Factors and Mechanism Affecting the Attractiveness of Public Transport: Macroscopic and Microscopic Perspectives

Abstract

Enhancing the attractiveness of public transport is a key objective for transit agencies that seek to achieve sustainable urban mobility. This study identifies the key factors that affect the attractiveness of public transport from the macroscopic and microscopic perspectives and establishes a comprehensive influence mechanism. Weighted least square-Structural equation model (WLS-SEM) is used to quantify the microscopic factors. Conclusions are summarized as follows: ① the macroscopic factors include urban spatial layout, land use pattern, geographical environment, level of economic development, demographic characteristics, public policy, and service characteristics of public transport; ② the microscopic factors include traveler’s individual characteristics, travel demand characteristics, travel characteristics and service level of public transport; ③ service level of public transport most positively impacts the attractiveness, while the travel characteristics of public transport impact it most negatively.

This work is licensed under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.

1. Introduction

Public transport plays an important role in the urban environment and contributes to the sustainability of urban development [1]. Public transport is a quasipublic product with positive externality, scalable economy, and fairness [2]. Public transport directly supports social welfare and livability [3], meanwhile providing other benefits such as air pollution mitigation and accessibility improvements [4]. The attention to public transport has been shown to directly affect people’s willingness to use public transport. Enhancing the level of the attractiveness of public transport in urban and intercity areas has therefore been recommended [5, 6].

Most developed countries have been through the negatives of urban development, such as traffic congestion and air pollution; therefore, most people prefer to use public transport [7, 8]. The ultimate goal of prioritizing public transport is to improve the quality of public transport services and enhance its attractiveness. Public transport system is a complicated system that involves three subjects: the public, the operators, and the government [9]. Which factors affect the attractiveness of public transport? Which factors are the key factors and do their impacts have differences? What is the relationship among the factors? To answer these questions, we explored the key factors and their interrelationship with the attractiveness of public transport. We identified the key influencing factors and proposed the policies and strategies to increase the attractiveness of public transport. The identified factors play an important role in improving the attractiveness of public transport. Recently, with the growth of the research efforts on the attractiveness of public transport, the factors affecting the attractiveness of public transport have been continually expanded [10–12]; Idris. et al., 2015; Chakrabarti S., 2017; [13, 14]. However, most research was conducted from a single perspective (either from a macroscopic or microscopic perspective). The state-of-the-art is unable to fully describe the key factors and their interrelationship with the attractiveness of public transport. This study comprehensively elaborates the attractiveness of public transport from both macroscopic and microscopic perspectives, analyzes the key factors affecting the attractiveness of public transport comprehensively, clarifies the interrelationship among the relevant factors, and establishes a comprehensive influence mechanism of the factors affecting the attractiveness of public transport. It is of great significance to promote the prioritization of public transport and the sustainability of urban mobility.

At the microscopic level, Discrete Choice Models (such as Binary Logit Model, Multinomial Logit Model, Mixed Logit Model, and Multinomial Probit Model) have been widely adopted to study travelers’ willingness-to-choose of public transport [3, 14–17]. Each independent variable is required to have no measurement error, which can not be avoided in practice. Nevertheless, there are also intricate relationships among the factors that affect the willingness-to-choose of public transport, yet Discrete Choice Models are difficult to identify the endogeneity of those factors. The great superiority is shown by the Weighted Least Square-Structural Equation Model (WLS-SEM) in this aspect. Therefore, the WLS-SEM model is adopted to explore the impact on the choice of public transport in this paper. This is the second contribution of this paper. Conclusions are summarized as follows: (1) the macroscopic factors include urban spatial layout, land use pattern, geographical environment, level of economic development, demographic characteristics, public policy, and service characteristics of public transport; (2) the microscopic factors include traveler’s individual characteristics, travel demand characteristics, travel characteristics of and service level of public transport; (3) service level of public transport most positively impacts the attractiveness, while the travel characteristics of public transport impact most negatively.

The remainder of the paper is organized as follows. Section 2 discusses the factors that affect the attractiveness of public transport from both macroscopic and microscopic perspectives. The methodology applied in this study is introduced in Section 3. In Section 4, we present our survey design in detail. An empirical analysis is then conducted in Section 5. Conclusions are given in Section 6.

2. Factors Affecting the Attractiveness of Public Transport: Macroscopic and Microscopic Perspectives

Psychologically, attractiveness refers to the power that guides people in one direction. The attractiveness of public transport refers to the “charm” that shifts people from other travel modes to public transport [18]. However, different agents interconnect with each other. We elaborate on the attractiveness of public transport from both macroscopic and microscopic perspectives, as shown in Figure 1. At the macroscopic level, the attractiveness of public transport refers to the demand for public transport, which is quantified by the share of public transport. Higher demand thus demonstrates better attractiveness or vice versa. At the microscopic level, the attractiveness of public transport is quantified by the willingness-to-choose of public transport. More willingness to choose indicates higher attractiveness of public transport or vice versa.

[figure(s) omitted; refer to PDF]

2.1. Macroscopic Level: The Demand of Public Transport

2.1.1. Influence Factors

Using aggregate models, the factors that impact the demand for public transport have been examined. These factors have a significant impact on the attractiveness of public transport. The literature revealed that the factors affecting the demand for public transport can be divided into six categories: spatial layout and land use pattern, level of economic development, public policy, geographical environment, demographic characteristics, and service characteristics of public transport.

(1) Spatial Layout and Land Use Pattern. Transportation and land use supplement and counteract each other. Specifically, spatial layout and land use pattern determine the source of the demands, which determine urban layout and travel mode. Furthermore, the scale and layout of cities are affected by traffic operating conditions. Accessibility determines the layout of urban residential, commercial, and cultural land. Their relationship is shown in Figure 2. As the scale-up of urban areas, the average trip length in the cities increases, leading to a decrease in the mode share of walking and an increase in motorization. This aggravates the competition between public and private transport. Consequently, the priority of public transport has been advocated to promote a compact and efficient urban development model that forms an urban center with public transport hubs and transport demand with mixed land use functions. This mode has improved the level of service including the accessibility and convenience of public transport. It has also promoted the supply of public transport services and increased the share of public transport.

[figure(s) omitted; refer to PDF]

Urban spatial layout and land use pattern have an impact on the demand for public transport [11, 19–22]. Taylor et al. [21] analyzed passenger demand in 265 cities in the United States in 2000 by using multiple regression methods and found that not only the service level of public transport affect the demand for public transport, but the spatial layout also significantly affected the demand. Chakraborty and Mishra [19] studied the connection between urban spatial layout, land use pattern, socioeconomic development level, and passenger demand in Maryland’s public transport system in 2000. It was found that passenger demand for public transport was significantly affected by land use characteristics, accessibility of public transport, residents’ income level, and population density. They also found the differences among the factors that affect the demand of public transport in urban, suburban, and rural areas. Sung and Oh [11] analyzed the relationship between transient-orient development (TOD) planning and the demand for public transport using passenger demand of Seoul Metro Station in South Korea in 2006. It was found that high-density cities can obviously increase the demand for public transport. However, specific TOD planning should pay attention to the mixed degree of land use, the design of a pedestrian-friendly environment around the subway station and the public transport hubs.

(2) Public Policy and Level of Economic Development. The demand of public transport is affected not only by service level but by factors such as private car ownership, fuel price, parking cost, and the image of public transport, and the subsidy for public transport played an important role on the demand of public transport [12, 21, 23]. The experience of mitigating traffic congestion shows that the regulations of private cars have an effect on enhancing the attractiveness of public transport [24, 25]. In addition to reducing private car ownership and promoting regional access to motor vehicles, some measures should be taken to encourage travelers to use public transport in the central urban area instead of private cars. Such measures include setting up bus lanes in corridors and roads with the large demand for public transport, Building Park and ride facilities in bus terminals and main transfer stations of rail transit outside of the city center, providing parking spaces for private cars and preferential public transport charging policy. These solutions could reduce the use of private cars in the central urban area and increase the share of public transport. Laa and Leth [26]; Mitra and Hess(2021); Nikiforiadis et al. [27] found that e-scooter sharing mainly replaced walking and public transport.

The level of economic development, characterized by employment level, income level, consumption level, and private car ownership, affect residents’ attitude and understanding of public transport [19–21, 23, 28]. With the increase of income, people may prefer driving themselves than taking public transport given the convenience, comfortableness, and easier access of their own cars [28].

Urban public policy and the level of economic development have an impact on the demand for public transport, as shown in Figure 3.

