Investigating the Multifaceted Impact of Urban Environment on Winter Green Travel in Cold Regions: An Empirical Study of Shenyang, China

Abstract

Green travel offers significant benefits, including reducing pollution and alleviating traffic congestion. Despite these advantages, green travel is less popular in cold regions, particularly during winter, due to comfort, convenience, and safety concerns. While climate factors are immutable, enhancing the urban environment can promote winter green travel. This paper examines the impact of urban environments on residents’ willingness to engage in green travel in cities located in cold regions during winter. Using data from Shenyang, China, a comprehensive structural equation model based on the stimulus–organism–response (S-O-R) theory was constructed, revealing the causal relationships and underlying structure between environmental factors and green travel willingness. The model demonstrates that social, built, natural, and travel environments collectively shape residents’ willingness to engage in green travel, with the social environment emerging as the most impactful factor. Additionally, this study identified two crucial mediating variables, travel perception evaluation and attitude towards green travel, which indirectly influence green travel willingness. This study also identifies nine critical factors—often underappreciated in traditional analyses—that should be prioritised in urban planning. These findings advance the understanding of green travel behaviour in winter cities by illustrating the complex interplay between environmental factors and individual attitudes while providing actionable guidance for fostering supportive social environments and strategically enhancing built and travel environments to promote green travel in cold climates.

1. Introduction

Urbanisation and economic growth have led to a rapid increase in urban road traffic. The proliferation of private automobiles has placed significant strain on urban transportation systems, exacerbating ecological and societal challenges such as air contamination, roadway congestion, and vehicular collisions. China’s primary carbon-emitting industries include energy supply, manufacturing, construction, and transportation, with transportation alone accounting for approximately 10% of the nation’s total carbon emissions. As per the Ministry of Public Security, China had 435 million motor vehicles by 2023, with 336 million of them being automobiles. Additionally, 94 cities in the country have over 1 million automobiles [1]. Therefore, regulating car travel is essential for reducing carbon emissions from transportation.

Since the 1970s and 1980s, the notion of sustainable development has progressively become prominent, including distinct meanings in numerous domains. Green transportation is an essential component for advancing the transportation industry within the sustainable development framework. Currently, green transportation is a popular and innovative approach to facilitate the transformation of urban transportation growth. Travel refers to the need for people to move from one place to another for a certain purpose or activity [2]. Green travel, derived from green transportation, categorises travel behaviour according to the mode of transportation used. The focus is mostly on the notion of human-orientated behaviour, which refers to the deliberate actions taken by individuals to mitigate or address environmental issues through their transportation choices [3]. The China Council for International Cooperation on Environment and Development (CICED) Policy Research Group defines green travel as a form of transportation that can substitute car travel and is appealing to various social strata groups. It is effective in mitigating urban traffic congestion and reducing air pollution. Green travel occurs continuously and has the potential to enhance the quality of the urban air environment, alleviate road congestion, and support the growth of urban public transportation systems. Furthermore, it can reduce the daily travel expenses of residents, promoting their overall health benefits. Hence, it is crucial to study green travel behaviour and investigate strategies to encourage such travel in order to preserve the urban environment and foster a healthy lifestyle.

A multitude of experts have performed extensive research on green travel behaviour, encompassing both its behavioural composition and interpretation. When discussing behavioural composition, green travel behaviour typically emphasises low-consumption and low-emission modes of transportation such as walking, cycling, using public buses, and taking the underground. Researchers commonly use indicators such as the origin points, modes of travel, routes, destinations, time, and costs to describe the characteristics of green travel. Regarding the interpretation of green travel behaviour, there are variations in the study of behaviour in the fields of psychology and behavioural sciences. As a result, theories have been developed from different perspectives to explain behaviour. Numerous studies have been conducted in relation to these theories. The theory of planned behaviour (TPB), developed by Professor Ajzen, has gained significant popularity in explaining human conduct because of its precise conceptual framework, standardised assessment methods, and strong explanatory capacity. This theory posits that behavioural willingness is influenced by three factors: personal attitudes, subjective norms, and perceived behavioural control, which in turn directly impact behaviour [4]. Incorporating the traditional TPB theory, researcher Xuemei Fu introduced the concept of affective-cognitive consistency of attitudes as a moderator to examine the green travel behaviour of residents. The study revealed that the more pronounced the affective-cognitive consistency of attitudes, the greater the impact of attitudes on the willingness to engage in green travel [5]. Unlike the theory of planned behaviour (TPB), the theory of interpersonal behaviour (TIB) developed by Harry integrates the influential factors of habit and emotion in explaining behaviours [6]. On the other hand, the value–belief–norm theory (VBN) proposed by Stern elucidates the relationship between values, beliefs, and norms in understanding the environment [7]. Hiratsuka et al. conducted a study in Japan to examine the validity of the VBN theory and found evidence supporting its ability to explain the decrease in car usage [8]. Lewin’s intrinsic factor-external environment model posits that individual behaviours result from the interplay between the individual and their surroundings [9], while Guagnano et al. introduced the A-B-C model (attitude–behaviour–context) for forecasting environmental behaviour [10]. Song et al. applied the Stimulus–Organism–Response (S-O-R) theory to explore how environmental emotions mediate travel behaviours [11]. Kaiser, Chen, Tongleta, and other scholars have highlighted that behavioural willingness refers to an individual’s psychological inclination and motivational drive prior to engaging in a specific behaviour. It is considered a crucial prerequisite for the actualisation of behaviours, with behavioural willingness being the primary determinant of behaviour occurrence. Furthermore, there exists a significant and positive correlation between the willingness to green travel and actual green travel behaviour [12,13,14]. Therefore, the above studies often interpret green travel behaviour by examining behavioural willingness. It can be said that the research of green travel behaviour has transitioned from a singular understanding to various explanations and from focusing solely on individual factors to considering both internal and external factors.

Urban space, as an external context for human living, has a direct or indirect impact on human activities. Pikora and other scholars have examined the correlation between the environment and travel behaviours, such as walking and cycling, specifically in terms of the functionality, safety, aesthetics, and destinations of urban environments [15]. Various urban development patterns can influence travel patterns, notably transit-orientated development (TOD), which reduces car dependency for non-work activities such as shopping by locating workplaces within TOD areas and promoting slower transport modes like walking [16]. Researchers Rodríguez and Joo conducted a study to examine how pavements and the frequency of bus service influence the travel mode choices of inhabitants [17]. Cervero highlighted that the arrangement of “compact neighbourhoods with a well-connected road network” promotes walking and cycling [18]. Additionally, Schwanen and other researchers discovered that the thoughtful design of bus stops and comprehensive transportation infrastructure greatly influence residents’ choice of travel mode, favouring public transportation options [19]. In conclusion, research has demonstrated that the urban environment has a substantial influence on the travel behaviour of individuals. Hence, urban planning is regarded as a pivotal strategy to encourage citizens to engage in walking, cycling, and utilising public transportation while discouraging reliance on motorised transportation [20].

Research on the relationships between environmental factors and green travel willingness has traditionally focused on several key areas, such as the accessibility and quality of urban green spaces [21], socio-economic influences [22], individual attitudes [23], and urban design [24]. However, the issue is complex and involves the interplay of multiple factors, suggesting that a holistic approach is necessary. This research aims to integrate these diverse aspects, offering a comprehensive analysis that can inform effective urban planning strategies to promote green travel, particularly in challenging environments like winter climates.

In recent years, urban planning research related to green travel has expanded, increasingly focusing on the psychological and attitudinal factors that influence transportation mode choices. This body of work has also begun to integrate various urban environmental elements to better understand and enhance the green travel experience [25]. Eva Heinen and colleagues, for instance, employed factor analysis to demonstrate the significant role of psychological factors in promoting cycling behaviour [26]. De Vos’s study utilised structural equation modelling to analyse the behaviour of individuals engaging in walking and cycling in Ghent, Belgium, revealing that these active modes of transportation are associated with higher levels of satisfaction [27]. Similarly, Kamargianni and colleagues found that satisfaction plays a crucial role in shaping travel preferences [28]. In Lithuania, Audronė Minelgaitė and her team highlighted how long-term changes in pro-environmental behaviour, driven by environmental education and public initiatives, contribute positively to environmental awareness and conduct [29]. Additionally, Bowei Zhong and colleagues explored the effects of travel time costs and feedback types on green travel behaviour, finding that environmental and health feedback can effectively encourage green travel when time costs are low [30]. Collectively, these studies underscore the importance of considering individual psychological factors, societal contexts, and urban infrastructure when understanding and promoting green travel.

Despite significant advancements in science and technology, individuals are still unable to fully mitigate the adverse effects of climatic differences due to various factors, including socio-economic and geographical conditions, presenting critical challenges for urban planners. Statistics indicate that about 600 million individuals reside in frigid urban areas worldwide. According to Lu Xinchao and other researchers, winter cities are a distinct category of cities found in the northern hemisphere. These cities are characterised by their severe winters and harsh environment, with an average temperature in January below −18 °C [31]. Construction circumstances in these regions are considerably more challenging in comparison to cities with milder climates, shorter building seasons, and extensive permafrost, resulting in increased construction challenges and decreased prospects for extensive development in specific areas.

