Next Article in Journal
Challenges in Achieving 1.5-Degree Lifestyle Mitigation Options—Insights from a Citizen-Participatory Household Experiment in Japan
Previous Article in Journal
Cultural Capital of Sea Salt Farming in Ban Laem District of Phetchaburi Province as per the Globally Important Agricultural Heritage Systems (GIAHS)
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Investigating the Impacts of Urban–Rural Bus Service Quality on Rural Residents’ Travel Choices Using an SEM–MNL Integration Model

1
School of Civil and Transportation Engineering, Hebei University of Technology, Tianjin 300401, China
2
School of Architecture and Art Design, Hebei University of Technology, Tianjin 300401, China
*
Author to whom correspondence should be addressed.
Sustainability 2023, 15(15), 11950; https://doi.org/10.3390/su151511950
Submission received: 19 June 2023 / Revised: 28 July 2023 / Accepted: 1 August 2023 / Published: 3 August 2023

Abstract

:
The high-quality development of urban–rural buses is an important way to narrow the gap between urban and rural areas. However, studies on the quality of bus services and its impact on bus travel-mode choice have mostly focused on urban buses, with less attention paid to urban–rural buses. This paper aims to explore how urban–rural bus service quality affects rural residents’ travel-mode-choice behavior based on survey data from rural residents in Henan Province, China. A structural equation model (SEM) is applied to examine the critical factors affecting rural residents’ satisfaction with urban–rural bus service quality and the heterogeneity of satisfaction among rural residents with different attributes. Furthermore, an integrated approach of SEM and the multinomial logit (MNL) model is proposed to identify the key factors that affect rural residents’ bus travel-mode choice. The results indicate that reliability and convenience are the key factors affecting satisfaction with urban–rural bus service quality. There are significant differences in rural residents’ satisfaction by gender, age, income, education level, occupation, and travel-mode attributes. Urban–rural bus service quality has a significant positive impact on rural residents’ bus travel-choice behavior, and its degree of influence is greater than all explicit variables except private car ownership. The findings of this study can help operators and governments formulate policies to improve the service quality of urban–rural buses and ultimately promote the use of buses.

1. Introduction

In rural areas of many countries, buses are the most affordable and widely provided mode of public transportation [1]. Urban–rural buses provide a link between urban and rural areas and are an important part of improving rural transportation services. China has entered a stage of rapid urbanization and new rural construction [2], which has led to an increasing demand for rural residents to travel between urban and rural areas. Providing rural bus services that meet the needs of rural residents is important [3]. Urban–rural buses should offer residents satisfactory travel services and accelerate the flow of people between urban and rural areas. However, there are many problems with the quality of urban–rural bus services in China, such as low punctuality, long waiting times, overloading, and high fares. This makes passengers feel dissatisfied, which in turn results in a decline in urban–rural bus passenger flow and the worsening of the urban–rural transportation system.
Currently, urban–rural travel is facing traffic congestion caused by the rapid growth of private cars, as well as traffic chaos and safety hazards caused by the mismanagement of electric bicycles. One of the reasons is that the service level of urban–rural buses cannot meet the needs of rural residents. For example, as a major agricultural province in China, Henan Province had an average of 39.25 private cars and 135.91 electric vehicles per hundred rural households in 2021, an increase of 47.7% and 9.5% compared to the same period last year [4]. In addition, the low level of urban–rural bus services will further lead to an increase in the number of private cars and electric vehicles in rural areas. Despite the rapid advancement of urbanization, numerous people still live in rural areas of China [5]. According to the China Statistical Yearbook 2022 [4], the rural population was 498.35 million in 2021, accounting for 35.28% of the country’s total population. Bus travel is a reasonable transportation option for rural residents to access critical services [6]. Offering high-quality bus services is important to encourage a shift from private modes to public transport [7]. Therefore, in order to solve the problem of urban–rural travel, it is necessary to improve the quality of urban–rural bus services. Public transport services are provided directly to customers, so their service quality is a result perceived by users [8]. In addition, user-satisfaction is a fundamental measure of quality improvement [9,10].
The main users of urban–rural buses are rural residents. Therefore, in order to solve the problem of urban–rural travel, it is necessary to understand rural residents’ satisfaction with the quality of urban–rural bus services and identify the shortcomings of urban–rural bus services to ensure that they can provide better services to rural residents. In addition, the quality of bus services has an impact on residents’ bus travel-choice behavior [11]. Understanding rural residents’ urban–rural travel behavior is an important basis for the scientific formulation of urban–rural transportation policies and the effective implementation of transportation management. The travel choices of residents in rural areas may be fundamentally different from those in large urban areas [12]. Thus, it is necessary to analyze the influencing factors and mechanisms of rural residents’ travel-mode-choice behavior, especially their choice of urban–rural buses. This can provide a theoretical basis for optimizing the urban–rural transportation structure and improving the sharing ratio of urban–rural buses. The purpose of this paper is to provide a model that can simultaneously consider rural residents’ satisfaction with the quality of urban–rural bus services and their urban–rural travel-mode-choice behavior.
In this paper, the subjective and objective factors that affect rural residents’ urban–rural travel-mode-choice behavior are considered. First, 22 observed indicators reflecting the comfort, convenience, reliability, safety, economy, and facility completeness of urban–rural bus services are identified. The satisfaction data of rural residents are collected through a questionnaire survey. Then, SEM is established to determine the key factors affecting rural residents’ satisfaction with urban–rural bus service quality and to explore the heterogeneity of satisfaction among rural residents with different attributes. Finally, the SEM–MNL integration model is developed to explore how different factors, especially satisfaction latent variables, affect the urban–rural travel-mode-choice behavior of rural residents. The structure of the paper is as follows. Section 2 provides the framework and calculation method of the SEM–MNL model. Section 3 describes the collection of questionnaire data and the statistical information of the data. Section 4 analyzes the estimation results of the model and raises some related policy implications. Section 5 summarizes the key findings and contributions of this study.

2. Literature Review

2.1. The Quality of Public Transportation Services

The quality of public transportation services reflects the level of service. Evaluating and improving service quality is an important element in reducing traffic congestion and enhancing transportation system infrastructure [13]. Public transportation service-quality evaluation indicators can be divided into two categories: objective and subjective indicators [3]. Objective indicators can provide clearer and less-biased information and are often evaluated by developing relevant criteria [14,15]. Objective evaluation indicators selected according to the attributes of bus services include four main categories: people, vehicles, roads, and facilities [16,17]. However, evaluating bus services from an objective perspective alone does not provide insight into the potential relationship between users and service quality. Subjective indicators that reflect passenger satisfaction are the basis of service quality evaluation [10]. To overcome this limitation, Stuart et al. [18] used New York as a study area to construct a structural equation model for service quality and passenger satisfaction and investigated the influencing factors in depth. In recent years, many scholars have conducted studies on how service attributes affect passenger satisfaction [19,20,21]. Passenger satisfaction is the gap between a passenger’s perception and expectation, and the subjective psychological state that passengers finally attain by comparing the service they experience with their psychological expectations before the experience. Many attributes have been used to assess the quality of bus services, such as punctuality, comfort, convenience, safety, fare, driver courtesy, and information availability [9,22,23]. Most studies use scales to measure customer satisfaction with bus service quality [24]. Among them, 5-point Likert scales are the most widely adopted, which range from 1 (the lowest) to 5 (the highest) and are defined as “very dissatisfied” to “very satisfied”, respectively [3]. These questionnaires ask users to rate their satisfaction with each key service attribute, thus identifying the priority of each service attribute. For example, Garrido et al. [25] used artificial neural networks to analyze the quality of service perceived by passengers in public transport systems and found that frequency was the attribute that had the greatest impact on service quality. Sinha et al. [26] used the TOPSIS method to analyze scale data from a questionnaire and found that commuters were dissatisfied with the punctuality, frequency of service, and bus information at stations. Eboli and Mazzulla [27] found that when service quality is poor, users are more interested in the basic attributes of the service, as quality improves, their demand for “higher-order” quality increases.
In addition, there is heterogeneity in the perception of the same service by passengers with different attributes, so some scholars have studied this heterogeneity [28,29]. The perceived differences among users are mainly influenced by demographic characteristics (e.g., income, gender, age, car ownership), geographical conditions, economics, travel habits, and other factors [30,31,32]. For example, Das and Pandit [33] found differences in the level of service scale values between developed and developing countries and between expert opinions and user perceptions. Echaniz et al. [34] found that passenger satisfaction with transit services varied over time and space. Zheng et al. [35] found gender differences in both user satisfaction and service-quality improvement priorities. In addition, some scholars evaluated the service quality according to the gap between users’ perceptions of the current and expected service quality [36,37].
In terms of research methodology, early scholars directly asked customers to rank the importance of service attributes through satisfaction surveys [38,39]. In recent years, many methods have been proposed to assess service quality, such as regression analysis [40], path analysis [37], principal component analysis [28], structural equation model [11,19], decision tree model [41], artificial neural networks (ANNs) [25] and the fuzzy evaluation method [42]. Among them, SEM is the most widely used method [23]. SEM is able to handle the interrelationships between multiple independent variables and multiple dependent variables simultaneously [21] and identify the measurement errors in both independent and dependent variables [43], which makes up for the shortcomings of traditional ordinal linear regression methods.
Most of the aforementioned studies have examined passenger satisfaction with the quality of bus services. However, the research subjects are mostly urban residents and urban buses. There are few studies on rural residents’ satisfaction with the service quality of urban–rural buses. The study of Ponrahono et al. [1] took into account the satisfaction of rural residents with rural buses, but it only investigated their satisfaction with the overall bus service and did not study the specific rural bus service attributes that determined their satisfaction.