[figure(s) omitted; refer to PDF]

(3) Geographical Environment and Demographic Characteristics. Urbanization stimulates the increase of population in cities, the reutilization of industrial structure, the improvement of economy, changes in residents’ lifestyle and the way of travel, affects residents’ attitude and understanding of public transport, and affects the demand for public transport [21, 28]. Chen et al. [20] used the panel data of 645 cities in China from 2002 to 2008. They found that the demand for public transport was influenced by urban geographical location, level of economic development, and service level of public transport, and the impact of each factor was significantly different. The impact of weather on the demand for public transport has also attracted the attention of many scholars. Weather conditions affect the service features of public transport, which in turn affect the demand for public transport, and the demand for public transport is directly affected by changing the utility of public transport [29, 30]. Passenger flow of public transport significantly decreased due to wind, rain, and temperature [14]. The relationship between weather and passenger demand provides important information for the daily operation of public transport.

In terms of demographics, the urban population directly determines the level of the demand for public transport [19, 28]. Population subgroups, such as the proportion of students and the proportion of poor people, reflect the people who are most likely to take public transport. These factors have a direct impact on the demand for public transport.

(4) Service Characteristics of Public Transport. The existing literature has in-depth research on public transport fares. The Simpson-Curtin rule was once used as the standard fare elasticity of the demand for public transport by the public transport sector in an era of lacking effective data. That is, the demand for public transport would be reduced by 0.33% for every 1% increase in fares of public transport [31]. However, differences emerged in fare elasticity in different environments. First, fare elasticity exhibits asymmetry, and passengers are more sensitive to the increase of fare than the decrease of fare. Secondly, fare elasticity increases with the increase of fare. Thirdly, fare elasticity varies with time range, travelers’ age, income, trip purpose, trip distance, and urban land use [32]. Despite variations, the impact of fares on the demand for public transport remains significant and inelastic. Gkritza et al. [33] studied the impact of fare structures of the multimodal public transport system and found that low fare was very important to attract more travelers to public transport. Moreover, the fare integration among different public transport modes can significantly increase the demand for public transport, and the differential fare system can effectively alleviate the competitive relationship among public transport modes. Besides, operation mileage, carriage allocation, hub station setting, and station coverage directly affect public transport operation, thus affecting the demand for public transport [34].

In addition, the impact of social values (such as public attitudes to public transport) and occasional events (such as strikes) on the demand for public transport has been verified [35].

On the basis of these research results, common indicators and influence directions of the significant factors affecting the demand for public transport are sorted out (as shown in Table 1).

Table 1

Summary of the tested and significant factors affecting the demand for public transport.

Category	Factors	Common index	Influence	Primary literature
Spatial layout and land use pattern	Urban spatial structure	Urban spatial structure type	Positive	[19, 21]
	Land use pattern	Density, mixed-use degree	Positive	[11, 19]
Public policy	Public transport priority	Whether to give subsidy, subsidy amount, or subsidy proportion, lane setting proportion	Positive	[12, 36]
	Restrictions on private cars	Parking fee, congestion pricing	Positive	[24]
	Fuel price	Fuel price	Positive	Kohn, 2000; [21, 23]
Level of economic development	Per capita GDP	Per capita GDP	Positive	[19, 21, 23]
	Residents’ income level	Per capita income or household income	Negative	[19, 21, 23]
	Private car ownership	Private car ownership	Negative	[28]
Demographic	Urban population	Total urban population or population density	Positive	[19, 28]
	Population type	Proportion of migrants, students or the poor	Positive	[19, 28]
Geographical environment	Urbanization rate	Urbanization rate	Positive	[21, 28]
	Urban area	Urban area	Negative	[20, 28]
	Weather condition	Weather type - sunny/Rainy, temperature, humidity, wind speed, air pressure	Negative	[14, 29]
Service characteristics of public transport	Fare of public transport	Per capita single trip fare	Negative	[31, 33]
	Service quantity	Operation mileage, number of vehicle, station coverage, service frequency, etc.	Positive	Currie and wallis, 2008
Social value	Public attitudes towards public transport	Whether public transport is considered as a “low-end commodity”	Negative	Ferguson, 1992
Occasional events	Strike	Whether or not a strike occurs	Negative	Ferguson, 1992

2.1.2. Interaction Mechanism

Referring to Taylor [37], we classified all of the factors that affect the demand of public transport into internal and external factors. External factors refer to factors that cannot be changed by public transport management departments, such as urban spatial layout and land use pattern, geographical environment characteristics, level of economic development, demographic characteristics, and public policies. Internal factors, on the other hand, are factors that can be changed by public transport management departments, such as service characteristics of public transport. Most scholars agree that external factors have a dominant impact on the demand for public transport and is more important than internal factors.

Based on the above literature review and the research of Taylor et al. [21], Figure 4 shows the interaction among the factors affecting the demand of public transport.

[figure(s) omitted; refer to PDF]

2.2. Microscopic Level: Willingness-to-Choose of Public Transport

With the development of disaggregate models, the research on the willingness-to-choose of public transport examines the factors that significantly affect the attractiveness of public transport from a microscopic perspective. A summary and analysis of existing research indicate that the factors affecting the willingness-to-choose of public transport can be divided into four groups of variables: individual characteristics of travelers, travel demand characteristics, travel characteristics of public transport, and service level of public transport.

2.2.1. Individual Characteristics of Traveler

Travels originate from the demand to participate in social and economic activities. In the study of travel mode choice, the individual characteristics of travelers are the most common and fundamental factors that can directly affect mode choice. Factors such as age, income, educational background, and private car ownership significantly affect travel mode choice [10, 13, 38, 39]. Many scholars have focused on the relationship between private cars and the use of public transport. Kitamura [40] acknowledged that private car ownership promoted the use of cars and thus decreased the use of public transport. Li et al. [41] reached the same conclusion. However, Chakrabarti [42] showed that mode choice could be reversed by a high-level public transport service even though car ownership could significantly reduce the use of public transport.

2.2.2. Travel Demand Characteristics

Travel demand is the set of characteristics that manifest in travel, such as trip purpose (work, school, shopping, and social), trip distance, and departure time (peak or flat peak period). Travelers may choose a proper mode according to their own characteristics and travel demand. Travel demand characteristics are one of the key factors that affect the willingness to travel by public transport [13, 43–45]. In addition, travel demand characteristics usually change the influence of other factors on the choice of traffic modes.

2.2.3. Travel Characteristics of Public Transport

Trip time and cost are the most commonly used measures of travel characteristics for public transport. The frequency of public transport could be increased significantly by reducing the travel time and trip cost of public transport [46, 47].

2.2.4. Service Level of Public Transport

Level of Service has the most direct impact on whether passengers are willing to travel by public transport [2, 20, 21] and 2019; [7]. When choosing public transport, passengers mainly have an expectation of public transport from the perspectives of economy, convenience, reliability, comfort, safety, and facility level. Passengers hope to complete their journey in a safe, convenient, and reliable environment. Travelers would have an actual travel experience after completing a trip by public transport. This experience would be compared with the travel expectation to form a satisfied or dissatisfied evaluation of the journey. Hence, the evaluation would directly affect their mode choice in the future (as shown in Figure 5). The impact of public transport service levels (such as reliability, comfort, safety, and convenience) on the willingness-to-choose of public transport has also been tested [15, 42, 43].

[figure(s) omitted; refer to PDF]

3. Methodology

Combining the research perspective of the attractiveness of public transport and the existing research results, we classified the influencing factors of the attractiveness into macroscopic and microscopic levels, as shown in Figure 6. The macroscopic-level factors mainly include spatial layout and land use pattern, geographical environment characteristics, level of economic development, demographic characteristics, public policy, and service characteristics of public transport. The microscopic-level factors can be summarized as individual characteristics of travelers, travel demand characteristics, travel characteristics of and service level of public transport. Considering the difficulty in performing quantitative analysis at the macroscopic level, this study chooses to quantify the key factors from the microscopic level.

[figure(s) omitted; refer to PDF]

Based on the results in Section 2.2, we analyzed the key factors at the microscopic level from four aspects: individual characteristics of travelers, travel demand characteristics, travel characteristics of public transport, and service level of public transport (subdivided into hardware facilities and software service). As these four variables are group variables that are not directly observable, they can be used as latent variables affecting the willingness-to-choose of public transport. Moreover, travel demand characteristics would significantly affect the individual characteristics of travelers and then indirectly affect the willingness-to-choose of public transport. The initial construction of the model and the specific description of the variables are shown in Figure 7 and Table 2, respectively.

[figure(s) omitted; refer to PDF]

Table 2

Description of initial model variables.