Winter cities face unique challenges due to harsh climatic conditions, which complicate urban planning and the promotion of green travel. These challenges include physical barriers such as snow and ice, as well as the psychological impacts of long winters, leading to decreased public space usage and increased social isolation [32]. Extreme cold significantly affects accessibility and livability, necessitating urban design that supports healthy lifestyles and social interactions [33]. Effective climate-sensitive urban planning must integrate local topography and microclimatic conditions to enhance resilience and infrastructure efficiency [34]. Infrastructure tailored for cold climates, such as well-designed subway stations, is essential to ensure safety, comfort, and increased public transit usage [35]. However, green travel initiatives like public transport and cycling are often hindered by prolonged cold weather, leading to low resident satisfaction, underscoring the need for infrastructure updates that meet seasonal demands [36]. Despite extensive research conducted since the 1980s by the Association of Winter Cities and various international campaigns, numerous thematic issues related to the impact of cold climates on urban development and design remain that require further in-depth discussion.

Promoting green transportation in winter cities poses greater challenges compared to other regions. However, winter cities have a more pressing need to reduce carbon emissions from transportation, as the combustion of fossil fuels for winter heating already places significant strain on the environment. Green travel is less popular in cold regions, particularly during winter, due to issues of comfort, convenience, and safety. While climate factors in these areas are immutable, enhancing the urban environment can facilitate and promote winter green travel. Despite this, existing research has yet to establish a comprehensive framework.

Using data from Shenyang, China, this study employs structural equation modelling (SEM) to investigate the impact of urban environments on residents’ willingness to engage in green travel in cities located in cold regions during winter. The study seeks to answer several key research questions:

• What are the potential influencing factors on residents’ willingness to engage in green travel in cold-region cities during winter from an urban design perspective?
• How can these potential factors be categorised into different dimensions that influence the willingness to engage in green travel?
• What are the pathways through which these different dimensions impact residents’ willingness to engage in green travel, and how do they interact with each other, both directly and indirectly?
• How do the influencing factors related to the urban environment exert a multifaceted on green travel willingness within this structured relationship?

This study innovatively introduces the stimulus–organism–response (S-O-R) theory to construct a structural equation model, examining the comprehensive interrelationships among the urban environment, residents’ attitudes and perceptions, and green travel willingness. By uncovering the influential patterns of travel willingness, this research provides theoretical insights and practical guidance for policymakers and urban planners to create urban environments that effectively encourage green travel behaviour among residents of winter cities during the winter season.

The subsequent sections of the paper are organised as follows: Section 2 centres on our research technique, including the identification of potential influencing factors and the process of constructing the theoretical model. Additionally, it covers the construction, distribution, and processing of the questionnaire. Section 3 presents the factor analysis results and structural equation modelling results. Section 4 discusses the findings of the investigation, while Section 5 provides a concise overview of the study.

2. Materials and Methods

The purpose of this study is to examine the relationships between urban environmental factors and green travel behaviours in winter climates. While correlation analysis and regression analysis are commonly used methods for such research, they often fall short in handling the complexity of multiple interacting variables [37]. Structural equation modelling (SEM) offers significant advantages over these methods, particularly when dealing with complex relationships among variables. SEM’s ability to simultaneously model multiple dependent and independent variables, including latent constructs, allows for a comprehensive examination of direct, indirect, and mediating effects within a single framework [38,39]. This is more efficient than regression analysis, where each mediator requires a separate model. Moreover, SEM provides essential fit indices, such as chi-square, RMSEA, and CFI, ensuring that theoretical models align well with observed data, which is crucial for theory testing and refinement [40]. SEM’s capacity to incorporate latent variables and test complex causal relationships makes it particularly powerful for developing and validating theoretical frameworks [41].

In this study, SEM is particularly suited to exploring the complex relationships influencing green travel behaviors. It allows for modelling factors like priming and social influences, which activate latent environmental attitudes and shape sustainable travel practices, especially in winter contexts [42]. Additionally, SEM captures the mediation and moderation effects of behaviour change processes, such as consciousness-raising and self-reevaluation, crucial for fostering green travel behaviours in challenging conditions [43].

Furthermore, integrating the stimulus–organism–response (S-O-R) theory into SEM provides a robust framework for analysing how environmental factors act as stimuli that shape cognitive and emotional responses, ultimately influencing sustainable travel behaviours [44,45,46]. This approach is particularly suitable to understanding green travel behaviours in winter climates, where the interplay of environmental stimuli and behavioural responses is complex and multifaceted.

2.1. Theoretical Model

The S-O-R theory posits that external environmental cues influence users’ reaction behaviour through subjective motivation and internal consciousness [47]. This study examines the willingness to green travel as a response to the environmental conditions. Individuals with different characteristics may exhibit varying responses within the same environmental context, and conversely, an individual may have distinct responses in different environmental contexts. The process of individuals responding to the external environment is not immediate but rather involves personal judgement followed by a response. Judgement refers to an individual’s perception of the environment and their attitude towards it. The theoretical model depicted in the figure below (Figure 1) can be developed as a foundation framework for investigating the various interconnections of green travel.

2.2. Potential Influencing Factors

The influencing factors of green travel vary in winter cities due to their unique weather characteristics. When studying winter cities, it is important to focus on the weather, climate conditions, and transportation aspects of these places. Chengxi Liu conducted a study using trip data of Swedish citizens and weather data modelling, revealing that weather factors have the most substantial influence among the various factors that affect travel [48]. In a study conducted by Gössling et al. in Zanzibar, Tanzania, it was found that specific meteorological factors, such as rainfall, storms, and humidity, have a detrimental impact on travel choices [49]. Similarly, in a study conducted by Meng et al., the researchers investigated the influence of weather conditions and weather forecasts on bicycle travel behaviour in Singapore, highlighting that cyclists were more inclined to alter their mode of transportation based on pre-rain weather forecast information [50]. A study conducted by Ahmed et al. analysed the travel patterns of commuter cyclists in relation to variations in weather conditions, emphasising the influence of weather on the cyclists’ daily decisions on cycling [51]. Consequently, we incorporated the inherent meteorological conditions of the city as a factor of influence. Urban areas in low-temperature circumstances necessitate additional cold protection measures. Therefore, we consider the spatial environment of waiting areas and connecting pathways as significant factors that determine the need for such measures. Mark J. Koetse conducted a study on the correlation between climate change, weather conditions, and transportation. Through a series of cases, he confirmed that climate change resulting from global warming and extreme weather events, such as high winds and snowstorms, had detrimental effects on both passenger and freight transport [52]. Thus, fluctuations in transportation efficiency due to seasonal changes in winter cities might also influence the travel decisions made by inhabitants.

As shown in

Table A1. Supporting materials of potential influencing factors.

