density designed to maximize internal space and revenue. Green parks and other public open spaces are often considered unnecessary, as such land use does not bring any direct economic benefits. As a result, there is a shortage of parks in metro station areas. The results of this study suggest that a more attractive environment with more green spaces could actually play a positive role in encouraging cycling as a transfer mode. Secondly, the role of exclusive bicycle lanes in promoting bicycle-metro integration in Beijing remains contentious. One of the interesting findings of this paper is that the number of exclusive bicycle lanes was not directly associated with a higher likelihood of cycling to and from metro stations. There could be three reasons for this. One is that the Road Construction Code Act in Beijing requires that all types of road except highways should either have exclusive bicycle lanes or have exclusive spaces for bicycle use. As a result, the differences in the availability of exclusive bicycle lanes between neighborhoods are small. Another reason may be that a large number of Beijing’s exclusive bicycle lanes are illegally occupied by cars. The problem is particularly bad near metro stations. As a result, the potential advantages these lanes offer to cyclists for bicycle-metro integration have not been realized (SCMP, 2015). Nevertheless, it is unquestionable that a better cycling environment, which gives priority to cyclists in the urban transport network, is of vital importance in achieving bicycle-metro integration. Protecting road rights and reserving road space for cyclists should be addressed in transport policies (Pucher and Buehler, 2008;

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

57

Pucher et al., 2011). Whether more strict policing of illegally-parked cars in Beijing can improve bicycle-metro integration should be investigated by future studies. Thirdly, bicycle-sharing systems could promote bicycle-metro integration. Many researchers argue that a well-integrated system combining public bicycle sharing and transit use could stimulate cycling rates (DeMaio, 2009; DeMaio and Gifford, 2004; Shaheen et al., 2010; Martens, 2007; Pucher and Buehler, 2008). The results of our Beijing case study provides new evidence for this argument. Bicycle-sharing systems have several advantages. They provide affordable access to bicycles for short-distance trips in metro or train station areas as an alternative to motorized transport, such as buses or cars. As mentioned in the Introduction, ‘‘bicycle + transit” could be a way to solve the ‘‘last mile” problem for train or metro services. Bicycle-sharing systems at metro stations could be an efficient way of promoting bicycle-transit integration. A wellspread network of public bicycle stations near users’ homes is also important for the encouragement of cycling transfers (Wang et al., 2015). A recent evaluation of the public bicycle system in Beijing reported that there was a mismatch between the demand for, and supply of, public bicycles (Liu et al., 2012). On the one hand, some public bicycle stations were rarely used. One the other hand, many residents complain that public bicycle stations are located too far from their neighborhoods. Therefore, public bicycle systems should address both the availability of bicycles within a well-spread network, and the location of transit stations. Fourthly, the demographic and socioeconomic influences on bicycle-metro integration should be given greater attention. The results of this study suggest that middle-aged and/or high-income passengers are more likely use cycling as a transfer mode. This differs from previous findings in US cases, in which the majority of cyclists were from low-income groups (Wang and Liu, 2013). In China, there has been a change in the lifestyle of younger people. The younger generation has a different attitude towards cycling than their parents. For most young people, cycling is no longer fashionable. It has been proven that the attraction of cycling to the younger generation has declined in China (SCMP, 2015). Thus, travel behavior education may be imperative for encouraging young people to cycle to and from metro stations. Moreover, it is also a challenge to encourage low-income earners to integrate bicycle and metro transport. Fifthly, personal preferences for travel mode do affect bicycle-metro integration. Our results show that people who prefer a cheaper travel mode are more likely to transfer by bicycle. The bicycle is generally considered to be an economic mode of travel in Beijing. This result is consistent with many empirical studies (Guo et al., 2015; de Dios Ortuzar et al., 2000; Nkurunziza et al., 2012; Vandenbulcke et al., 2011). However, it should be recognized that the bicycle’s reputation of being a ‘cheap mode’ may in fact have negative effects on some people’s decisions about whether to cycle or not, especially for lowincome metro passengers. This could be an important reason why low-income metro passengers are found to be less likely to cycle to and from metro stations. The negative image of cycling is a key obstacle to bicycle-metro integration in Beijing, as it is in other countries (Nkurunziza et al., 2012). Therefore, policies designed to change people’s perceptions about cycling are very important. It is important for communities and workplaces to encourage bicycle-metro integration and rebuild respect for cycling. A positive image of cycling may play an important role in encouraging bicycle-metro integration. Sixthly, the effect of environmental awareness on cycling rates may be limited in China, while it is widely believed be an important influence on cycling rates in North American and European cities (Dill and Voros, 2007; Heinen and Handy, 2012). However, in the case of Beijing, cycling transfer rates were not correlated with rates of environmental awareness. One reason may be that people’s environmental awareness in relation to transport is weak in China (Wong and Chan, 1996). This means that behavioral education aimed to arouse passengers’ environmental awareness of travel modes should be given consideration. However, it may be difficult to encourage people to give up using cars once they have started using them. Our results suggest that people who prefer driving are less likely to cycle. In fact, Beijing’s car owners are much more car-dependent than those of other Asian cities such as Tokyo, Singapore and Hong Kong. It has been reported that the average number of daily trips by car in Beijing is three times higher than in these other Asian cities. Moreover, 44% of these car trips were of distances less than 5 km (Lu et al., 2011). Accordingly, methods of discouraging private car owners from driving should be addressed. Restricting parking spaces and road capacities, and increasing the costs of driving (e.g. petrol prices, car registration fees, congestion tolls, driver licensing fees; Pucher and Buehler, 2008; Rietveld and Daniel, 2004) may motivate metro passengers who prefer driving to cycle instead.

