IntroductionThe last mile between metro stations and the homes or workplaces of commuters has become a central topic in relation to metro ridership in megacities. Regarding the different origins and destinations of different commuters and various travel preferences, it is difficult but necessary for governments and city planners to develop applicable solutions for the last-mile problem. The last-mile problem requires more attention and discussion for three main reasons. First, transport services for the last-mile problem would directly affect the accessibility of the metro system (Zuo et al. 2020). At the same time, how to motivate people to use public transport such as metro to control and reduce air pollution is always important but challenging (Chen and Wang 2018). Second, in many cities, especially growing megacities of developing countries, road congestion has been worsening, which makes commuting inefficient (Zhao and Hu 2019). Thus, minimizing time and monetary cost on the last mile could alleviate congestion and improve the efficiency of transport systems. Third, many commuters in megacities may choose to live far away from metro stations in order to reduce housing costs, which leads to a long-distance commute (Zhao et al. 2020). Under such circumstances, providing a better last-mile service for long-distance commuters is conducive to social welfare. However, at present, experts and scholars have mainly studied the last-mile issue from the logistics distribution perspective (Halldórsson and Wehner 2020; Janjevic and Winkenbach 2020; Kapser and Abdelrahman 2020), only a few studies have dealt with this issue from the perspective of individual travel behavior (Li et al. 2019; Zuo et al. 2020).The public bicycle is widely believed to be one of new ways to solve the last-mile problem (Liu et al. 2012; Ma et al. 2015; Ni and Chen 2020). It can improve the accessibility to public transportation systems by directly connecting the origin/destination and metro stations (Liu et al. 2012; Zhang et al. 2019). Compared with other feeder modes, such as walking, electromobiles, and private cars, public bicycles have significant advantages. For example, a study in Hamilton County, Ohio, showed that the transit access distance covered by bicycling may be tripled compared with walking, such that bicycling enhances transit equity (Zuo et al. 2020). A study in Beijing pointed out that people prefer to use dockless bike sharing instead of taxis as a transfer mode, especially in less economically developed areas (Ni and Chen 2020). Furthermore, promoting public cycling provides direct health benefits at the population level (Mueller et al. 2015), along with substantial environmental and economic advantages (Woodcock et al. 2009; Handy and Xing 2011; Handy et al. 2014). Thus, promoting the integration of public bicycles and the metro system has the potential to be a great solution for the last-mile problem (Ji et al. 2018; Liu et al. 2020).Among the many studies on bicycle use, previous research has focused mainly on which kind of people tend to use this mode and which factors influence their choice. People who are younger and more educated are thought to use cycling as the primary mode to travel to work (Daniel et al. 2011). Various factors have been examined to study the determinants of public bicycle ridership (Fishman et al. 2015; Verma et al. 2016; Scott and Ciuro 2019), which can be divided into four main types. These include the built environment (El-Assi et al. 2017), individual characteristics (Titze et al. 2008), travel distance (Zhao and Li 2017), and some natural conditions, such as weather and temperature (An et al. 2019), which show mixed effects on travel behavior in relation to public bicycle use. However, the determinants of commuters’ use of public bicycle as a feeder mode are scarcely discussed. Moreover, since the knowledge on public bicycle use is not transferable among transnational cities (Lin et al. 2018), there is still a lack of research on Chinese megacities.This paper aims to contribute to this topic by looking at Beijing as a case study. As a travel mode with high efficiency and good patency, the metro has become an important travel choice for most commuting and non-work trips in megacities of China. In 2018, the average daily passenger volume of the 16 metro lines in Beijing was 8.537 million and the maximum daily passenger volume was 10.907 million on August 17, 2018. Up to 16.2% of travelers took the metro as their travel mode, and this proportion had increased by 0.8 percentage points from 2017 (BMCT 2019). Policymakers and urban planners are seeking strategies to develop an integrated public transit system that allows riders to travel between their final destinations and metro stations easily. Promoting public bicycles may provide a possible solution to last-mile problem. Despite the benefits of cycling, it is difficult to encourage riders to use bicycles. Evidence has shown that the percentage of taking bicycle as a travel mode in mega cities in China is quite low. In Beijing, only 12.40% of work-based trips (excluding walking) were by bicycle (BTI 2016), and in 2018, only 11.50% of trips were by bicycle (BTI 2019).On June 16, 2012, the launching ceremony of the pilot of Beijing’s public bicycle service system was held in Dongcheng District, one of the two pilot areas in Beijing. The government has regarded the promotion and development of urban public bicycle services as an important part of optimizing the public transportation structure and improving the public transportation service quality (BMCT 2012). According to an interview from the Parking Management Office of the Beijing Municipal Commission of Transport in 2018, there were at least 110,000 docks in Beijing. The average turnover was 2.7, and there were about 200,000 orders per day (BMCT 2018). In recent years, another type of bicycle services, called share bicycles, have grown rapidly in Beijing. However, shared bicycles have long suffered from illegal parking, potential safety risks, damage and disposal, occupation of roads, and deposits difficult to be sent back (JinwanNews 2020). Compared with shared bicycles, public bicycles have advantages in parking management, and subsidies from the government make public bicycles more affordable than shared bicycles.This paper intends to explore how to solve the last-mile problem by using public bicycle as a feeder mode. By applying cluster analysis and binary logit model, this paper aims to find out who are the users of public bicycle and how to encourage people to take public bicycles as a feeder mode. The results will enhance our understanding of public bicycle as a feeder mode to solve the last-mile problem and provide new evidence for policymaking on promoting integrated public transit systems in megacities.The remainder of this paper is organized as follows. The next section, “Literature Review: Factors Involved in Cycling”, reviews earlier studies on this subject and the contribution that the current study makes. The section “Research Method” provides a description of how the survey data were collected, followed by the “Results” section, which outlines the modeling methodology applied for the analysis. The next section presents the sample analysis and modeling results, and the final section, “Discussion”, summarizes the findings and concludes with recommendations for policy.Literature Review: Factors Involved in CyclingThe built environment is generally considered to have effects on cycling behavior. For example, a high-density design is significantly related to a high level of cycling (Guo et al. 2007; Parkin et al. 2008). Diverse land use types also increase the use of public bicycles (Litman 2008; Zhang et al. 2017). These may entail a mixture of significant environmental variables, including the presence of stores, offices, restaurants, and hospitals (Moudon et al. 2005). Most cases from US cities (Dill and Carr 2003), Toronto (El-Assi et al. 2017), and Beijing (Zhao 2014) have shown that a higher level of bicycle infrastructure (e.g., bicycle lanes) resulted in greater use of bicycles. In addition, the impact of one single built environment factor is often associated with that of another, such as land use and bicycle infrastructure (Mitra and Nash 2018; Cervero et al. 2019; Liu and Lin 2019). However, some studies have drawn different conclusions in which transportation infrastructure and built environment have influence on ridership (Cervero et al. 2009; Scott and Ciuro 2019). Since the impacts of the built environment on ridership remain unclear, together with the fact that most studies have discussed general cycling instead of public bicycle systems, the role of the built environment in public bicycle use requires more exploration.Previous studies have suggested that many individuals frame their commutes based in part on the perceived safety of the environment (Riggs 2019). A study using