Factor		Support	Source
Travel perception evaluation	Security	1. The sense of security and perceived accessibility of public transport considered to be highly related to the quality of public transport services [60].	1. Friman, M.; Lättman, K.; Olsson, L.E.
		2. The increase of safety is related to the increase or greater decrease of driving volume [61].	2. Ha, J.; Lee, S.; Ko, J.
	Comfort	Sinha puts forward the factors that affect the perception of public transport users’ service quality, including the attributes of cost, time, trip, comfortable and relaxed quality attributes and the availability of information. The arrival and departure time, waiting time and transfer convenience are equally important to the overall satisfaction research [62].	Sinha S, Shivanand Swamy HM, Modi K.
	Physical cost perception	Scholars Rotter and Levenson divide the source of psychological control into internal control source and external control source, and point out that individuals with internal control source tend to think that their behaviour affects the behaviour results in specific situations, while individuals with external control source tend to think that their behaviour ability is weak and the behaviour results in specific situations are beyond their control [63,64].	1. Thøgersen, J.; Møller, B.
			2. Rotter, J.
	Money cost perception	Congestion charging and parking charging can effectively urge travelers to change their travel modes, and congestion charging may be more effective in alleviating traffic congestion during peak hours [65];	Shiftan Y, Golani A.
	Perception of built environment	Urban construction is related to planning and residents’ health behavior [20].	Giles-Corti, B.; Vernez-Moudon, A. et al.
	loyalty	Customer loyalty is considered to lead to repeated purchases, positive attitude, intention to continue to join and intention to actively recommend [66,67].	1. Zins A H.
			2. Davis-Sramek, B.; Mentzer, J.T.; Stank, T.P.
	Satisfaction degree of operating environment	Passenger satisfaction is a key driver of the positive process of recommending to others [68].	Ingvardson, J.B.; Nielsen, O.A.
	Satisfaction of decision
	Overall satisfaction of travel
Attitude towards Green travel	Personal physiological significance	One of the main motivations for residents to participate in active commuting is that walking or cycling is good for their health [69].	Anable, J., Gatersleben, B.
	Personal social significance	The opportunity to work while traveling may be important to some people, while social interaction may be important to others [70].	Ettema, D.; Friman, M.; Gärling, T.
	Personal spiritual significance	European scholars Thøgersen and Tanner pointed out more clearly that in some rich Nordic countries, citizens’ environmental behaviour largely stems from their own moral responsibility [71,72].	1. Thøgersen, J.
			2. Tanner, C
	Green traffic cognition	The importance of expected emotion and emotion as the intrinsic motivation of sustainable action. They are both the result of pro-environmental behavior and the premise of behavior [73].	Brosch, T.
	Health condition	Roger Mackett found that the elderly in Britain have a stronger willingness to participate in society, but it is difficult for the elderly to travel easily because of travel obstacles [74].	Mackett R.
	Living habits	1. Take automobile drivers as an example to explore whether travel behaviour is habitual decision-making or conscious choice. Many car drivers habitually choose travel modes, but considering the price-quality relationship of various choices at present, their final choice is consistent with their informed preferences [63].	1. Thøgersen, J.; Møller, B.
		2. Previous behaviour is a powerful regression factor for future behaviour [75].	2. Garling, T., Axhausen, K.W.
Urban environment	Green traffic construction	Davis et al. and O ‘Donovan & McCarthy’s research also show that consumers’ green consumption behaviour is restricted by the availability of green products, which mainly refers to whether the types of green products are complete or not [76,77].	1. Davies, A., Titterington, A.J., Cochrane, C.
	Identification degree of urban construction		2. O‘Donovan, P., McCarthy, M.
	Relevance of urban construction	The built environment is defined as five core dimensions, or “5d”: density, diversity, design, transit distance and destination accessibility [18].	Cervero, R.
	Road network density	Hedel and Vance studied the influence of street density on per capita VKT [78].	Vance, C., Hedel, R.
	Supporting facilities	The choice of travel mode is related to the design of public transportation stations and the completeness of other supporting facilities, that is, reasonable bus station design and complete transportation facilities are conducive to residents’ choice of public transportation [19].	Schwanen T, Dieleman F M, Dijst M.
	Policy encouragement	Individuals who own subsidized housing may have a lower demand for owning cars because the expected use of cars is low [59].	Huang, A.; Levinson, D.
	Social atmosphere	Heath and Gifford (2002) applied for TPB to explain the use of public transport by college students before and after the implementation of the bus pass program. It is found that two subjective norms, that is, what important people do, and descriptive norms, that is, what most people do, significantly explain the passenger flow [79].	Heath, Y.; Gifford, R.
	General mood of society
	Weather conditions	The weather will affect the choice of traveler’s travel mode, and the influence of weather on non-commuters is much greater than that on commuters [48].	Liu, C.; Susilo, Y.O.; Karlström, A.
	Air environment	When individuals choose the travel mode, they will consider the surrounding air pollution and greening degree [15].	Pikora T, Giles-Corti B, Bull F, et al.
	Natural greening
	Topography	Mao Hai-ji, a scholar in China, has analyzed the different factors that affect the travel modes of residents in mountain cities and plain cities, and thinks that the terrain particularity of mountain cities is one of the main factors [80].	Mao Hai-ji
	Seasonal climate	Seasonal changes in climate will have a great impact on the transportation sector [81].	Jaroszweski, D.; Chapman, L.; Petts, J.
Travel environment	Path support	Cervero and Kockelman studied the influence of sidewalk width on the mileage of each household. Fan believes that the length of sidewalk is an effective factor affecting the per capita mileage [82,83].	1. Cervero, R.; Kockelman, K.
			2. Fan, Y.
	Guidance of path	Swanson & Lewis’s research points out that the quality of environmentally friendly products, such as ease of use [84].	Swanson, R.B.; Lewis, C.E.
	Accessibility of starting and ending points	Accessibility based on destination distance is related to more walking [85,86,87].	1. Ewing, R.
			2. Humpel N, Owen N, Leslie E.
			3. Mcmillan, T.
	The convenience of site connection	According to the review by Ding C, Wang D, Liu C, et al., in terms of foreign research, Boarnet and others’ research on Baltimore found that if it is close to the bus stop, residents’ willingness to buy private cars will decrease, which will drive the use of bus commuting [88]	Ding C, Wang D, Liu C, et al.
	Operational efficiency	Road congestion and its duration will affect commuters’ choice of travel modes [89].	Palma, A.d.; Rochat, D.
	Operational technical support	This paper studies the influence of Smart traveler information service system in Boston, USA on people’s travel behaviour. It is considered that advanced traveler information system (ATISs) will significantly affect people’s travel path choice, and the design of system service will also affect people’s behaviour of using Smart traveller [90].	Polydoropoulou, A.; Ben-Akiva, M.E.

in Appendix A, we compiled a comprehensive list of 34 observed variables by summarising prior studies and incorporating the unique characteristics of winter cities. While we did not introduce entirely new impact factors exclusively for cold cities, we have adapted and contextualised existing factors based on relevant literature and our disciplinary expertise. For example, factors such as natural greening, seasonal climate, path support, and the convenience of site connection, although not unique to cold cities, exhibit distinctive characteristics or attributes in these environments. This makes them particularly relevant and significant in the context of cold-region cities. There are 9 variables for evaluating travel perception, 6 variables for measuring attitude towards green travels, and 19 variables for assessing the external environment. These external environment variables consist of 13 variables related to the macro-urban environment and 6 variables related to the micro travel environment.

2.3. Questionnaire Design

The influence of subjective perceptions on subjective choice outcomes is well-grounded in established theories of behavioural science and cognitive psychology. For instance, the theory of planned behaviour (TPB) posits that individual attitudes and perceived behavioural control, both of which are subjective perceptions, significantly influence behavioural intentions and actions [4]. Similarly, expectancy theory highlights the role of individual expectations and perceived outcomes in decision-making processes. Therefore, analysing the impact of subjective perceptions on subjective choices is both theoretically relevant and logically coherent [53].

In this study, we focus on subjective perceptions as key variables influencing residents’ willingness to engage in green travel. This approach is based on the understanding that individuals’ subjective perceptions of their environment and the potential outcomes of their choices are critical determinants of their behaviour. By examining these perceptions, we aim to gain a deeper understanding of the factors that drive green travel decisions, particularly in the unique context of cold-region cities.

We designed a subjective questionnaire based on the listed observed variables to assess the evaluation of travel perception, attitude towards green travel, perception of the overall urban environment, and the perception of specific travel environments. Respondents were asked to rate a list of statements using a five-point Likert scale ranging from “Strongly Disagree” (1) to “Strongly Agree” (5) (

Table A2. Five-level questionnaire.

Code	Question	Grade
PG1	I feel safe whenever I travel.	1	2	3	4	5
PG2	My daily travel environment is comfortable.	1	2	3	4	5
PG3	I feel physically and mentally exhausted during my daily trip.	1	2	3	4	5
PG4	My daily travel expenses are reasonable.	1	2	3	4	5
PG5	The traffic environment in Shenyang is good.	1	2	3	4	5
PJ1	Even if there are other options, I am still willing to stick to the current mode of travel.	1	2	3	4	5
PJ2	I am satisfied with the traffic management in Shenyang.	1	2	3	4	5
PJ3	I am satisfied with my daily travel mode now.	1	2	3	4	5
PJ4	Traveling in Shenyang is something that makes me satisfied.	1	2	3	4	5
TC1	Public transportation, such as subway, can provide guarantee for my trip in winter.	1	2	3	4	5
TC2	Cycling, walking and public transportation can provide me with opportunities to exercise.	1	2	3	4	5
TC3	Cycling, walking and public transportation can provide me with opportunities to get in touch with others.	1	2	3	4	5
TC4	Riding, walking and public transportation can bring me a sense of accomplishment.	1	2	3	4	5
TC5	My physical mobility is enough for me to travel by public transport.	1	2	3	4	5
TC6	Traveling by public transport is an important part of my life.	1	2	3	4	5
HJ1	The number of bus/subway stations in Shenyang is large and covers a wide area.	1	2	3	4	5
HJ2	Every area of the city in Shenyang has different scenery.	1	2	3	4	5
HJ3	In Shenyang, the traffic links between different areas are close.	1	2	3	4	5
HJ4	I feel that there are many street intersections and it is convenient to turn.	1	2	3	4	5
HJ5	When I travel, I can meet all kinds of convenient facilities (public toilets, seats, etc.)	1	2	3	4	5
HS1	The government has promoted various policies to encourage green travel.	1	2	3	4	5
HS2	Most people in Shenyang hope to make efforts for the urban environment.	1	2	3	4	5
HS3	Low-carbon life is popular in Shenyang recently.	1	2	3	4	5
HZ1	The weather in Shenyang has always been good and it is convenient to travel.	1	2	3	4	5
HZ2	Shenyang has an excellent air environment suitable for travel.	1	2	3	4	5
HZ3	Shenyang has a flat terrain and convenient travel.	1	2	3	4	5
HZ4	Shenyang’s natural environment is suitable for outdoor recreation.	1	2	3	4	5
HZ5	Every season, the traffic is smooth and the vehicles pass in an orderly way.	1	2	3	4	5
HW1	Most of the pedestrian walkways where I go are flat and spacious.	1	2	3	4	5
HW2	The traffic signs during my trip are clear and easy to read.	1	2	3	4	5
HW3	In my daily life, there are bus/subway stations near my destination.	1	2	3	4	5
HW4	When I wait at the station, the waiting space and waiting time are reasonable.	1	2	3	4	5
HW5	I feel convenient when I change to public transportation, and I won’t walk for long.	1	2	3	4	5
HW6	Real-time bus arrival information can be easily found.	1	2	3	4	5
Y1	I am willing to choose public transportation, walking and cycling and other green modes of travel.	1	2	3	4	5
Y2	I plan to do more green trips in the future.	1	2	3	4	5
Y3	I will recommend the green mode of travel to the people around me.	1	2	3	4	5

in Appendix A). Furthermore, we measure the willingness to green travel with ’Intention to travel green’, ’Planning to travel green’, and ’Recommending to travel green’ as the observed variables. Additionally, the questionnaire included a series of inquiries about demographic information such as gender, age, and household composition. A committee of specialists was established to thoroughly review the substance of the questionnaire via multiple iterations. Prior to administering the formal survey, a preliminary survey was undertaken to establish the reliability of the final questionnaire. We administered the initial survey questionnaires to individuals of various age groups in Shenyang City. Subsequently, we made revisions to the questions based on the feedback, resulting in the final version of the formal questionnaire.