6. Conclusions Integration of bicycle and metro systems could be a successful way to achieve efficient and sustainable urban transport. This study can assist planners and politicians who are often confronted with the question: How do we promote the integration of cycling and metro systems? This paper aimed to examine the determinants of people’s choices to use cycling as a transfer mode to and from metro stations, using Beijing as a case study. Travel distances between home and transit stations was found to be the most important factor influencing people’s decisions to cycle or not. Of the respondents, 80% made transfers that were within 5 km of their home. We also found that a cycling-friendly environment is vital to the promotion of cycling transfers. A higher level of mixed land use and more green spaces could also play positive roles in encouraging cycling to and from metro stations. More public bicycle-sharing opportunities in metro station areas also increase the likelihood of bicycle-metro integration. Thus, public bicycle-sharing programs integrated with metro stations could be an efficient way to encourage cycling while reducing car use in metro

58

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

station areas. In addition, there are some spatial variations in cycling rates in metro station areas—metro passengers living in suburban areas are less likely to cycle than take a bus, compared with those living in the city center. Socio-economic attributes and individual attitudes are also important determinants of cycling transfer rates. The younger generation in Beijing seems to be less likely to cycle than take buses. Low-income earners are also less likely to cycle. However, high-income earners or those who own cars do exhibit a tendency to cycle rather than drive. People’s personal preferences regarding travel modes are also important. Those who prefer cheaper travel modes are more likely to choose cycling, and people who prefer to drive are less likely to choose cycling. In the future, education initiatives should be integrated with land use policies to encourage bicycle-metro integration. Taking all of these factors into account, metro passengers who transfer by bicycle can be classified into three categories: (1) low-income earners living far from transit stations and having few alternatives; (2) those in higher socioeconomic groups pursuing a healthy lifestyle and quality of life; (3) people attracted to bicycle-sharing systems located near metro stations. In order to gain knowledge about the determinants of cycling transfer rates, the demographic, socioeconomic and attitudinal factors of different groups should be given greater consideration in the future. There are several limitations to the current study which can be considered starting points for further research. Firstly, difficulty in accessing small-scale spatial data was an obstacle to exploring more detailed information about the built environment. Secondly, the proportion of driving of the samples is relatively low transfers recorded, though the share of bicyclemetro integrators was much larger than that reported by previous studies. Finally, although personal preferences for travel modes were considered, personal preferences for residential location, which are also known as residential self-selection effects, were neglected in this study because of a shortage of data. This effect is important because cyclists might choose to live in neighborhoods which are suited to bicycle-metro integration (Xing et al., 2010). If this is the case, policies designed to promote cycling in metro station areas should take into consideration individual preferences for transport modes and living environments. Future policies should consider ways to encourage people to use bicycles and reside in neighborhoods designed for bicycle-metro integration. This largely requires changes in lifestyle and behavioral education. Acknowledgments This research is