2.2. Travel-Mode-Choice Behavior

Earlier studies analyzed residents’ travel-mode-choice behavior by constructing discrete choice models, such as the multinomial logit model [44], the nested logit model [45], and the cross-nested logit model [46]. However, traditional discrete choice models ignore certain latent variables of psychological factors that are not directly observed, which makes the model estimation results deviate from real travel behavior. To overcome this shortcoming, Ben-Akiva et al. [47] developed a discrete choice model framework with the inclusion of latent variables and constructed a hybrid choice model (HCM) to explore travel-choice behavior. By constructing a hybrid choice model, many scholars found that latent variables have a significant effect on choice behavior and that the explanatory power of the model increases accordingly [48,49]. Han et al. [50] found that the level of service in terms of the safety and convenience of public transportation had a significant effect on the choice of public and private transport by constructing an integrated approach of SEM and the nested logit (NL) model. Si et al. [51] developed an SEM–logit model to explore the effect of passengers’ attitude perception on the mode-choice behavior between taxis and online cars. The results of the study found that economy, comfort, and convenience had a significant effect on passengers’ choice behavior.
In studies of rural residents’ travel-mode-choice behavior, some scholars focused on vulnerable groups in rural areas, such as elderly people [52], women [53], and low-income individuals [54]. Early research methods focused on the descriptive statistical analysis of survey data [55]. In recent years, some scholars have shifted from descriptive statistical analysis to a more multifaceted and in-depth analysis of impact mechanisms. However, compared to urban areas, there is still a lack of research on rural travel behavior [12]. The main studies analyzed the influence of personal attributes, travel attributes, and transportation attributes on the travel-mode-choice behavior of rural residents by using the discrete choice model. For example, Shirgaokar et al. [56] used logistic models to study the differences in the travel behavior of older adults between rural and urban areas. Uduak and Risako [57] used a multinomial logit model to analyze the factors affecting the choice of transportation mode for smallholder farmers. The study found that respondents’ travel-mode-choice behavior was mainly influenced by the attributes of the means of transportation. Mattson et al. [58] developed a hybrid logit model to estimate the travel-mode choice of rural residents. The results of this study showed that all three aspects—personal, travel, and transportation characteristics—had a significant impact on travel choice. In addition, some scholars explored the impact of the built environment in rural areas on travel-mode-choice behavior [59,60,61]. A summary of representative studies is highlighted in Table 1. It can be seen from the aforementioned studies and Table 1 that many studies of urban areas have considered the impact of passengers’ perceptions of service quality on travel-mode-choice behavior. However, the influencing factors considered in previous studies of rural areas are mostly directly observable variables. This ignored the influence of rural residents’ psychological latent factors (e.g., their psychological perceptions and attitudes toward bus services), which may make the model estimation results deviate from real travel behavior.
In conclusion, the existing research rarely considers rural residents’ satisfaction with the service quality of urban–rural buses and overlooks the effect of satisfaction on bus-choice behavior. Accordingly, in this study, an SEM–MNL integrated model is constructed to analyze rural residents’ satisfaction with each service attribute of urban–rural buses; identify the main factors that affect rural residents’ overall satisfaction; and explore the effect of rural residents’ satisfaction on their bus travel-choice behavior.

3. Methodology

3.1. Questionnaire Design

A questionnaire survey was used to collect information on rural residents’ socio-demographic characteristics, urban–rural travel characteristics, and satisfaction with the quality of urban–rural bus services. The questionnaire was designed based on existing findings in the literature [23,50,51] and the characteristics of urban–rural buses. Before the formal survey, a village was selected for a presurvey of 20 rural residents. The purpose of the presurvey was to determine whether the number of questions, the reasonableness of the options, and the difficulty of the questions were within the acceptable range for rural residents. The final composition of this survey questionnaire is as follows:
  • The socio-demographic characteristics of rural residents include gender, age, occupation, education level, monthly income, and private car ownership.
  • The urban–rural travel characteristics survey mainly includes rural residents’ urban–rural travel mode, urban–rural travel distance, and urban–rural travel frequency.
  • To evaluate rural residents’ satisfaction, the survey covered six latent variables reflecting the comfort, convenience, safety, reliability, economy, and facility completeness of urban–rural bus service quality. Twenty-two corresponding observed variables were selected to quantify the latent variables. The specific measurement scale is shown in Table 2. All the indicators were measured with a 5-point Likert scale (1 = very dissatisfied, 2 = dissatisfied, 3 = neutral, 4 = satisfied, and 5 = very satisfied).
  • The actual service level of urban–rural buses and the expectations of rural residents were investigated, including the ticket price, waiting time, and walking time from home to the stop. In addition, rural residents’ current monthly urban and rural bus trips and future travel intentions were also investigated. It should be noted that these surveys were designed to better understand the subjective intentions of rural residents and were not considered in the model.

3.2. Study Area and Data Collection

Existing studies on the travel behavior of rural residents focus on whole areas of the city [57,59,60]. By contrast, nearly no research has been conducted on the travel behavior of rural residents in county areas. This study was conducted in Yanjin County, Henan Province, China. Located in central-eastern China, Henan is a large, agricultural province with 42.39 million rural residents [4], ranking first in the country. Yanjin is a county located in the north of Henan Province covering an area of 1480 km2. There are three sub-districts, 11 townships, and 339 villages. It is worth noting that the rural areas of the county cover 96.04% of the total area. Due to the small size of the county’s urban area, there is no urban bus service. Therefore, urban–rural buses are the only mode of public transport linking county and rural areas. Yanjin has a population of 447,800, with the rural population accounting for 61.8% of the county’s population, or 276,700 [4]. Due to the large rural population in the survey area, it would have been difficult to survey the whole population. The current study investigated rural residents in 11 townships of Yanjin by using the simple random sampling method to explore their satisfaction with urban–rural buses and their travel-mode-choice behavior. Furthermore, in order to avoid excessive concentrations of respondents, we conducted random sampling surveys in each township separately to ensure the broad representativeness of the samples. The geographical locations of sample townships are shown in Figure 1.
In order to ensure that the sample could accurately reflect the whole, a certain sample size needed to be met. The sample size can be affected by various factors such as population size, population proportion, and degree of precision [23]. The sample size was calculated as follows [64]:
n = X 2 N P ( 1 P ) M E 2 N + X 2 P ( 1 P )
where n is the required sample size; M E is the desired margin of error (5%); N is the total size of the survey population; P is the population proportion (0.5); X 2 is the value of chi square at 95% confidence level (3.841).
The total population of the survey area is 276,700. The value of the sample size calculated from the formula was 384.
Sample surveys of rural residents in 11 townships were conducted by using a Chinese online survey platform (https://www.wjx.cn/) from 10 June to 15 June 2022, and 692 questionnaires were collected. Questionnaires with a response time of less than two minutes [65] and consecutive selections of the same value equal to or greater than half the length were considered invalid [66]. Since most of the respondents were rural residents with low education levels and there were many questions on the scale, after eliminating invalid questionnaires, a total of 578 valid questionnaires were obtained, with an effective rate of 83.53%.

3.3. Framework of the SEM–MNL Integration Model

To comprehensively describe the influence of subjective and objective factors on rural residents’ urban–rural travel-mode-choice behavior, an SEM–MNL model considering the psychological latent variables was constructed. The structure of the SEM–MNL integration model in this paper is shown in Figure 2.
The model consists of the structural equation model and the MNL discrete choice model. SEM can solve the problem of quantifying variables that are not directly observable and address the interrelationship between multiple independent variables and multiple dependent variables simultaneously [67]. Therefore, it is used to quantify the satisfaction latent variables through observed variables and describe the causal relationship between latent variables and explicit variables of socio-demographics and travel characteristics. It is important to note that observed variables cannot affect individual choice behavior and can only be used to measure latent variables [68]. The MNL choice model is used to explore the probability of rural residents’ choice of urban–rural travel modes and the nonlinear functional relationship with the latent and explicit variables that influence this decision.