Variable type	Category		Variable	Symbol	Description
Exogenous variable	Passengers’ individual characteristics (PC)		Gender	Gender	Male = 1, Female = 0
			Age	Age	≤18 = 1, 19–40 = 2, 41–55 = 3, ≥56 = 4
			Occupation	Occupation	Student = 1, teacher or civil servants = 2, employee = 3, freelancer = 4, others = 5
			Monthly income	Income	≤¥2000 = 1, ¥2001-¥5000 = 2, ¥5001-¥8000 = 3, ＞¥8000 = 4
			Family numbers	Famine	Take “person” as the unit
			Car ownership	Car	1 or more = 1, None = 0
	Service level (SL)	Hardware facilities（HF）	Vehicle condition	HF1	Very unimportant = 1, ···, very important = 5
			Route planning	HF2	Same as above
			Station setting	HF3	Same as above
		Software service (SS	Running speed	SS1	Same as above
			Waiting time	SS2	Same as above
			Cleaning condition	SS3	Same as above
			Congestion degree	SS4	Same as above
			Driving stability	SS5	Same as above
			Transfer convenience	SS6	Same as above
			Traffic information acquisition	SS7	Same as above
Endogenous variable	Travel demand characteristics (TC)		Departure time	Departure time	Peak period = 1, flat peak period = 0
			Trip purpose	Trip purpose	Commuting (work, school) = 1, others = 0
			Trip distance	Trip distance	Near residence = 1, Origin and destination in the same administrative area = 2, else = 3
	Travel characteristics of public transport (PTC)		Trip time of public transport	Trip time	Below 30 min = 1,31–60 min = 2, 61–90 min = 3, above 90 min = 4
			Trip cost of public transport	Trip cost	Very low = 1, low = 2, normal = 3, high = 4, very high = 5
	Willingness-to-choose of public transport (WPT)		Bus use frequency	Use frequency	0∼3 times/week = 1, 4∼10 times/week = 2, above 10 times/week = 3

Figure 7 shows that these two group variables including the individual characteristics of travelers and service level of public transport (refined into hardware facilities and software services) are exogenous variables, whereas the two group variables of travel demand characteristics and travel characteristics of public transport are endogenous variables.

Discrete choice models, such as Binary Logit Model, Multinomial Logit Model, Mixed Logit Model, and Multinomial Probit Model, were commonly adopted in traditional research [15, 16, 36]. Each independent variable is required to have no measurement error, which can not be avoided in practice. Nevertheless, there are also intricate relationships among the factors affecting the willingness-to-choose of public transport, such as individual characteristics, travel demand characteristics, travel characteristics of and service level of public transport, yet Discrete Choice Models are difficult to identify the endogeneity of those factors. However, SEM shows great superiority in this aspect. It cannot only deal with measurement errors but also explores the structural relationship among influencing factors. Therefore, we use SEM to try to explore the influence mechanism of the willingness-to-choose of public transport.

SEM consists of two basic models: the structural and measured model. The general equation is as follows:$$\begin{array}{}\text{(1)}& E\frac{\eta }{\eta ,\xi }=\text{B}\eta +\Gamma \xi +\zeta ,\end{array}$$where $\eta ={\eta }_1,{\eta }_2,\dots ,{\eta }_m$ is a potentially endogenous variable and $\xi ={\xi }_1,{\xi }_2,\dots ,{\xi }_{\text{n}}$ is a potentially exogenous variables; $m$ and $n$ are the number of potentially endogenous and exogenous variables. $\text{B}$ and $\mathrm{\Gamma }$ denote the parameter coefficient matrix of $\eta$ and $\xi$, respectively. Three exogenous latent variables are used in this study, namely $n=3$, which are passengers’ individual characteristics, the level of hardware facilities, and the level of software services of public transport. The three endogenous latent variables in this model include travel demand characteristics, travel characteristics of public transport, and the willingness-to-choose of public transport, that is, $m=3$. $\zeta$ denotes the unexplained part, that is, the residual term. The calculation formula of the structural model applied in this model is shown in the following:$$\begin{array}{}\text{(2)}& \begin{array}{c}TC\hfill \\ PTC\hfill \\ WPT\hfill \end{array}=\begin{array}{ccc}0\hfill & 0\hfill & 0\hfill \\ 0\hfill & 0\hfill & 0\hfill \\ {\beta }_{31}\hfill & {\beta }_{32}\hfill & 0\hfill \end{array}\begin{array}{c}TC\hfill \\ PTC\hfill \\ WPT\hfill \end{array}+\begin{array}{ccc}{\Gamma }_{11}\hfill & 0\hfill & 0\hfill \\ 0\hfill & {\Gamma }_{22}\hfill & {\Gamma }_{23}\hfill \\ {\Gamma }_{31}\hfill & {\Gamma }_{32}\hfill & {\Gamma }_{33}\hfill \end{array}\begin{array}{c}PC\hfill \\ HF\hfill \\ SS\hfill \end{array}+\begin{array}{c}{\zeta }_1\hfill \\ {\zeta }_2\hfill \\ {\zeta }_3\hfill \end{array},\end{array}$$where PC, TC, PTC, HF, SS, and WPT represent individual characteristics, travel demand characteristics, travel characteristics of public transport, the level of hardware facilities, the level of software services, and the willingness-to-choose of public transport, respectively. Three group variables, HF, SS, and WPT, which are equivalent to three latent variables, have obvious measurement relationships with their subvariables. However, three group variables, PC, TC, and PTC, have no significant measurement relationships with their sub-variables. Thus, we treat each subvariable of these three categorical group variables as a latent variable.

The measured model is shown as follows:$$\begin{array}{}\text{(3)}& \begin{array}{c}X={\Lambda }_x\xi +\delta ,\\ Y={\Lambda }_y\eta +\epsilon ,\end{array}\end{array}$$where $X=x_1,x_2,\dots ,x_p$ and $Y=y_1,y_2,\dots ,y_q$ are the manifest variables of potentially exogenous variables and endogenous variables, respectively. $p$ and $q$ are the number of the manifest variables of potentially exogenous variables and endogenous variables, respectively. $\delta$ and $\epsilon$ are the measuring errors of $X$ and $Y$, respectively. ${\mathrm{\Lambda }}_x$ and ${\mathrm{\Lambda }}_y$ refer to the relationship between $X$ and $\xi$ and $Y$ and $\eta$, respectively.

4. Data Collection

4.1. Survey Object

Based on the findings of previous research, a questionnaire of “travel intention survey of public transport” was designed for urban residents in Hangzhou, Zhejiang Province. The questionnaire is intended to analyze the factors affecting the willingness-to-choose of public transport. A 5-point scale, with “1” corresponding to “tremendously unimportant” and “5” corresponding to “tremendously important,” was used to measure the importance of the factors in the category item of service level.

Public transport stations and areas within and outside of buses are the major testing locations. According to local conditions, bus lines are made up of general lines, express lines, night lines, and tourist lines, which are the four common public transport routes in Hangzhou. To reflect the real willingness-to-choose of public transport in Hangzhou, we adopt a stratified sampling of the bus lines. Stratified sampling is a powerful and flexible method that is extensively utilized in practice [7]. The sample bus routes could cover the urban area as much as possible and was divided into ordinary bus routes and special bus routes (including express lines, tour lines, night lines), which could cover well the operation lines of bus enterprises and reflect the real willingness-to-choose of public transport in Hangzhou. For each class of the bus lines, the sampling rate is set at 10%. In stratified sampling, suppose J was the number of the bus line strata and is equal to 4 in this study. N_j was the number of the bus line sampling stratum j, j = 1, 2, J. N was the number of sampling units in the bus lines, where N = N₁ + N₂ + N_J. To obtain the full benefit from stratification, the values of N_j must be known. When the bus line strata had been determined, a sample set was drawn from each bus line, the drawings being made independently in the different bus lines. The sample sizes within the bus line were denoted by n₁, n₂, n_J, respectively, where n₁ + n₂,  +   + n_J = n, and n was the total sample size of bus lines. In this paper, the sampling rate σ is also known and is equal to 10%. Thus, the sample sizes n₁, n₂, n_J can be obtained.

4.2. Survey Content

A structured questionnaire survey of public transport choice intention is used to identify the critical factors affecting residents’ willingness-to-choose of public transport. The main contents include the following:

(1) Individual characteristics

  This part collected sociodemographic information, such as age, gender, occupation, income, car ownership, and family members.

(2) Travel demand characteristics

  Respondents’ travel demand characteristics include departure time, trip purpose, trip distance, etc.

(3) Travel characteristics of public transport

  Respondents’ time consumption and trip cost of public transport are also surveyed.