funded by the National Natural Science Foundation of China project No. 41571147. References Advani, M., Tiwari, G., 2006 Bicycle – as a feeder mode for bus service. In: Velo Mondial Conference: Third Global Cycling Planning Conference, Cape Town. Retrieved at (accessed April 1st, 2016). Akar, G., Clifton, K., 2009. Influence of individual perceptions and bicycle infrastructure on decision to bike. Transport. Res. Rec.: J. Transport. Res. Board 2140, 165–172. http://dx.doi.org/10.3141/2140-18. Bachand-Marleau, J., Larsen, J., El-Geneidy, A., 2011. Much-anticipated marriage of cycling and transit: How Will It Work? Transport. Res. Rec.: J. Transport. Res. Board 2247, 109–117. http://dx.doi.org/10.3141/2247-13. Beijing Daily, 2014. Public transportation trip rate approaches 50 percent and bicycle trip rate decreases. Retrieved at (in Chinese) (accessed November 25th, 2015). Beijing Morning, 2015. Beijing: bicycles could be parked at the stations of new transit lines. Retrieved at (in Chinese) (accessed November 25th, 2015). Beijing News, 2013. Five problems exist on the bicycle lanes in Beijing. Cars are parking on the exclusive bicycle lanes. Retrieved at (in Chinese) (accessed December 25th, 2015). Beijing Municipal Bureau of Statistics (BMBS), 2015. Beijing Statistical Yearbook 2015. China Statistics Press, Beijing (in Chinese). Beijing Municipal Commission of Transport (BMCT), 2015. Public Bicycle Map of Beijing. Retrieved at (in Chinese) (accessed November 25th, 2015). Beijing Transportation Research Center (BTRC), 2014. 2013 Beijing Annual Transport Development Report. Retrieved at (in Chinese) (accessed December 10th, 2015). Bergström, A., Magnusson, R., 2003. Potential of transferring car trips to bicycle during winter. Transport. Res. Part A: Policy Practice 37 (8), 649–666. http:// dx.doi.org/10.1016/S0965-8564(03)00012-0. Buehler, R., Pucher, J., Gerike, R., Götschi, T., 2016. Reducing car dependence in the heart of Europe: lessons from Germany, Austria, and Switzerland. Transport Rev. http://dx.doi.org/10.1080/01441647.2016.1177799. Buehler, R., Pucher, J., 2012. Cycling to work in 90 large American cities: new evidence on the role of bike paths and lanes. Transportation 39 (2), 409–432. http://dx.doi.org/10.1007/s11116-011-9355-8. Cerin, E., 2011. Statistical approaches to testing the relationships of the built environment with resident-level physical activity behavior and health outcomes in cross-sectional studies with cluster sampling. J. Plan. Liter. 26 (2), 151–167. http://dx.doi.org/10.1177/0885412210386229. Cervero, R., Duncan, M., 2003. Walking, bicycling, and urban landscapes: evidence from the San Francisco Bay Area. Am. J. Public Health 93 (9), 1478–1483. http://dx.doi.org/10.2105/AJPH.93.9.1478. DeMaio, P., 2009. Bike-sharing: history, impacts, models of provision, and future. J. Public Transport. 12 (4), 41–56. http://dx.doi.org/10.5038/23750901.12.4.3. DeMaio, P., Gifford, J., 2004. Will smart bikes succeed as public transportation in the United States? J. Public Transport. 7 (2), 1–15. http://dx.doi.org/ 10.5038/2375-0901.7.2.1. Dill, J., Carr, T., 2003. Bicycle commuting and facilities in major US cities: if you build them, commuters will use them. Transport. Res. Rec.: J. Transport. Res. Board 1828, 116–123. http://dx.doi.org/10.3141/1828-14. Dill, J., Voros, K., 2007. Factors affecting bicycling demand: initial survey findings from the Portland, Oregon, region. Transport. Res. Rec.: J. Transport. Res. Board 2031, 9–17. http://dx.doi.org/10.3141/2031-02. Dill, J., McNeil, N., 2013. Four types of cyclists? Examination of typology for better understanding of bicycling behavior and potential. Transport. Res. Rec.: J. Transport. Res. Board 2387, 129–138. http://dx.doi.org/10.3141/2387-15. de Dios Ortuzar, J., Iacobelli, A., Valeze, C., 2000. Estimating demand for a cycle-way network. Transport. Res. Part A: Policy Practice 34 (5), 353–373. http:// dx.doi.org/10.1016/S0965-8564(99)00040-3.