an electronic survey showed that perceived crash risk, adverse weather conditions, and lack of safety are the most relevant discouraging factors for cycling (Useche et al. 2019). Another study based on a survey in Amsterdam found statistically significant positive associations between safety and cycling, especially for women (Timmermans et al. 2019). However, researches have also pointed out that the measurement of safety mainly comes from individuals’ subjective perceptions related to the cycling infrastructure (Ng et al. 2017; Riggs 2019). This means that there is no universal measurement for “safety” (Riggs 2019). Therefore, our paper applies measurable and objective indicators, such as traffic flow and the ratio of bicycle lanes, to reflect the safety level of the route environment. In this way, the interference caused by the cognitive differences among survey respondents can be largely alleviated.Individual characteristics also have significant impacts on cycling. In areas with adequate bicycle infrastructure, individual determinants even outperform the role of the built environment (de Bas et al. 2008). Gender and age are the two most widely discussed factors. Generally, men (Stinson and Bhat 2004; Ogilvie and Goodman 2012; Julii et al. 2019) and young people (Fishman et al. 2015; Dill and McNeil 2013) are thought to be more likely to cycle. For example, a study in London found that females made 1.63 fewer bicycle trips per month than males (Ogilvie and Goodman 2012). As age increases, the preference for participating in active travel such as cycling will decline (Julii et al. 2019). Income level also has an influence on cycling, although it has different influences among cities and research conclusions may even be contrasting (Ogilvie and Goodman 2012; Mitra 2013; Goodman and Cheshire 2014; Handy et al. 2014; Julii et al. 2019; Wang and Akar 2019). For example, in the Netherlands, people of low income were found more likely to be bicycle users (Shelat et al. 2018), while in the United States (Zahran et al. 2008) and the United Kingdom (Parkin et al. 2008), people of higher income were associated with a higher probability of cycling. A study in Australia found that those with relatively high incomes had increased odds of public bicycle membership (Fishman et al. 2015). Individual factors such as the attitude toward cycling and personal experience also have some impact, although they are difficult to measure and less discussed in the literature. Those who prefer affordable travel are more likely to cycle (Zhao and Li 2017), while commuters who encountered personal bicycle stolen tend to use public bicycles more often (Ji et al. 2016). Furthermore, gender also indirectly affect cycling behavior through influencing people’s sensitivity toward the built environment. For example, a study in Toronto found that the effects of built environment-related factors were different between male and female riders (Mitra and Nash 2018). Another study in New York also found that installing more bicycle racks affects women more than men (Wang and Akar 2019).Travel distance is another important factor influencing choosing bicycles or other transport modes (Zhao and Li 2017; Kang and Fricker 2018). Evidence shows that bicycles are more likely to be used for short and home-based trips. For example, a survey in Beijing found that 80% of respondents made cycling transfers within 5 km from home (Zhao and Li 2017), while a study in Spain found that the threshold distance of university students who cycled was 5.1 km (Chillón et al. 2016). Travel distance around 1.5–2 km has been found to have the highest potential for people to use bicycles to replace other public transport modes (Sun and Zacharias 2017). Within the threshold distance, people’s attitudes toward cycling may become more active and they would cycle more when the commute journey intensifies (Heinen et al. 2011). Furthermore, there are some differences among bicycle types. A comparison between public and private bicycles found that the main journey length travelled by private bicycles was 700–800 m longer than public bicycles (Castillo-Manzano et al. 2016).Cycling as a feeder mode has been attracting an increasing amount of research interests (Bachand-Marleau et al. 2011; Ji et al. 2016; Zhao and Li 2017; Ma et al. 2018). Along with walking, cycling is found to function as a significant access and egress mode for transit travelers (Shelat et al. 2018). Thus, an integration of cycling and transit such as the metro has become a hot topic of research (Krygsman et al. 2004; Bachand-Marleau et al. 2011; Zhao and Li 2017; Shelat et al. 2018; Zhang et al. 2019). Some studies in China have suggested that bike sharing as a feeder mode to a metro station could help solve the last-mile problem. For example, a study using smartcard data in Nanjing explored the general characteristics of metro–bicycle transfer trips. Results showed that commuting on workdays is a significant and common purpose of metro–bicycle integration, and that improving the route environment for metro–bicycle integration is of great significance to alleviate traffic congestion in megacities (Ma et al. 2018). Promoting cycling as a feeder mode would bring benefits to both cycling and transit (Ji et al. 2018; Liu et al. 2020). For example, a quantitative analysis of the pioneering large-scale bicycle sharing system called Vélib’ in Paris, France, confirmed this positive effect. The results showed that close coupling of transit and vehicle-sharing systems is beneficial through improvement in the utilization rate of the public transit (Nair et al. 2013). A successful integration between the two will most likely result in an increase in the subsequent ridership of transit, the efficiency of transit, and the overall demand for cycling (Krizek and Stonebraker 2010; Liu et al. 2020; Zuo et al. 2020).The use of bicycle as a feeder mode in metro–bicycle integration is influenced by many factors. As for the built environment, mixed land use, parks and public squares, and transportation facilities such as bicycle lanes have been found to be positively related to integrated use (Guo and He 2020). Areas that are less economically developed with fewer signalized intersections are more favored by those who use metro–bicycle integration (Ni and Chen 2020). However, research conclusions have differed among cities. In cities in China, such as Beijing and Nanjing, the combined use with bicycle as a feeder mode occurs more in high-density areas, while in cities in the United States, such as Minneapolis, MN, and Washington, DC, areas with low population density attract more ridership (Martin and Shaheen 2014; Lin et al. 2018). As for individual factors, people with a high school degree or even below may use public bicycles as a feeder mode more frequently (Liu et al. 2020). Female, older, and low-income commuters are less likely to use public bicycles to access transit (Ji et al. 2016; Yang et al. 2016). The lack of a second car in a household may be one reason for people choosing the metro–bicycle integration (Liu et al. 2020). As for travel-related characteristics, travel purpose shows a significant impact. A study in Nanjing, China, found that integration occurs more frequently for non-time-sensitive trip purposes, such as visiting friends and shopping (Chen et al. 2012). Moreover, travel with long time and distance tends to have a higher possibility of metro–bicycle integration (Yang et al. 2016; Liu et al. 2020).However, there are still some research gaps in studies on public bicycles. First, although there are many studies investigating the adoption of public bikes, there is little discussion on how a public bicycle system could better integrate with a metro system. Moreover, further analysis is needed for planners and policymakers to identify which factors can influence the metro–bicycle integration. For example, many studies have paid attention to public bicycle usage and noted that several factors could influence people’s choice, including infrastructure such as bicycle lanes (Dill and Carr 2003; Faghih-Imani et al. 2014; Verma et al. 2016; Assunçao-Denis and Tomalty 2019; Cauwenberg et al. 2019) and the connectivity of streets (Moudon et al. 2005; Dill and Voros 2007; Titze et al. 2008; Beenackers et al. 2012). In addition to built-environment factors, studies have also focused on the association between people’s individual characteristics and their cycling behavior (Cervero et al. 2009; Scott and Ciuro 2019). It seems that there is already some consensus on the value of public bicycles, as well as which factors may influence their usage. However, the use of a public bicycle as a feeder mode is much less discussed in the literature. Although metro–bicycle integration has shown great potential (Zhang et al. 2019), there is a lack of agreement on how to promote this integration in megacities. Thus, further studies are needed to provide