(4) Service level of public transport

  The service level of public transport is described in terms of hardware and software services. The hardware facilities of public transport that are mainly investigated are route planning, station setting, and vehicle condition. On the basis of the evaluation indexes of public transport service quality [3,48–50], the software services of public transport that are mainly investigated are operation speed, waiting time, level of congestion, level of cleanliness, driving stability, transfer convenience, and traffic information acquisition.

(5) Willingness-to-choose of public transport

The main travel mode and the usage frequency of public transport for respondents were investigated.

4.3. Survey Data Collection

Before the formal survey, a small-scale pilot survey was conducted to ensure the efficiency of the survey results. The participants were randomly selected by age and gender and provided with the questionnaire after obtaining informed consent. Participants were given a general overview of the questionnaire and its completion procedure beforehand. Anonymity of participants was assured, and interviewers recorded responses to each question in the case of illiterates.

To guarantee the quality of the survey data, we employed the college students of Zhejiang Sci-Tech University to perform the survey. A training meeting was held to instruct the surveyors on how to perform the survey. The travel data was collected in a two-stage survey. From March to May 2019, a small-scale presurvey was carried out. According to the respondents’ feedback and the problems found in the survey process, the initial questionnaire was tested and modified, and then the formal investigation scheme was adjusted. However, due to the change of the Principal investigator of the questionnaire design and data processing, and the impact of COVID-19 at the beginning of 2020, the second phase of investigation started later. The second phase of the survey was conducted at the end of April 2020. Nearly 700 questionnaires were issued in Hangzhou. After removing incomplete questionnaires, 658 survey responses were available for data analysis. The effective questionnaire recovery rate was 94.0%.

Table 3 provides a summary of the sociodemographic characteristics of the survey sample. Males accounted for a larger proportion of the sample (52.5% compared 47.5% females), with almost half of respondents under 40 years old (59.6%). More than one-third of respondents were students and company employees (38.9% and 41.5, respectively). Respondents with a private car accounted for a larger proportion of the sample (57.5% compared with 42.5%).

Table 3

Sociodemographic characteristics of the survey sample.

Characteristics	Number of respondents	Percentage (%)
Gender
Male	345	52.5
Female	313	47.5
Age
Under 18 years	47	7.2
19–40 years	345	52.4
41–55years	103	15.7
Above 56 years	163	24.7
Occupation
Student	256	38.9
Company employee	273	41.5
Teacher or civil servant	48	7.3
Freelancer	45	6.8
Others	36	5.5
Monthly income
Under ¥2000	263	40.0
¥2001-¥5000	163	24.7
¥5001-¥8000	128	19.5
Above ¥8000	104	15.8
Car ownership
No	280	42.5
Yes	378	57.5

According to “Statistical Communiqué of Hangzhou on the 2019 and 2020 National Economic and Social Development”, the minimum wage for employees of Hangzhou in 2019 and 2020 are both ¥2010, the annual per capita disposable income in Hangzhou in 2019 and 2020 is ¥59261 and ¥61879, respectively. Therefore, we design the first gear of monthly income as Under ¥2000 and the second gear as ¥2001-¥5000. According to “Statistical Communiqué of Annual Average Wages of Employed Persons of Units in Hangzhou,” the annual average wages of employed persons of nonprivate units in Hangzhou in 2019 were ¥102289. The data for 2020 has not been released. Therefore, we designed the third grade of monthly income as ¥5001-¥8000, and the fourth gear as above ¥8000.

5. Empirical Analysis

5.1. Normality Test

The manifest variables are assumed to fit normal distribution in SEM. Therefore, we tested the normality of the manifest variable and determined whether the manifest variables conformed to the normal distribution. Mardia [51] proposed that the normality test was achieved by the skewness and kurtosis coefficient in the multi-dimensional situation. Mardia and Foster [52] further proposed that the absolute value of the coefficient of skewness and kurtosis were both less than 2. The variable was considered to pass the normality test. We use SPSS21.0 to test the normality of the manifest variables. The results are shown in Table 4.

Table 4

Normality test result of the normal questionnaire.

Variable	Skewness		Kurtosis
	Coefficient	S.E.	Coefficient	S.E.
Gender	0.349	0.126	−1.888	0.252
Age	1.279	0.126	3.125	0.252
Occupation	0.743	0.126	−0.578	0.252
Income	0.601	0.126	−1.192	0.252
Famine	−0.174	0.126	−0.290	0.252
Car	0.442	0.126	−0.754	0.252
HF1	−2.090	0.126	1.732	0.252
HF2	−0.700	0.126	0.087	0.252
HF3	−0.612	0.126	−0.170	0.252
SS1	−0.010	0.126	−0.153	0.252
SS2	−0.281	0.126	−0.498	0.252
SS3	−0.490	0.126	−0.392	0.252
SS4	−0.374	0.126	−0.565	0.252
SS5	−0.497	0.126	−0.388	0.252
SS6	−2.771	0.126	2.242	0.252
SS7	−0.433	0.126	−0.270	0.252
Departure time	−1.043	0.126	−2.009	0.252
Trip purpose	0.573	0.126	−0.676	0.252
Trip distance	0.516	0.126	−0.881	0.252
Trip time	0.703	0.126	0.073	0.252
Trip cost	0.255	0.126	−0.458	0.252
Use frequency	2.291	0.126	1.463	0.252

S.E. denotes standard error. The meaning of the bold values indicates that these manifest variables did not conformed to the normal distribution.

As can be seen from Table 4, the absolute value of the coefficient of skewness of HF1, SS6, and Use frequency are both more than 2, and the absolute value of the coefficient of kurtosis of Age, SS6, and departure time are both more than 2. The manifest variables did not all conform to the normal distribution. Hence, the use of ordinary multiple regression would produce a large error. The great superiority is shown by the WLS-SEM in this aspect. On the one hand, WLS can effectively solve the measuring errors of latent variables. On the other hand, it can effectively solve the nonnormal problems of variables encountered in the construction of traditional SEM to obtain better regression results. Thus, WLS–SEM is used in this study to estimate the model parameters.

5.2. Reliability and Validity Test

Reliability and validity tests should be carried out before the investigation data are brought into the structural equation model for the goodness-of-fit test. In the structural model, only the three group variables, namely the level of hardware facilities, the level of software services, and willingness-to-choose of public transport, are treated as latent variables. In other group variables, each subvariable is equivalent to a latent variable. Additionally, the willingness-to-choose of public transport is represented by only one manifest variable. Thus, the reliability and validity tests are conducted only for the level of hardware facilities and the level of software services.

Cronbach's alpha coefficient (α), which tests the internal consistency of the scale, is also used to assess reliability [53]. DeVellis [54] reckoned that the scale or questionnaire with a good reliability coefficient is generally above 0.80. The reliability of variables was tested by Cronbach's α reliability coefficient in this paper, and the results are shown in Table 5. As shown in Table 5, Cronbach's alpha reliability coefficients of variables are all greater than 0.8, demonstrating that a good internal consistency of this scale.

Table 5

Results of reliability analysis and validity analysis.

Latent variable	Manifest variable	Standard load	$p$	CR	AVE	Cronbach’s α
Software facilities	HF1	0.65	$\mathrm{\ast \ast \ast }$	0.798	0.574	0.806
	HF2	0.80	$\mathrm{\ast \ast \ast }$
	HF3	0.80	$\mathrm{\ast \ast \ast }$
Software service	SS1	0.61	$\mathrm{\ast \ast \ast }$	0.859	0.511	0.847
	SS2	0.66	$\mathrm{\ast \ast \ast }$
	SS3	0.69	$\mathrm{\ast \ast \ast }$
	SS4	0.76	$\mathrm{\ast \ast \ast }$
	SS5	0.72	$\mathrm{\ast \ast \ast }$
	SS6	0.74	$\mathrm{\ast \ast \ast }$
	SS7	0.68	$\mathrm{\ast \ast \ast }$

${}^{\ast }p$ ＜ 0.1; ${}^{\ast \ast }p$ ＜ 0.05; ${}^{\ast \ast \ast }p$ ＜ 0.01.

To achieve the validity of the questionnaire, the convergence validity test and content validity test of each variable would be essential. According to the actual situation, we modified the existing mature scale, so the content validity of this scale was considered to meet the standard. Then, we carried out the Confirmatory Factor Analysis (CFA), Composite Reliability (CR), and Average Variance Extracted (AVE), used them to test the convergence validity. Table 5 shows that the standard factor load of each variable is above 0.6, which is significant; the CR and AVE of each latent variable are more than 0.7 and 0.5, respectively. This indicates that the convergence validity of this scale is good. To sum up, the scale in this paper has good reliability and validity.

5.3. Results Analysis

When SEM is used to verify a theoretical model, good model fit is necessary for SEM analysis. The better the goodness of fit is, the closer the model matrix is to the sample matrix. Table 6 shows the value of the goodness of fit indices.