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

59

Du, Y., 2015. Metro lines grow longer for eight years, and the total length will reach 554 km in the end of the year. Retrieved at (in Chinese) (accessed December 10th, 2015). Fan, X., Niu, X., Liao, S., 2013. Study on development of bicycle transportation in Beijing. Traffic Eng. 1 (1), 83–86 (in Chinese). Garrard, J., Rose, G., Lo, S.K., 2008. Promoting transportation cycling for women: the role of bicycle infrastructure. Prev. Med. 46 (1), 55–59. http://dx.doi.org/ 10.1016/j.ypmed.2007.07.010. Gatersleben, B., Uzzell, D., 2007. Affective appraisals of the daily commute comparing perceptions of drivers, cyclists, walkers, and users of public transport. Environ. Behavior 39 (3), 416–431. http://dx.doi.org/10.1177/0013916506294032. Gatersleben, B., Appleton, K.M., 2007. Contemplating cycling to work: attitudes and perceptions in different stages of change. Transport. Res. Part A: Policy Practice 41 (4), 302–312. http://dx.doi.org/10.1016/j.tra.2006.09.002. Green, L.W., Richard, L., Potvin, L., 1996. Ecological foundations of health promotion. Am. J. Health Promotion 10 (4), 270–281. Guo, J., Bhat, C., Copperman, R., 2015. Effect of the built environment on motorized and nonmotorized trip making: substitutive, complementary, or synergistic? Transport. Res. Rec.: J. Transport. Res. Board 2010, 1–11. http://dx.doi.org/10.3141/2010-01. Handy, S., Cao, X., Mokhtarian, P., 2005. Correlation or causality between the built environment and travel behavior? Evidence from Northern California. Transport. Res. Part D: Transport Environ. 10 (6), 427–444. http://dx.doi.org/10.1016/j.trd.2005.05.002. Handy, S.L., Xing, Y., 2011. Factors correlated with bicycle commuting: a study in six small US cities. Int. J. Sustainable Transport. 5 (2), 91–110. http://dx.doi. org/10.1080/15568310903514789. Heinen, E., Maat, K., Van Wee, B., 2011. The role of attitudes toward characteristics of bicycle commuting on the choice to cycle to work over various distances. Transport. Res. Part D: Transport Environ. 16 (2), 102–109. http://dx.doi.org/10.1016/j.trd.2010.08.010. Heinen, E., van Wee, B., Maat, K., 2010. Commuting by bicycle: an overview of the literature. Transport Rev. 30 (1), 59–96. http://dx.doi.org/10.1080/ 01441640903187001. Heinen, E., Maat, K., van Wee, B., 2013. The effects of work-related factors on the bicycle commute mode choice in the Netherlands. Transportation 40 (1), 23–43. http://dx.doi.org/10.1007/s11116-012-9399-4. Heinen, E., Handy, S., 2012. Similarities in attitudes and norms and the effect on bicycle commuting: evidence from the bicycle cities Davis and Delft. Int. J. Sustainable Transport. 6 (5), 257–281. http://dx.doi.org/10.1080/15568318.2011.593695. Heinen, E., Bohte, W., 2014. Multimodal commuting to work by public transport and bicycle: attitudes toward mode choice. Transport. Res. Rec. J. Transport. Res. Board 2468, 111–122. http://dx.doi.org/10.3141/2468-13. Huang, S., Guan, H., Yan, H., 2009. A study on urban railway connection mode selections in Beijing. China Civil Eng. J. 42 (7), 126–130 (in Chinese). Huang, H., 2014. Exclusive bicycle lanes will be extended from transit stations to neighborhoods. Retrieved at (in Chinese) (accessed November 25th, 2015). Kim, S., Ulfarsson, G.F., Hennessy, J.T., 2007. Analysis of light rail rider travel behavior: impacts of individual, built environment, and crime characteristics on transit access. Transport. Res. Part A: Policy Practice 41 (6), 511–522. http://dx.doi.org/10.1016/j.tra.2006.11.001. Keijer, M.J.N., Rietveld, P., 2000. How do people get to the railway station? The Dutch experience. Transpo. Plan. Technol. 23 (3), 215–235. http://dx.doi.org/ 10.1080/03081060008717650. Kenworthy, J., Hu, G., 2002. Transport and urban form in Chinese cities: an international comparative and policy perspective with implications for sustainable urban transport in China. Plan. Rev. 38 (151), 4–14. http://dx.doi.org/10.1080/02513625.2002.10556818. Krizek, K.J., Roland, R.W., 2005. What is at the end of the road? Understanding discontinuities of on-street bicycle lanes in urban settings. Transport. Res. Part D: Transport Environ. 