more indications for planners and policymakers. Second, although many studies have explored the correlation between the usage of public bicycles and the built environment attributes around the destination or the origin, few of them have paid attention to the factors that could influence the whole journey (such as tree shade along the route or the directness of the trip). Thus, more explorations are needed to see how built environment along the route affects travel behavior for metro–bicycle integration and whether different attributes influence each other.Research MethodCity ContextBeijing is the capital of China, with a population of 21.54 million and a land area of 16,410.54 km² in 2019 (BeijingDaily 2020). The rapidly growing population and urban development have brought many transportation challenges. Beijing has the highest level of traffic congestion index in the country. During the rush hour of commuting, the reported congestion index may rise to above 8.0, which indicates a “severe congestion” level (BeijingNews 2018). According to an analysis of vehicle quantity and usage characteristics by Beijing Transport Institute, the number of vehicles in Beijing had reached 6.084 million, with an annual growth rate of 3.0% by the end of 2018 (BTI 2019). Although the government has implemented some corresponding regulatory policies, the dependency on private cars is still strong. To solve congestion problems caused by the increasing population and high level of private vehicle usage, planners have proposed the promotion of public transport and the green travel mode.Cycling used to be the dominant travel mode in Beijing during the 1980s, when China was known as the “Kingdom of Bicycles.” However, the use of bicycles decreased rapidly with the development of the economy and the acceleration of urbanization. In 2018, only 11.50% of trips in central urban areas were by cycling, while 27% of trips were by cars (BTI 2019). Under this circumstance, the government has tried to promote the use of bicycles, including investments in providing more public bicycles and infrastructure, as well as new policies and plans. Since 2012, Beijing has invested in the construction of a public bicycle program. By the end of 2018, there were 3,575 public bicycle service stations in Beijing, which is a 9% increase over the previous year (BTI 2019). However, the proportion of trips by bicycle has remained at a low level in the past decade (Fig. 1).In addition to the public bicycle system provided by the government, dockless bike-sharing systems have grown rapidly in China. Dockless bike-sharing entered Beijing in September 2016, and a large quantity of dockless shared bicycles have been put into the city. By the end of 2018, there were nine bike-sharing enterprises and the number of operating bicycles was 1.91 million (BTI 2019). Previous studies and empirical cases have shown the huge potential of bicycles to solve last-mile problems (Liu et al. 2012; Zuo et al. 2020). Although multiple shared bicycle systems have grown rapidly, shared bicycles have long suffered from illegal parking, potential safety risks, damage and disposal, occupation of roads, and deposits difficult to be sent back. Compared with shared bicycles, public bicycles have advantages in parking management, and receive more governmental subsidies to keep affordable. Public bicycles thus have greater potential to provide an efficient way to improve the utilization rate of public transportation and keep the use of private cars under control.Survey and DataThis paper concentrates on the determinants of cycling behavior as a feeder mode to metro systems in Beijing. Three dependent variables are discussed: travel mode choice, travel route choice, and travel distance distribution. Mode choice refers to travelers’ preferences between cycling and walking. Route choice refers to their preferences among several types of routes, which are categorized by clustering analysis of the built environment. Travel distance distribution refers to the distribution characteristics of the riding distance when travelers take bicycle as a feeder mode, such as how far most travelers ride. To explore and explain the underlying mechanisms that lead to these aspects of travel, both built environment and individual socioeconomic factors were controlled and analyzed.To investigate the commuting trips with public bicycle as a feeder mode to metro transit, two metro stations with a high level of traffic volume, Chaoyangmen and Liangmaqiao, were selected (Fig. 2). These two metro stations are both located in flat, dense inner-city areas. Their surroundings are characterized by mixed land use of office buildings, restaurants, retailing, amenities, and multifamily dwellings, to keep the other built environment factors under control. Fig. 3 shows the distribution of main roads and secondary roads around the two metro stations. To accurately collect information of people’s travel behavior during commuting, we conducted a survey in May 2016. Commuters’ mode choice, route choice, as well as individual socioeconomic attributes are all based on their statements through this survey. Since weather has a significant influence on travel behavior along with temperature (Böcker et al. 2013; Corcoran et al. 2014; Faghih-Imani et al. 2014; El-Assi et al. 2017), the survey also controlled natural environment factors. Rainy days were excluded from the survey, and the average temperature was between 8°C and 20°C.Our sample included passengers leaving or entering metro stations for commuting trips. Questionnaires were sent out on weekdays around the two metro stations. Information on random travelers’ behavior and their individual socioeconomic attributes was collected, including gender, age, income, education, and type of work. Invalid questionnaires, such as incomplete or inconsistent surveys, were removed. Finally, a total of 403 commuters provided valid responses. Among those respondents, up to 94.29% of respondents chose riding public bicycles or walking as a feeder mode. And the proportion of other modes (including bus, private car, taxi, and motorcycle) was less than 6%. Since this study focuses on the usage of public bicycle as a feeder mode and aims to discuss the contributions of public bicycle to cover the last mile, we select the respondents choosing walking or riding public bicycles as our samples. There are 380 valid respondents (47.0% female) aged from 16 to 60 years in further analysis. Educational level might be a measure of socioeconomic status and was categorized into three groups: 9.5% are noncollege educated (secondary or elementary school as highest degree), 68.9% are college educated (college or university as highest degree), and 21.6% have graduate-level education (master or doctorate as highest degree). In terms of income, 27.1% can be categorized as low-income earners (below 5,000 CNY per month), 47.6% as middle-income earners, and 25.3% as high-income earners (above 10,000 CNY per month). In this study, types of employment are categorized based on the People’s Republic of China Occupational Classification, and up to 84.2% belong to two major categories (technical professional 57.9% and commercial service professional 26.3%) (Fig. 4).In addition to data from the survey questionnaires, this study utilizes data from relevant databases and published documents. The data for land use and public facilities were obtained from the Beijing Municipal Institute of City Planning and Design. The data for travel distance were calculated according to the route choices reported by respondents in the questionnaire.Method and ModelThis paper focuses on three aspects of travelers’ transfer behaviors: mode choice, route choice, and travel distance. The first part of the study explores the influence of individual socioeconomic attributes, travel distance, and route type on the mode choice for accessing metro stations. The second part shifts to determine which kind of route cyclists of different characteristics prefer. The third part examines the most popular distance by public bicycle. According to the survey results, modes of short-distance transfer travel around the two metro stations mainly consisted of two categories: public cycling and walking. Respondents choosing other modes (taxi, bus, and motorbike) as a feeder mode accounted for less than 10%. This distribution of travel mode supports the evidence from earlier studies (Krygsman et al. 2004; Bachand-Marleau et al. 2011; Shelat et al. 2018). Therefore, this paper focuses on the two main mode choices (cycling and walking) to discuss the determinants. Based on the data characteristics and previous research (Ling et al. 2017; Zuo et al. 2020), this study selected the binary logit model for regression analysis. The construction model is realized by(1) ln(pi1−pi)=α+∑bjxj+ε(2) where pi = probability when the dependent variable yi = 1, that is, the probability that respondents choose