2.4. Data Collection

We selected Shenyang, a representative city known for its frigid climate as the subject of our study. Shenyang, being a prominent city in the northeast region of China, is experiencing an influx of population from nearby cities, resulting in increased transportation demands. In the city core of Shenyang, the primary means of transportation include buses as the main method of transit, a metro network with five lines serving as the central transportation framework, and taxis providing additional support to meet the travel needs of city inhabitants. Nevertheless, challenges such as the suboptimal design of public transport stations, inadequate winter road safety measures, and unfavourable winter urban landscape all impede the progress of green travel.

As shown in Figure 2, we selected seven locations within the central city of Shenyang based on high pedestrian flow and diverse population composition. Our research started on 23 November and ended on 26 December, a total of seven times. On average, each survey takes five hours. These locations are situated within different municipal districts: Heping District, Dadong District, Huanggu District, Tiexi District, Shenhe District, Sujiatun District, and Hunnan District. The chosen sites encompass a variety of busy areas such as shopping malls, parks, and markets. It is important to note that the data collection was conducted in specific locations within the central city of Shenyang and was not randomised across the entire population. As such, the data may not fully represent the broader population of Shenyang and should be viewed as illustrative rather than definitive.

To ensure the statistical validity of our study, we determined the appropriate sample size using the Raosoft sample size calculator (Raosoft, Inc.: Seattle, WA, USA) (http://www.raosoft.com/samplesize.html, accessed on 16 July 2024). Given the population size of 7,620,000 for the urban area of Shenyang and the margin of error of 5%, the recommended sample size is 271 for a 90% confidence level and 384 for a 95% confidence level. After eliminating invalid questions, a total of 306 valid questionnaires were gathered, which exceeds the recommended sample size for a 90% confidence level. Although this does not fully meet the requirement for a 95% confidence level, the sample size is still statistically sound and supports the robustness of our findings within the scope of this research.

2.5. Data Analysis Method

Our primary data analysis method is structural equation modelling (SEM), which is effective in evaluating the connection between endogenous and exogenous latent variables [54]. SEM also allows for the analysis of relationships between multiple independent and dependent variables, making it a crucial analytical tool in social science research. SEM comprises two components: the measurement model and the structural model. The measurement model represents the causal relationship between latent variables and their measures and is frequently employed to ascertain the relative importance of each indicator. The structural model is sensitive to the causal relationships between the variables. We utilised exploratory factor analysis to ascertain the number of factors associated with the observed variables. Subsequently, we developed the chained multiple mediation effect model, drawing upon the S-O-R theory. The model’s validity is assessed by confirmatory factor analysis and further refined and processed to ultimately determine the influence link between each variable.

2.6. Addressing Potential Endogeneity

In addressing the potential endogeneity of subjective perceptions influencing green travel willingness, we recognise the risks posed by factors such as omitted variables, measurement error, or reverse causality. To mitigate these concerns and ensure the robustness of our findings, we conducted exclusion tests, systematically removing variables to verify the stability of the results [55]. Additionally, within the SEM framework, we accounted for measurement errors by explicitly modelling latent variables, a process that helps to alleviate some endogeneity issues. Where feasible, we also explored the application of instrumental variables (IVs) to reinforce the causal interpretation of our model [56], although this approach is inherently more challenging within SEM. These methodological steps collectively provide a more accurate estimation of the relationships between subjective perceptions and green travel willingness, thereby enhancing the reliability and validity of our conclusions.

3. Results

3.1. Sample Descriptive Statistical Analyze Results

As shown in

Table 1. Demographic and Household Composition of Survey Respondents.

Category		Number of Respondents	Proportion (%)
Gender	male	160	52.30
	female	146	47.70
Age	18–25 years old	68	22.20
	26–35 years old	57	18.60
	36–45 years old	37	12.10
	46–60 years old	38	12.40
	Over 60 years old	106	34.60
Household Composition	live alone	60	19.60
	live with a partner	113	36.90
	a family of three or four with children	98	32.00
	three generations under one roof	35	11.40

, out of the 306 valid questionnaires, 52.3% of respondents were male and 47.7% were female. Respondents spanned all age groups from 18 to over 60, with the largest proportion being over 60 years old. Only 19.6% of respondents lived alone, while the majority lived with partners or in family units, including 11.4% residing in multi-generational households. As of the end of 2023, the registered population of Shenyang aged 60 and above exceeded 2.24 million, accounting for 29.4% of the total population. In our study, the proportion of elderly individuals in the sample is 34.6%, which is slightly higher than the actual population proportion. This over-representation is consistent with the ongoing trend of population ageing in Northeast China and reflects the demographic reality of Shenyang, where an ageing population is coupled with a decline in the working-age population.

To provide a more granular understanding of the differences in green travel willingness among various demographic groups, we conducted a differentiated analysis using independent samples t-tests and one-way ANOVA, focusing on gender, age, and household composition as independent variables potentially affecting the dependent variable of green travel willingness.

Given that gender consists of two categories (male and female), we utilised an independent samples t-test to assess differences. A significant difference at the p < 0.05 level would indicate a substantial effect of this variable on green travel willingness. As depicted in

Table A5. Differential analysis of willingness to green travel of different gender groups.

Independent Variables	Gender (Mean ± Standard Deviation)		p	T
Willingness to green travel	Male (n = 160)	Female (n = 146)	0.588	−0.542
	−0.030 ± 1.001	0.032 ± 1.001

in Appendix C, the p-value for gender was 0.588, which exceeds the threshold for significance (p > 0.05), suggesting no notable disparity in green travel willingness between males and females.

For age and household composition, which encompass multiple categories, a one-way ANOVA was applied to determine if there were significant differences. A significant result would be indicated by a p-value less than 0.05.

Table A6. Differential analysis of willingness to green travel of different age groups.

	Number of Cases	Average Value	Standard Deviation	F	p	Multiple Comparisons
18–25 years old	68	0.2605489	0.949845	2.976	0.020	18–25 years old > Over 60 years old, 26–35 years old > Over 60 years old
26–35 years old	57	0.1644074	0.889289
36–45 years old	37	0.0092644	0.955452
46–60 years old	38	−0.125389	0.88716
Over 60 years old	106	−0.213835	1.09838
(grand) total	306	0	1

The F-value is used in ANOVA to estimate the ratio of variances, while the p-value indicates the probability of observing such a ratio if there were no effect, helping to determine statistical significance.

in Appendix C illustrates that the p-value for age was 0.02, which is below the significance threshold (p < 0.05), indicating a significant effect of age on green travel intention. Post-hoc tests using the Least Significant Difference (LSD) method revealed that the 18–25 and 26–35 age groups exhibit significantly higher green travel willingness compared to the group aged 60 and above. This suggests a pronounced inclination towards green travel among younger age brackets, possibly due to greater environmental consciousness and adaptability to sustainable practices. Conversely, for household composition, the p-value obtained was 0.393, which is greater than 0.05, as shown in

Table A7. Differential analysis of willingness to green travel of different household composition groups.

	Number of Cases	Average Value	Standard Deviation	F	p
Live alone	60	0.1153032	0.990965	1.028	0.393
Live with a partner	113	−0.113098	1.078385
A family of three or four with children	98	0.1097732	0.941949
Three generations under one roof	35	−0.1398821	0.885188
(grand) total	306	0	1

in Appendix C. This indicates no significant differences in green travel intention across various family structures, implying that household composition may not be a strong determinant of one’s preference for green travel modes.

3.2. Explanation of Factor Analysis Results

Prior to analysing the data, reliability and validity tests were performed to ascertain the soundness of the scale structure and the extent to which the indicators in the questionnaire accurately represented the underlying factors.

Initially, we utilised the SPSS 26.0 program to conduct the KMO and Bartlett’s test in order to assess the suitability of the data for factor analysis. The results (

Table 2. KMO and Bartlett test.

KMO Sampling Suitability Measure		0.925
Bartlett Test of Sphericity	Approximate chi-square	6438.114
	Degrees of freedom	666
	Significance	0 (p < 0.001)

) indicate that the questionnaire has a strong overall KMO value of 0.925, suggesting a strong correlation between variables and suitability for factor analysis.