10 (1), 55–68. http://dx.doi.org/10.1016/j.trd.2004.09.005. Krizek, K.J., El-Geneidy, A., Thompson, K., 2007. A detailed analysis of how an urban trail system affects cyclists’ travel. Transportation 34 (5), 611–624. http://dx.doi.org/10.1007/s11116-007-9130-z. Krizek, K., Stonebraker, E., 2011. Assessing options to enhance bicycle and transit integration. Transport. Res. Rec.: J. Transport. Res. Board 2217, 162–167. http://dx.doi.org/10.3141/2217-20. Krygsman, S., Dijst, M., Arentze, T., 2004. Multimodal public transport: an analysis of travel time elements and the interconnectivity ratio. Transp. Policy 11 (3), 265–275. http://dx.doi.org/10.1016/j.tranpol.2003.12.001. La Paix, L., Geurs, K.T., 2016. Train station access and train use: A joint stated and revealed preference choice modelling study. In: Geurs, K.T., Patuelli, R., Dentinho, T. (Eds.), Accessibility, Equity and Efficiency. Challenges for Transport and Public Services. Edward Elgar, Northampton, USA, pp. 1–20. La Paix, L., Geurs, K., 2015. Modelling observed and unobserved factors in cycling to railway stations: application to transit-oriented-developments in the Netherlands. Eur. J. Transport Infrastruct. Res. 15 (1), 27–50. Lawlor, D.A., Ness, A.R., Cope, A.M., Davis, A., Insall, P., Riddoch, C., 2003. The challenges of evaluating environmental interventions to increase population levels of physical activity: the case of the UK National Cycle Network. J. Epidemiol. Community Health 57 (2), 96–101. http://dx.doi.org/ 10.1136/jech.57.2.96. Lee, I., Park, H., Sohn, K., 2011. Increasing the number of bicycle commuters. Proc. ICE-Transport 165 (1), 63–72. Li, J., 2010. Beijing will Open the Road for Bicycles. Beijing Technology Newspaper. Retrieved at (in Chinese) (accessed November 25th, 2015). Liu, W., 2011. Survey on the environmental awareness and behavior in Beijing. Modern Business 30, 285–286 (in Chinese). Liu, Z., Jia, X., Cheng, W., 2012. Solving the last mile problem: ensure the success of public bicycle system in Beijing. Proc. - Social Behavioral Sci. 43, 73–78. Litman, T., 2002. Evaluating transportation equity. World Transport Policy Practice 8 (2), 50–65. Long, Y., Liu, X.J., 2013. Automated Identification and Characterization of Parcels (AICP) with OpenStreetMap and Points of Interest. arXiv. preprint arXiv: 1311.6165. Lu, X., Tang, J., Zhang, J., 2011. Annual Report on Analysis of Beijing Society Building. Social Science Academic Press, Beijing (in Chinese). Martens, K., 2004. The bicycle as a feedering mode: experiences from three European countries. Transport. Res. Part D: Transport Environ. 9 (4), 281–294. http://dx.doi.org/10.1016/j.trd.2004.02.005. Martens, K., 2007. Promoting bike-and-ride: the Dutch experience. Transport. Res. Part A: Policy Practice 41 (4), 326–338. http://dx.doi.org/10.1016/ j.tra.2006.09.010. Moudon, A.V., Lee, C., Cheadle, A.D., Collier, C.W., Johnson, D., Schmid, T.L., Weather, R.D., 2005. Cycling and the built environment, a US perspective. Transport. Res. Part D: Transport Environ. 10 (3), 245–261. http://dx.doi.org/10.1016/j.trd.2005.04.001. Nkurunziza, A., Zuidgeest, M., Brussel, M., Van Maarseveen, M., 2012. Examining the potential for modal change: Motivators and barriers for bicycle commuting in Dar-es-Salaam. Transp. Policy 24, 249–259. http://dx.doi.org/10.1016/j.tranpol.2012.09.002. O’Sullivan, S., Morrall, J., 1996. Walking distances to and from light-rail transit stations. Transport. Res. Rec.: J. Transport. Res. Board 1538, 19–26. http://dx. doi.org/10.3141/1538-03. Parkin, J., Wardman, M., Page, M., 2008. Estimation of the determinants of bicycle mode share for the journey to work using census data. Transportation 35 (1), 93–109. http://dx.doi.org/10.1007/s11116-007-9137-5. Parra, D.C., Hoehner, C.M., Hallal, P.C., Ribeiro, I.C., Reis, R., Brownson, R.C., Simoes, E.J., 2011. Perceived environmental correlates of physical activity for leisure and transportation in Curitiba, Brazil. Prevent. Med. 52 (3), 234–238. http://dx.doi.org/10.1016/j.ypmed.2010.12.008. People’s Government of Beijing Municipality (PGBM), 2009. Action plan for humanity transportation, scientific transportation and green Transportation (2009-2020). Retrieved at (in Chinese) (accessed November 25th, 2015). People’s Government of Beijing Municipality (PGBM), 2010. Action plan for green Beijing (2010-2012). Retrieved at (in Chinese) (accessed November 25th, 2015). Pucher, J., Buehler, R., 2008. Making cycling irresistible: lessons from the Netherlands, Denmark and Germany. Transport Rev. 28 (4), 495–528. http://dx.doi. org/10.1080/01441640701806612.