3.4. Solution Method of the SEM–MNL Model

The SEM–MNL model adds the latent variables to the fixed utility term of the MNL discrete choice model [51]. Therefore, the utility function of the MNL model includes not only explicit variables of urban–rural travel attributes and socio-demographic attributes but also the latent variables of rural residents’ satisfaction. The improved utility function U i n can be expressed as follows [69]:
U i n = V i n + ε i n .
V i n = l a i l s i l n + q b i q z i q n + k c i k η i k n .
where V i n is the fixed utility term; ε i n is the random utility term; s i l n is explicit variable of socio-demographics; z i q n is the urban–rural travel characteristic explicit variable; η i k n is the latent variable of rural residents’ satisfaction with bus service quality; l is the number of socio-demographic variables; q is the number of urban–rural travel characteristic variables; k is the number of satisfaction latent variables; and a i l ,   b i q ,   c i k are parameters to be estimated.
In this paper, the urban–rural travel modes are divided into three options: non-motorized vehicles, private cars and urban–rural buses, so the multinomial logit model is chosen for the discrete choice model part. We assume that the set of urban–rural travel-mode options for rural residents n is G n . The MNL model assumes that travelers follow the utility maximization principle when making travel choices [70]. According to utility maximization theory, rural residents n will choose i when option i brings higher utility than all other options in the set of options G n . Therefore, the condition for rural residents n to choose mode i is expressed as [48]:
U i n U j n , i j , j G n
Then the probability P i n that rural residents n choose mode i can be expressed as [68]:
P i n = P r o b ( U i n > U j n , i j , j G n ) = P r o b ( V i n + ε i n > V j n + ε j n , i j , j G n ) = P r o b ( ε j n < V i n + ε i n V j n , i j , j G n )
Prob denotes probability.
If the random term ε i n is assumed to obey a Gumbel type-1 distribution [48], the probability P i n obtained by derivation is as follows:
P i n = P U i n U j n ; i j , i , j = 1 , 2 , 3 = exp ( l a i l s i l n + q b i q z i q n + k c i k η i k n ) j exp ( l a j l s j l n + q b j q z j q n + k c j k η j k n ) .
To estimate the parameter values of the latent variables, it is necessary to establish the relationship between the latent variables and the measured variables. A structural equation modeling approach is used to explore this relationship. SEM comprises a structural model and a measurement model [50]. The structural model explains the relationship between latent and explicit variables [65]. The measurement model quantifies latent variables by exploring the relationship between each directly perceived observed variable and its corresponding latent variable [71]. SEM can be expressed as:
η i k n = r λ i k n x i r n + ζ i k n .
y i t n = k γ i k t η i k n + ξ i t n .
where r is the number of explicit variables that have influential relationships with the latent variables; t is the number of observed variables corresponding to the latent variables; x i r n is the explicit variable affecting the latent variables; y i t n is the observed variable corresponding to the latent variable; λ i k n , γ i k t are parameters to be estimated; ζ i k n , ξ i t n are the error terms and ζ i k n ~ N ( 0 , 1 ) , ξ i t n ~ N ( 0 , 2 ) .
The load factors Λ X n of the latent variables can be obtained by constructing the measurement model of satisfaction with urban–rural bus service quality [51]. Take a latent variable η X in bus services as an example. Suppose the observed variables corresponding to this latent variable are X 1 , X 2 X n , then Equation (8) is converted to vector form and expressed as [50]:
X 1 X 2 X n = Λ X 1 Λ X 2 Λ X n η X
The factor loading coefficients Λ X 1 , Λ X 2 , , Λ X n between the observed and latent variables are used as the weights of each observed variable. Then, the load factors are standardized to obtain the standardized coefficient values r X 1 , r X 2 , , r X n of each observed variable.
r X 1 = Λ X 1 Λ X 1 + Λ X 2 + + Λ X n r X 2 = Λ X 2 Λ X 1 + Λ X 2 + + Λ X n r X n = Λ X n Λ X 1 + Λ X 2 + + Λ X n
Finally, the value of the latent variable is expressed as:
η X = r X 1 X 1 + r X 2 X 2 + + r X n X n

4. Results

4.1. Descriptive Statistical Analysis

The statistical results of the socio-demographic and urban–rural travel characteristics of rural residents are shown in Table 3. Approximately 66.8% of the respondents were female. This is because most men living in rural areas of China work outside the home, and most people who stay at home are female. The age of the respondents was mainly between 30 and 50 (74.2%). The majority of respondents were peasants, accounting for 47.8% of the total sample. The monthly income of the respondents was mostly below CNY 2000 (53.3%), whereas the per capita monthly disposable income of urban residents is CNY 3951 [4]. The overall education level of rural residents is low, with only 14.2% of respondents having received higher education (junior college or above). Respondents who owned a private car accounted for 58.5% of the total sample. The proportion of respondents who made more than five urban–rural trips per month was only 7.6%, and 38.6% of respondents had a monthly urban–rural trip frequency of zero. The majority of respondents (40.5%) traveled 20–30 km between urban and rural areas, followed by less than 20 km (34.3%) and more than 30 km (25.3%).
Figure 3a shows that 27% of respondents walked more than 30 min to the bus stop. However, 69% of them had an acceptable walking time of less than 20 min. Moreover, 50% of the respondents waited more than 20 min for a bus, as shown in Figure 3b. However, 77% of respondents had an acceptable waiting time of between 0 and 20 min. As shown in Figure 3c, the actual fares of urban–rural buses were mainly CNY 7 to 8 (34%), and 47% were above CNY 7. However, 76% of respondents could accept fares of less than CNY 6. In general, there is a gap between the current level of urban–rural bus services and the acceptable level for respondents. This has contributed to the low number of respondents taking urban–rural buses.
As shown in Figure 4, the percentage of rural residents in the survey area who took the urban–rural buses zero times per month was 65.7%, and only 2.7% took it more than five times per month. However, when the quality of urban–rural bus services is improved, 70% of rural residents were willing to take the bus for travel between urban and rural areas, as shown in Figure 5. This shows that there is a great potential for rural residents to demand urban–rural buses, so it is necessary to fully address the actual needs of rural residents and meet their expectations.

4.2. Analysis of Reliability and Validity

Before analyzing the structural equation model, the scale data obtained from the survey needed to be tested for reliability and validity. This statistical approach is used to understand the structure of the observed variables and their relationship to the underlying variables [62]. Confirmatory factor analysis (CFA) was conducted to test the fit of the sample data to the model. First, a first-order CFA model was constructed by plotting the path diagram and calculating the path coefficients between each variable by AMOS 24.0 software. The model structure is shown in Figure 6 and e1~e28 are the errors of each observable.
Four indices, namely Cronbach’s α, construct reliability (CR), and average variance Extracted (AVE), were applied to measure reliability and validity [49]. The results are shown in Table 4. The Cronbach’s α values and CR values for each latent variable are greater than 0.7, indicating a high degree of internal consistency between the factors [50,72]. The AVE values are all greater than 0.5, indicating that all six latent variables could be well represented by the corresponding observed variables [68]. Overall, the survey data have good reliability and validity.
However, it can be seen from Table 5 that the correlation coefficients of the six latent variables are positive and greater than 0.5, indicating that there is a significant positive interaction between the latent variables of urban–rural bus service quality. Therefore, improving the key indicators that affect rural residents’ satisfaction will increase their overall satisfaction with buses. This also suggests that the model may have a higher level of factor structure [73].
Therefore, a second-order structural equation model of satisfaction with urban–rural bus service quality was constructed. The variance inflation factor (VIF) was employed to test for the presence of multi-collinearity among the latent variables [48]. The test results indicate that the VIF values are all below 5.0, indicating that there is no multi-collinearity among the latent variables [74]. The second-order model was tested for goodness-of-fit. The corresponding test results are shown in Table 6. The chi-square/degrees of freedom ( χ 2 / d f ) , root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker–Lewis Index (TLI), and standardized root mean square residual (SRMR) indicators are within acceptable limits [62,75], indicating that the overall fitness of this structural equation model is good.