3.2.1. Exploratory Factor Analysis

We conducted an exploratory factor analysis (EFA) to evaluate the number of factors and the structural validity of the questionnaire. The analysis of the total variance explained showed that seven factors with eigenvalues greater than 1 account for 67.275% of the variance among the 37 items (

Table A3. Total variance explained.

Component	Initial Eigenvalues			Extracting Sum of Square Loadings			Rotation Sums of Squared loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	11.35	30.677	30.677	11.35	30.677	30.677	6.052	16.357	16.357
2	3.687	9.964	40.641	3.687	9.964	40.641	4.072	11.006	27.363
3	2.947	7.964	48.604	2.947	7.964	48.604	3.748	10.13	37.494
4	2.36	6.378	54.982	2.36	6.378	54.982	3.419	9.241	46.735
5	1.822	4.924	59.906	1.822	4.924	59.906	3.362	9.086	55.821
6	1.466	3.962	63.868	1.466	3.962	63.868	2.13	5.757	61.579
7	1.261	3.407	67.275	1.261	3.407	67.275	2.108	5.696	67.275
8	0.677	1.83	69.105
9	0.634	1.713	70.818
10	0.61	1.648	72.466
11	0.591	1.597	74.063
12	0.56	1.514	75.578
13	0.528	1.427	77.005
14	0.504	1.363	78.368
15	0.499	1.348	79.716
16	0.484	1.307	81.023
17	0.467	1.263	82.286
18	0.463	1.252	83.538
19	0.447	1.209	84.746
20	0.435	1.175	85.922
21	0.41	1.108	87.03
22	0.406	1.097	88.127
23	0.39	1.053	89.181
24	0.374	1.011	90.192
25	0.367	0.993	91.184
26	0.355	0.959	92.143
27	0.342	0.923	93.066
28	0.319	0.863	93.929
29	0.312	0.844	94.774
30	0.298	0.805	95.578
31	0.277	0.749	96.327
32	0.269	0.727	97.054
33	0.247	0.669	97.723
34	0.236	0.639	98.362
35	0.223	0.602	98.964
36	0.198	0.534	99.498
37	0.186	0.502	100

Extraction method: principal component analysis.

, Appendix B), indicating these factors effectively represent the dataset.

Upon examining these seven factors (

Table A4. Component matrix after rotation.

Component	1	2	3	4	5	6	7
Factor type	Travel perception evaluation	Travel environment	Attitude towards green travel	Natrural environment	Built environment	Social environment	Willingness to green travel
PG1	0.83	0.052	0.114	0.072	0.072	0.054	0.06
PJ1	0.8	0.054	0.125	0.033	0.033	0.114	0.011
PJ4	0.786	0.032	0.163	0.083	0.06	0.012	0.016
PJ3	0.781	0.131	0.066	0.116	0.078	0.074	0.072
PG3	0.78	0.027	0.043	0.037	0.143	−0.021	0.068
PG5	0.768	0.138	0.187	0.084	0.088	0.132	−0.037
PG4	0.762	0.154	0.184	0.119	0.029	0.105	0.101
PG2	0.755	0.137	0.148	0.075	0.151	0.158	0.038
PJ2	0.754	0.091	0.17	0.109	0.021	0.11	0.183
HW1	0.059	0.796	0.15	0.098	0.101	−0.034	−0.045
HW5	0.156	0.789	0.092	0.055	0.086	0.051	0.093
HW2	0.073	0.781	0.179	0.049	0.145	0.067	0.072
HW3	0.127	0.777	0.148	0.123	0.106	0.072	0.063
HW4	0.122	0.775	0.089	−0.027	0.056	0.093	−0.009
HW6	0.08	0.768	0.055	0.063	0.076	0.12	0.119
TC1	0.213	0.126	0.754	0.108	0.143	0.132	0.106
TC5	0.125	0.093	0.736	0.151	0.172	0.113	0.1
TC6	0.229	0.131	0.704	0.164	0.222	0.121	0.118
TC4	0.174	0.134	0.7	0.19	0.056	0.105	0.182
TC3	0.206	0.195	0.69	0.158	0.194	0.087	0.14
TC2	0.237	0.223	0.655	0.165	0.159	0.15	0.103
HZ2	0.152	0.073	0.183	0.799	0.129	0.057	0.039
HZ3	0.116	0.051	0.057	0.795	0.129	0.044	0.057
HZ1	0.084	0.014	0.124	0.793	0.107	0.113	0.182
HZ4	0.096	0.095	0.135	0.786	0.1	−0.033	0.052
HZ5	0.087	0.098	0.235	0.731	0.029	0.092	−0.018
HJ1	0.131	0.099	0.117	0.092	0.783	0.059	0.107
HJ3	0.106	0.066	0.224	0.066	0.763	0.116	−0.038
HJ2	0.072	0.12	0.126	0.148	0.748	0.086	0.192
HJ4	0.054	0.118	0.189	0.138	0.746	0.054	0.082
HJ5	0.162	0.172	0.089	0.076	0.719	0.157	0.211
HS1	0.144	0.133	0.158	0.042	0.18	0.795	0.097
HS3	0.213	0.074	0.168	0.147	0.135	0.77	0.12
HS2	0.185	0.143	0.241	0.07	0.128	0.752	0.19
Y3	0.114	0.103	0.173	0.099	0.131	0.099	0.802
Y2	0.158	0.026	0.225	0.117	0.104	0.131	0.752
Y1	0.054	0.126	0.176	0.066	0.302	0.157	0.71

Extraction method: principal component analysis. Rotation method: Kaiser normalization maximum variance method. A rotation has converged after 6 iterations.

, Appendix B), four corresponded to our proposed classifications: travel perception evaluation, green travel attitude, micro travel environment, and green travel willingness. The other three factors were categorised under macro-urban environment factors, specifically as urban built environment, urban social environment, and urban natural environment.

The rotated component matrix confirmed that all questionnaire dimensions had loadings greater than 0.65 in their respective components, demonstrating strong structural validity without requiring any adjustments (Table A4, Appendix B).

3.2.2. Confirmatory Factor Analysis

Following the exploratory factor analysis (EFA), we conducted confirmatory factor analysis (CFA) to validate the factor structure and ensure that the measurement model accurately reflects the relationships between latent factors and observed variables. The CFA assessed model fit, convergent validity, composite reliability, and discriminant validity.

The results (

Table 3. Fitting index results of the CFA model.

Fitting Index	Reference Standard	Model Fitting
CMIN/DF	1–3 is excellent, 3–5 is good	1.138
RMSEA	<0.05 is excellent, <0.08 is good	0.021
IFI	>0.9 is excellent, >0.8 is good	0.986
TLI	>0.9 is excellent, >0.8 is good	0.985
CFI	>0.9 is excellent, >0.8 is good	0.986

) show a CMIN/DF ratio of 1.144, within the acceptable range of 1–3, and an RMSEA of 0.022, which is excellent as it is below 0.05. The IFI, CFI, and TLI values all exceeded the commendable threshold of 0.9, indicating a strong model fit.

With the CFA model confirmed, we evaluated convergent validity and composite reliability. We calculated the average variance extracted (AVE) and composite reliability (CR) for each dimension, with AVE values exceeding 0.5 and CR values exceeding 0.7, indicating strong convergent validity and composite reliability [57,58] (refer to

Table 4. Test of convergence validity and composite reliability.

Observed Variable		Latent Variable	Estimate	AVE	CR
HJ5	<---	Built environment	0.756	0.558	0.863
HJ4	<---	Built environment	0.731
HJ3	<---	Built environment	0.725
HJ2	<---	Built environment	0.762
HJ1	<---	Built environment	0.761
TC6	<---	Attitude towards green travel	0.78	0.571	0.889
TC5	<---	Attitude towards green travel	0.729
TC4	<---	Attitude towards green travel	0.725
TC3	<---	Attitude towards green travel	0.766
TC2	<---	Attitude towards green travel	0.753
TC1	<---	Attitude towards green travel	0.778
HW6	<---	Travel environment	0.737	0.586	0.894
HW5	<---	Travel environment	0.781
HW4	<---	Travel environment	0.728
HW3	<---	Travel environment	0.789
HW2	<---	Travel environment	0.793
HW1	<---	Travel environment	0.761
HZ5	<---	Natural environment	0.711	0.584	0.875
HZ4	<---	Natural environment	0.746
HZ3	<---	Natural environment	0.754
HZ2	<---	Natural environment	0.82
HZ1	<---	Natural environment	0.784
Y3	<---	Willingness to green travel	0.738	0.542	0.78
Y2	<---	Willingness to green travel	0.728
Y1	<---	Willingness to green travel	0.742
HS3	<---	Social environment	0.756	0.6	0.818
HS2	<---	Social environment	0.822
HS1	<---	Social environment	0.744
PJ1	<---	Travel perception evaluation	0.786	0.614	0.935
PJ2	<---	Travel perception evaluation	0.781
PJ3	<---	Travel perception evaluation	0.78
PJ4	<---	Travel perception evaluation	0.768
PG5	<---	Travel perception evaluation	0.793
PG4	<---	Travel perception evaluation	0.795
PG3	<---	Travel perception evaluation	0.734
PG2	<---	Travel perception evaluation	0.787
PG1	<---	Travel perception evaluation	0.825

).