public bicycles to travel; xj = various influencing factors studied that may affect the preference for bicycles; bj = regression coefficient of independent variables that represents the possible influence of each variable on the dependent variable; α = constant term and ɛ = the error term; and ui1 = utility of sample i to choose public bikes and ui2 = utility of sample i to choose the other mode, which refers to walking here.Table 1 lists the definition and measurement unit of each variable in the model. Among the variables, individual characteristics, such as age, gender, income level, profession, and education level, were obtained directly from the survey. Data about transfer travel were obtained according to the trip endpoints (origins and destinations) of the respondents. Clustering analysis was applied to divide routes with different characteristics, which resulted in four categories.Table 1. Definitions and measurements of variables adopted in the binary logit modelTable 1. Definitions and measurements of variables adopted in the binary logit modelVariableDefinitionUnitAgeRespondent’s ageYearGenderMale (=1) or female (=0)—Low-incomeMonthly income under 5,000 (=1)CNYMiddle-incomeMonthly income between 5,000 and 10,000 (=2)CNYHigh-incomeMonthly income over 10,000 (=3)CNYProfessionalBased on the People’s Republic of China Occupational Classification—EducationHigh school education or below (=1), college degree (=2), graduate degree or above (=3)—Travel distanceActual travel distance between a metro station and the trip endpointmRoute categoryResults of the K-means clustering (1, 2, 3, 4)—In terms of the clustering analysis, cross-section data of the built environment of each route were consulted and integrated. Survey attributes and clustering methods were selected based on our focus on the mode choice (walking or cycling) and route choice. Table 2 lists the definitions and measurement units of the selected routes’ attribute indicators. The K-means clustering analysis method was selected to classify a total of seven indicators. By repeatedly adjusting the number of initial categories and comparing the clustering results, routes were ultimately divided into four categories. Table 3 presents the final clustering center (standardized) and provides a qualitative description of the route characteristics in the four different categories, based on the clustering results.Table 2. Definitions and measurements of the route characteristicsTable 2. Definitions and measurements of the route characteristicsIndicatorDefinitionUnitA. Street intersectionNumber of street intersections/length of routeIntersection/mB. Arterial intersectionNumber of arterial intersections/length of routeIntersection/mC. Ratio of bicycle laneLength of bike lane/length of routem/mD. DirectnessRatio of straight-line distance to actual distance%E. Tree shadeNumber of shade trees/length of routeTree/mF. Street lampNumber of streetlamps/length of routeLamp/mG. Traffic flowPassenger car unit/length of route/an hourPCU/m/hTable 3. Results of the K-means clustering analysisTable 3. Results of the K-means clustering analysisIndicatorRoute 1 (26.84%)Route 2 (12.64%)Route 3 (20.26%)Route 4 (40.26%)Higher directness, less traffic flowMore arterial intersection, higher ratio of bicycle lane, less street intersection, fewer streetlamps, and less tree shadeMore tree shade, lower directnessMore traffic flow, more street intersection, higher directnessStreet intersection0.10941−1.79837−0.021410.49607Arterial intersection−0.141161.86757−0.73438−0.12318Ratio of bicycle lane−0.480291.622550.03921−0.21004Directness0.57702−0.80258−1.200770.47580Tree shade0.13519−1.465780.668400.02147Street lamp0.24061−1.572480.269600.19877Traffic flow−1.05368−0.461130.218910.74210ResultsCycling Mode ChoiceThe binary logit model of accuracy of fit reached up to 94.494. Variables with coefficient-significances below the confidence level of 1−α = 90% were removed from the estimations. Table 4 gives the final variables, including age, income level, and travel distance.Table 4. Results of the binary logit modelTable 4. Results of the binary logit modelVariableBS.E.WalsdfSig.Exp(B)Age0.0660.01616.55110.0001.068Income——11.16320.004—Low-income0.7330.4093.21610.0732.081Middle-income1.1160.33910.83510.0013.053Travel distance0.0010.00022.19310.0001.001Constant−4.8080.92427.06810.0000.008According to the modeling results, age is positively associated with the preference for bicycle riding, which differs from some previous studies (Cervero et al. 2009). Calculating the rate of cycling in different age groups, the results show 29.6% in 16–30 years, 45.5% in 30–45 years, and 68.8% in 45–60 years. An explanation may be that in the age range of 16–60 years, people’s travel behavior still reflects flexibility and mobility, without too much concern about being too old to ride a bicycle. Among the population aged 16–60 years, older travelers pay more attention to the potential benefits that bicycling can attain. Cycling is more than a mode for commuting but also an active lifestyle to keep them healthy.The high-income group was set as the reference group. The modeling result shows that, compared with the high-income group, travelers with lower income are more likely to utilize public bicycles as the transfer mode for commuting. The income effect of the middle-income group is more significant (p ≈ 0.001). Exp(B) value shows that the odds of choosing public bicycles for the low-income group are nearly 2.08 times greater than for the high-income group, and the odds of the middle-income group are 3.05 times greater than for the high-income group. It seems that people with a higher income are less likely to ride a public bicycle. A possible explanation may be the difference of lifestyle and habits among groups with low income, middle income, and high income. As for the high-income group, they may be more critical in relation to the comfort and safety of the trips. What is more, although gender has not shown a significant influence on riding a public bicycle, there may still be some gender difference. For example, for those females wearing a dress or taking an umbrella against the sun, walking seems much more convenient and comfortable than cycling. Among the surveyed respondents, the proportion of females in the high-income group (69.79%) is much higher than that in the low-income group (51.46%) and the middle-income group (45.30%). Thus, the different lifestyles and habits, along with some gender difference may be the reasons that people with higher income are less likely to take public bicycle as a feeder mode.One of the most important factors associated with cycling is distance (Heinen et al. 2010), which is confirmed by the results of this study. The exp(B) value suggests that when the travel distance increases by 1 m, the odds of choosing a public bicycle as the transfer mode increase by 0.1%, and when distance increases from 1 to 2 km, the odds of cycling increase up to 1.71 times. Longer travel distance makes it more inconvenient to walk and raises the time cost, while public bicycles provide a better choice. In addition to the association between travel distance and transfer mode choice, our survey results even showed differences in the travel distance distribution between different categories of routes. This might indicate a relationship between travel distance and route preference, which will be discussed later in the paper.According to the model, three factors of age and travel distance were proven to have a positive influence on travelers’ decision to use a bicycle to go to metro stations. Income also has an influence, being that people with low and medium income are more likely to take public bicycle as a feeder mode. Other factors, such as gender, education level, and route types, did not have a significant influence on mode choice. However, this does not mean the route characteristics have nothing to do with travel behavior. Since individuals are faced with different situations and options, more discussion is needed for a better understanding. In the next section, the paper will turn to address how commuters choose their cycling route from the several categories of possible options.Cycling Route ChoiceTo understand commuters’ route preferences while riding to access metro stations, the samples will be classified based on which type of route the commuter takes. For commuters with different individual characteristics (e.g., male, female), Table 5 presents the distribution of the different groups’ route choices when using public bicycles as their feeder mode.Table 5. Percentages of four categories of routes chosen by each groupTable 5. Percentages of four categories of routes chosen by each groupGroupRoute 1Route 2Route 3Route 4Male30.776.4120.5142.31Female38.244.4111.7645.59Aged between 16 and 3034.384.6917.1943.75Aged between 31 and 4528.335.0018.3348.33Aged between 46 and 6050.009.099.0931.82Technical