4.3. SEM Estimation Results and Analysis

The relationship paths between the variables obtained after correction and normalization are shown in Figure 7. The numbers on the paths reflect the influence degree of the observed and explicit variables on the corresponding latent variables [48]. A larger absolute value of the path coefficient means a higher degree of correlation between the two. The structural model path coefficient estimates and test values are shown in Table 7. It can be seen from Table 7 that the load factor coefficients of each variable are at the 0.05 significance level.
Gender in socio-demographic characteristics is positively correlated with satisfaction with urban–rural bus service quality, that is, women are more satisfied than men. This is inconsistent with the findings of urban bus satisfaction studies [31,76]. This is probably because most rural women do not have a driver’s license and use non-motorized vehicles for their daily travels. Therefore, when they travel longer distances between urban and rural areas, urban–rural buses will provide them with a more comfortable trip than non-motorized vehicles. The older the rural residents are, the less satisfied they are with urban–rural bus services. Similar results have been found in other studies [29]. However, older residents are the main users of urban–rural buses [12], so it is necessary to focus on the travel needs of elderly people and provide services oriented toward them. The higher the respondents’ average monthly income, the lower their satisfaction. The reason for this finding is that rural residents with high incomes mostly own private cars [2] and have higher requirements for all aspects of travel.
Rural residents who commute to work show stronger dissatisfaction with the overall bus service. Similar conclusions can be found in other studies [1]. This is because they have a high demand for the timeliness of travel, and urban–rural buses fail to meet their expectations. However, their urban–rural trips are more frequent and more regular, so the more economical urban–rural buses should become their preferred mode of travel. Therefore, it is worth exploring how to provide appropriate bus services that meet their travel needs. The higher the respondents’ education level, the lower their travel satisfaction. The travel mode in urban–rural travel characteristics is positively correlated with satisfaction. This indicates that villagers who choose bus travel are more satisfied with urban–rural buses than those who choose non-motorized vehicles and private car travel. Similar results have been found in studies of other public-transportation-related policies [65]. However, Dandapat et al. [29] found that people who always use the urban bus are less satisfied than those who use other modes. Travel distance and frequency did not pass the significance test. This finding suggests that rural residents with different characteristics perceive the quality of public transportation services differently.
The measurement model path coefficient estimates and test values are shown in Table 8. In terms of the path coefficient between latent variables, the most critical factor that affects rural residents’ satisfaction with the quality of urban–rural bus services is reliability, with the largest factor loading coefficient (0.939), followed by convenience, economy, completeness of facilities, safety, and comfort. This is similar to the theory of essential and non-essential attributes of service quality proposed by Eboli and Mazzulla [27]. In terms of the path coefficients between the latent and measured variables, among the measurement factors of reliability, the punctuality of urban–rural buses is the most critical factor. This is consistent with earlier studies of urban buses, which found that punctuality is the most important service attribute for all user groups [3,28]. In addition, bus reliability was cited as a major problem affecting wait times [77]. It is also evident from Figure 3b that rural residents in the survey area have long waiting times for buses. Therefore, the low punctuality of local urban–rural buses is a key issue that urgently needs to be addressed.
Convenience from the bus station to the destination is an important factor that affects satisfaction with convenience. The convenience of accessing buses for residents is considered a key attribute in most studies on urban buses [25,30]. This is because the urban area of the surveyed counties is small (approximately 3.96% of the total area) and does not have urban buses, so rural residents have to walk to other places after getting off the urban–rural bus. In terms of the economy, bus fare concessions have a greater impact than fares. However, Shabani et al. [42] found that urban residents are more concerned about the direct costs they must pay for using buses. This is because most urban areas have implemented fare discount policies, while rural areas lag behind in this regard. Rural residents are more concerned about the adequacy of facilities inside the bus in terms of facility completeness. This is consistent with the findings of Zheng et al. [31], who found that passengers are more concerned about the comfort inside the bus.
The safety of bus stops is a key factor affecting the safety of urban–rural buses. Many rural residents do not have bus shelters, so the safety of bus stops is difficult to ensure. The same relationships have also been found in previous studies [23,76]. The indicator that has the greatest impact on satisfaction with bus comfort is the attitude of staff service. This indicates that the service consciousness of the existing urban–rural bus staff is not sufficient. The above findings provide a basis for government and urban–rural bus enterprises to improve their service quality from the perspective of rural residents’ subjective perceptions.
Each latent variable of SEM can be calculated according to Equations (10) and (11). The adaptation values of each latent variable are as follows:
T E = 0.491 T E 1 + 0.509 T E 2
T S = 0.315 T S 1 + 0.342 T S 2 + 0.343 T S 3
T R = 0.325 T R 1 + 0.354 T R 2 + 0.321 T R 3
C F = 0.326 C F 1 + 0.348 C F 2 + 0.326 C F 3
B C = 0.212 B C 1 + 0.214 B C 2 + 0.240 B C 3 + 0.165 B C 4 + 0.169 B C 5
T C = 0.133 T C 1 + 0.180 T C 2 + 0.179 T C 3 + 0.179 T C 4 + 0.161 T C 5 + 0.168 T C 6
The overall satisfaction degree η can be formulated as Equation (18).
η = 0.142 B C + 0.150 T S + 0.174 T E + 0.187 T R + 0.180 T C + 0.167 C F

4.4. Estimation Results of the SEM–MNL Model

The fitted value η of the latent variable of overall satisfaction was obtained by the measurement model, together with rural residents’ socio-demographic characteristics and urban–rural travel characteristics, as the influencing variable of the MNL choice model to construct the SEM–MNL model. STATA 17.0 software was used to calculate the model. In addition, this paper compares the SEM–MNL model with latent variables to the MNL model without latent variables. Table 9 shows the model estimation results with the private car as the reference term and gives the odds ratio (OR) value.
The OR is used to show the effect of a one-unit change in the explanatory variable on the decision outcome [48]. The goodness of fit R 2 is the index of the evaluation of the MNL model, ranging from 0 to 1. The accuracy of the model is generally considered high when R 2 is higher than 0.2 [78]. The R 2 values of the two types of models are 0.2890 and 0.2953, indicating that the two models have high fitting accuracy. The SEM–MNL model is better than the traditional MNL model without latent variables. Similar results have been found in other studies [49,65]. This indicates that the SEM–MNL model can better describe the urban–rural travel-mode-choice behaviors of rural residents.
In the fitting results of the MNL model without latent variables, the constant terms for non-motorized vehicles and buses are 2.45 and 1.34, respectively. This shows that, holding other factors fixed, rural residents are 2.45 times more likely to choose non-motorized vehicles than private cars, and 1.34 times more likely to choose buses than private cars. Therefore, non-motorized vehicles are the dominant mode of transportation for urban–rural travel. This is consistent with the findings of Yu et al. [12] and Kong and Yao [79]. This may be closely related to the relatively higher convenience and affordability of non-motorized vehicles.
From the estimation results of the SEM–MNL model, we can find that male rural residents are more likely to choose the urban–rural bus for travel than females. Higher-educated rural residents are more inclined to choose private car travel compared to non-motorized vehicles. The same relationships have also been found in previous studies [80]. However, Guo et al. [65] found that urban residents with higher levels of education will choose public transit. In addition, the results of Ao et al. [2] indicated that rural residents who believe that riding motorcycles or electric bicycles is environmentally friendly emit less CO2. Accordingly, this result may be related to the general lack of environmental awareness among rural residents. Rural residents with private cars and high income levels are more likely to choose private cars than urban–rural buses. Earlier studies have also confirmed these relationships [57,59]. In terms of travel attributes, travel distance has a negative effect on the choice of both non-motorized vehicles and public transportation, indicating that rural residents prefer to travel by private car for long-distance trips. This is in line with the findings of other studies on rural residents’ travel [12]. The higher the frequency of rural residents’ urban–rural trips, the higher their probability of choosing non-motorized vehicles, followed by buses and private cars. On the one hand, residents with a high frequency of urban–rural trips generally live in villages closer to urban areas, so non-motorized vehicles are more convenient and economical for them. On the other hand, the higher frequency of trips means that they have to pay more for their trips, thus leading them to choose more economical non-motorized vehicles or buses.
The latent variable of rural residents’ satisfaction with urban–rural bus service quality has a significant positive effect on their choice of bus. From the OR value, we can see that when villagers’ satisfaction with the quality of bus services increases by one indicator value, the likelihood of choosing buses relative to private cars increases by 1.99 times. The latent variable of satisfaction has a greater impact on rural residents’ urban–rural bus travel-choice behavior than all explicit variables except private car ownership. Therefore, improving rural residents’ satisfaction with the service quality of urban–rural buses is a key factor in promoting their choice of urban–rural bus travel.

4.5. Policy Implications

Improving the quality of urban–rural bus services and providing travel services to satisfy passengers is important for improving the share of urban–rural buses. We have identified the current issues that need urgent improvement based on the perceived satisfaction of passengers. This will allow the relevant authorities to allocate limited resources to more important aspects to improve the efficiency of resource utilization. According to the descriptive statistical analysis of the survey data and the estimation results of the SEM–MNL model, we propose some policy implications for promoting the development of urban–rural buses as follows:
Local transport authorities and bus companies should recognize the importance of passengers in the evaluation of urban–rural buses and always investigate, analyze, and give feedback on passengers’ opinions. Our study found that passengers’ satisfaction has a significant impact on their bus travel-choice behavior. Therefore, the relevant authorities should provide appropriate platforms, such as citizen hotlines, online service platforms, and social software, to collect residents’ suggestions on urban–rural buses. To encourage active participation, appropriate incentives can also be given to residents who provide suggestions. In addition, it is important to provide correct and prompt feedback on residents’ opinions and make improvements.
In terms of improving the level of urban–rural bus infrastructure construction, the completeness of the facilities inside the bus and the safety of the waiting stop are key factors that affect rural residents’ satisfaction with the safety and facility-completeness of buses. This implies that the government should accelerate the construction of bus shelters in rural areas to ensure the safety of rural residents at waiting stops. Furthermore, old bus vehicles should be renovated and replaced, and the facilities inside the vehicles should be improved to enhance the comfort of rural residents in the vehicles.
Improving the operation and management level of urban–rural buses is another recommendation. The SEM results show that reliability is the most critical factor that affects the overall satisfaction of urban and rural bus services, with punctuality having the greatest impact on it. Therefore, this is an urgent issue that needs to be addressed. Urban–rural bus operators should first develop vehicle operating schedules and supervise the operation of the vehicles. Second, the frequency of vehicle departures should be increased appropriately according to passenger flow. These approaches will reduce rural residents’ waiting times and increase their satisfaction with bus travel. In addition, convenience to the destination, the service attitude of the driver, and fare discounts are key factors that affect residents’ satisfaction with the convenience, comfort, and economy of urban–rural buses. For counties similar to the survey area (without urban bus service due to a small urban area), the transport authorities can consider setting up shared public transportation, such as shared bicycles and shared electric bicycles, to solve the “last mile” problem of bus travel for rural residents. Local governments should reasonably optimize ticket prices and develop preferential fare policies to reduce travel costs for rural residents. To improve the service level of drivers, they need to be trained and assessed regularly.