60

P. Zhao, S. Li / Transportation Research Part A 99 (2017) 46–60

Pucher, J., Garrard, J., Greaves, S., 2011. Cycling down under: a comparative analysis of bicycling trends and policies in Sydney and Melbourne. J. Transp. Geogr. 19 (2), 332–345. http://dx.doi.org/10.1016/j.jtrangeo.2010.02.007. Pucher, J., Komanoff, C., Schimek, P., 1999. Bicycling renaissance in North America? Recent trends and alternative policies to promote bicycling. Transport. Res. Part A: Policy Practice 33 (7), 625–654. http://dx.doi.org/10.1016/S0965-8564(99)00010-5. Rastogi, R., Krishna Rao, K.V., 2003. Travel characteristics of commuters accessing transit: Case study. J. Transport. Eng. 129 (6), 684–694. http://dx.doi.org/ 10.1061/(ASCE)0733-947X(2003) 129:6(684). Rietveld, P., Daniel, V., 2004. Determinants of bicycle use: do municipal policies matter? Transport. Res. Part A: Policy Practice 38 (7), 531–550. http://dx.doi. org/10.1016/j.tra.2004.05.003. Rietveld, P., 2000a. Non-motorised modes in transport systems: a multimodal chain perspective for The Netherlands. Transport. Res. Part D: Transport Environ. 5 (1), 31–36. http://dx.doi.org/10.1016/S1361-9209(99)00022-X. Rietveld, P., 2000b. The accessibility of railway stations: the role of the bicycle in The Netherlands. Transport. Res. Part D: Transport Environ. 5 (1), 71–75. http://dx.doi.org/10.1016/S1361-9209(99)00019-X. Shaheen, S., Guzman, S., Zhang, H., 2010. Bikesharing in Europe, the Americas, and Asia: past, present, and future. Transport. Res. Rec.: J. Transport. Res. Board 2143, 159–167. http://dx.doi.org/10.3141/2143-20. Sallis, J.F., Cervero, R.B., Ascher, W., Henderson, K.A., Kraft, M.K., Kerr, J., 2006. An ecological approach to creating active living communities. Annu. Rev. Public Health 27, 297–322. http://dx.doi.org/10.1146/annurev.publhealth.27.021405.102100. Sigurdardottir, S.B., Kaplan, S., Møller, M., Teasdale, T.W., 2013. Understanding adolescents’ intentions to commute by car or bicycle as adults. Transport. Res. Part D: Transport Environ. 24, 1–9. http://dx.doi.org/10.1016/j.trd.2013.04.008. Singleton, P.A., Clifton, K.J., 2014. Exploring synergy in bicycle and transit use: empirical evidence at two scales. Transport. Res. Rec. J. Transport. Res. Board 2014, 92–102. http://dx.doi.org/10.3141/2417-10. South China Morning Post (SCMP), 2015. Beijing Attempts to Revitalize Bicycles. Retrieved at (in Chinese) (accessed November 25th, 2015). Stokols, D., Allen, J., Bellingham, R.L., 1996. The social ecology of health promotion: implications for research and practice. Am. J. Health Promotion 10 (4), 247–251. Stinson, M., Bhat, C., 2004. Frequency of bicycle commuting: internet-based survey analysis. Transport. Res. Rec.: J. Transport. Res. Board 1878, 122–130. http://dx.doi.org/10.3141/1878-15. Taylor, D., Mahmassani, H., 1996. Analysis