professional38.037.0416.9038.03Commercial service professional34.255.4816.4443.84Low income26.325.265.2663.16Middle income34.255.4816.4443.84High income26.094.3534.7834.78Education level 131.250.000.0068.75Education level 232.715.6117.7643.93Education level 343.488.7021.7426.09According to the clustering results, Route 1 tends to have higher directness and less traffic flow, Route 2 tends to have more arterial intersection and bicycle lanes, but with less tree shade, Route 3 tends to have more tree shade but lower directness, and Route 4 tends to have greater directness but more traffic flow. Among the four types of routes, Route 1 and Route 4 are in an obvious priority, as they are both of higher directness according to the clustering analysis. Evidence has shown that street connectivity is positively correlated with cycling (Dill and Voros 2007; Beenackers et al. 2012), and network connectivity works in favor of bicycle travel. Since street connectivity generally increases the directness level, the result confirms the effect. There is no significant difference between males and females, as they both take Route 4 as the favorite choice, with more traffic flow but more street intersection and higher directness.In terms of age, travelers under 45 years old prefer to take Route 4, while those aged between 46 and 60 show a significant preference for Route 1, which has the same high directness but much lower traffic flow. As age increases, although direct health benefits of cycling are mentioned (Mark and Yoichi 2008), safety is also of particular concern. Older travelers would like to take a convenient and safe route for their cycling. As for the two major groups of professionals, namely technical and commercial services, the latter show a greater preference for Route 4 (43.84%). Similar to the model result, it seems that the type of employment has little effect on travelers’ commuting behavior, and what really matters may be the location which directly determines travel distance.In the binary logit model, the quantitative results proved the influence of income on mode choice. In this section, income also influences travel route choice. The low-income group with less than 5,000 CNY per month (income 1) takes Route 4 as the favorite (63.16%), while the middle-income group (income 2) shows a higher preference for Route 1 and Route 3, and high-income group (income 3) shows the same degree of preference for Route 3 (34.78%) and Route 4 (34.78%). Preference for Route 3 increases and for Route 4 decreases when income level rises. According to the clustering analysis, Route 4 represents a built environment with higher traffic efficiency but less security, while Route 3 has lower efficiency but a higher comfort and security level. An explanation may be that travelers with a high level of income tend to be more concerned about comfort and security. A similar result occurs when it comes to educational level. For the group with a lower educational level (junior high or even below), 68.75% chose Route 4, and no one chose Routes 2 and 3. When the education level rises, preference for Route 4 decreases sharply, while preferences for Routes 1, 2, and 3 all increase. For the group with a master’s or even higher degree, 43.38% chose Route 1, 8.70% chose Route 2, 21.74% chose Route 3, and only 26.09% chose Route 4. This might be explained by the notion that people with a higher education level tend to have other expectations beyond only efficiency, thus preferring routes with less traffic flow, higher ratio of bicycle lane, and more tree shade.Cycling DistanceIn addition to mode choice and route choice, analysis of travel distance also contributes to understanding the travel characteristics while using public bicycle as a feeder mode for metro stations. Fig. 5 shows the histographs of travel distance when travelers cycle to metro stations.The histographs indicate that travel distance in the four categories of routes concentrated significantly at around 1,000 m. More specifically, 78.0% of all the trips on Route 1 were within 1.5 km, while the proportions for Route 2, Route 3, and Route 4 were 62.5%, 75.0%, and 73.4%, respectively. In addition, 88.0% of all the trips on Route 1 were within 2.0 km, and the proportions for Route 2, Route 3, and Route 4 were 75.0%, 95.8%, and 95.3%, respectively. As the figure illustrates, whichever types of route travelers take, travel distance when using public bicycle as a feeder mode is all concentrated within 2.0 km. The results confirm that public bicycles play an important role in solving the last-mile problem. By acting as a feeder mode, public bicycles help save the time cost of transfer and thus improve the efficiency of public transport.Among the four types of routes, there are still some differences between Route 1 and Route 4 versus Route 2 and Route 3. The average travel distances of Route 1 and Route 4 are much longer than those of Route 2 and Route 3. Half of the trips in Route 2 and Route 3 are within 1 km, while the proportion is much smaller in Route 1 (30.0%) and Route 4 (45.3%). It is likely that when travel distance is not so long and time cost is still acceptable, travelers tend to prefer more comfortable and secure routes (with more tree shade and a higher ratio of bicycle lane). However, when the travel distance is longer and time cost is higher, directness becomes the dominant factor.To conclude, the distribution of the travel distance suggests two viewpoints. First, it is confirmed that public bicycle system has provided an alternative mode choice for short trips, especially for accessing metro stations. Second, the mode choice and route choice may be influenced not only by the built environment and the individual factors but may also be related to how long distance the trips will cover.DiscussionBased on a survey at two metro stations with high traffic, high density, and mixed land use in Beijing, this research investigated the travel behavior of commuters who use public bicycle as a feeder mode to metro stations. We asked three questions: (1) Which factors influence the mode choice between cycling and walking? (2) What types of routes do commuters prefer when cycling to a metro station? (3) For what distance are public bicycles primarily used as feeder mode, and do they apply to solving the last-mile problem? Based on the binary logit modeling results and statistical analysis, three points are discussed to understand the travel behavior and preference of commuters when connecting their origins/destinations with metro stations. Furthermore, the discussion provides a reference for policymakers on promoting public bicycle as a feeder mode to cover the last mile.First, the results from the binary logit model confirm the influence of socioeconomic individual factors when choosing a feeder mode between bicycling and walking. Among the sampled travelers, older people showed a higher likelihood to use public bicycles as a feeder mode to gain access to/from metro stations, while young people may prefer walking. This result appears to be in contrast with some previous studies that determined that young people and low-income groups were more likely to cycle (Dill and McNeil 2013; Guo and He 2020). Considering the direct health benefits of bicycling, this physical activity makes a contribution to lower the odds of being overweight or obese (Ming and Chris 2008) and increases those of having a higher level of cardiovascular fitness, as well as an overall reduction in cardiovascular risk (Mark and Yoichi 2008). There is also evidence that commuters who take up cycling for transport can reduce body fat (Møller et al. 2011). Thus, a possible explanation could be that, in China, many people aged between 45 and 60 years old may be more likely to do cycling as an exercise activity. Thus, the differences between cities should be considered when drawing research conclusions. Moreover, since the sampled travelers were all under 60 years old, this result does not mean that elderly people cycle more when they are over age 60. In fact, due to the need to be physically fit in order to ride a bicycle, people over 60 may turn to other alternative modes. As for income, the middle-income group tend to be the main user of the public bicycle system around metro stations in Beijing, while the low-income group and the high-income group are less likely to take public bicycle as a feeder mode. This finding indicates that planners and managers of public bicycle system should adopt appropriate charging strategies according to the consumption capacity and preferences of middle-income groups to achieve a larger user base for public bicycling. However, the influence of income differs among cities. In the United States, bicycle usage was reported to be negatively associated with income levels, which may indicate higher usage among people with more-limited transfer options (Krizek and Stonebraker 2010). In the Netherlands, the opposite