Table 6

Goodness of fit indices.

Indices	Reference value	Test results
X2/df	<5	2.158
RMSR	<0.05	0.046
RMSEAGFI	<0.08>0.9	0.0590.89
NFI	>0.9	0.92
IFI	>0.9	0.93
CFI	>0.9	0.93

Incremental fit indices and absolute fit indices were used to test the fitting degree of the model [55,56]. Incremental fit indices include the normed fit index (NFI), comparative fit index (CFI), and incremental fit index (IFI). Table 6 shows that NFI = 0.92, CFI = 0.93, and IFI = 0.93. They are all greater than 0.9 (reference value), and approach to 1, which indicates a perfect fit. Absolute fit indices consist of the χ²/df, the root mean square error index (RMSE), the root mean square error of approximation index (RMSEA), and the goodness of fit index (GFI). Table 6 shows that χ²/df = 2.158, which is less than 5. The absolute fit indices are as follows: RMSE = 0.046 (less than 0.05), RMRSEA = 0.059 (less than 0.08), GFI = 0.89 (close to the recommended value of 0.9). These results indicate that the fitting is good. Thus, subsequent analysis can be conducted. Tables 7–9 show the impact of exogenous variables on endogenous variables, as well as the interaction among endogenous variables in the SEM. The effect of Occupation, Famine, and Trip Purpose on other variables is not significant; hence, the effect of these three variables is not discussed in this study.

Table 7

Influence of travelers’ individual characteristics on the endogenous variables.

Dependent variables		Independent variables (travelers’ individual characteristics)
		Gender	Age	Income	Car
Travel demand characteristics	Departure time	0.15${}^{\mathrm{\ast \ast \ast }}$	0.14${}^{\mathrm{\ast \ast }}$	0.16${}^{\mathrm{\ast \ast }}$	0.28${}^{\mathrm{\ast \ast \ast }}$
	Trip distance	0.08${}^{\mathrm{\ast \ast }}$	0.04	0.12${}^{\mathrm{\ast \ast }}$	0.28${}^{\mathrm{\ast \ast \ast }}$
Willingness-to-choose of public transport	Willingness of PT	0.12${}^{\mathrm{\ast \ast }}$	−0.15${}^{\mathrm{\ast \ast \ast }}$	−0.19${}^{\mathrm{\ast \ast }}$	−0.23${}^{\mathrm{\ast \ast \ast }}$

${}^{\ast }p$ ＜ 0.1; ${}^{\mathrm{\ast \ast }}p$ ＜ 0.05; ${}^{\mathrm{\ast \ast \ast }}p$ ＜ 0.01.

Table 8

Effect of the service level of public transport on the endogenous variables.

Dependent variables		Independent variables (service level of public transport)
		Hardware facilities	Software services
Travel characteristics of public transport	Trip time of PT	−0.11${}^{\mathrm{\ast \ast }}$	−0.58${}^{\mathrm{\ast \ast \ast }}$
	Trip cost of PT	−0.073${}^{\ast }$	−0.014${}^{\mathrm{\ast \ast }}$
The willingness-to-choose of public transport	Willingness of PT	0.21${}^{\mathrm{\ast \ast \ast }}$	0.31${}^{\mathrm{\ast \ast \ast }}$

${}^{\ast }p$＜0.1; ${}^{\mathrm{\ast \ast }}p$＜0.05; ${}^{\mathrm{\ast \ast \ast }}p$＜0.01.

Table 9

Effect of travel demand characteristics and travel characteristics of public transport on the willingness-to-choose of public transport.

Independent variables		Dependent variables (the willingness-to-choose of public transport
Travel demand characteristics	Departure time	0.22${}^{\mathrm{\ast \ast }}$
	Trip distance	0.18${}^{\mathrm{\ast \ast \ast }}$
Travel characteristics of public transport	Trip time of PT	−0.34${}^{\mathrm{\ast \ast \ast }}$
	Trip cost of PT	−0.32${}^{\mathrm{\ast \ast \ast }}$

${}^{\ast }p$ ＜ 0.1; ${}^{\mathrm{\ast \ast }}p$ ＜ 0.05; ${}^{\mathrm{\ast \ast \ast }}p$ ＜ 0.01.

5.3.1. Individual Characteristics have a Significant Effect on Travel Demand Characteristics and the Willingness-to-Choose of Public Transport

Gender has an impact on travel demand characteristics and a significant effect on the willingness-to-choose of public transport. Moreover, gender has a relatively significant influence on departure time and trip distance in travel demand characteristics, with coefficients of 0.15 and 0.08, respectively. These results indicate that more male travelers commonly commute and travel longer distances. Gender has a significant positive effect of 5% on public transport choice intention. This influence is caused by the different social responsibilities between men and women. As women undertake certain life activities, they tend to make more noncommuting trips so that trip distances are short, and they are less dependent on public transport.

Age has a certain effect on public transport choice intention, but its effect on travel demand characteristics is nonsignificant. Table 7 shows that the influence coefficient of age on public transport choice intention is -0.15, which is significant at the 1% level. This result shows that age has a negative impact on public transport choice intention. As age increases, the willingness-to-choose of public transport as the main travel mode decreases. Moreover, the younger the age, the lower the probability of shifting to travel modes (such as cars or electric bicycles) and the higher the convenience offered by public transport, resulting in a higher willingness-to-choose of public transport as their main travel mode.

Income has a significant effect on travel demand characteristics and the willingness-to-choose of public transport. The influence coefficients of income on departure time and trip distance in travel demand characteristics are 0.16 and 0.12, respectively. Both are significant at the 5% level, indicating that the higher the traveler’s income is, the more likely that commuting is chosen and the longer the trip distance is. The effect of traveler’s income on public transport choice intention is negative, with a coefficient of -0.19, which is significant at the 5% level. This result indicates that the higher the traveler’s income is, the lower the willingness-to-choose of public transport is. This is because the larger the income of travelers, the higher the economic independence, the greater the probability of using other travel modes (such as cars), and the lower the dependence on public transport. Hence, the willingness-to-choose of public transport as the main travel mode is lower.

Private car ownership has a significant impact on travel demand characteristics and the willingness-to-choose of public transport. Owing to the high accessibility and convenience of owning a car, private car owners’ travels are usually via commuting, and their trip distances are relatively long. Private car ownership has a significantly negative impact on public transport travel, with an impact coefficient of -0.23. Therefore, to improve the attractiveness of public transport, implementing a reasonable driving restriction policy is necessary to reduce car trips. However, private car ownership has a positive impact on travel demand characteristics. Table 9 shows that travel demand characteristics also have a positive impact on the willingness-to-choose of public transport. This finding indicates that private car ownership can indirectly and positively affect the willingness-to-choose of public transport through travel demand characteristics. Private car ownership can promote the use of public transport, but the premise is to provide adequate parking and transfer space for private car owners.

5.3.2. Service Level of Public Transport has a Significant Effect on the Travel Characteristics of Public Transport and the Willingness-to-Choose of Public Transport

Service level is divided into two aspects in this study: the level of hardware facilities and the level of software services. As seen in Table 8, the level of hardware facilities and the level of software services have a significant impact on the travel characteristics of public transport and public transport choice intention.

The level of hardware facilities has a negative influence on trip time and cost in the travel characteristics of public transport. The influence coefficients are -0.11 and -0.073, respectively, which are significant at the 5% and 10% levels, respectively. This result indicates that the higher the level of hardware facilities, the lower the trip time and the lower the trip cost. The level of hardware facilities has a positive impact on public transport choice intention and is significant at the level of 1%. This result shows that the higher the level of hardware facilities, the greater the willingness-to-choose of public transport. As seen in Tables 2 and 5, the level of hardware facilities is characterized by vehicle condition, route planning, and station setting. Their loading coefficients are 0.65, 0.80, and 0.80, respectively, which demonstrate that improving the route planning and station setting is conducive to increasing the willingness-to-choose of public transport and thus enhancing the attractiveness of public transport.

The impact of the level of software services on trip time and trip cost of public transport is negative, with the influence coefficient of -0.58 and -0.014, respectively, which are significant at the level of 1% and 5%, respectively. This result indicates that the higher the level of software services, the shorter the trip time and the lower the trip cost of public transport are, and its influence on the time consumption of public transport is greater. The level of software services has a positive impact on public transport choice intention, which is significant at the 1% level. Compared with the level of hardware facilities, the level of software services has a greater impact on public transport choice intention. Tables 2 and 5 show that operation speed, waiting time, level of cleanliness, level of congestion, driving stability, transfer convenience, and traffic information acquisition represent the level of software services in this study. Their loading coefficients are 0.61, 0.66, 0.69, 0.76, 0.72, 0.74, and 0.68, respectively. These values show that alleviating the congestion level of public transport, improving the convenience of public transport transfer, driving stability, and the degree cleanliness are conducive to enhancing the willingness-to-choose of public transport, and thus improving the attractiveness of public transport.