of stated preferences for intermodal bicycle-transit interfaces. Transport. Res. Rec.: J. Transport. Res. Board 1556, 86–95. http://dx.doi.org/10.3141/1556-11. Tilahun, N.Y., Levinson, D.M., Krizek, K.J., 2007. Trails, lanes, or traffic: Valuing bicycle facilities with an adaptive stated preference survey. Transport. Res. Part A: Policy Practice 41 (4), 287–301. http://dx.doi.org/10.1016/j.tra.2006.09.007. Vandenbulcke, G., Dujardin, C., Thomas, I., de Geus, B., Degraeuwe, B., Meeusen, R., Panis, L.I., 2011. Cycle commuting in Belgium: spatial determinants and ‘re-cycling’strategies. Transport. Res. Part A: Policy Practice 45 (2), 118–137. http://dx.doi.org/10.1016/j.tra.2010.11.004. Wahlgren, L., Schantz, P., 2012. Exploring bikeability in a metropolitan setting: stimulating and hindering factors in commuting route environments. BMC Public Health 12 (1), 1. http://dx.doi.org/10.1186/1471-2458-12-168. Wang, R., Liu, C., 2013. Bicycle-Transit Integration in the United States, 2001–2009. J. Public Transport. 16 (3), 95–119. http://dx.doi.org/10.5038/23750901.16.3.6. Wang, X., Lindsey, G., Schoner, J.E., Harrison, A., 2015. Modeling bike share station activity: effects of nearby businesses and jobs on trips to and from stations. J. Urban Plan. Dev. 142 (1), 13–19. http://dx.doi.org/10.1061/(ASCE)UP.1943-5444.0000273. Witlox, F., Tindemans, H., 2004. Evaluating bicycle-car transport mode competitiveness in an urban environment. An activity-based approach. World Transport Policy Practice 10 (4), 32–42. Winters, M., Brauer, M., Setton, E.M., Teschke, K., 2013. Mapping bikeability: a spatial tool to support sustainable travel. Environ. Plan. B: Plan. Des. 40 (5), 865–883. http://dx.doi.org/10.1068/b38185. Wong, K.K., Chan, H.S., 1996. Environmental awareness of environmental protection bureaucrats in the People’s Republic of China. Environmentalist 16 (3), 213–219. Xing, Y., Handy, S.L., Mokhtarian, P.L., 2010. Factors associated with proportions and miles of bicycling for transportation and recreation in six small US cities. Transport. Res. Part D: Transport Environ. 15 (2), 73–81. http://dx.doi.org/10.1016/j.trd.2009.09.004. Zahran, S., Brody, S.D., Maghelal, P., Prelog, A., Lacy, M., 2008. Cycling and walking: explaining the spatial distribution of healthy modes of transportation in the United States. Transport. Res. Part D: Transport Environ. 13 (7), 462–470. http://dx.doi.org/10.1016/j.trd.2008.08.001. Zhao, P., 2014. The impact of the built environment on bicycle commuting: evidence from Beijing. Urban Stud. 51 (5), 1019–1037. http://dx.doi.org/10.1177/ 0042098013494423. Zhao, P., Lü, B., De Roo, G., 2011. Impact of the jobs-housing balance on urban commuting in Beijing in the transformation era. J. Transp. Geogr. 19 (1), 59–69. http://dx.doi.org/10.1016/j.jtrangeo.2009.09.008.