5. Discussion and Conclusions

This study conducted an empirical analysis of rural residents’ satisfaction and the perceived heterogeneity in the quality of urban–rural bus services and explored the subjective and objective factors affecting rural residents’ urban–rural travel-mode choice. Several conclusions drawn from this study have implications for developing urban–rural buses that satisfy riders.
First, the results show that rural residents with higher incomes, older age, and better education and those who commute to work are less satisfied with urban–rural bus services. The satisfaction of male and rural residents who choose buses for travel is higher than that of other residents. Therefore, when planning urban–rural buses, the government should take into account the differences in the socio-economic attributes of the inhabitants of each village to meet the needs of different groups of people.
Second, the study found an interesting conclusion—that the correlation coefficients of the six service quality latent variables are positive and greater than 0.5—indicating that there is a significant positive interaction between rural residents’ satisfaction with the quality of each urban–rural bus service. Therefore, improving the key service indicators that affect rural residents’ satisfaction will increase their overall satisfaction with bus services. However, similar conclusions have not been reached in previous studies of urban buses [23,37]. This implies that there are differences in the demand for bus services between rural and urban residents. Therefore, the government should understand the actual needs of rural residents and not just pursue the complete unification of urban–rural buses. For the survey area, the key factors that affect current urban–rural bus satisfaction are reliability and convenience, with the punctuality rate and convenience from bus stops to the destination having the greatest degree of influence on both.
Third, the results revealed that rural residents’ satisfaction with urban–rural bus service quality has a significant positive effect on their choice of bus, and its degree of influence is greater than all explicit variables except private car ownership. Therefore, improving rural residents’ travel satisfaction is a key initiative to promote their choice of urban–rural bus travel. In addition, this study found that rural residents in the survey area mainly used non-motorized vehicles to travel between urban and rural areas. For long-distance trips, private cars replaced buses as the preferred mode. This indicates that urban–rural bus services in the survey area do not satisfy villagers. Therefore, the relevant authorities should focus on the needs of rural residents for urban–rural bus services and introduce more refined improvement strategies for different types of services to enhance the satisfaction of rural residents.
Nonetheless, it is necessary to recognize that the residents’ travel choices will change in different travel environments (e.g., rain, snow, and wind speed) and at different times (e.g., weekends, holidays, and workdays). This was not considered in the present study. Additionally, due to the limitation of the survey conditions, an online survey was applied in this study. This affected the validity of the sample, and the sample group cannot cover all types of rural residents, such as the elderly without smartphones and those with lower literacy levels. Also, the service quality model and model results formed in this study are specific to the area context. In future studies, the applicability of the findings should be validated by analyzing more geographically similar areas. Moreover, this work could be extended by obtaining data from other cities to examine and cross-compare the difference in the quality of urban–rural bus services and rural residents’ travel-mode-choice behavior.

Author Contributions

Conceptualization, H.C. and M.L.; methodology, H.C. and M.L.; software, M.L.; validation, H.C., M.Z. and X.M.; formal analysis, M.L.; investigation, M.L.; resources, H.C.; data curation, M.Z.; writing—original draft preparation, M.L.; writing—review and editing, H.C.; visualization, X.M.; supervision, M.Z.; project administration, H.C.; funding acquisition, M.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Social Science Foundation of Hebei province (HB22YJ040).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