5.3.3. Travel Demand Characteristics and Travel Characteristics of Public Transport Have a Significant Effect on the Willingness-to-Choose of Public Transport

Table 9 shows that departure time and trip distance in travel demand characteristics have a significantly positive influence on public transport choice intention. The influence coefficients are 0.26 and 0.18, respectively, which are significant at the 5% and 1% levels, respectively. This result indicates that people who travel during peak commuting periods and travel on long distances are dependent on public transport and choose public transport at a high frequency.

Trip time and cost have a relatively large negative impact on the willingness-to-choose of public transport. Their influence coefficients are -0.42 and -0.36, respectively, which are significant at the 1% level. This result indicates that the higher the trip time of public transport, the higher the trip cost of public transport, the lower the frequency of travelers’ choice of public transport travel, and the lower the willingness-to-choose of public transport. Hence, the attractiveness of public transport is low. Therefore, reducing the trip time and trip cost of public transport can greatly improve the willingness-to-choose of public transport, and thus enhance the attractiveness of public transport.

The impact of the four group variables is sorted out, including travelers’ individual characteristics, travel demand characteristics, travel characteristics of and service level of public transport on public transport choice intention, as shown in Table 10.

Table 10

Path coefficients among group variables.

Path relation		Total effect	Direct effect	Indirect effect	$p$
PC	Gender--- > wpt	0.17	0.12	0.05	${}^{\ast \ast }$
	Age--- > WPT	−0.11	−0.15	0.04	—
	Income--- > WPT	−0.13	−0.19	0.06	${}^{\ast \ast }$
	Car--- > WPT	−0.12	−0.23	0.11	${}^{\ast \ast \ast }$
	Sum	-0.19	−0.45	0.26	—
TC	Departure time--- > WPT	0.22	0.22	—	${}^{\ast \ast }$
	Trip distance--- > WPT	0.18	0.18	—	${}^{\ast \ast \ast }$
	Sum	0.40	0.40	—	—
PTC	Trip time of PT--- > WPT	−0.34	−0.34	—	${}^{\ast \ast \ast }$
	Trip cost of PT --- > WPT	−0.32	−0.32	—	${}^{\ast \ast \ast }$
	Sum	−0.66	−0.66		—
SL	HF--- > WPT	0.27	0.21	0.06	${}^{\ast \ast \ast }$
	SS--- > WPT	0.51	0.31	0.20	${}^{\ast \ast \ast }$
	Sum	0.78	0.48	0.26	—

${}^{\ast }p$ ＜ 0.1; ${}^{\ast \ast }p$ ＜ 0.05; ${}^{\ast \ast \ast }p$ ＜ 0.01.

As shown in Table 10, the total impact of the four group variables, namely, traveler’s individual characteristics, travel demand characteristics, travel characteristics of and service level of public transport, on the willingness-to-choose public transport is -0.19, 0.40, -0.66, and 0.78, respectively. This result shows that service level has the greatest impact on the willingness-to-choose of public transport, with a coefficient of 0.78, which is positive and significant at the 1% level. The service level of public transport is subdivided into the level of hardware facilities and the level of software services. Compared with the level of hardware facilities, software services have a greater impact on the willingness-to-choose of public transport (the coefficients are 0.27 and 0.51, respectively). Therefore, improving the level of software services can significantly increase the willingness-to-choose of public transport and consequently enhance the attractiveness of public transport. Travel characteristics of public transport have a large negative impact on the willingness-to-choose of public transport, with a coefficient of −0.66, which is significant at the 1% level. Nevertheless, the trip time and cost of public transport are adopted in this study to represent the travel characteristics of public transport. Therefore, reducing the trip time and cost of public transport can enhance the willingness-to-choose of public transport and improve the attractiveness of public transport. Travel demand characteristics have a large positive impact on public transport choice intention, with an influence coefficient of 0.4. This result indicates that travelers who travel during the peak commuting time and on long distances have a great dependence on public transport and a high probability to choose public transport as their travel mode. Passengers’ individual characteristics have a large direct negative impact on the willingness-to-choose of public transport (a coefficient of -0.45). However, it indirectly and positively affects the willingness-to-choose of public transport by influencing travel demand characteristics (a coefficient of 0.26).

6. Discussion

Although many factors analyzed above affect the attractiveness of public transport, policy makers can only use some of them to improve the attractiveness of public transport based on the principle of maximizing social welfare. Based on the results of this paper, it can be seen that the available factors include land use characteristics, characteristics of private car service, public policy, travel characteristics of public transport, service level of public transport, social values, and travel time. Based on the different emphases of the seven factors on enhancing the attractiveness of public transport, the policies to improve the attractiveness of public transport can be divided into three categories: reducing travel demand, inhibiting private car travel, and encouraging public transport travel. The detailed strategies are shown in Table 11.

Table 11

Policies to enhance the attractiveness of public transport.

Policies	Factors	Strategies
Reducing travel demand	Land use	Increase land use density, improve land use mix and balance; develop land planning models that are conducive to public transport, such as the TOD model
	Travel time.	Flexible work hours, such as staggering work hours and working from home
Inhibiting private car travel	Private car ownership	Purchase limit; increase purchase tax and license tax
	Characteristics of private car service	Peak congestion charges; license plate restrictions; higher fuel prices and taxes; parking restrictions; higher parking fees
Encouraging public transport travel	Public policy	Public transport priority policy; reasonable subsidies for public transport
	Service level of public transport	Improve the level of hardware facilities, such as perfect station facilities, increasing bus coverage, improving the density of bus network, regularly updating vehicles; improve the level of software services, such as alleviating congestion in bus, improving the convenience of transfer, training drivers and conductors, driving stability, keeping the carriage clean, developing intelligent bus, improving the information level; provide personalized and customized bus
	Travel characteristics of public transport	Reduce fares; increase speed and reduce travel time
	Social values	Carry out various forms of public transport travel cultural activities; create an urban public transport culture of “low-carbon transport and green travel”

We found that the service level of public transport and private car ownership has a positive and negative impact on the attractiveness of public transport, respectively. It is more effective to improve urban transportation when the policies of improving public transport service and inhibiting private car travel are simultaneously implemented, which was consistent with Casello [57]. Policy instruments interact with each other in different ways. Based on the effect of policy packages, the effects of policies are divided into four terms: complementarity, additivity, synergy, and perfect substitutability (May et al., 2006). Complementarity exists when the use of two instruments gives greater total benefits than the use of either alone. Additivity exists when the welfare gain from the use of two or more instruments in a policy package is equal to the sum of the welfare gains of using each alone. Synergy occurs when the simultaneous use of two or more instruments gives a greater benefit than the sum of the benefits of using either one of them alone. Perfect substitutability exists when the use of one instrument eliminates entirely the welfare gained from using another instrument. Policies to inhibit private car travel and policies to encourage public transport travel have complementary effects on enhancing the attractiveness of public transport. However, Georgina [58] found synergy between them. Therefore, in the process of formulating policies, management departments should try their best to avoid perfect substitutability and pursue synergy.

7. Conclusion

This study analyzes the key factors that affect the attractiveness of public transport from both macroscopic and microscopic perspectives. This study establishes a comprehensive mechanism for the factor. Considering the difficulty of carrying out quantitative analysis at the macroscopic level, we only design and collect questionnaire data from the microscopic level using a five-level scale and stratified sampling method. We use WLS-SEM to quantify the key factors influencing the attractiveness of public transport. Questions such as “what is the attractiveness of public transport,” “what factors affect the attractiveness of public transport,” and “how to affect the attractiveness of public transport” were answered in this study. The main findings are as follows.

At the macroscopic level, the main factors include spatial layout and land use pattern, geographical environment characteristics, level of economic development, demographic characteristics, public policy, and service characteristics of public transport. At the microscopic level, the main factors are travelers’ individual characteristics, travel demand characteristics, travel characteristics of and service level of public transport, with influence coefficients of -0.19, 0.40, -0.66, and 0.78, respectively. The total impact of service level on the willingness-to-choose of public transport is the largest. Its coefficient is 0.78, which is positive and significant at the 1% level. Furthermore, the travel characteristics of public transport have a significant negative impact on the willingness-to-choose of public transport, with a coefficient of −0.66, which is significant at the 1% level.