To assess discriminant validity, we examined cross-factor correlations to ensure that different constructs are distinct. The standardised correlation coefficients between dimensions were lower than the square root of their respective AVE values, confirming strong discriminant validity (

Table 5. Factor Distinction Validity Tests.

Variable	Built Environment	Travel Perception Evaluation	Attitude Towards Green Travel	Travel Environment	Natural Environment	Willingness to Green Travel	Social Environment
Built environment	0.558
Travel perception evaluation	0.326	0.614
Attitude towards green travel	0.536	0.515	0.571
Travel environment	0.367	0.314	0.453	0.586
Natural environment	0.369	0.317	0.494	0.247	0.584
Willingness to green travel	0.528	0.333	0.572	0.303	0.347	0.542
Social environment	0.460	0.439	0.571	0.352	0.316	0.532	0.600
Square root of AVE value	0.747	0.784	0.756	0.766	0.764	0.736	0.775

The square root of AVE values is provided for each variable to indicate the threshold for discriminant validity.

).

At this point, we have completed the establishment and validation of the measurement model (Figure 3), laying the foundation for subsequent analysis and interpretation of the structural model.

3.3. Structural Equation Modeling Results and Hypothesis Verification

Factor analysis demonstrates that each indicator provides strong evidence for the model’s reliability and validity. In our theoretical model, we posit that the external environment has a direct impact on the willingness to green travel. Additionally, it can influence this inclination by shaping individuals’ perceptions and attitudes of green travel. Based on this premise, we have formulated fifteen hypotheses regarding the paths of influence (as presented in

Table 6. Impact Relationship Hypothesis.

Path	Hypothesis
Ha1	Built environment has an impact on travel perception evaluation.
Ha2	Social environment has an impact on travel perception evaluation.
Ha3	The natural environment has an impact on the evaluation of travel perception.
Ha4	Travel environment has an impact on travel perception evaluation.
Hb1	Built environment has an impact on attitude towards green travel.
Hb2	Social environment has an impact on attitude towards green travel.
Hb3	The natural environment has an impact on the attitude of green travel.
Hb4	Travel environment has an impact on attitude towards green travel.
Hc1	The built environment has an impact on the willingness to travel green.
Hc2	Social environment has an impact on green travel willingness.
Hc3	The natural environment has an impact on the willingness to travel green.
Hc4	Travel environment has an impact on green travel willingness.
Hd1	Travel perception evaluation has an impact on green travel intention.
Hd2	Attitude towards green travel has an impact on green travel intention.
He1	Travel perception evaluation has an impact on attitude towards green travel.

and Figure 4). We then compute the hypothesised paths of the constructed structural equation model, obtaining initial path coefficients.

Table 7. Test results of path relationship of initial SEM.

Path Relation			Estimate	S.E.	C.R.	p	Test Result
Travel perception evaluation	<---	Built environment	0.082	0.077	1.073	0.283	Not support
Travel perception evaluation	<---	Social environment	0.324	0.078	4.13	0.000 ***	Support
Travel perception evaluation	<---	Natural environment	0.178	0.072	2.482	0.013 *	Support
Travel perception evaluation	<---	Travel environment	0.158	0.072	2.19	0.029 *	Support
Attitude towards green travel	<---	Built environment	0.187	0.056	3.319	0.000 ***	Support
Attitude towards green travel	<---	Social environment	0.227	0.059	3.834	0.000 ***	Support
Attitude towards green travel	<---	Natural environment	0.219	0.054	4.085	0.000 ***	Support
Attitude towards green travel	<---	Travel environment	0.159	0.053	3.013	0.003 **	Support
Attitude towards green travel	<---	Travel perception evaluation	0.18	0.048	3.772	0.000 ***	Support
Willingness to travel	<---	Travel environment	−0.014	0.067	−0.209	0.834	Not Support
Willingness to travel	<---	Natural environment	0.041	0.069	0.593	0.553	Not Support
Willingness to travel	<---	Social environment	0.236	0.078	3.006	0.003 **	Support
Willingness to travel	<---	Built environment	0.244	0.074	3.292	0.000 ***	Support
Willingness to travel	<---	Travel perception evaluation	−0.015	0.061	−0.238	0.811	Not support
Willingness to travel	<---	Attitude towards green travel	0.301	0.1	3.004	0.003 **	Support

“Estimate” is the estimated value of the model parameter, indicating the magnitude or direction of the independent variable’s impact on the dependent variable. “S.E.” refers to the standard error of the parameter estimate, which measures the precision of the estimate. A smaller standard error suggests that the estimate is closer to the true value. “C.R.” is a standardised value used to assess the statistical significance of the parameter estimate, which is calculated using the formula: C.R. = Estimate/S.E. The p-values are rounded to three decimal places to indicate the level of statistical significance. The *** means p < 0.001, which indicates that the observed effect is highly unlikely to have occurred by chance, and we can be very confident in rejecting the null hypothesis. The ** means 0.001< p < 0.01, which suggests that the effect is quite strong and unlikely to be a random occurrence. The * means 0.01< p < 0.05, which still represents a significant effect that is unlikely to be due to random variation.

demonstrates that the urban social environment, natural environment, and micro travel environment each have an impact on travel perception evaluation, whereas the urban built environment does not have a significant influence. These four factors are causally related to attitudes towards green travel. The impact of urban social environment, built environment, and attitude towards green travel on willingness to green travel is significant. However, the micro travel environment, natural environment, and travel perception evaluation do not have a significant effect on willingness to green travel. Therefore, hypotheses Ha2, Ha3, Ha4, Hb1, Hb2, Hb3, Hb4, Hc1, Hc2, Hd2, and He1 are supported, while hypotheses Ha1, Hd1, Hc3, and Hc4 are not supported.

We used AMOS 24 software to analyse the path coefficients and significance test results derived from the linear regression for the purpose of refining the model. This refinement is based on empirical evidence, ensuring that the model is supported by both real data and theoretical foundations. Based on the analysis results, we progressively eliminated paths that were not significant, leading to a well-fitting model. The refined model fit is presented in

Table 8. Fitting index results of the refined model.

Fitting Index	Reference Standard	Model Fitting
CMIN/DF	1–3 is excellent, 3–5 is good	1.133
RMSEA	<0.05 is excellent, 0.08 is good	0.021
IFI	>0.9 is excellent, >0.8 is good	0.986
TLI	>0.9 is excellent, >0.8 is good	0.985
CFI	>0.9 is excellent, >0.8 is good	0.987

, and after refining, we obtained a total of 11 paths, as illustrated in Figure 5.

4. Discussion

4.1. Interpretation of Results

4.1.1. Causal Insights for Individual Dimensions: Analyzing Every Influence Path from Urban Environment to Green Travel Willingness

We identified four influencing factors and two mediating variables that affect the willingness to green travel of Shenyang residents. These factors can explain most of the variance in residents’ willingness to green travel and are consistent with the characteristics of the S-O-R theory [47]. Based on the results of the structural equation model, the following nine paths that influence willingness to green travel have been sorted out, and the influential effects of these nine paths have been calculated by multiplying the local path loads (Figure 6). The empirical research results show that the natural environment, social environment, built environment, and travel environment all positively influence the willingness to green travel. Among them, Path 5 and Path 7 have the strongest impact on the willingness to travel green, with effects of 0.24 and 0.26, respectively. This suggests that enhancing the built environment and the social environment has the most noticeable effect on residents’ willingness to travel green. The natural environment and travel environment do not directly impact willingness to green travel. However, they can indirectly influence willingness to green travel by affecting travel perception evaluation and attitude towards green travel, revealing the deep logic behind the phenomenon “the cold weather has people hibernating”. While the travel perception evaluation does not directly affect the willingness to green travel, it does have a positive effect on this intention through the attitude towards green travel, which is different from previous research results that emphasised that both travel perception and attitude have direct effects [4]. This suggests that promoting green travel in cold regions requires more than just a satisfactory urban environment; it also necessitates a shift in residents’ attitudes towards embracing green travel as a preferred mode of transportation.

4.1.2. Aggregated Causality Assessment for Urban Environment Dimensions: Synthesizing Influence Paths on Attitudinal, Perceptual, and Behavioral Outcomes

Based on the paths described above, we integrated the paths of each of the four influencing factors to obtain the total effects of each dimension on the willingness to green travel and the two mediating variables (

Table 9. Total effect of Multiple Influence Relationships for Green Travel.