Data used in this study is unavailable due to privacy or ethical restrictions.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Ponrahono, Z.; Bachok, S.; Ibrahim, M.; Osman, M.M. Assessing Passengers’ Satisfaction Level on Bus Services in Selected Urban and Rural Centres of Peninsular Malaysia. Procedia-Soc. Behav. Sci. 2016, 222, 837–844. [Google Scholar] [CrossRef] [Green Version]
  2. Ao, Y.; Yang, D.; Chen, C.; Wang, Y. Effects of rural built environment on travel-related CO2 emissions considering travel attitudes. Transp. Res. Part D Transp. Environ. 2019, 73, 187–204. [Google Scholar] [CrossRef]
  3. Porter, G. Transport services and their impact on poverty and growth in rural sub-saharan Africa: A review of recent research and future research needs. Transp. Rev. 2014, 34, 25–45. [Google Scholar] [CrossRef] [Green Version]
  4. Jiang, A.; Zhong, Q.; Wang, Y.; Ao, Y.; Chen, Y. Influencing factors of commercial energy consumption intention of rural residents: Evidence from rural Chengdu. Energies 2021, 14, 1010. [Google Scholar] [CrossRef]
  5. National Bureau of Statistics. Statistical Yearbook of China; China Statistics Press: Beijing, China, 2022.
  6. Qu, X.; Wang, S.; Niemeier, D. On the urban-rural bus transit system with passenger-freight mixed flow. Commun. Transp. Res. 2022, 2, 100054. [Google Scholar] [CrossRef]
  7. De Oña, J.; De Oña, R. Quality of service in public transport based on customer satisfaction surveys: A review and assessment of methodological approaches. Transp. Sci. 2015, 49, 605–622. [Google Scholar] [CrossRef] [Green Version]
  8. Das, S.; Pandit, D. Importance of user perception in evaluating level of service for bus transit for a developing country like India: A review. Transp. Rev. 2013, 33, 402–420. [Google Scholar]
  9. Eboli, L.; Mazzulla, G. Service quality attributes affecting customer satisfaction for bus transit. J. Public Transp. 2007, 10, 21–34. [Google Scholar] [CrossRef] [Green Version]
  10. Hernandez, S.; Monzon, A. Key factors for defining an efficient urban transport interchange: Users’ perceptions. Cities 2016, 50, 158–167. [Google Scholar] [CrossRef] [Green Version]
  11. Fu, X.M.; Juan, Z.C. Understanding public transit use behavior: Integration of the theory of planned behavior and the customer satisfaction theory. Transportation 2017, 44, 1021–1042. [Google Scholar] [CrossRef]
  12. Yu, Z.; Li, P.; Schwanen, T.; Zhao, P.; Zhao, Z. Role of rural built environment in travel mode choice: Evidence from China. Transp. Res. Part D Transp. Environ. 2023, 117, 103649. [Google Scholar]
  13. Tumsekcali, E.; Ayyildiz, E.; Taskin, A. Interval valued intuitionistic fuzzy AHP-WASPAS based public transportation service quality evaluation by a new extension of SERVQUAL Model: P-SERVQUAL 4.0. Expert Syst. Appl. 2021, 186, 115757. [Google Scholar]
  14. EN 13816; Transport–Logistics and Services–Public Passenger Transport–Service Quality Definition Target and Measurement. European Committee for Standardization (CEN): Brussels, Belgium, 2002.
  15. Transportation Research Board. Transit Capacity and Quality of Service Manual, 2nd ed.; Transit Cooperative Research Program (TCRP) Report 100; Transportation Research Board: Washington, DC, USA, 2004. [Google Scholar]
  16. Hassan, M.N.; Hawas, Y.E.; Ahmed, K. A Multi-dimensional framework for evaluating the transit service performance. Transp. Res. Part A Policy Pract. 2013, 50, 47–61. [Google Scholar]
  17. Hensher, D.A.; Stopper, P.; Bullock, P. Service quality-developing a service quality index in the provision of commercial bus contracts. Transp. Res. Part A Policy Pract. 2003, 37, 499–517. [Google Scholar] [CrossRef]
  18. Stuart, K.R.; Mednick, M.; Bockman, J. Structural equation model of customer satisfaction for the New York City subway system. Transport. Res. Rec. 2000, 1735, 133–137. [Google Scholar] [CrossRef] [Green Version]
  19. De Oña, J.; De Oña, R.; Eboli, L.; Mazzulla, G. Perceived service quality in bus transit service. A structural equation approach. Transp. Policy. 2013, 29, 219–226. [Google Scholar] [CrossRef]
  20. Hernandez, S.; Monzon, A.; De Oña, R. Urban transport interchanges: A methodology for evaluating perceived quality. Transp. Res. Policy Pract. 2015, 84, 31–43. [Google Scholar]
  21. Allen, J.; Eboli, L.; Forciniti, C.; Mazzulla, G.; de Dios Ortúzar, J. The role of critical incidents and involvement in transit satisfaction and loyalty. Transp. Policy 2019, 75, 57–69. [Google Scholar]
  22. Wang, Y.; Ao, Y.; Zhang, Y.; Liu, Y.; Zhao, L.; Chen, Y. Impact of the built environment and bicycling psychological factors on the acceptable bicycling distance of rural residents. Sustainability 2019, 11, 4404. [Google Scholar] [CrossRef] [Green Version]
  23. Chauhan, V.; Gupta, A.; Parida, M. Demystifying service quality of Multimodal Transportation Hub (MMTH) through measuring users’ satisfaction of public transport. Transp. Policy 2021, 102, 47–60. [Google Scholar] [CrossRef]
  24. Hu, K.C.; Jen, W. Passengers’ perceived service quality of city buses in Taipei: Scale development and measurement. Transp. Rev. 2006, 26, 645–662. [Google Scholar]
  25. Garrido, C.; De Oña, R.; De Oña, J. Neural networks for analyzing service quality in public transportation. Expert Syst. Appl. 2014, 41, 6830–6838. [Google Scholar] [CrossRef]
  26. Sinha, S.; Sadhukhan, S.; Priye, S. The role of quality assessment for development of sustainable bus service in mid-sized cities of India: A case study of Patna. Procedia Eng. 2017, 198, 926–934. [Google Scholar]
  27. Eboli, L.; Mazzulla, G. A stated preference experiment for measuring service quality in public transport. Transp. Plann. Technol. 2008, 31, 509–523. [Google Scholar]
  28. Bose, T.; Pandit, D. Heterogeneity in perception of service quality attributes of bus transit across various user categories- A case of Kolkata. Transp. Res. Procedia 2020, 48, 2784–2799. [Google Scholar] [CrossRef]
  29. Dandapat, S.; Maitra, B. Preference heterogeneity in trip makers’ perception and policy Issues: A study with reference to bus services in Kolkata. Case Stud. Transp. Policy 2020, 8, 1504–1517. [Google Scholar]
  30. Dell’Olio, L.; Ibeas, A.; Cecin, P. Modelling user perception of bus transit quality. Transp. Policy 2010, 17, 388–397. [Google Scholar] [CrossRef]
  31. Zheng, Y.H.; Kong, H.; Petzhold, G.; Barcelos, M.M.; Zegras, C.P.; Zhao, J.H. Gender differences in the user satisfaction and service quality improvement priority of public transit bus system in Porto Alegre and Fortaleza, Brazil. Travel Behav. Soc. 2022, 28, 22–37. [Google Scholar]
  32. Wong, R.C.P.; Yang, L.; Szeto, W.Y. Comparing passengers’ satisfaction with fixed-route and demand-responsive transport services: Empirical evidence from public light bus services in Hong Kong. Travel Behav. Soc. 2023, 32, 100583. [Google Scholar]
  33. Das, S.; Pandit, D. Determination of level-of-service scale values for quantitative bus transit service attributes based on user perception. Transp. A Transp. Sci. 2015, 11, 1–21. [Google Scholar]
  34. Echaniz, E.; Cordera, R.; Rodriguez, A.; Nogués, S.; Coppola, P.; Dell’Olio, L. Spatial and temporal variation of user satisfaction in public transport systems. Transp. Policy 2022, 117, 88–97. [Google Scholar] [CrossRef]
  35. Zheng, Y.H.; Kong, H.; Petzhold, G.; Barcelos, M.M.; Zegras, C.P.; Zhao, J.H. User satisfaction and service quality improvement priority of bus rapid transit in Belo Horizonte, Brazil. Case Stud. Transp. Policy 2021, 9, 1900–1911. [Google Scholar] [CrossRef]
  36. Sam, E.F.; Hamidu, O.; Daniels, S. SERVQUAL analysis of public bus transport services in Kumasi metropolis, Ghana: Core user perspectives. Case Stud. Transp. Policy 2017, 6, 25–31. [Google Scholar] [CrossRef]
  37. Deb, S.; Ahmed, M.A.; Das, D. Service quality estimation and improvement plan of bus Service: A perception and expectation based analysis. Case Stud. Transp. Policy 2022, 10, 1775–1789. [Google Scholar] [CrossRef]
  38. Eboli, L.; Mazzulla, G. A new customer satisfaction index for evaluating transit service quality. J. Public Transp. 2009, 12, 21–37. [Google Scholar] [CrossRef] [Green Version]
  39. Eboli, L.; Mazzulla, G. A methodology for evaluating transit service quality based on subjective and objective measures from the passenger’s point of view. Transp. Policy 2011, 18, 172–181. [Google Scholar] [CrossRef]
  40. Lois, D.; Monzon, A.; Hernandez, S. Analysis of satisfaction factors at urban transport interchanges: Measuring travellers’ attitudes to information, security and waiting. Transp. Policy 2018, 67, 49–56. [Google Scholar] [CrossRef]
  41. De Oña, J.; De Oña, R.; Calvo, F.J. A classification tree approach to identify key factors of transit service quality. Expert Syst. Appl. 2012, 39, 11164–11171. [Google Scholar] [CrossRef]
  42. Shabani, A.; Shabani, A.; Ahmadinejad, B.; Salmasnia, A. Measuring the customer satisfaction of public transportation in Tehran during the COVID-19 pandemic using MCDM techniques. Case Stud. Transp. Policy 2022, 10, 1520–1530. [Google Scholar] [CrossRef]