In this study, the service level is divided into the level of hardware and software services. Compared with the level of hardware facilities, the level of software services has a greater impact on choice intention (the coefficients of 0.27 and 0.51, respectively). Thus, improving the level of software services can significantly enhance the willingness-to-choose of public transport. As a result, the attractiveness of public transport can be improved, thus encouraging residents to travel in a green way.

In this study, operation speed, waiting time, level of cleanliness, level of congestion, driving stability, transfer convenience, and traffic information acquisition represent the level of software services. Their loading coefficients are 0.61, 0.66, 0.69, 0.76, 0.72, 0.74, and 0.68, respectively, indicating that the most important factors for residents in choosing public transport are comfort, safety, and convenience. Improving the convenience of public transport transfer, driving stability, cleanliness, and alleviating the level of congestion is conducive to increasing the willingness-to-choose of public transport, thus enhancing the attractiveness of public transport.

Our research identifies the key factors that affect the attractiveness of public transport. We also propose the corresponding management suggestions, which can be divided into three categories: reducing travel demand, inhibiting private car travel, and encouraging public transport travel. These suggestions provide a theoretical basis for encouraging residents to travel by public transport and building the urban development mode dominated by public transport.

However, this research has limitations. First, this study does not carry out quantitative analysis at the macroscopic level. Second, the factors are very complex at the microscopic level. Factors such as residents’ satisfaction with the supply of parking spaces in public areas and road traffic conditions would also affect the willingness-to-choose of public transport. These factors are greatly affected by the level of urban economic development and population demand, with significant regional heterogeneity. The follow-up studies will explore these in-depth. Third, different measures should be considered in the future including different levels of costs and the cost-effectiveness of the proposed measures.

Authors’ Contributions

CZ proposed and designed the research ideas and framework, wrote, and revised the manuscript; MW analyzed the data and revised the manuscript; JD designed the survey, collected, and analyzed the data; WL provided some comments and helped to edit the manuscript. LY & AN conducted discussions and proofread the manuscript; XY proposed the research framework, conducted discussions, and proofread the manuscript. All authors reviewed and approved the final version of the manuscript.

Acknowledgments

The authors thank the managers and respondents at the sector of public transport in Hangzhou for providing data and information that were essential for this work. This paper was largely supported by Humanities and Social Sciences Project, Ministry of Education in China (Grant no. 19YJCZH238), National Natural Science Foundation of China (Grant no. 71901196), Philosophy and Social Science Project of Zhejiang Province, China (Grant no. 19NDQN361 YB), Major Humanities and Social Sciences Project of Universities in Zhejiang Province, China (Grant no. 2018QN028), and Project of Shanghai Science and technology innovation action plan (Grant no. 20511101800).

References

[1] C. Morton, B. Caulfield, J. Anable, "Customer perceptions of quality of service in public transport: evidence for bus transit in Scotland," Case Stud. Transport Policy, vol. 4, pp. 199-207, DOI: 10.1016/j.cstp.2016.03.002, 2016.

[2] C. Q. Zhang, G. N. Xiao, L Yong, Y Feng, "The relationship between organizational forms and comprehensive effectiveness for public transport services in China?," Transportation Research Part A: Policy and Practice, vol. 118, pp. 783-802, DOI: 10.1016/j.tra.2018.10.025, 2018.

[3] C. Q. Zhang, Y. Liu, L Weite, X Guangnian, "Evaluating passenger satisfaction index based on PLS-SEM model: evidence from Chinese public transport service," Transportation Research Part A: Policy and Practice, vol. 120, pp. 149-164, DOI: 10.1016/j.tra.2018.12.013, 2019.

[4] P. Zhao, "Sustainable urban expansion and transportation in a growing megacity: consequences of urban sprawl for mobility on the urban fringe of Beijing," Habitat International, vol. 34 no. 2, pp. 236-243, DOI: 10.1016/j.habitatint.2009.09.008, 2010.

[5] J. De Oña, R. De Oña, G. López, "Transit service quality analysis using cluster analysis and decision trees: a step forward to personalized marketing in public transportation," Transportation, vol. 43, pp. 725-747, DOI: 10.1007/s11116-015-9615-0, 2016.

[6] C. H. Wen, S. C. Lai, "Latent class models of international air carrier choice," Transportation Research Part E: Logistics and Transportation Review, vol. 46 no. 2, pp. 211-221, DOI: 10.1016/j.tre.2009.08.004, 2010.

[7] A. N. Ni, C. Q. Zhang, H Yuting, L Weite, L Hongwei, "Influence mechanism of the corporate image on passenger satisfaction with public transport in China," Transport Policy, vol. 94, pp. 54-65, DOI: 10.1016/j.tranpol.2020.04.014, 2020.

[8] Y. Yang, M. Li, Y Jiaying, H Fang, "Expressway bottleneck pattern identification using traffic big data-The case of ring roads in Beijing, China," Journal of Intelligent Transportation Systems: Technology, Planning, and Operations, vol. 24 no. 1, pp. 54-67, DOI: 10.1080/15472450.2019.1579091, 2020.

[9] H. Wang, M. E. Warner, T Yuanhong, Z Dajian, "Public owner with business delivery mode in China: case study of the Shanghai public bus system," Annals of Public and Cooperative Economics, vol. 85 no. 1, pp. 147-164, DOI: 10.1111/apce.12030, 2014.

[10] M. V. Johansson, T. Heldt, J Per, "The effects of attitudes and personality traits on mode choice," Transportation Research Part A: Policy and Practice, vol. 40 no. 6, pp. 507-525, DOI: 10.1016/j.tra.2005.09.001, 2006.

[11] H. Sung, J. T. Oh, "Transit-oriented development in a high-density city: identifying its association with transit ridership in Seoul, Korea," Cities, vol. 28 no. 1, pp. 70-82, DOI: 10.1016/j.cities.2010.09.004, 2011.

[12] H. T. Van, K. Choocharukul, F Satoshi, "The effect of attitudes toward cars and public transportation on behavioral intention in commuting mode choice—a comparison across six Asian countries," Transportation Research Part A: Policy and Practice, vol. 69 no. 69, pp. 36-44, DOI: 10.1016/j.tra.2014.08.008, 2014.

[13] C. Liu, E. Bardaka, P Ravikumar, T Li-Wei, "Analysis of travel characteristics and access mode choice of elderly urban rail riders in Denver, Colorado," Travel Behaviour and Society, vol. 19, pp. 194-206, DOI: 10.1016/j.tbs.2019.11.004, 2020.

[14] J. Wu, H. Liao, "Weather, travel mode choice, and impacts on subway ridership in Beijing," Transportation Research Part A: Policy and Practice, vol. 135, pp. 264-279, DOI: 10.1016/j.tra.2020.03.020, 2020.

[15] M. A. Abdel-Aty, "Using ordered probit modeling to study the effect of ATIS on transit ridership," Transportation Research Part C: Emerging Technologies, vol. 9 no. 4, pp. 265-277, DOI: 10.1016/s0968-090x(00)00037-1, 2001.

[16] M. Frondel, C. Vance, "Rarely enjoyed? A count data analysis of ridership in Germany's public transport," Transport Policy, vol. 18 no. 2, pp. 425-433, DOI: 10.1016/j.tranpol.2010.09.009, 2011.

[17] J. P. Li, L. Q. Xu, Y Di, M Yanwu, "Impacts of symbolic value and passenger satisfaction on bus use," Transportation Research Part D: Transport and Environment, vol. 72, pp. 98-113, DOI: 10.1016/j.trd.2019.04.012, 2019.

[18] D. Yao, L. Q. Xu, J. P. Li, "Does technical efficiency play a mediating role between bus facility scale and ridership attraction? Evidence from bus practices in China," Transportation Research Part A: Policy and Practice, vol. 132, pp. 77-96, DOI: 10.1016/j.tra.2019.11.002, 2020.

[19] A. Chakraborty, M. Sabyasachee, "Land use and transit ridership connections: implications for state-level planning agencies," Land Use Policy, vol. 30 no. 1, pp. 458-469, DOI: 10.1016/j.landusepol.2012.04.017, 2013.

[20] X. Chen, X. Wang, Z Hua, L Jia, "The diversity and evolution process of bus system performance in Chinese cities: an empirical study," Sustainability, vol. 6 no. 11, pp. 7751-7767, DOI: 10.3390/su6117751, 2014.

[21] B. D. Taylor, D. Miller, I Hiroyuki, F Camille, "Nature and/or nurture? Analyzing the determinants of transit ridership across US urbanized areas," Transportation Research Part A: Policy and Practice, vol. 43 no. 1, pp. 60-77, DOI: 10.1016/j.tra.2008.06.007, 2009.