Impact Factor	Dependent Variable	Number of Paths	Total Effect
Social environment	Willingness to green travel	3	0.3326
Built environment	Willingness to green travel	2	0.318
Natural environment	Willingness to green travel	2	0.0603
Travel environment	Willingness to green travel	2	0.0584
Social environment	Attitude towards green travel	2	0.3193
Natural environment	Attitude towards green travel	2	0.2078
Travel environment	Attitude towards green travel	2	0.2015
Built environment	Attitude towards green travel	1	0.2
Social environment	Travel perception evaluation	1	0.33
Natural environment	Travel perception evaluation	1	0.18
Travel environment	Travel perception evaluation	1	0.15

). First, in terms of influencing travel willingness, the social environment factor has the highest impact on the willingness to green travel, with a total effect of 0.3326 over three distinct paths. The built environment factor, on the other hand, influences the willingness to green travel through two paths, with a total effect of 0.3180. In comparison, the natural environmental factor and the micro-travel environment factor have a total effect about one-sixth of the former, but it is still not negligible. Second, when analysing attitude towards green travel, the social environment factor has the most substantial impact, with a total effect of 0.3193 across two paths. The natural environmental factor, travel environment factor, and built environmental factor all have a total effect around 0.2, not much different from the former, showing their potential in shaping travel attitudes. Last, in terms of residents’ travel perception evaluation, there is a direct impact from the social environment, natural environment, and travel environment, with the social environmental factor’s effect being twice that of the others. A total effect of N means that if a one-unit change is made to a certain influencing factor, a change of N units in the dependent variable is expected. A larger total effect indicates that the output generated per unit input is higher. Therefore, under conditions of limited resources, we can prioritise optimising those environmental factors with the highest total effects. Alternatively, over the long term, we can systematically enhance the urban environment in a sequence based on the ranking of total effects. This will help change residents’ attitudes towards travelling, enhance their perception of travelling, and strengthen their willingness to green travel.

4.1.3. Considering Factor Interactions Between Urban Environment Dimensions: Uncovering Optimal Combinations for Green Travel Willingness Enhancement

Based on the SEM model results, it is evident that there are different standardised correlation coefficients among the four environmental factors, as depicted in

Table 10. Standardized correlation coefficients among the four environmental factors.

Variable	Built Environment	Travel Environment	Natural Environment	Social Environment
Built environment	0.558
Travel environment	0.367	0.586
Natural environment	0.369	0.247	0.584
Social environment	0.46	0.352	0.316	0.6

. This implies that a unit change in one environmental factor will similarly result in corresponding changes in the correlation strengths of related environmental factors. Therefore, a greater correlation strength indicates stronger synergistic effects among the environmental factors. Therefore, if sufficient resources are available, we can incorporate the correlation strength of the four external environmental factors and attempt to optimise the combination of these factors in a more comprehensive manner. This will enable us to achieve the efficient allocation of resources in prompting green travel. Taking the combination of the social environment with other factors to strengthen willingness to green travel as an example, the strongest correlation is found between the built environment and the social environment. This suggests that the combined effect of these two factors is more likely to be greater than the sum of their individual effects. By examining the correlation between these factors and their respective effects, we can conclude that the combination of the built environment and the social environment is the most effective. The combination of the social environment and the travel environment is the next most efficient, while the combination of the social environment and the natural environment is the least efficient.

4.1.4. Intra-Cluster Analysis: Prioritizing Tangible Elements Within Urban Environment Dimensions for Green Travel

Based on the CFA results, each category of environmental factors (latent variables) involves various specific environmental elements (manifest variables). In practical terms, adjusting these specific environmental elements is necessary to achieve corresponding changes in environmental factors, thereby promoting green travel. Factor loadings reflect the contribution of each manifest variable to the latent variable. Therefore, based on the factor loadings, we can propose environmental optimisation strategies suitable for winter cities. Among the factors related to the social environment, the social atmosphere has the highest factor loading, followed by social climate and policy encouragement, with factor loadings of 0.82, 0.76, and 0.74, respectively. This suggests that a green travel culture and implementing policies that incentivise green travel are the most effective ways to increase the willingness of residents in winter cities to travel in an environmentally friendly manner. These findings align with conclusions drawn by scholars such as A. Huang and D. Levinson regarding policy factors influencing travel modes [59]. Within the built environment, the most influential factors are regional compatibility, recognition of urban construction, and green transportation infrastructure, all of which have loadings of 0.76. This aligns with the suggestion made by Schwanen T. and other scholars that the selection of travel modes is connected to the design of public transport stations and the level of completeness of other supporting facilities [19]. In other words, well-designed public transport stations and comprehensive transport facilities make it easier for residents to choose public transportation. The air quality and weather conditions exhibit the highest factor loadings among the natural environmental elements, suggesting that a favourable air environment in a winter city encourages residents to opt for environmentally-friendly means of transportation. In the context of the travel environment, path guidance and the accessibility of starting and ending points both have the highest factor loading. To enhance residents’ willingness towards green travel, we can enhance the urban traffic’s signage system for guidance and augment accessibility by incorporating public transport stations. Operational efficiency represents another key factor. Enhancing the operational efficiency of public transportation, such as augmenting bus frequency and accelerating metro speed, positively influences citizens’ willingness to green travel.

4.1.5. Comprehensive Factor Evaluation: Manifesting the Influence of All Potential Influencing Factors on Green Travel Willingness

In this study, we identified 34 potential influencing factors on residents’ willingness to green travel in cold-region cities. To assess the strength of these associations, we calculated two correlation coefficients: the Pearson correlation coefficient, which measures direct linear relationships, and the structural correlation coefficient (SCC), a measure we propose that accounts for both direct and indirect effects within the SEM framework.

The SCC is calculated by multiplying the factor loadings of observed variables by the total path coefficients between latent variables and green travel willingness in the SEM. This approach captures the complex interplay of multiple factors, providing a more nuanced understanding of their contributions. As shown in

Table A8. The structural correlation coefficient analysis between obvious variables and willingness to green travel.

Rank	Observed Variable		Latent Variable	Estimate	Latent Variable Path Coefficient with Willingness	Structural Correlation Coefficient
1	Social atmosphere	HS2	Social environment	0.822	0.3326	0.273
2	General mood of society	HS3	Social environment	0.756	0.3326	0.251
3	Policy encouragement	HS1	Social environment	0.744	0.3326	0.247
4	Identification degree of urban construction	HJ2	Built environment	0.762	0.318	0.242
5	Green traffic construction	HJ1	Built environment	0.761	0.318	0.242
6	Road network density	HJ5	Built environment	0.756	0.318	0.240
7	Supporting facilities	HJ4	Built environment	0.731	0.318	0.232
8	Relevance of urban construction	HJ3	Built environment	0.725	0.318	0.231
9	Living habits	TC6	Attitude towards green travel	0.78	0.29	0.226
10	Personal spiritual significance	TC1	Attitude towards green travel	0.778	0.29	0.226
11	Personal social significance	TC3	Attitude towards green travel	0.766	0.29	0.222
12	Personal physiological significance	TC2	Attitude towards green travel	0.753	0.29	0.218
13	Health condition	TC5	Attitude towards green travel	0.729	0.29	0.211
14	Green travel cognition	TC4	Attitude towards green travel	0.725	0.29	0.210
15	Security	PG1	Travel perception evaluation	0.825	0.0609	0.050
16	Air environment	HZ2	Natural environment	0.82	0.0603	0.049
17	Money cost perception	PG4	Travel perception evaluation	0.795	0.0609	0.048
18	Perception of construction situation	PG5	Travel perception evaluation	0.793	0.0609	0.048
19	Comfort	PG2	Travel perception evaluation	0.787	0.0609	0.048
20	Loyalty	PJ1	Travel perception evaluation	0.786	0.0609	0.048
21	Satisfaction degree of operating environment	PJ2	Travel perception evaluation	0.781	0.0609	0.048
22	Satisfaction of decision	PJ3	Travel perception evaluation	0.78	0.0609	0.048
23	Weather conditions	HZ1	Natural environment	0.784	0.0603	0.047
24	Overall satisfaction of travel	PJ4	Travel perception evaluation	0.768	0.0609	0.047
25	Guidance of path	HW2	Travel environment	0.793	0.0584	0.046
26	Accessibility of starting and ending points	HW3	Travel environment	0.789	0.0584	0.046
27	The convenience of site connection	HW5	Travel environment	0.781	0.0584	0.046
28	Topography	HZ3	Natural environment	0.754	0.0603	0.045
29	Seasonal climate	HZ4	Natural environment	0.746	0.0603	0.045
30	Physical cost perception	PG3	Travel perception evaluation	0.734	0.0609	0.045
31	Path support	HW1	Travel environment	0.761	0.0584	0.044
32	Operational technical support	HW6	Travel environment	0.737	0.0584	0.043
33	Natural greening	HZ5	Natural environment	0.711	0.0603	0.043
34	Operational efficiency	HW4	Travel environment	0.728	0.0584	0.043

in Appendix D, the top 14 factors demonstrate significantly higher SCC values compared to the remaining 20, suggesting these are the most critical factors when considering the multifaceted impacts within urban environments on green travel.

Conversely, the Pearson correlation coefficients, listed in

Table A9. The Pearson correlation coefficient analysis between obvious variables and willingness to green travel.