result was reported, stating that people with high income levels are more likely to use bicycles and rail transit in combination (Krygsman and Dijst 2001; Bachand-Marleau et al. 2011).Second, the built environment represented by the four types of routes in our study did not have a significant effect on mode choice between bicycling and walking. Since multiple attributes of the built environment have been integrated into several categories of routes, the result indicates that, although each single factor has an effect [e.g., more bicycle lanes are found to be associated with higher use of bicycles (Dill and Carr 2003; El-Assi et al. 2017) and mixed land use often increases the use of public bicycles (Zhang et al. 2017)], when the built environment appears as an integrated factor, the effects become weakened. In a real-world situation, travelers are facing with a built environment composed of various attributes rather than a single element, and the offsetting of the positive and negative effects could lead to a failure of some current efforts for promoting cycling. Planning and designing around metro stations should be done from an overall perspective. Instead of only building bicycle lanes, street intersection, tree shade, and streetlamps should also be taken into consideration to work together to create an environment that promotes cycling as a feeder mode. What is more, there exists a difference in built-environment preferences among different groups of travelers. In general, the built environment that improves travel efficiency through higher directness is more attractive for cyclists, while bicycle lanes seem not to be crucial. Previous studies have mainly shown the effects of income and education level on deciding whether to cycle (Handy et al. 2014; Julii et al. 2019), but pay little attention to their effect on which route to choose. The results in this paper indicate that as income and education level grow higher, travelers shift toward a preference for a built environment with higher security and more comfort (such as lower traffic and increased tree shade). It seems that different groups have different expectations about transfer travel. People who are older or have a higher income level or a higher education level pay more attention to comfort and safety in the travel experience, while those who are younger, with a lower income or lower education level, may give priority to a highly efficient built environment.Third, travel distance had a significant influence on mode choice for accessing metro stations. As previous studies have shown that, travel distance could be an important predictor for mode choice (Handy et al. 2014; Chillón et al. 2016). In our study, we found that longer transfer travel distance makes cycling more attractive than walking for commuters. Meanwhile, among commuters who choose a public bicycle as a feeder mode, the distribution of travel distance presents some specific characteristics: 67.9% of cycling trips were between 500 and 1,000 m, and up to 91.8% of cycling trips were under 2,000 m. The survey results also demonstrated that trips for accessing metro stations are mainly around 1 km, and that public bicycle has made a great contribution to cover this short distance. It also confirms that cycling could be one of the most direct, convenient, and pleasant options for many short-distance intra-urban trips (Woodcock et al. 2009). With the substantial improvement of the efficiency between metro stations in megacities like Beijing, the last mile from the origin to a metro station or from a station to the destination provides an opportunity for promoting public bicycles. By integrating public bicycle and metro as a combination, commuters could significantly reduce their travel time, while simultaneously benefitting the environment and their physical health.For future policy, promoting public bicycle for short-distance travel and especially for accessing metro stations could be a target of policy efforts in China. To achieve this, socioeconomic individual characteristics of the potential users and the built-environment features along the transfer route should be studied. On the one hand, since older people (from 46 to 60 years old in our sample) and the middle-income group showed significantly higher preferences for cycling, the design of bicycle facilities and charging strategies could be tailored to the needs of that group. At present, in Beijing the public bicycle service implements a membership card for pricing and cycling within 1 hour is free and 1 CNY per hour if longer. Since the average income in Beijing reached 7,706 CNY per month in 2016, the cost of riding a public bicycle is affordable. On the other hand, the comfort and convenience of public bicycle as a feeder mode must be further improved. In addition to constituting the largest group of users, younger people and the lower or higher income groups could also be encouraged to use a public bicycle as a feeder mode through some targeted services, such as different types of charging rules.Furthermore, promotional measures should be implemented not only around the public bicycle service station, but also along the whole route from metro stations to the bicycle stations nearby. All the built characteristics should be taken into consideration, including bicycle lanes, street intersection, tree shade, and so on, to jointly contribute to a bicycle-friendly built environment. Since 2018, Beijing has started the construction of bicycle lanes and management of a 928 km slow-traffic system in key areas (BTI 2019). A 6.5 km bicycle lane located between the Huilongguan Community and Shangdi District was put into operation in 2019. The construction of bicycle lanes does help to encourage residents to use bicycles more, but this measure alone is not enough, and the optimization of the built environment for cycling along the route is also of great importance. More trees, higher level of directness, more streetlamps, and fewer intersections all maybe useful to promoting the use of public bicycles. Moreover, to better promote the integration of the bicycle system and the metro system, public bicycle service stations should be located near metro stations or bus stations.ConclusionsUsing public bicycle as a feeder mode for access to/from metro stations has been shown to have great potential. Promoting the integration of public bicycle and the metro system provides an effective solution for the last-mile problem. By taking Beijing as a case study, this paper answers the questions that often confuse urban planners and politicians: which groups of people are more likely to use public bicycle as a feeder mode to metro stations, and how can we promote metro–bicycle integration by designing the built environment? Differently from most of the previous studies, this research considered the built environment along the route instead of only in isolated spots. A method of classification was adopted because the travelers faced several types of built environment rather than independent single factors. Therefore, this study provides a new perspective for investigating the influence of the built environment and individual factors on travel behavior.Commuters who are older (the oldest respondent in our sample is aged 60) and with a middle-income level were found to be more likely to use public bicycle systems, whereas the subgroups of gender, occupational class, and education level did not show significant differences. In addition to mode choice, people with different individual socioeconomic characteristics also demonstrated different preferences in terms of route choice. With higher education and income levels, commuters were more concerned about comfort and safety of the route rather than efficiency. Although built-environment characteristics do have an influence on transfer route choice, when the factors appear as a “whole environment” instead of single and separate elements, positive effects may be offset by some negative effects. It is thus important to consider not only more bicycle lanes, but also higher directness, more streetlamps, and more green spaces with trees to jointly build a bicycle-friendly environment. Since up to 91.8% of the respondents used a public bicycle to access metro stations ride within a distance of 2 km, promoting a public bicycle system in Beijing would absolutely make a contribution to solve the last-mile problem. Thus, the public bicycle rental stations and supporting facilities could be located near metro stations and other public transport, such as bus stations. Moreover, pricing for the public bicycle system may be integrated with the fare system of subways, which will offer some incentives to attract commuters to use a bicycle for transfers. Taking all these effects into account, a public bicycle system has great potential to be developed as a green, efficient, and healthy beneficial feeder mode to solve the last-mile problem. Political