[22] C. Vincent, E. Naveen, "Examining the influence of urban form and land use on bus ridership in Montreal," Procedia Social & Behavioral Sciences, vol. 104, pp. 875-884, 2013.

[23] C. Chen, D. Varley, C Jason, "What effects transit ridership? A dynamic analysis involving multiple factors, lags and asymmetric behavior," Urban Studies, vol. 48 no. 9, pp. 1893-1908, DOI: 10.1177/0042098010379280, 2011.

[24] L. De Grange, R. Troncoso, "Impacts of vehicle restrictions on urban transport flows: the case of Santiago, Chile," Transprot Policy, vol. 18, pp. 862-869, DOI: 10.1016/j.tranpol.2011.06.001, 2011.

[25] H. Yang, G. Zhai, L Xiaohan, Y Linchuan, L Yugang, Y Quan, "Determinants of city-level private car ownership: effect of vehicle regulation policies and the relative price," Transport Policy, vol. 115, pp. 40-48, DOI: 10.1016/j.tranpol.2021.10.025, 2022.

[26] B. Laa, U. Leth, "Survey of E-scooter users in Vienna: who they are and how they ride," Journal of Transport Geography, vol. 89,DOI: 10.1016/j.jtrangeo.2020.102874, 2020.

[27] A. Nikiforiadis, E. Paschalidis, S Nikiforos, R Alexandra, K Athanasia, B Socrates, "Analysis of attitudes and engagement of shared e-scooter users," Transportation Research Part D: Transport and Environment, vol. 94,DOI: 10.1016/j.trd.2021.102790, 2021.

[28] K. S. Lee, K. E Jin, Y. Y So, M Jaehong, Y. Y Keun, "An empirical study on the relationship between urban railway ridership and socio-economic characteristics," Procedia Computer Science, vol. 52, pp. 106-112, DOI: 10.1016/j.procs.2015.05.037, 2015.

[29] P. Arana, S. Cabezudo, M. Peñalba, "Influence of weather conditions on transit ridership: a statistical study using data from smartcards," Transportation Research Part A: Policy and Practice, vol. 59 no. 1,DOI: 10.1016/j.tra.2013.10.019, 2014.

[30] W. S. Victor, D. Edward, "The impact of weather on bus ridership in pierce county, Washington," Journal of Public Transportation, vol. 15 no. 1, pp. 95-110, DOI: 10.5038/2375-0901.15.1.6, 2012.

[31] J. F. Curtin, "Effect of fares on transit riding," Highway Research Record, vol. 213, 1968.

[32] R. Cervero, "Transit pricing research," Transportation, vol. 17 no. 2, pp. 117-139, DOI: 10.1007/bf02125332, 1990.

[33] K. Gkritza, M. G. Karlaftis, F. L. Mannering, "Estimating multimodal transit ridership with a varying fare structure," Transportation Research Part A: Policy and Practice, vol. 45 no. 2, pp. 148-160, DOI: 10.1016/j.tra.2010.12.005, 2011.

[34] M. A. Kemp, "Some evidence of transit demand elasticities," Transportation, vol. 2 no. 1, pp. 25-52, DOI: 10.1007/bf00172614, 1973.

[35] E. Ferguson, "Transit ridership, incident effects and public policy," Transportation Research Part A: Policy and Practice, vol. 26 no. 5, pp. 393-407, DOI: 10.1016/0965-8564(92)90003-p, 1992.

[36] C. Zhang, Z. Juan, L Weite, X Guangnian, "Do the organizational forms affect passenger satisfaction? Evidence from Chinese public transport service," Transportation Research Part A: Policy and Practice, vol. 94, pp. 129-148, DOI: 10.1016/j.tra.2016.09.007, 2016.

[37] B. D. Taylor, C. N. Y. Fink, "The factors influencing transit ridership: a review and analysis of the ridership literature," University of California transportation center working papers, 2003.

[38] J. Ji, X. Gao, "Analysis of people's satisfaction with public transportation in Beijing," Habitat International, vol. 34 no. 4, pp. 464-470, DOI: 10.1016/j.habitatint.2009.12.003, 2010.

[39] J. Scheiner, C. Holz-Rau, "Travel mode choice: affected by objective or subjective determinants?," Transportation, vol. 34, pp. 487-511, DOI: 10.1007/s11116-007-9112-1, 2007.

[40] R. Kitamura, "A causal analysis of car ownership and transit use," Transportation, vol. 16 no. 2, pp. 155-173, DOI: 10.1007/bf00163113, 1989.

[41] M. Li, D. Lau, S Daniel, "Car ownership and urban transport demand in Singapore," International Journal of Transport Economics, vol. 38 no. 1, pp. 47-70, 2011.

[42] S. Chakrabarti, "How can public transit get people out of their cars? An analysis of transit mode choice for commute trips in Los Angeles," Transport Policy, vol. 54, pp. 80-89, DOI: 10.1016/j.tranpol.2016.11.005, 2017.

[43] D. Birago, S. O. Mensah, S Somesh, "Level of service delivery of public transport and mode choice in Accra, Ghana," Transportation research F: Traffic Psychology and Behaviour, vol. 46, pp. 284-300, DOI: 10.1016/j.trf.2016.09.033, 2017.

[44] E. Heinen, "Identity and travel behaviour: a cross-sectional study on commute mode choice and intention to change," Transportation Research F: Traffic Psychology and Behaviour, vol. 43, pp. 238-253, DOI: 10.1016/j.trf.2016.10.016, 2016.

[45] S. Kaffashi, M. N. Shamsudin, S. C Maynard, F. S Shaufique, B Abdullatif, R Alias, U. A Shehu, A. R Khalid, "Are Malaysians eager to use their cars less? Forecasting mode choice behaviors under new policies," Land Use Policy, vol. 56, pp. 274-290, DOI: 10.1016/j.landusepol.2016.05.005, 2016.

[46] A. O. Idris, K. M. Nurul Habib, S Amer, "An investigation on the performances of mode shift models in transit ridership forecasting," Transportation Research Part A: Policy and Practice, vol. 78, pp. 551-565, DOI: 10.1016/j.tra.2015.06.012, 2015.

[47] N. Sharaby, Y. Shiftan, "The impact of fare integration on travel behavior and transit ridership," Transport Policy, vol. 21,DOI: 10.1016/j.tranpol.2012.01.015, 2012.

[48] J. Allen, J. C. Muñoz, J. d. D. Ortúzar, "Modelling service-specific and global transit satisfaction under travel and user heterogeneity," Transportation Research Part A: Policy and Practice, vol. 113, pp. 509-528, DOI: 10.1016/j.tra.2018.05.009, 2018.

[49] C. De Gruyter, G. Currie, T. T Long, N Farhana, "A meta-analysis and synthesis of public transport customer amenity valuation research," Transport Reviews, vol. 39, pp. 261-283, DOI: 10.1080/01441647.2018.1461708, 2019.

[50] J. De Oña, R. De Oña, "Quality of service in public transport based on customer satisfaction surveys: a review and assessment of methodological approaches," Transportation Science, vol. 49 no. 3, pp. 605-622, DOI: 10.1287/trsc.2014.0544, 2015.

[51] K. V. Mardia, "Measures of multivariate skewness and kurtosis with applications," Biometrika, vol. 57, pp. 519-530, DOI: 10.1093/biomet/57.3.519, 1970.

[52] K. V. Mardia, K. Foster, "Omnibus tests of multinormality based on skewness and kurtosis," Communications in Statistics - Theory and Methods, vol. 12, pp. 207-212, DOI: 10.1080/03610928308828452, 1983.

[53] J. L. Price, "Handbook of organizational measurement," International Journal of Manpower, vol. 18, pp. 305-558, DOI: 10.1108/01437729710182260, 1997.

[54] R. F. DeVellis, Scale Development: Theory and Applications, 2016.

[55] R. P. Mcdonald, M. H. Ho, "Principles and practice in reporting structural equation analyses," Psychological Methods, vol. 7 no. 1, pp. 64-82, DOI: 10.1037/1082-989x.7.1.64, 2002.

[56] J. Miles, M. Shevlin, "A time and a place for incremental fit indices," Personality and Individual Differences, vol. 42 no. 5, pp. 869-874, DOI: 10.1016/j.paid.2006.09.022, 2007.

[57] J. M. Casello, "Transit competitiveness in polycentric metropolitan regions," Transportation Research Part A: Policy and Practice, vol. 41 no. 1, pp. 19-40, DOI: 10.1016/j.tra.2006.05.002, 2007.

[58] S. Georgina, B. Hannah, T. Alexander, "Part II: policy instruments for sustainable road transport," Research in Transportation Economics, vol. 28 no. 1, pp. 46-91, 2010.