Rank	Observed Variable	Willingness of Green Travel
1	Regional compatibility	Pearson correlation	0.418 **
2	Social atmosphere	Pearson correlation	0.406 **
3	Personal spiritual significance	Pearson correlation	0.402 **
4	Living habits	Pearson correlation	0.393 **
5	Identification degree of urban construction	Pearson correlation	0.388 **
6	Green travel cognition	Pearson correlation	0.385 **
7	Personal physiological significance	Pearson correlation	0.380 **
8	Personal social significance	Pearson correlation	0.378 **
9	Health condition	Pearson correlation	0.351 **
10	General mood of society	Pearson correlation	0.345 **
11	Green traffic construction	Pearson correlation	0.330 **
12	Policy encouragement	Pearson correlation	0.329 **
13	Road network density	Pearson correlation	0.328 **
14	Satisfaction degree of operating environment	Pearson correlation	0.326 **
15	Weather conditions	Pearson correlation	0.320 **
16	Money cost perception	Pearson correlation	0.273 **
17	Comfort	Pearson correlation	0.258 **
18	Relevance of urban construction	Pearson correlation	0.257 **
19	Air environment	Pearson correlation	0.244 **
20	Operational technical support	Pearson correlation	0.238 **
21	Guidance of path	Pearson correlation	0.237 **
22	Accessibility of starting and ending points	Pearson correlation	0.230 **
23	Operational efficiency	Pearson correlation	0.229 **
24	Decision satisfaction	Pearson correlation	0.228 **
25	Security	Pearson correlation	0.227 **
26	Topography	Pearson correlation	0.207 **
27	Physical cost perception	Pearson correlation	0.201 **
28	Perception of construction situation	Pearson correlation	0.196 **
29	Natural greening	Pearson correlation	0.196 **
30	Seasonal climate	Pearson correlation	0.188 **
31	Loyalty	Pearson correlation	0.187 **
32	Overall satisfaction of travel	Pearson correlation	0.184 **
33	The convenience of site connection	Pearson correlation	0.147 **
34	Path support	Pearson correlation	0.128 *

** At the level of 0.01 (double tail), the correlation is significant. * At the level of 0.05 (two-tailed), the correlation is significant.

in Appendix D, assess the bivariate relationships between each of the 34 potential influencing factors and green travel willingness. All factors are significant at the 0.05 level or above; their correlations are more uniformly distributed, making it harder to classify them as clearly as with the SCC.

However, by comparing the rankings from both coefficients, we identify nine factors—the convenience of site connection, overall satisfaction of travel, loyalty, perception of built environment, security, relevance of urban construction, road network density, policy encouragement, and general mood of society—where SCC rankings are noticeably higher. This implies that these nine factors, often underemphasised in traditional linear analyses, might play a substantial role through their interactions with other environmental factors in influencing green travel willingness. Given that complex interactions reflect real-life scenarios more accurately and align with the principles of complex adaptive systems theory, these nine factors, though individually less impactful, could collectively exert a significant influence on green travel. Future research and urban design should prioritise these factors to enhance green travel adoption effectively.

4.2. Implications of Findings

This study reaffirms the practicality of using the S-O-R theory in the context of decision-making in travel, particularly in winter cities like Shenyang. The findings provide evidence that the urban natural, social, built, and micro traveling environments have a significant influence on residents’ willingness to engage in green travel, consistent with the environmental influence theory [9,10]. Additionally, the study supports the attitude–behaviour congruence theory, indicating that changes in the urban environment can alter residents’ travel perceptions, evaluations, and attitudes, thereby influencing their green travel intentions.

In this research, certain concepts, such as the natural environment, social environment, built environment, and travel environment, may appear abstract. These categories serve as frameworks for analysing various factors influencing green travel willingness in cold regions. The natural environment, social environment, and built environment were derived from an exploratory factor analysis (EFA) of observed variables with high factor loadings, while the travel environment was identified by summarising prior studies and incorporating unique characteristics relevant to winter cities. The natural environment includes factors like weather conditions, air quality, topography, seasonal climate, and natural greening, which impact the comfort and feasibility of sustainable travel. Improvements here could involve developing weather-resilient pathways, enhancing greenery along routes, and adapting infrastructure for seasonal changes, such as snow removal in winter. The social environment encompasses policy encouragement, social atmosphere, and societal mood, which reflect cultural and regulatory influences on travel behaviors. Supporting green travel within this environment may involve implementing policy incentives, promoting environmental awareness, and fostering a pro-sustainability culture. The built environment refers to the physical infrastructure, including green traffic construction, urban construction identification, infrastructure relevance, supporting facilities, and road network density. Enhancements could include expanding public transportation options, designing pedestrian-friendly streets, and increasing connectivity between key urban areas to reduce car dependency. Last, the travel environment covers factors such as path support, route guidance, accessibility of starting and ending points, site connection convenience, operational efficiency, and technical support. Improving this environment involves maintaining clear and safe sidewalks, optimizing transit stop accessibility, managing road congestion effectively, and integrating smart travel information systems that guide users toward eco-friendly travel choices. By clarifying these concepts and providing actionable improvement strategies, this study contributes a more holistic understanding of how the urban environment can be optimised to encourage sustainable travel behaviours, particularly in winter cities.

The path analysis conducted in the study reveals that the social environment is the most influential factor in changing residents’ attitudes towards green travel, enhancing their perception of travel, and strengthening their willingness to green travel [5]. This suggests that promoting a culture of green travel and implementing supportive policies can be effective strategies for improving green travel behaviours, particularly in winter cities.

While the built environment and travel environment are not the most dominant factors, they still play a crucial role. These factors, which are closely linked to urban design, can enhance the willingness to engage in green travel through both direct and indirect means. The combined effect of these two factors is particularly noteworthy, suggesting that urban planners should prioritise the design and advancement of the built environment and transportation infrastructure to foster green travel practices.

Given the high costs and long durations associated with constructing infrastructure in winter cities, it is important to strategically optimise urban spaces, increase public recognition of urban construction efforts, and develop green transportation infrastructure. By doing so, cities can encourage more residents to adopt green travel practices during the challenging winter months.

These findings offer valuable insights for urban planners and policymakers, providing a foundation for developing targeted strategies that enhance green travel behaviours in cold climates. However, it is important to note that while these recommendations are grounded in the study’s findings, further research with more representative data would be necessary to validate and refine these strategies.

4.3. Limitation of Research

The focus of this study is on the central city of Shenyang. However, to develop a comprehensive theoretical framework for understanding the factors influencing green travel behaviour in winter cities, it is essential to also examine the travel behaviours of individuals in counties and suburban areas where urban development is less advanced. The attitudes and behaviours of urban residents may vary significantly across different locations and cultural contexts, which suggests that our findings might not be universally applicable to all cold-climate cities.

Moreover, this study specifically examines the city’s external environment and residents’ attitudes as a collective entity, placing less emphasis on individual residents’ personal characteristics. Future research could incorporate additional variables to enhance the comprehensiveness of the theoretical framework.

Although the statistical analysis conducted in this study is methodologically rigorous, the non-random nature of the data collection limits the generalisability of the results. Therefore, the analysis should be viewed as illustrative of potential trends within the sampled population rather than as conclusive evidence applicable to the entire population of Shenyang. Future research with a more representative sample would be necessary to validate these findings.

5. Conclusions

This study employs the stimulus–organism–response (S-O-R) theory to explore how urban environments influence residents’ willingness to engage in green travel during winter in cold-region cities, using data from Shenyang, China. Through structural equation modelling (SEM), we identified four key environmental factors—social environment, built environment, natural environment, and travel environment—and two mediating variables, travel perception evaluation and attitude towards green travel, that collectively shape residents’ willingness to engage in green travel.

The analysis reveals nine distinct paths through which these factors influence green travel willingness, with the social and built environments having the most significant direct impacts. The natural and travel environments, while not directly affecting green travel willingness, exert crucial indirect effects by shaping travel perceptions and attitudes, illustrating the deep-seated interplay between these environmental factors. These findings highlight that promoting green travel in cold regions requires not only improvements to the urban environment but also a shift in residents’ attitudes towards embracing green travel as a preferred mode of transportation.

Further analysis of the total effects of each environmental factor shows that the social environment has the strongest overall impact on green travel willingness, followed closely by the built environment. The natural and travel environments, though less impactful, still contribute meaningfully, particularly through their influence on attitudes and perceptions. Additionally, the interaction between these factors indicates that combining enhancements in the social and built environments can produce synergistic effects, making this combination particularly effective for encouraging green travel.

To provide a more nuanced understanding of these influences, we introduced the structural correlation coefficient (SCC), which captures both direct and indirect effects. This analysis identified nine critical factors—such as site connection convenience, overall satisfaction with travel, and policy encouragement—that should be prioritised in urban planning. These factors, often underappreciated in traditional linear analyses, are shown to have a stronger influence on green travel willingness when considered within the broader context of interacting environmental influences.

The findings of this study advance the understanding of green travel behaviour in winter cities by illustrating the complex interplay between environmental factors, mediating variables, and individual attitudes. They offer actionable insights for policymakers and urban planners, suggesting that fostering supportive social environments and strategically enhancing the built and travel environments can effectively promote green travel practices in cold climates. Moreover, this research lays a foundation for future studies on sustainable urban development in winter cities, encouraging further exploration of the factors that influence environmentally friendly travel behaviours and contributing to the creation of livable, sustainable urban environments even in the harshest winter conditions.