  43. Goldberger, A.S. Structural equation methods in the social sciences. Econometrica 1972, 40, 979–1001. [Google Scholar] [CrossRef]
  44. Zhou, Y.; Yuan, Q.; Ding, F.; Chen, M.; Yang, C.; Guo, T. Demand, mobility, and constraints: Exploring travel behaviors and mode choices of older adults using a facility-based framework. J. Transp. Geogr. 2022, 102, 103368. [Google Scholar] [CrossRef]
  45. Huang, Y.; Gao, L.; Ni, A.; Liu, X. Analysis of travel mode choice and trip chain pattern relationships based on multi-day GPS data: A case study in Shanghai, China. J. Transp. Geogr. 2021, 93, 103070. [Google Scholar] [CrossRef]
  46. Kwigizie, V.; Valerian, D.; Sand, O.T. Across nensted logit model for trip type mode choice: An application. Adv. Transp. Stud. 2011, 23, 29–40. [Google Scholar]
  47. Ben-Akiva, M.; Mcfadden, D.; Train, K.; Walker, J.; Bhat, C.; Bierlaire, M. Hybrid Choice Model: Progress and Challenge. Mark. Lett. 2002, 13, 163–175. [Google Scholar] [CrossRef] [Green Version]
  48. Li, X.; Han, C.; Huang, H.; Pervez, A.; Xu, G.M.; Hu, C.; Jiang, Q.S.; Wei, Y.L. Pursuing higher acceptability and compliance for electric two-wheeler standardization policy in China: The importance of socio-demographic characteristics, psychological factors, and travel habits. Transp. Res. Part A Policy Pract. 2023, 167, 103557. [Google Scholar] [CrossRef]
  49. Mou, Z.H.; Liang, W.W.; Chen, Y.Y.; Zhang, H.; Yan, K.L.; Li, X. The effects of carpooling on potential car Buyers’ purchasing Intention: A case study of Jinan. Case Stud. Transp. Policy 2020, 8, 1285–1294. [Google Scholar] [CrossRef]
  50. Han, Y.; Li, W.; Wei, S.; Zhang, T. Research on passenger’s travel mode choice behavior waiting at bus station based on SEM-Logit integration model. Sustainability 2018, 10, 1996–2018. [Google Scholar] [CrossRef] [Green Version]
  51. Si, Y.; Guan, H.; Cui, Y. Research on the choice behavior of taxis and express services based on the SEM-logit model. Sustainability 2019, 11, 2974. [Google Scholar] [CrossRef] [Green Version]
  52. Plazinić, B.R.; Jović, J. Mobility and transport potential of elderly in differently accessible rural areas. J. Transp. Geogr. 2018, 68, 169–180. [Google Scholar] [CrossRef]
  53. Plazinić, B.R.; Jović, J. Women and transportation demands in rural Serbia. J. Rural Stud. 2014, 36, 207–218. [Google Scholar] [CrossRef]
  54. Singh, S.S.; Sarkar, B. Transport accessibility and affordability as the determinant of non-motorized commuting in rural India. Transp. Policy 2022, 118, 101–111. [Google Scholar] [CrossRef]
  55. Nutley, S. Monitoring rural travel behaviour: A longitudinal study in Northern Ireland 1979–2001. J. Transp. Geogr. 2005, 13, 247–263. [Google Scholar] [CrossRef]
  56. Shirgaokar, M.; Dobbs, B.; Anderson, L.; Hussey, E. Do rural older adults take fewer trips than their urban counterparts for lack of a ride? J. Transp. Geogr. 2020, 87, 102819. [Google Scholar] [CrossRef]
  57. Akpan, U.; Morimoto, R. Motorized travel mode choices of smallholder farmers in Akwa Ibom State, Nigeria. Transp. Res. Interdiscip. Perspect. 2022, 13, 100569. [Google Scholar] [CrossRef]
  58. Mattson, J.; Hough, J.; Varma, A. Estimating demand for rural intercity bus services. Res. Transp. Econ. 2018, 71, 68–75. [Google Scholar] [CrossRef]
  59. Ao, Y.; Zhang, Y.T.; Wang, Y.; Chen, Y.; Yang, L. Influences of rural built environment on travel mode choice of rural residents: The case of rural Sichuan. J. Transp. Geogr. 2020, 85, 102708. [Google Scholar] [CrossRef]
  60. Li, H.M.; Zhang, Y.T.; Ao, Y.B.; Wang, Y.; Wang, T.; Chen, Y.F. Built environment impacts on rural residents’ daily travel satisfaction. Front. Ecol. Evol. 2022, 10, 931118. [Google Scholar] [CrossRef]
  61. Ao, Y.; Yang, D.; Chen, C.; Wang, Y. Exploring the effects of the rural built environment on household car ownership after controlling for preference and attitude: Evidence from Sichuan, China. J. Transp. Geogr. 2019, 74, 24–36. [Google Scholar]
  62. Shah, B.A.; Zala, L.B.; Desai, N.A. An integrated estimation approach to incorporate latent variables through SEM into discrete mode choice models to analyze mode choice attitude of a rider. Transport. Res. Interdiscip. Perspect. 2023, 19, 100819. [Google Scholar] [CrossRef]
  63. Chen, J.; Li, S. Mode choice model for public transport with categorized latent variables. Math. Probl. Eng. 2017, 2017, 7861945. [Google Scholar] [CrossRef] [Green Version]
  64. Krejcie, R.V.; Morgan, D.W. Determining sample size for research activities. Educ. Psychol. Meas. 1970, 30, 607–610. [Google Scholar] [CrossRef]
  65. Guo, D.B.; Yao, E.J.; Liu, S.S.; Chen, R.S.; Hong, J.Y.; Zhang, J.Y. Exploring the role of passengers’ attitude in the integration of dockless bike-sharing and public transit: A hybrid choice modeling approach. J. Clean Prod. 2023, 384, 135627. [Google Scholar] [CrossRef]
  66. Curran, P.G. Methods for the detection of carelessly invalid responses in survey data. J. Exp. Soc. Psychol. 2016, 66, 4–19. [Google Scholar] [CrossRef] [Green Version]
  67. Ye, X.; Sui, X.; Wang, T.; Yan, X.; Chen, J. Research on parking choice behavior of shared autonomous vehicle services by measuring users’ intention of usage. Transp. Res. Part F Traffic Psychol. Behav. 2022, 88, 81–98. [Google Scholar] [CrossRef]
  68. Xia, X.; Jiang, H.; Wang, J. Analysis of user satisfaction of shared bicycles based on SEM. J. Ambient Intell. Humaniz. Comput. 2022, 13, 1587–1601. [Google Scholar] [CrossRef]
  69. Yáñez, M.F.; Raveau, S.; Ortúzar, J.D. Inclusion of latent variables in mixed logit models: Modelling and forecasting. Transp. Res. Part A Policy Pract. 2010, 44, 744–753. [Google Scholar] [CrossRef]
  70. Zhang, W.; Wang, K. Parking futures: Shared automated vehicles and parking demand reduction trajectories in Atlanta. Land Use Policy 2020, 91, 103963. [Google Scholar] [CrossRef]
  71. Huan, N.; Hess, S.; Yao, E. Understanding the effects of travel demand management on metro commuters’ behavioural loyalty: A hybrid choice modelling approach. Transportation 2022, 49, 343–372. [Google Scholar] [CrossRef]
  72. Hair, J.; Alamer, A. A Partial Least Squares Structural Equation Modeling (PLS-SEM) in second language and education research: Guidelines using an applied example. Res. Methods Appl. Linguist. 2022, 1, 100027. [Google Scholar] [CrossRef]
  73. Marsh, H.W.; Hocevar, D. Application of confirmatory factor analysis to the study of self-concept: First-and higher order factor models and their invariance across groups. Psychol. Bull. 1985, 97, 562–582. [Google Scholar] [CrossRef]
  74. Hair, J.F.; Ringle, C.M.; Sarstedt, M. PLS-SEM: Indeed a silver bullet. J. Mark. Theory Pract. 2011, 19, 139–152. [Google Scholar] [CrossRef]
  75. Ayaz, A.; Waheed, A.; Saleem, H.; Abid, M.M. Travelers’ attitude towards carpooling in Islamabad. J. Eng. Appl. Sci. 2021, 68, 1–24. [Google Scholar] [CrossRef]
  76. Namgung, M.; Akar, G. Role of gender and attitudes on public transportation use. Transp. Res. Rec. 2014, 2415, 136–144. [Google Scholar] [CrossRef]
  77. Sinha, S.; Swamy, H.M.S.; Modi, K. User Perceptions of Public Transport Service Quality. Transp. Res. Procedia. 2020, 48, 3310–3323. [Google Scholar] [CrossRef]
  78. Qin, H.; Gao, J.; Zhang, G.; Chen, Y.; Wu, S. Nested logit model formation to analyze airport parking behavior based on stated preference survey studies. J. Air Transp. Manag. 2017, 58, 164–175. [Google Scholar] [CrossRef]
  79. Kong, R.; Yao, Z. Comparison of travel characteristics between urban and rural residents in Haining city. Transp. Enterp. Manag. 2015, 30, 17–19. [Google Scholar]
  80. Gupta, P.D. What affects bicycle use in rural and semi-urban settlements: Evidence from selected regions in India. Case Stud. Transp. Policy 2022, 10, 625–636. [Google Scholar] [CrossRef]
Figure 1. The geographical location of the survey area.
Figure 1. The geographical location of the survey area.
Sustainability 15 11950 g001
Figure 2. Framework of the SEM–MNL integration model.
Figure 2. Framework of the SEM–MNL integration model.
Sustainability 15 11950 g002
Figure 3. Comparison of actual and acceptable levels of urban–rural bus services. (a) Comparison of walking time to the bus stop; (b) Comparison of waiting time; (c) Comparison of ticket price.
Figure 3. Comparison of actual and acceptable levels of urban–rural bus services. (a) Comparison of walking time to the bus stop; (b) Comparison of waiting time; (c) Comparison of ticket price.
Sustainability 15 11950 g003
Figure 4. Frequency of rural residents taking urban–rural buses per month.
Figure 4. Frequency of rural residents taking urban–rural buses per month.
Sustainability 15 11950 g004
Figure 5. Rural residents’ willingness to travel by urban–rural bus in the future.
Figure 5. Rural residents’ willingness to travel by urban–rural bus in the future.
Sustainability 15 11950 g005
Figure 6. First-order confirmatory factor analysis model.