measures should be taken to specifically encourage metro–bicycle integration. Through the guarantee of infrastructure, the design and adjustment of charging rules, and the advocacy and promotion of a social atmosphere, the integration of bicycles and metro could attract more commuters as well as travelers for other purposes.The study contributes to the urban planning literature through three main aspects. First, the findings and conclusions can enhance our existing understanding of public bicycle as a feeder mode to solve the last-mile problem. We find that commuters who are older (the oldest respondent in our sample is aged 60) and with a middle-income level were found to be more likely to use public bicycle systems. However, the subgroups of gender, occupational class, and education level did not show significant differences. What is more, up to 91.8% of the respondents who used a public bicycle to access metro stations ride within a distance of 2 km. Thus, promoting the integration of public bicycle and a metro system would be beneficial in megacities such as Beijing. Second, the findings provide a view of how to build a friendly environment for better usage of public bicycle in megacities. Our study shows that comfortable and safe routes can encourage commuters to use public bicycle as a feeder mode, especially for those with higher educational and income levels. Furthermore, to build a friendly environment for metro–bicycle integration, multiple factors should be considered jointly. For example, a route with more bicycle lanes may not encourage people to cycle if the directness level is quite low. Third, the findings provide new policy implications that promoting the integration of public bicycles and public transit should be adapted to the local conditions. Since the results of our study in Beijing differ from those in other cities, an effective policy must consider the unique travel preferences of local people. The incentives that worked in the United States, such as increasing bicycles lanes, may not work that well in China if other factors, such as streetlamps and tree shade, are ignored. Local survey and study are crucial for policymaking and urban planning.In terms of future research in this area, there remain some limitations of this study that could be addressed. First, the survey area focused on two similar metro stations, which could represent only one type of land use with similar high density and fixed usage. Therefore, the conclusions may not apply to other districts and cities. And due to the limited scope and time range of the survey, valid samples may not cover all kinds of travelers. In this study, our sample is skewed toward young and middle-aged commuters. Second, the difficulty of collecting spatial data makes it challenging to explore all the built environment elements. Factors that were not included in our study might also influence cycling behavior. For example, the accessibility of public bike stations and the scale of the bicycle supply may also influence commuters’ behavior. Third, the travel situation faced by travelers differs, and there are differences in personal habits, lifestyle, and preferences that may also have an influence on their travel decisions. Further investigations should study more metro station areas with different land use and even in different cities, and comparisons should be made to determine whether the influence of age and income varies across areas. In addition, reasons and explanations for the influence of individual factors and different preferences between groups must be further examined.References Assunçao-Denis, M-È, and R. Tomalty. 2019. “Increasing cycling for transportation in Canadian communities: Understanding what works.” Transp. Res. Part A Policy Pract. 123: 288–304. Bachand-Marleau, J., J. Larsen, and A. M. El-Geneidy. 2011. “Much-anticipated marriage of cycling and transit: How will it work?” Transp. Res. Rec. 2247 (1): 109–117. https://doi.org/10.3141/2247-13. Beenackers, M. A., S. Foster, C. Kamphuis, S. Titze, M. Divitini, M. Knuiman, F. van Lenthe, and B. Giles-Corti. 2012. “Taking up cycling after residential relocation: Built environment factors.” Am. J. Preventive Med. 42 (6): 610–615. https://doi.org/10.1016/j.amepre.2012.02.021. BeijingDaily. 2020. “Statistical communique on Beijing’s national economic and social development in 2019.” May 2, 2020. BeijingNews. 2018. “Traffic in Beijing has reached severe congestion levels during the morning rush hour on Monday.” May 2, 2020. Böcker, L., M. Dijst, and J. Prillwitz. 2013. “Impact of everyday weather on individual daily travel behaviours in perspective: A literature review.” Transp. Rev. 33 (1): 71–91. https://doi.org/10.1080/01441647.2012.747114. BTI (Beijing Transport Institute). 2016. 2016 Beijing transport annual report, 25–29. Beijing: BTI. BTI (Beijing Transport Institute). 2019. 2019 Beijing transport annual report, 29–51. Beijing: BTI. Castillo-Manzano, J. I., L. López-Valpuesta, and A. Sánchez-Braza. 2016. “Going a long way? On your bike! Comparing the distances for which public bicycle sharing system and private bicycles are used.” Appl. Geogr. 71: 95–105. https://doi.org/10.1016/j.apgeog.2016.04.003. Cauwenberg, J. V., I. De Bourdeaudhuij, P. Clarys, B. De Geus, and B. Deforche. 2019. “Older adults’ environmental preferences for transportation cycling.” J. Transp. Health 13: 185–199. https://doi.org/10.1016/j.jth.2019.03.014. Cervero, R., S. Denman, and Y. Jin. 2019. “Network design, built and natural environments, and bicycle commuting: Evidence from British cities and towns.” Transp. Policy 74: 153–164. https://doi.org/10.1016/j.tranpol.2018.09.007. Cervero, R., O. L. Sarmiento, E. Jacoby, L. F. Gomez, and A. Neiman. 2009. “Influences of built environments on walking and cycling: Lessons from bogotá.” Int. J. Sustainable Transp. 3 (4): 203–226. https://doi.org/10.1080/15568310802178314. Chen, L., A. J. Pel, X. Chen, D. Sparing, and I. A. Hansen. 2012. “Determinants of bicycle transfer demand at metro stations: Analysis of stations in Nanjing, China.” Transp. Res. Rec. 2276 (1): 131–137. https://doi.org/10.3141/2276-16. Chillón, P., J. Molina-García, I. Castillo, and A. Queralt. 2016. “What distance do university students walk and bike daily to class in Spain.” J. Transp. Health 3 (3): 315–320. https://doi.org/10.1016/j.jth.2016.06.001. Corcoran, J., T. Li, D. Rohde, E. Charles-Edwards, and D. Mateo-Babiano. 2014. “Spatio-temporal patterns of a public bicycle sharing program: The effect of weather and calendar events.” J. Transp. Geogr. 41: 292–305. https://doi.org/10.1016/j.jtrangeo.2014.09.003. Daniel, F., G. Lise, K. Yan, D. Mark, F. Michel, M. Patrick, and D. Louis. 2011. “Use of a new public bicycle share program in Montreal, Canada.” Am. J. Preventive Med. 41 (1): 80–83. https://doi.org/10.1016/j.amepre.2011.03.002. de Bas, G., I. De Bourdeaudhuij, C. Jannes, and R. Meeusen. 2008. “Psychosocial and environmental factors associated with cycling for transport among a working population.” Health Educ. Res. 23 (4): 697–708. Dill, J., and T. Carr. 2003. “Bicycle commuting and facilities in major U.S. cities: If you build them, commuters will use them.” Transp. Res. Rec. 1828 (1): 116–123. https://doi.org/10.3141/1828-14. Dill, J., and N. McNeil. 2013. “Four types of cyclists?: Examination of typology for better understanding of bicycling behavior and potential.” Transp. Res. Rec. 2387 (1): 129–138. https://doi.org/10.3141/2387-15. Dill, J., and K. Voros. 2007. “Factors affecting bicycling demand: Initial survey findings from the Portland, Oregon, Region.” Transp. Res. Rec. 2031 (1): 9–17. https://doi.org/10.3141/2031-02. El-Assi, W., M. S. Mahmoud, and K. N. Habib. 2017. “Effects of built environment and weather on bike sharing demand: A station level analysis of commercial bike sharing in Toronto.” Transportation 44 (3): 589–613. https://doi.org/10.1007/s11116-015-9669-z. Faghih-Imani, A., N. Eluru, A. M. El-Geneidy, M. Rabbat, and U. Haq. 2014. “How land-use and urban form impact bicycle flows: Evidence from the bicycle-sharing system (BIXI) in Montreal.” J. Transp. Geogr. 41: 306–314. https://doi.org/10.1016/j.jtrangeo.2014.01.013. Fishman, E., S. Washington, N. Haworth, and A. Watson. 2015. “Factors influencing bike share membership: An analysis of Melbourne and Brisbane.” Transp. Res. Part A Policy Pract. 71: 17–30. Goodman, A., and J. Cheshire. 2014. “Inequalities in the London bicycle sharing system revisited: Impacts of extending the scheme to poorer areas but then doubling prices.” J. Transp. Geogr. 41: 272–279. https://doi.org/10.1016/j.jtrangeo.2014.04.004. Guo, J. Y., C. R. Bhat, and R. B. Copperman. 2007. “Effect of the built environment on motorized and nonmotorized trip making: Substitutive, complementary, or synergistic?” Transp. Res. Rec. 2010 (1): 1–11. https://doi.org/10.3141/2010-01. Guo, Y., and S. Y. He. 2020. “Built environment effects on the integration of dockless bike-sharing and the metro.” Transp. Res. Part D Transp. Environ. 83: 102335. https://doi.org/10.1016/j.trd.2020.102335. Heinen, E., K. Maat, and B. van Wee. 2011. “The role of attitudes toward characteristics of bicycle commuting on the choice to cycle to work over various distances.” Transp. Res. Part D Transp. Environ. 16 (2): 102–109. https://doi.org/10.1016/j.trd.2010.08.010. Janjevic, M., and M. Winkenbach. 2020. “Characterizing urban last-mile distribution strategies in mature and emerging e-commerce markets.” Transp. Res. Part A Policy Pract. 133: 164–196. https://doi.org/10.1016/j.tra.2020.01.003. Ji, Y., Y. Fan, A. Ermagun, X. Cao, W. Wang, and K. Das. 2016. “Public bicycle as a feeder mode to rail transit in China: The role of gender, age, income, trip purpose, and bicycle theft experience.” Int. J. Sustainable Transp. 11 (4): 308–317. https://doi.org/10.1080/15568318.2016.1253802. Ji, Y., X. Ma, M. Yang, Y. Jin, and L. Gao. 2018. “Exploring spatially varying influences on metro-bikeshare transfer: A geographically weighted poisson regression approach.” Sustainability 10 (5): 1526. https://doi.org/10.3390/su10051526. JinwanNews. 2020. “Has shared bicycle led to serious problems?” July 25, 2021. Julii, B., C. Rachel, L. Kathleen, and S. Charlotte. 2019. “Age, sex and other correlates with active travel walking and cycling in England: Analysis of responses to the active lives survey 2016/17.” Preventive Med. 123: 225–231. https://doi.org/10.1016/j.ypmed.2019.03.043. Kapser, S., and M. Abdelrahman. 2020. “Acceptance of autonomous delivery vehicles for last-mile delivery in Germany—Extending UTAUT2 with risk perceptions.” Transp. Res. Part C Emerging Technol. 111: 210–225. https://doi.org/10.1016/j.trc.2019.12.016. Krizek, K. J., and E. W. Stonebraker. 2010. “Bicycling and transit: A marriage unrealized.” Transp. Res. Rec. 2144 (1): 161–167. https://doi.org/10.3141/2144-18. Krygsman, S., and M. Dijst. 2001. “Multimodal trips in the Netherlands: Microlevel individual attributes and residential context.” Transp. Res. Rec. 1753 (1): 11–19. https://doi.org/10.3141/1753-02. Krygsman, S., M. Dijst, and T. Arentze. 2004. “Multimodal public transport: An analysis of travel time elements and the interconnectivity ratio.” Transp. Policy 11 (3): 265–275. https://doi.org/10.1016/j.tranpol.2003.12.001. Lin, J.-J., P. Zhao, K. Takada, S. Li, T. Yai, and C.-H. Chen. 2018. “Built environment and public bike usage for metro access: A comparison of neighborhoods in Beijing, Taipei, and Tokyo.” Transp. Res. Part D Transp. Environ. 63: 209–221. https://doi.org/10.1016/j.trd.2018.05.007. Litman, T. 2008. Land use impacts on transport how land use factors affect travel behavior. Victoria, BC: Victoria Transport Policy Institute 9-25. Liu, Y., Y. Ji, T. Feng, and Z. Shi. 2020. “Use frequency of metro-bikeshare integration: Evidence from Nanjing, China.” Sustainability 12 (4): 1426. https://doi.org/10.3390/su12041426. Ma, T., C. Liu, and S. Erdoğan. 2015. “Bicycle sharing and public transit: Does capital bikeshare affect metrorail ridership in Washington, D.C.?” Transp. Res. Rec. 2534 (1): 2–19. Ma, X., Y. Ji, M. Yang, Y. Jin, and T. Xu. 2018. “Understanding bikeshare mode as a feeder to metro by isolating metro-bikeshare transfers from smart card data.” Transp. Policy 71: 57–69. https://doi.org/10.1016/j.tranpol.2018.07.008. Ming, W. L., and R. Chris. 2008. “Inverse associations between cycling to work, public transport, and overweight and obesity: Findings from a population based study in Australia.” Preventive Med. 46 (1): 29–32. https://doi.org/10.1016/j.ypmed.2007.08.009. Mitra, R., and S. Nash. 2018. “Can the built environment explain gender gap in cycling? An exploration of university students’ travel behavior in Toronto, Canada.” Int. J. Sustainable Transp. 13 (2): 138–147. https://doi.org/10.1080/15568318.2018.1449919. Møller, N. C., L. Østergaard, J. R. Gade, J. L. Nielsen, and L. B. Andersen. 2011. “The effect on cardiorespiratory fitness after an 8-week period of commuter cycling—A randomized controlled study in adults.” Preventive Med. 53 (3): 172–177. https://doi.org/10.1016/j.ypmed.2011.06.007. Moudon, A. V., C. Lee, A. D. Cheadle, C. W. Collier, D. Johnson, T. L. Schmid, and R. D. Weather. 2005. “Cycling and the built environment, a US perspective.” Transp. Res. Part D Transp. Environ. 10 (3): 245–261. https://doi.org/10.1016/j.trd.2005.04.001. Mueller, N., R.-R. David, C.-H. Tom, d. N. Audrey, D. Evi, G. Regine, G. Thomas, I. P. Luc, K. Sonja, and N. Mark. 2015. “Health impact assessment of active transportation: A systematic review.” Preventive Med. 76: 103–114. https://doi.org/10.1016/j.ypmed.2015.04.010. Ng, A., A. K. Debnath, and K. C. Heesch. 2017. “Cyclist’ safety perceptions of cycling infrastructure at un-signalised intersections: Cross-sectional survey of Queensland cyclists.” J. Transp. Health 6: 13–22. https://doi.org/10.1016/j.jth.2017.03.001. Ni, Y., and J. Chen. 2020. “Exploring the effects of the built environment on two transfer modes for metros: Dockless bike sharing and taxis.” Sustainability 12 (5): 2034. https://doi.org/10.3390/su12052034. Ogilvie, F., and A. Goodman. 2012. “Inequalities in usage of a public bicycle sharing scheme: Socio-demographic predictors of uptake and usage of the London (UK) cycle hire scheme.” Preventive Med. 55 (1): 40–45. https://doi.org/10.1016/j.ypmed.2012.05.002. Parkin, J., M. Wardman, and M. Page. 2008. “Estimation of the determinants of bicycle mode share for the journey to work using census data.” Transportation 35 (1): 93–109. https://doi.org/10.1007/s11116-007-9137-5. Riggs, W. 2019. “Perception of safety and cycling behaviour on varying street typologies: Opportunities for behavioural economics and design.” Transp. Res. Procedia 41: 204–218. https://doi.org/10.1016/j.trpro.2019.09.039. Scott, D. M., and C. Ciuro. 2019. “What factors influence bike share ridership? An investigation of Hamilton, Ontario’s bike share hubs.” Travel Behav. Soc. 16: 50–58. https://doi.org/10.1016/j.tbs.2019.04.003. Stinson, M. A., and C. R. Bhat. 2004. “Frequency of bicycle commuting: Internet-based survey analysis.” Transp. Res. Rec. 1878 (1): 122–130. https://doi.org/10.3141/1878-15. Timmermans, E. J., E. M. Veldhuizen, T. Mäki-Opas, M. B. Snijder, J. Lakerveld, and A. E. Kunst. 2019. “Associations of neighbourhood safety with leisure-time walking and cycling in population subgroups: The HELIUS study.” Spatial Spatio-Temp. Epidemiol. 31: 100300. https://doi.org/10.1016/j.sste.2019.100300. Titze, S., W. J. Stronegger, S. Janschitz, and P. Oja. 2008. “Association of built-environment, social-environment and personal factors with bicycling as a mode of transportation among Austrian city dwellers.” Preventive Med. 47 (3): 252–259. https://doi.org/10.1016/j.ypmed.2008.02.019. Useche, S. A., L. Montoro, J. Sanmartin, and F. Alonso. 2019. “Healthy but risky: A descriptive study on cyclists’ encouraging and discouraging factors for using bicycles, habits and safety outcomes.” Transp. Res. Part F Traffic Psychol. Behav. 62: 587–598. https://doi.org/10.1016/j.trf.2019.02.014. Verma, M., T. M. Rahul, P. V. Reddy, and A. Verma. 2016. “The factors influencing bicycling in the Bangalore city.” Transp. Res. Part A Policy Pract. 89: 29–40. Yang, M., X. Liu, W. Wang, Z. Li, and J. Zhao. 2016. “Empirical analysis of a mode shift to using public bicycles to access the suburban metro: Survey of Nanjing, China.” J. Urban Plann. Dev. 142 (2): 05015011. https://doi.org/10.1061/(ASCE)UP.1943-5444.0000299. Zahran, S., S. D. Brody, P. Maghelal, A. Prelog, and M. Lacy. 2008. “Cycling and walking: Explaining the spatial distribution of healthy modes of transportation in the United States.” Transp. Res. Part D Transp. Environ. 13 (7): 462–470. https://doi.org/10.1016/j.trd.2008.08.001. Zhang, Y., T. Thomas, M. Brussel, and M. van Maarseveen. 2017. “Exploring the impact of built environment factors on the use of public bikes at bike stations: Case study in Zhongshan, China.” J. Transp. Geogr. 58: 59–70. https://doi.org/10.1016/j.jtrangeo.2016.11.014. Zhang, Z., C. Qian, and Y. Bian. 2019. “Bicycle–metro integration for the “last mile”: Visualizing cycling in Shanghai.” Environ. Plann. A Econ. Space 51 (7): 1420–1423. https://doi.org/10.1177/0308518X18816695. Zhao, P., and S. Li. 2017. “Bicycle-metro integration in a growing city: The determinants of cycling as a transfer mode in metro station areas in Beijing.” Transp. Res. Part A Policy Pract. 99: 46–60. Zhao, P., D. Liu, Z. Yu, and H. Hu. 2020. “Long commutes and transport inequity in China’s growing megacity: New evidence from Beijing using mobile phone data.” Travel Behav. Soc. 20: 248–263. https://doi.org/10.1016/j.tbs.2020.04.007. Zuo, T., H. Wei, N. Chen, and C. Zhang. 2020. “First-and-last mile solution via bicycling to improving transit accessibility and advancing transportation equity.” Cities 99: 102614. https://doi.org/10.1016/j.cities.2020.102614.