Figure 6. First-order confirmatory factor analysis model.
Sustainability 15 11950 g006
Figure 7. Standardized path coefficient of the second-order structural equation model.
Figure 7. Standardized path coefficient of the second-order structural equation model.
Sustainability 15 11950 g007
Table 1. Summary of the literature review on the travel-mode-choice behavior of urban and rural residents.
Table 1. Summary of the literature review on the travel-mode-choice behavior of urban and rural residents.
AuthorsStudy AreaObservable VariablesLatent Variables
(Users’ Perceptions)
Analysis Method
Han et al.
[50]
UrbanPersonal attributes (Gender, Age, Occupation, Education level, Income, Car ownership); Travel attributes (Travel time, Trip distance, Waiting time, Travel frequency)Public transport service attributes (Convenience, Safety, Flexibility, Comfort, and Economy)SEM–NL
Si et al.
[51]
UrbanPersonal attributes (Gender, Age, Occupation, Education level, Income); Travel attributes (Trip purpose, Travel time, Trip distance, Waiting time, Place of departure) Taxi service attributes (Convenience, Reliability, Safety, Comfort, and Economy)SEM–Logit
Shah et al.
[62]
UrbanPersonal attributes (Gender, Age, Income, Vehicle ownership); Travel attributes (Distance to the mode, Trip purpose, Trip time, Trip cost) Public transport and private vehicles service attributes (Comfort and Convenience, Safety and Security, Service and Facilities, Attraction and quality of riding)Integrated Choice and Latent Variable
(ICLV) model
Chen and Li
[63]
UrbanPersonal attributes (Gender, Occupation, Vehicle ownership); Mode attributes (Travel cost)Public transport service attributes (Convenience, Personal safety, Modal comfort, Service environment, and Waiting feelings)Integrated SEM and Discrete Choice Model (SEM–DCM)
Akpan et al.
[57]
RuralPersonal attributes (Gender, Age, Income, Household size); Ownership and use of means of transportation; Travel days and cost of tripsMNL
Mattson et al.
[58]
RuralPersonal attributes (Gender, Age, Income); Travel attributes (Disability, Trip purpose, Party size); Mode attributes (Travel time, Travel price, Access distance, Service frequency)Mixed Logit Model
Ao et al.
[59]
RuralPersonal attributes (Gender, Age, Hukou type, Education level); Family attributes (Number of people working, Total household income, Number of vehicles); Travel attributes (Travel distance, Travel time, Departure time, Daily activity); Built environment (Building density, Road density, Distance to transit, Destination accessibility) Built environment perception and preferenceMNL
Table 2. Measurement scale of latent variables.
Table 2. Measurement scale of latent variables.
Latent VariablesMeasurement Variables
Comfort (BC)BC1Stable running status of bus
BC2Comfortable interior environment of bus
BC3Friendly service attitude of driver
BC4Follows the prescribed route
BC5Crowdedness of the carriage
Convenience (TC)TC1Convenience from home to bus stop
TC2Convenience from bus stop to destination
TC3Ease of transfer to other transportation in the city
TC4Accessibility to other villages
TC5Convenient to carry large items
TC6Time of service provision
Economy (TE)TE1Bus fares
TE2Bus fare discounts
Safety (TS)TS1Safety of vehicle operation
TS2Safety of the interior facilities
TS3Safety of bus stop
Reliability (TR)TR1Waiting time
TR2Punctuality rate
TR3Travel time on bus
Facility Completeness (CF)CF1Adequacy of bus stop facilities
CF2Adequacy of bus interior facilities
CF3Release and inquiry of bus operation information
Table 3. Demographic and traveling characteristics of respondents.
Table 3. Demographic and traveling characteristics of respondents.
Attribute CategoryVariableCategoryPercentageVariableCategoryPercentage
Socio-demographic
characteristics
GenderMale33.2%OccupationStudent14.5%
Female66.8%Peasant47.8%
Age<1811.1%Migrant worker12.1%
18–307.6%Commuter9.5%
30–5074.2%Others16.1%
≥507.1%Education
level
Junior high school and below59.7%
Average monthly income (CNY)<200053.3%Senior high school26.1%
2000–400031.7%Junior
College or above
14.2%
4000–600011.4%Private car ownershipNo41.5%
≥60003.6%Yes58.5%
Urban–rural travel characteristicsMonthly travel frequency038.6%Travel distance<20 km34.3%
1–241%20–30 km40.5%
3–412.8%≥30 km25.3%
>57.6%
Note: CNY 1000 ≈ USD 145.6.
Table 4. Reliability and validity test results of latent variables.
Table 4. Reliability and validity test results of latent variables.
VariablesItemsCronbach αCRAVEVariablesItemsCronbach αCRAVE
BCBC10.7870.7980.523TETE10.7280.7290.573
BC2TE2
BC3TSTS10.8350.8370.631
BC4TS2
BC5TS3
TCTC10.8280.8320.534TRTR10.7790.7830.547
TC2TR2
TC3TR3
TC4CFCF10.8180.8210.605
TC5CF2
TC6CF3
Table 5. Standardized path coefficients between latent variables.
Table 5. Standardized path coefficients between latent variables.
PathStandardized Path CoefficientPathStandardized Path Coefficient
BC ⟷ TS0.588TS ⟷ TE0.652
BC ⟷ TE0.617TS ⟷ TR0.734
BC ⟷ CF0.545TS ⟷ CF0.784
BC ⟷ TC0.669TS ⟷ TC0.609
BC ⟷ TR0.638TE ⟷ TR0.791
TR ⟷ CF0.819TE ⟷ CF0.734
TR ⟷ TC0.828TE ⟷ TC0.797
CF⟷ TC0.718
Table 6. Fit statistics for structural equation models.
Table 6. Fit statistics for structural equation models.
Evaluation Index χ 2 / d f RMSEACFITLISRMR
Fit Criteria<3<0.08>0.9>0.9<0.08
Test Results2.6750.0540.9440.9360.046
Table 7. Structural model path coefficient estimates and test values.
Table 7. Structural model path coefficient estimates and test values.
PathStandardized CoefficientsStandard Error
(S.E.)
Critical Ratio
(C.R.)
p-Value
OS ← Gender0.532---
OS ← Age−0.1490.042−4.394***
OS ← Income−0.1870.038−5.831***
OS ← Occupation−0.1370.024−3.814***
OS ← Education level−0.1150.042−2.968***
OS ← Travel mode0.0830.0382.4210.015 **
Notes: (1) *** means p < 0.01 and ** means p < 0.05; (2) OS: Overall Satisfaction.
Table 8. Measurement model path coefficient estimates and test values.
Table 8. Measurement model path coefficient estimates and test values.
PathStandardized CoefficientsStandard Error (S.E.)Critical Ratio (C.R.)p-Value
BC ← OS0.712---
TS ← OS0.7530.09811.024***
TE ← OS0.8710.10911.577***
TR ← OS0.9390.11611.918***
CF ← OS0.8350.13511.683***
TC ← OS0.9020.1149.716***
BC1 ← BC0.702---
BC2 ← BC0.7060.06814.791***
BC3 ← BC0.7940.06816.177***
BC4 ← BC0.5450.06911.698***
BC5 ← BC0.5590.07011.963***
TS1 ← TS0.750---
TS2 ← TS0.8150.06018.655***
TS3 ← TS0.8170.05718.702***
TE1 ← TE0.744---
TE2 ← TE0.7710.06815.974***
CF1 ← CF0.761---
CF2 ← CF0.8130.05019.054***
CF3 ← CF0.7600.04917.831***
TC1 ← TC0.527---
TC2 ← TC0.7150.10111.679***
TC3 ← TC0.7130.10311.598***
TC4 ← TC0.7110.10011.641***
TC5 ← TC0.6400.09811.010***
TC6 ← TC0.6670.09011.267***
TR1 ← TR0.720---
TR2 ← TR0.7850.06017.266***
TR3 ← TR0.7110.05515.757***
Note: *** means p value < 0.01.
Table 9. Estimation results of choice models with private car as the reference term.
Table 9. Estimation results of choice models with private car as the reference term.
VariableMNLSEM–MNL
Non-Motorized
Vehicles
Urban–Rural BusNon-Motorized VehiclesUrban–Rural Bus
CoefficientORCoefficientORCoefficientORCoefficientOR
Constant2.447 ***11.551.341 *3.8242.284 **9.815−1.0080.365
Gender0.0751.078−0.507 *0.6020.0741.077−0.520 *0.594
Age0.0661.068−0.0610.9400.0761.080−0.0060.994
Occupation−0.0740.9290.0051.005−0.0730.9290.0161.016
Income−0.1560.856−0.368 **1.445−0.1480.863−0.378 **1.460
Education level−0.494 **0.6100.0431.044−0.492 **0.6110.1371.147
Private car ownership−0.2690.764−3.384 ***0.034−0.2770.758−3.437 ***0.032
Travel distance−1.435 ***0.238−0.643 ***0.526−1.452 ***0.234−0.672 ***0.511
Travel frequency 0.631 ***1.8800.465 ***1.5920.624 ***1.8670.437 **1.548
Overall satisfaction ----0.0551.0570.690 **1.993
LR chi2365.26373.23
Prob > chi20.00000.0000
Pseudo R 2 R0.28900.2953
Log likelihood−449.354−445.369
Note: *** means p value < 0.01, ** means p value < 0.05, and * means p value < 0.1.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Cui, H.; Li, M.; Zhu, M.; Ma, X. Investigating the Impacts of Urban–Rural Bus Service Quality on Rural Residents’ Travel Choices Using an SEM–MNL Integration Model. Sustainability 2023, 15, 11950. https://doi.org/10.3390/su151511950

AMA Style

Cui H, Li M, Zhu M, Ma X. Investigating the Impacts of Urban–Rural Bus Service Quality on Rural Residents’ Travel Choices Using an SEM–MNL Integration Model. Sustainability. 2023; 15(15):11950. https://doi.org/10.3390/su151511950

Chicago/Turabian Style

Cui, Hongjun, Mingzhi Li, Minqing Zhu, and Xinwei Ma. 2023. "Investigating the Impacts of Urban–Rural Bus Service Quality on Rural Residents’ Travel Choices Using an SEM–MNL Integration Model" Sustainability 15, no. 15: 11950. https://doi.org/10.3390/su151511950

APA Style

Cui, H., Li, M., Zhu, M., & Ma, X. (2023). Investigating the Impacts of Urban–Rural Bus Service Quality on Rural Residents’ Travel Choices Using an SEM–MNL Integration Model. Sustainability, 15(15), 11950. https://doi.org/10.3390/su151511950

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop