Do Unprecedented Gasoline Prices Affect the Consumer Switching to New Energy Vehicles? An Integrated Social Cognitive Theory Model

Abstract

From 2020 to 2022, the price of gasoline in China rose sharply, which may cause consumers to adopt new energy vehicles. However, the government subsidies in the yearly retreat for 2023 were completely abolished, which could hinder consumers’ switching behavior. The combination of these factors may affect consumer decision-making, making accurate analysis of consumer willingness to switch to new energy vehicles crucial for the development of this industry. The current study aims to investigate consumers’ vehicle switching intentions affected by multiple factors such as fuel prices, and we attempt to analyze these by combining personal and environmental factors from social cognitive theory. Data were collected through an online platform survey using developed reliable scales. The 464 responses were then synthesized using structural equation modeling and Bayesian networks, and the results showed that approximately 51% of consumers had high-level switching intentions to new energy vehicles; attitude, self-efficacy, environmental consciousness, and infrastructure barriers had the strongest effect on consumers’ vehicle switching intentions. This study assists in identifying the psychological demands of consumers switching to new energy vehicles and provides ideas for vehicle manufacturers and governments in terms of marketing strategies and policy formulation at crucial stages when new energy vehicles are in accelerated development.

1. Introduction

Petroleum extraction and the application of combustion engines, allowing both power and economy of the gasoline engine, have become an essential automotive power, automobile manufacturing technology is widely popular and applied across the world, and the automobile has become a significant means of transportation. Based on statistics from China’s Ministry of Public Security, China’s automobile fleet grew to 307 million by end of March 2022 [1]. Increasing numbers of automobiles create enormous petroleum demand. In 2022, the China Energy Research Institute released a report showing that China’s apparent petroleum consumption reached 715 million tons in 2021 [2]. This enormous demand resulted in much focus on its price. International petroleum prices also rose sharply from 2020 to June 2022, subject to political military conflicts, new coronavirus epidemics, and supply–demand conflicts [3], and domestic petroleum prices rose by 70%from 6800 per ton in November 2020 to 11,600 per ton in June 2022 [4,5]. The significant increase in petroleum prices raises the cost of motorized travel and induces travel behavior or vehicle choice changes [6,7]. Additionally, air pollution, the greenhouse effect, and respiratory diseases caused by heavy automotive usage are increasingly prominent.

With the breakthrough of the automotive sector restructuring the industry to tackle the rising gasoline prices, energy crisis, environmental pollution, and other issues, new energy vehicles (NEVs) with economic, energy-saving, and environmental protection features have emerged as new trends. The advent of NEVs has given individuals more alternatives under fluctuating fuel prices. According to the “New Energy Vehicle Industry Development Plan (2021–2035)”, NEVs in China’s passenger car market mainly include pure electric vehicles and plug-in hybrid (including extended-range) vehicles (PHVs). The goal is to reach 20% sales by 2025 and mainstream the sales of pure electric vehicles by 2035 [8]. Therefore, against the backdrop of enormous development opportunities, numerous scholars have conducted empirical research on whether consumers are willing to purchase NEVs, further exploring the audience of NEVs in the passenger car market. These studies have mainly explored the factors that affected consumer purchase intention, and gasoline price is one of the numerous factors that have been focused on by scholars. For example, Shafiei et al. [9] found that when gasoline prices rise, NEV prices decrease, and there is no range anxiety, so consumers are more inclined to purchase NEVs. Sun et al. [10] found that when gasoline prices rise and the price increase of NEVs does not exceed CNY 27,000, car owners are more inclined to purchase pure electric vehicles. With gasoline prices reaching an unprecedented high in February–March 2022, a total of 6200 comments were collected by trawling the comment data from online media. A text network analysis was plotted through text co-word matrix analysis and is presented in Figure 1. The following outcomes surface: topics such as electric vehicles, gasoline prices, and charging time become hot topics of user and media attention, and the charging infrastructure facilities for NEVs are also an essential factor considered by consumers. Thus, consumers will consider more factors when considering whether to purchase NEVs. Current studies generally focus on factors such as gasoline prices’ impact on NEV purchase intentions, but empirical research on individuals’ willingness to switch mode choices is lacking; research has shown that switching intentions are not equivalent to purchase intentions and that their intrinsic mechanisms are distinct [11,12,13]. Moreover, research on gasoline prices has focused on objective facts, but without understanding the changes in consumer perception of gasoline prices from psychological perspectives. Against the background of the sharp rise in gasoline prices reaching high levels, there are significant theoretical implications and practical value in testing the willingness of individuals to switch from fuel vehicles to NEVs.

The sharp rise in gasoline prices may drive switching to NEVs. Government subsidies for vehicle purchases are retreating each year and will be eliminated by the end of 2022, which may hinder people from switching to NEVs. Through text network analysis, it has been found that charging time and charging infrastructure are also key factors of concern for consumers. Promoting NEVs has become an important measure for the automotive industry to reduce emissions against the backdrop of “dual carbon development”. During the critical period when multiple factors including high gasoline prices, subsidy policy removal, and low-carbon development are at play, it is urgent and crucial to investigate individual psychological needs for NEVs. This will help predict individuals’ switching travel behavior and environment protection attitudes and effectively predict the future market share of NEVs, assisting companies in developing appropriate marketing plans and providing a basis for policymakers. This paper empirically studies the intention of individuals aged 18 and above to adopt NEVs from consumers’ perspectives on factors including rising gasoline prices and subsidy policy removal. Considering that NEVs are not individuals’ preferred means of transportation, this study focuses on behavior changes and exploring whether they are willing to switch from fuel vehicles to NEV.

Social cognitive theory (SCT) is among the crucial theories to illustrate individual behavior change, highlighting the interaction between behavior, the individual, and the environment [14], and is widely applied to knowledge sharing behavior [15,16]. Boateng et al. [17] applied SCT to internet banking technology adoption studies and showed that trust and social characteristics significantly affect the intention to use online banking systems. SCT can provide theoretical foundations for effectively assessing individuals’ decisions to switch to NEVs under the effects of gasoline prices. Nevertheless, few studies have investigated this theory in NEV adoption behavior. Therefore, the model was constructed from the environment and personal factors, and we added perceived risks and infrastructure barriers to deeply analyze the crucial factors of personal intention towards switching to NEVs. In addition, structural equation modeling can test the correlation between variables based on theoretical knowledge by fitting assumptions and data information, but it cannot directly determine causal relationships between variables. Moreover, if the relationship between the independent and dependent variables is non-linear, SEM may lead to a decrease in prediction accuracy. To address these limitations, the nonlinear relationships are further analyzed by employing Bayesian networks (BN) to test for causal relationships between variables and provide strong inferential knowledge evidence for dealing with uncertain relationships in SEM. Therefore, this study integrates structural equation modeling (SEM) and Bayesian network (BN) research methods to verify the causal relationships between elements, which provides decision-making support to reveal the factors affecting individuals’ switching intention to NEVs.

The studies contribute to:

• Analyzing the mechanisms that affect consumer switching to NEVs despite record high fuel prices and the imminent elimination of vehicle purchase subsidies, and providing insights into green behavior in the context of “dual carbon” development goals;
• The first application of SCT to construct a theoretical framework of consumer switching intention to NEVs, integrating structural equation modeling and Bayesian network causality modeling methods to diagnose key factors and provide effective countermeasures for the future formulation of relevant policies and measures to increase the market share and usage of NEVs.

The remainder of this paper is structured as follows. Section 2 provides the theoretical framework and research hypotheses. Section 3 shows the questionnaire design, data collection process, and research methods. Section 4 presents the results of the research. Section 5 contain discussion and policy implications.

2. Theoretical Framework and Research Hypotheses

According to SCT, human behavior is primarily driven by environmental and personal influences [14]. Bandura indicates that environmental factors are categorized into the micro-environment of family, friends, and neighbors and the macro-environment of social, economic, cultural, and legal aspects [18]. Subjective norms are categorized as significant environmental factors in numerous SCT works [19,20,21]. Subjective norms enable individuals to be more likely oriented to healthy behaviors, including knowledge sharing activities [22], social networking posts [21], video course participation [19], etc. Against the background of SCT, Guo [23] believed that the natural and social environments constitute the environmental factors that influence people’s perceptions and behaviors toward green consumption, while environmental consciousness is perceived as a social attribute of individuals [24]. In addition, Zhao et al. [25] proposed new insights based on SCT that social face consciousness can influence engaging in corrupt behavior. Drawing on previous studies, this study chose subjective norms reflecting the effect of surrounding relatives, friends, and news, environmental consciousness indicating individual social attributes and pro-environmental behaviors, and face consciousness representing social status as significant environmental factors that may influence individuals’ intention to switch to NEVs.

In terms of individual factors, outcome expectations and self-efficacy are two representative factors. Outcome expectations are beliefs about expected outcomes that individuals believe arise under specific conditions [26]. Lin and Chang [27] argue that outcome expectations can be categorized into different types by scenario characteristics. Within this, the perceived benefits and rewards of initial expectations are considered the most common individual psychological determinants in SCT [28]. Both are variables whose remarkable effects on users’ behavioral intentions have been confirmed in numerous studies [15,16,21,22]. Bandura indicated that attitudes could serve as personal internal beliefs that drive and regulate behavior. This was also confirmed in Zhou and Hsiao’s studies; individuals’ attitudes toward behavior directly influence learning behavior using YouTube and continuous participation in courses using videoconferencing, while attitudes can mediate between environmental factors and individual actions [19,29]. As such, this study seeks to explore the interaction between initial expectations, perceived monetary benefits, self-efficacy, and attitudes as significant personal cognitive factors and environmental factors, along with the impact on individuals’ intention towards switching to NEVs. Meanwhile, infrastructure barriers and perceived risks will be incorporated into the model to examine whether people’s perceptions regarding the quality of NEVs and the associated charging facilities will prevent them from switching to NEVs.

In conclusion, based on SCT, this research focuses on two major perspectives, environmental factors and personal factors, to explore the important factors that may affect individuals’ intention to switch to NEVs. The theoretical framework is shown in Figure 2.

2.1. Environmental Factors

2.1.1. Subjective Norms

Subjective norms (SN) are an important environmental factor in SCT that refer to the degree of social pressure individuals feel when implementing or not implementing a behavior [30,31]. Setiawan et al. [32] believe that this social pressure originates from friends, family members, and others in the community. Battacherjee [33] suggests that subjective norms contain influences from external sources such as news reports, mass media, etc. It has been confirmed in studies by Screen et al. [34] that when consumers are uncertain about the outcome of their behavior, they might decide based on media reports or seek others’ opinions. Yen [21] explores the influence of subjective norms as a significant factor in SCT on posting behavior on social networking sites. Van Tonder et al.’s [35] studies indicated that subjective norms positively influence consumers’ green attitudes, thereby generating positive green consumption intentions.

Subjective norms refer to the pressure that household residents perceive from their surrounding relatives or friends and news reports when choosing NEVs in this study (for example, the opinions of relatives and media reports about NEVs will have substantial effects on my decision). Wang et al. [36] indicated that subjective norms positively and significantly affect individuals’ intentions for green product purchases. Subjective norms representing societal pressures were confirmed to affect consumer intentions toward purchasing NEVs in a study of NEV purchase intentions [37]. Therefore, our study considered that individuals who are skeptical about new technological product quality, among other things, are more susceptible to news reports and relatives or friends when they consider adopting NEVs. The following assumptions are proposed based on the above.

Hypothesis (H1).

Subjective norms have a positive impact on self-efficacy.

Hypothesis (H2).

Subjective norms have a positive impact on attitudes.

Hypothesis (H3).

Subjective norms have a positive impact on switching intentions.

2.1.2. Environmental Consciousness

Environmental consciousness (EC) refers to the formation of the conservation concept through direct attention to issues and learning about environmental protection, which will affect people’s knowledge, attitude, intention, and action [38,39,40]. According to Cerri and Tan’s study [41,42], environmental consciousness is a significant factor in exploring consumers’ pro-environmental behavior. Darvishmotevali and Altinay [43] explore the effect of environmental consciousness as a factor in SCT on proactive support for environmental behavior and indicate that environmental consciousness plays a mediating role and significantly influences proactive support for environmental behavior. Zhang et al.’s [44] studies indicate that environmental consciousness significantly affects consumers’ attitudes toward purchasing energy-efficient appliances and indirectly results in consumers’ willingness to pay premiums for energy-efficient appliances.

People with stronger social responsibility are more willing to implement green consumption behaviors, those who believe that green consumption brings positive outcomes are more inclined to consume green, and environmentally conscious people have a higher propensity to implement green consumption behaviors [45,46,47]. NEVs have green product label characteristics for their cleaner power sources, higher energy efficiency conversions, and stronger emission reduction and environmental benefits [48,49]. Degirmenci and Breitner [50] concluded that environmental consciousness has stronger effects on attitudes and purchase intentions than price value and mileage confidence. This study defines environmental consciousness as the environmental value contribution and green identity embodiment that residents expect from switching to NEVs, advantages that fuel vehicles do not have. Hypotheses are proposed below.

Hypothesis (H4).

Environmental consciousness has a positive impact on attitudes.

Hypothesis (H5).

Environmental consciousness has a positive impact on switching intentions.

2.1.3. Face Consciousness

Chinese people with a background of collectivist values care about others’ perceptions and are more interested in showing others their identity, namely face consciousness. Face consciousness (FC) refers to the desire of individuals to enhance, maintain, and avoid losing their self-image when interacting socially [51]. In this study, face consciousness is defined as selecting NEVs as a behavior to highlight status and identity. In SCT research, Zhao et al. [25] found that social face consciousness affects individuals’ intention to engage in corrupt behavior. Influenced by Confucian culture, Chinese people’s consumption values are more easily shaped by face consciousness [52,53]. Highly face-conscious people are more eager to gain praise and recognition from society members and prefer products that highlight their status in the consumption process, such as buying branded products, etc. [54,55]. Regarding face consciousness and pro-environmental behavior, since people who perform pro-environmental behavior show their environmental literacy and cast positive images on society and others, people with high face consciousness are more willing to perform pro-environmental behavior to obtain respect from others [56,57]. Compared with fuel car users, people who use NEVs may feel “superior” in their social circle, which may derive from a positive image and social prestige. In summary, Hypotheses 6 and 7 are proposed.

Hypothesis (H6).

Face consciousness has a positive impact on attitudes.

Hypothesis (H7).

Face consciousness has a positive impact on switching intentions.

2.2. Personal Factors

2.2.1. Perceived Monetary Benefit

Perceived monetary benefit (PMB) refers to the possibility that an individually perceived action will produce a positive outcome in terms of economic benefits [58]. In detail, individuals behave in ways they believe will generate the expected results and concentrate on aspects that will increase their economic benefits. When people are convinced that their behavior benefits them financially, they strengthen their behavioral intentions [59]. Davenport and Prusak [60] conclude that individuals compare the costs of their behavior with the external rewards of their behavior before sharing knowledge; individuals are willing to share knowledge only when they expect the rewards to outweigh the costs. In SCT, Wang et al. [24] indicated that perceived monetary benefits significantly affect changes in behavioral patterns.

In this study, the perceived monetary benefit refers to the perceived economic cost savings that consumers experience when switching from gasoline to new energy vehicles. Electric vehicles (EVs) powered by electricity are more affordable to travel with compared to fuel vehicles [61]. The perceived reduction in travel costs from EVs is more pronounced when gasoline prices rise sharply and remain elevated. The higher the desired cost reduction, the higher the perceived monetary benefit [44]. Wang and Zhang et al. [62,63] showed that perceived monetary benefits have a significant positive impact on consumers’ purchase intentions and are the crucial path that affects consumer purchase intentions. The most direct manifestation of the benefits is not only gasoline prices but also vehicle prices, government purchase subsidies, and tax incentives. Consumer perception of the benefits before moving to new energy vehicles is an essential prerequisite for the switching intention to occur. The following Hypotheses, 8 and 9, are proposed.

Hypothesis (H8).

Perceived monetary benefit has a positive impact on self-efficacy.

Hypothesis (H9).

Perceived monetary benefit has a positive impact on switching intentions.

2.2.2. Self-Efficacy

Self-efficacy (SE) refers to speculation and judgment by action subjects on whether the self can achieve goals [64,65]. In SCT, self-efficacy is an effective driver of individual behavior. Tsai et al.’s studies indicated that self-efficacy directly affects user intention to share knowledge [16]. Some studies also concluded that individuals with higher self-efficacy have more confidence in their abilities, are more committed to their behavior in the face of difficulties, and, therefore, are better able to control the implementation of their behavior [66].

This study defines self-efficacy as consumers being able to directly and effectively perceive whether they have the strength and ability to switch to NEVs. Lin et al. [67] showed that self-efficacy promotes green consumption behavior positively and that enhanced self-efficacy in green consumption is a core element in determining whether green consumption is implemented. Consumers with high self-efficacy are confident in their abilities and will try hard to complete their purchases [68]. In this studied scenario, even if consumers tend towards pro-environmental behavior, assessments on the degree of difficulty in switching to NEVs should be at the top of the list, and these assessments cover convenience, economic status, etc. Especially when gasoline prices rise, subsidies are removed, vehicle prices fluctuate, etc., consumers’ evaluation of their economic situation will heavily determine whether or not to purchase NEVs and switch their travel mode. Thus, we propose the following hypotheses.

Hypothesis (H10).

Self-efficacy has a positive impact on attitude.

Hypothesis (H11).

Self-efficacy has a positive impact on switching intentions.

2.2.3. Attitude

Attitude (ATT) is the psychological state that reflects an individual’s preference or dispreference for specific things [18,69]. This research refers to the favorability of individuals switching to NEVs. The SCT suggests that an individual’s attitude toward behavior can accurately predict their intention to engage in that behavior [69,70]. Lin and Lee found that knowledge sharing attitudes were significantly associated with encourage knowledge sharing intentions [22].

A study of consumers’ intentions to switch from BS4- to BS6-compliant vehicles with emission standards in India found that consumers’ attitudes towards BS6 vehicles with emission standards were associated with environmental protection preferences, and consumers holding positive attitudes towards environmental protection would prefer BS6 vehicles [71]. This study hypothesizes that individual attitudes regarding the environment and fuel prices will reflect their psychological attitudes toward using NEVs, with favorable feelings toward NEVs motivating them to abandon their existing fuel vehicles in favor of NEVs. Therefore, we propose Hypothesis 12.

Hypothesis (H12).

Attitude has a positive impact on switching intentions.

2.3. Perceived Risk

Perceived risk (PR) refers to the unpredictability of the outcome perceived by individuals when accepting new technology or using new services [72]. It is strongly subjective; if an individual perceives risks in this behavior, then his final decision-making may be affected. Currently, scholars prefer to link “perceived risk” to consumer psychological activities during the purchase process. Kamal and Chen et al. [73,74] indicated that consumers affected by perceived risk largely change, delay, or cancel their purchase decision making. Mitchell [75] concluded that consumers are more inclined to risk aversion rather than maximizing their benefits when purchasing products, which makes the explanation of perceived risk in consumer behavior stronger and more powerful. Multiple dimensions of perceived risk have been confirmed in studies of NEV purchase intentions to significantly negatively influence vehicle purchase intentions [76,77,78]. Since NEVs involve many technological innovations, they are often accompanied by multiple dimensions of risk. In this research context, perceived risk is defined as the subjective assessment by individuals of negative effects that may be triggered by the adoption of NEVs, including fears about performance safety, long charging times, operational complexity, and other risks. These concerns will prevent them from choosing NEVs. Hence, Hypothesis 13 is proposed below.

Hypothesis (H13).

Perceived risk has a negative impact on switching intentions.

2.4. Infrastructure Barrier

The infrastructure barrier (IB) refers to the degree of completeness in charging facilities for NEVs [79]. This research refers to the individual perceived charging facility status of NEVs. As an essential ancillary facility for electric vehicles, the number and availability of charging stations are significant factors affecting the promotion of NEVs. The imperfection of charging facilities will lead to mileage anxiety among consumers, which, in turn, will affect the adoption of NEVs [80,81]. Numerous studies show that consumers are highly focused on the timely charging of NEVs; even individuals interested in NEVs will fear running out of power halfway through journeys if no sufficient charging infrastructure is provided within reasonable driving distances. This so-called “range anxiety” will inhibit the market spread of NEVs [82,83,84]. In SCT research, Schade and Schuhmacher [85] examined digital infrastructure as an external factor to explore its impact on entrepreneurial behavior and showed that digital infrastructure significantly drives entrepreneurial behavior. Wang et al.’s research indicates that improved infrastructure positively influences the intentions to purchase NEVs [62]. In this research, the favorable perception of new energy vehicles indirectly reflects the demand for charging infrastructure, which will prompt individuals to take into account the infrastructure implications before switching to NEVs. Hence, Hypothesis 14 is proposed.

Hypothesis (H14).

Infrastructure barriers have a positive impact on switching intentions.

3. Research Design and Methods

3.1. Measurement

A targeted survey for this study was conducted to understand the intentions of Chinese residents switching to NEVs under the influence of multiple factors, such as high gasoline prices and extended government subsidies for NEVs. Based on the research hypotheses and review, a questionnaire-based survey was designed for collecting empirical data on Chinese residents’ willingness to switch to NEVs. The content of the questionnaire is composed of three segments. The first segment describes the context, intention, and NEV definition, and finally states that the questionnaire will have no involvement in the privacy of the participants and that their participation is appreciated. The second segment of the questionnaire mainly contains the socioeconomic characteristics of the respondents. The final segment of the questionnaire focuses on the factors in SCT that may impact people’s intentions to switch to NEVs. To ensure the validity and credibility of the questionnaire content, a professor-led team of questionnaire designers was formed. By reviewing the relevant research literature, such as SCT and the theory of planned behavior, reference sources for the question items of the questionnaire in this research were identified, and through extensive discussions and modifications by the team members, the question items for the questionnaire constructs were determined. Nine constructs were measured in the questionnaire, each with more than three items. All items were measured on a seven-point Likert scale ranging from 1 to 7 (1 = strongly disagree; 7 = strongly agree).

Table A1. Measurement items.

Construct	Items	Sources
Subjective norm (SN)	SN1: If my friend buys new energy vehicles, I will also switch to new energy vehicles.	[71]
	SN2: If my family members and relatives buy new energy vehicles, I will also switch to new energy vehicles.
	SN3: If my colleagues and leaders buy new energy vehicles, I will also switch to new energy vehicles.
Environmental consciousness (EC)	EC1: I think environmental problems are very important.	[103]
	EC2: I think everyone has a responsibility to protect the environment.
	EC3: I think we should care about environmental problems.
Face consciousness (FC)	FC1: Compared with traditional oil vehicles, switching to new energy vehicles can improve people’s impression of me.	[104]
	FC2: Compared with traditional oil vehicles, switching to new energy vehicles will bring me much more social approval.
	FC3: Compared with traditional oil vehicles, the brands of new energy vehicles are more well-known.
Perceived monetary benefit (PMB)	PMB1: With gas prices remaining high and continuing to rise, driving a new energy vehicle will help me save on travel costs.	[105]
	PMB2: Compared with traditional fuel vehicles, the daily use and maintenance cost of new energy vehicles is lower.
	PMB3: Compared with traditional fuel vehicles, the price of new energy vehicles is lower.
Self-efficacy (SE)	SE1: If I want to buy, I can buy a pure electric vehicle.	[106]
	SE2: I have no difficulty deciding whether to buy a pure electric vehicle.
	SE3: I decide whether to buy a pure electric vehicle free from public opinion.
Attitude (ATT)	ATT1: With oil prices staying high and continuing to rise, I think it is necessary to switch from traditional fuel vehicles to new energy vehicles.	[103]
	ATT2: With oil prices staying high and continuing to rise, I think switching from traditional fuel vehicles to new energy vehicles is a good choice.
	ATT3: With oil prices staying high and continuing to rise, I think it’s a good decision to switch from traditional fuel vehicles to new energy vehicles.
Infrastructure barrier (IB)	IB1: New energy vehicles is available at infrastructure at shopping malls, restaurants and entertainment places.	[79]
	IB2: New energy vehicles infrastructure is available at work.
	IB3: New energy vehicles infrastructure is available at home.
	IB4: New energy vehicle infrastructure is available in the highway service area.
Perceived risk (PR)	PR1: New energy vehicles are more likely to have accidents.	[48]
	PR2: New energy vehicle performance and quality may not meet expectations.
	PR3: New energy vehicles’ batteries are more likely to be damaged and burned.
Switch intention (SI)	SI1: With fuel prices staying high and continuing to rise, I will prefer to switch from traditional fuel vehicles to new energy vehicles.	[107]
	SI2: With fuel prices staying high and continuing to rise, I will prefer to try to switch from traditional fuel vehicles to new energy vehicles.
	SI3: With fuel prices staying high and continuing to rise, I will surely switch from traditional fuel vehicles to new energy vehicles.

(see Appendix A) presents the nine constructs and their items in this study.

3.2. Sample and Data Collection

This research mainly collected data through a questionnaire survey, which was provided by the largest professional online survey website in China, “www.wjx.cn, (accessed on 10 March 2022)”. The website has 6.2 million registered subscribers in China, with an average of more than 10 million users filling out questionnaires daily, and the subscribers are in each provincial administrative region within China [10,86,87]. Because of the vast and uniformly distributed group of completed questionnaires, the requirements of this research were met. The survey and data collection were divided into two phases. The first phase was a pre-survey conducted by distributing questionnaires on the website and offline on 22 March 2022. In total, 118 questionnaires were collected, which were modified based on the feedback from the pre-survey. A second-stage formal survey was conducted through an online questionnaire posted on the website between 12 April and 16 April, 2022. Participants who completed the questionnaire received a reward of CNY 6, and a total of 500 responses were collected. Invalid responses with incomplete or illogical answers were excluded from the analysis. Notably, residents who already owned an NEV were excluded from the study as they were not the target population. Ultimately, a total of 464 valid responses were included in the analysis, resulting in an effective response rate of 80%.

Table 1. Description of respondents (N = 464).

Demographic		Frequency	Percentage (%)
Gender	Male	249	53.66%
	Female	215	46.34%
Age group (years)	<25	87	18.75%
	26–35	302	65.09%
	36–45	57	12.28%
	46–55	14	3.02%
	>56	3	0.56%
Marital status	Married	337	72.63%
	Unmarried	127	27.37%
Education	Junior college or below	92	19.83%
	Bachelor’s degree	342	73.71%
	Master’s degree or above	30	6.47%
Monthly income (RMB)	<3000	46	9.91%
	3000–5000	63	13.58%
	5001–10,000	220	47.41%
	10,001–15,000	94	20.26%
	15,001–20,000	23	4.96%
	>20,000	18	3.88%
Has a driver’s license	Yes	437	94.18%
	No	27	5.82%
Household car ownership	0	66	14.22%
	1	367	79.09%
	2	29	6.25%
	≥3	2	0.43%

shows the socioeconomic characteristics of the valid sample, with 249 males and 215 females surveyed in a ratio of 1.15:1, which is close to the data counted in the “China Statistical Yearbook-2020” [88]. These respondents are mostly in the age range of 26–35 years old, accounting for 65.09%, and the proportion of married people is 72.63%. Regarding education level, these respondents generally have a bachelor’s degree or higher education, with a proportion of 80.17%. Almost 90% have a driver’s license, less than 10% do not have a driver’s license, and more than 80% of households own at least one vehicle.

3.3. Data Analysis

Structural equation modeling is a research tool that comprehensively examines the role of psychological and behavioral effects based on questionnaire data [89,90,91,92]. To examine the crucial factors affecting people switching to NEVs and test the interactions between them, a structural equation model was employed to analyze the data, testing the relationship between variables through covariance-based structural equation modeling (CB-SEM). There were two reasons for choosing CB-SEM to test the hypothesized model. Firstly, CB-SEM is frequently applied as the prime option for theoretical studies with large sample sizes and low dimensionality and is the common method to explore causal relationships between latent variables. Secondly, CB-SEM highlights the total fitness of models and tests the applicability of models, which is suitable for testing theoretical models [93]. Furthermore, the analysis of the SCT model proposed in previous studies typically adopts CB-SEM.

CB-SEM allowed us to test the reasonableness of hypothesized paths and analyze the effect relationships of various factors. Bayesian networks are based on probabilistic inference to graphically model the causal relationships between variables and consist of two main components: qualitative and quantitative [94,95]. In the qualitative part, the Bayesian network is a directed acyclic graph representing the relationship between variables, while in the quantitative part, every node indicates the conditional probability distribution of each variable. Bayesian joint probability distribution can be represented as:

$$P\left(X_1\dots X_n\right)={\prod }_{i=1}^{n}p(X_i|Pa\left(X_i\right)),$$

(1)

where

• Pa(X_i) is the parent of X_i.
• p(X_i|Pa(X_i)) is the conditional probability of each observed variable under the condition of its parent node.

The empirical power of CB-SEM in empirical validation combined with the predictive and diagnostic power of Bayesian networks contributes to effective support from causality identification to research decisions. After structural equation modeling, Bayesian network analysis was employed to infer and diagnose the significant factors.

4. Results

4.1. Measurement Model

For constructing the structural model in the following step, an evaluation of the reliability and validity of the questionnaire data is necessary. The reliability and validity outcomes of the data are listed in Table 2 and Table 3. Firstly, the reliability of the data was jointly reflected through Cronbach’s alpha coefficients and the composite reliability (CR). All constructed Cronbach coefficients ranged from 0.752 to 0.856, which was over the commonly recommended value of 0.7 [96]. Moreover, the composite reliability was above 0.7, indicating that the questionnaire data have better reliability [97]. Secondly, validity for data consists of convergent and discriminant validity. Item factor loadings and average variance extracted values (AVE) are effective indicators to appraise the convergent validity of data. Findings calculated through confirmatory factor analysis using AMOS V21.0 are shown in Table 2, and all factor loading coefficients are over 0.6. Since AVE and CR are not available through AMOS software directly, the CR and AVE values in this research were calculated using the following formula.

$$CR=\frac{{(\sum \lambda )}^2}{{(\sum \lambda )}^2+\sum \epsilon },$$

(2)

$$AVE=\frac{{(\sum \lambda )}^2}{N},$$

(3)

where:

• λ indicates the standardized factor loading coefficient.
• ε shows the error value of the measured variable.
• N refers to the number of measurements of the variable.

The AVE values were calculated to reside between 0.505 and 0.620, all above the threshold of 0.5, showing that the data have favorable convergent validity [98]. Table 3 presents the results of the discriminant validity, which shows that the square root of the variable AVE is stronger than the correlation coefficient of other variables, resulting in favorable discriminant validity of the data [98]. In summary, our analysis indicates that the data have good reliability and validity.

Table 2. Reliability and convergent validity analysis.

Construct	Item	Factor Loading	AVE	Cronbach’s α	Composite Reliability
SI	SI1	0.840	0.620	0.826	0.830
	SI2	0.796
	SI3	0.721
SN	SN1	0.737	0.567	0.796	0.797
	SN2	0.735
	SN3	0.786
EC	EC1	0.812	0.539	0.774	0.777
	EC2	0.711
	EC3	0.673
FC	FC1	0.714	0.671	0.856	0.859
	FC2	0.883
	FC3	0.851
ATT	ATT1	0.687	0.505	0.752	0.753
	ATT2	0.718
	ATT3	0.725
SE	SE1	0.772	0.611	0.824	0.825
	SE2	0.823
	SE3	0.749
PMB	PMB1	0.844	0.571	0.794	0.799
	PMB2	0.700
	PMB3	0.715
PR	PR1	0.755	0.617	0.827	0.828
	PR2	0.827
	PR3	0.772
IB	IB1	0.799	0.585	0.848	0.849
	IB2	0.779
	IB3	0.770
	IB4	0.709

Table 2. Reliability and convergent validity analysis.

Construct	Item	Factor Loading	AVE	Cronbach’s α	Composite Reliability
SI	SI1	0.840	0.620	0.826	0.830
	SI2	0.796
	SI3	0.721
SN	SN1	0.737	0.567	0.796	0.797
	SN2	0.735
	SN3	0.786
EC	EC1	0.812	0.539	0.774	0.777
	EC2	0.711
	EC3	0.673
FC	FC1	0.714	0.671	0.856	0.859
	FC2	0.883
	FC3	0.851
ATT	ATT1	0.687	0.505	0.752	0.753
	ATT2	0.718
	ATT3	0.725
SE	SE1	0.772	0.611	0.824	0.825
	SE2	0.823
	SE3	0.749
PMB	PMB1	0.844	0.571	0.794	0.799
	PMB2	0.700
	PMB3	0.715
PR	PR1	0.755	0.617	0.827	0.828
	PR2	0.827
	PR3	0.772
IB	IB1	0.799	0.585	0.848	0.849
	IB2	0.779
	IB3	0.770
	IB4	0.709

Table 3. Discrimination validity analysis.

	PR	FC	EC	SN	PMB	SE	SI	ATT	IB
PR	0.785
FC	0.081	0.819
EC	−0.131	−0.015	0.734
SN	−0.022	0.486	0.358	0.753
PMB	−0.106	0.413	0.411	0.719	0.756
SE	−0.165	0.383	0.459	0.690	0.741	0.782
SI	−0.190	0.333	0.540	0.744	0.775	0.760	0.787
ATT	−0.320	0.350	0.525	0.597	0.647	0.626	0.705	0.711
IB	−0.123	0.475	0.258	0.506	0.579	0.562	0.598	0.498	0.765

Note: The bold numbers indicate the square root of the AVE.

Table 3. Discrimination validity analysis.

	PR	FC	EC	SN	PMB	SE	SI	ATT	IB
PR	0.785
FC	0.081	0.819
EC	−0.131	−0.015	0.734
SN	−0.022	0.486	0.358	0.753
PMB	−0.106	0.413	0.411	0.719	0.756
SE	−0.165	0.383	0.459	0.690	0.741	0.782
SI	−0.190	0.333	0.540	0.744	0.775	0.760	0.787
ATT	−0.320	0.350	0.525	0.597	0.647	0.626	0.705	0.711
IB	−0.123	0.475	0.258	0.506	0.579	0.562	0.598	0.498	0.765

Note: The bold numbers indicate the square root of the AVE.

4.2. Structural Model

Following the determination regarding the reliability and validity of the questionnaire data, structural equation modeling was initiated to analyze the factors influencing households’ intention towards switching from careless or fuel vehicles to NEVs while examining the interactions of the factors. The SEM model was established using AMOS V21.0 software, shown in Figure 3 below. With the evaluation of the model fit quality by multiple indicators such as CMIN/DF, GFI, RMSEA, CFI, IFI, and PGFI, evaluation indicator results shown in

Table 4. Results of model fit texting.

	Indicators	Criterion	Results	Judgment
Absolute fit measures	CMIN/DF	<3.0	1.680	Yes
	GFI	>0.9	0.924	Yes
	AGFI	>0.9	0.904	Yes
	SRMR	<0.05	0.047	Yes
	RMSEA	<0.08	0.038	Yes
Incremental fit measures	NFI	>0.9	0.917	Yes
	CFI	>0.9	0.965	Yes
	IFI	>0.9	0.965	Yes
	RFI	>0.9	0.903	Yes
Parsimonious fit measures	PGFI	>0.5	0.731	Yes
	PNFI	>0.5	0.779	Yes
	PCFI	>0.5	0.819	Yes

Note: Source: [99] CMIN/DF = discrepancy divided by degree of freedom, GFI = goodness-of-fit statistic, AGFI = adjusted goodness-of-fit statistic, SRMR = standardized root mean square residuals, RMSEA = root mean square error approximation, NFI = normative fit index, CFI = comparative fit index, IFI = incremental fit index, RFI = relative fit index, PGFI = parsimony goodness-of-fit index, PNFI = parsimony normed fit index, PCFI = parsimony comparison fitting index.

, all the indicators yielded results above the standard values; hence, we can assume that the fit of the model to the survey data is within the acceptable range. The model can be utilized to evaluate the implications of switching intentions.

The analysis of the relevant path coefficients and significance was utilized to determine whether the hypothesis was valid. Figure 3 shows the structural equation model with standardized path coefficients test results. Among the 14 hypotheses, all of them are valid except that perceived risk influences switching intention negatively (H13; β = −0.028, p > 0.05). Among the personal factors, the path coefficient of attitude accounted for the largest share, followed by perceived monetary benefits and self-efficacy. This indicates that among the personal factors, attitude has an essential effect on consumers’ switching intentions. Perceived monetary benefits also play a considerable role in consumers’ switching intentions. In environmental factors, subjective norms representing news coverage and the influence of relatives and friends have the largest path coefficient of affecting consumers’ vehicle switching intentions. Finally, the external factor infrastructural barriers is also a factor affecting consumer’s switching willingness decision making, with details shown in

Table 5. Results of hypotheses testing.

Hypothesis	Path	Path Coefficient	Results
H1	SN → SE	0.278 ***	Supported
H2	SN → ATT	0.209 *	Supported
H3	SN → SI	0.262 ***	Supported
H4	EC → ATT	0.341 ***	Supported
H5	EC → SI	0.107 *	Supported
H6	FC → ATT	0.143 *	Supported
H7	FC → SI	−0.100 *	Supported
H8	PMB → SE	0.577 ***	Supported
H9	PMB → SI	0.238 **	Supported
H10	SE → ATT	0.298 ***	Supported
H11	SE → SI	0.156 *	Supported
H12	ATT → SI	0.240 ***	Supported
H13	PR → SI	−0.028	Rejected
H14	IB → SI	0.140 **	Supported

Note: * p < 0.05, ** p < 0.01, *** p < 0.001.

.

4.3. Bayesian Network Analysis

Based on Thomas Bayes’ theorem Bayesian network (BN) approach, this research adopts Bayesian networks for inference from currently available evidence by visualizing the relationships between constructs through directed acyclic graphical situations to examine the factors that influence residential households switching into NEVs. Having identified the causal relationships between the various psychological factors influencing residents switching to NEVs in the prior structural equation modeling, this section utilizes Bayesian networks to predict and diagnose these causal relationships.

4.3.1. Bayesian Detection Prior Probabilities

Prior probability estimation is the foundation for diagnosis and inference in Bayesian networks. In this research, prior probabilities were acquired using Netica software learning. Netica software automatically determines the learning process through directed acyclic graphs created to calculate the conditional probability values for each variable. The determination of the probabilities mentioned above is based on the construction of the Bayesian network structure and, thus, a structural model with significant path analysis was employed as the Bayesian network structure for subsequent probability calculations and learning. Since the variables within the network structure are latent variables, variable-specific values are required. By weighting the factor loadings to calculate the latent variable scores as the raw data for probability estimation, each latent variable was classified into low, medium, and high states with the K-means clustering method. Ultimately, the prior probabilities were calculated the network structure created by Netica, as shown in

Table 6. Prior probabilities of psychological latent variables.

State	Variable
	SI	ATT	SN	SE	FC	PMB	IB	EC
Low	0.283	0.235	0.371	0.193	0.542	0.244	0.184	0.081
Medium	0.308	0.419	0.383	0.241	0.205	0.420	0.270	0.118
High	0.409	0.346	0.246	0.566	0.253	0.336	0.546	0.801

. Table 6 shows that individuals generally have high environmental consciousness, and the proportion of individuals who intend to switch to NEVs reaches 40.9%.

The determination of conditional probabilities is the most crucial issue for Bayesian network models. As the Bayesian network constructed in this research involves psychological variables, the expectation maximization (EM) algorithm was applied to automatically update the Bayesian model [100]. Figure 4 presents the computed conditional probability distribution results. Combining the interaction between all the factors, the forecasts indicate about 51 out of every 100 household residents will switch from conventional fuel vehicles to NEVs.

Following the construction of the model and acquisition of relevant parameters, the predictive performance of Bayesian network models needs to be evaluated. Error rates and confusion tables are metrics often utilized to assess Bayesian network performance. The results are listed in

Table 7. Bayesian network confusion matrix and error rate results.

Error Rates and Confusion Tables				Error Rate	Total Error Rate
Forecasting results			Actual
Low	Medium	High
6	0	1	Low	14.29%	14.89%
2	6	1	Medium	33.34%
1	2	28	High	8.68%

. We extracted 10% (47) of the total sample at random as the test set, while the others were used as the training set for testing the prediction effectiveness of the model. Among the samples drawn, the structural equation model predicted residents switching to NEVs at low levels, with an accuracy of 85.71%. The prediction accuracy was 66.66% and 91.32% for the medium and high level situations of switching intention, respectively. As such, it was deemed that the prediction accuracy of the Bayesian network model constructed in this research was within an acceptable range.

4.3.2. Bayesian Inference and Diagnostics

The prediction function of Bayesian networks is an essential technique for the application of Bayes’ theorem. Bayesian networks form statistical data into a priori probability distributions based on Bayes’ theorem, then conditional probabilities are integrated into the model, which combines the prior knowledge and posterior data for inference and diagnosis, thus realizing the prediction function. The main idea is that each child node recalculates the probabilities of the other nodes as it acquires new information; specifically, with prior probabilities of the nodes known, inference from cause to effect is described as Bayesian inference, and inference from effect to cause is defined as Bayesian diagnosis. Based on the prior probabilities of each variable, the conditional probabilities of the three states (low, medium, and high) were obtained by adjusting the prior probabilities for each variable individually. For instance, if individuals have low attitudes toward NEVs (the probability of “low” attitudes is set to 1), the probability of “low” switching intentions is 0.324 through Bayesian inference. Figure 5 reflects the inference results affecting switching intentions. The horizontal coordinates indicate the three states at low, medium, and high levels of the psychological factors that influence individuals switching to NEVs, while the vertical coordinates indicate the probability change in the occurrence of the switching intention as the states are switched. All variables are positive for switching intention except face consciousness, which has a negative effect on switching intention. Among them, attitude and self-efficacy have the greatest influence. When self-efficacy changed from the low to high state, the switching intention in high states grew from 35.9% to 58.8%, while when the attitude was at the high level, the individual’s intention about switching to NEVs reached very high compared to the influence of all variables, rising to 61.6%.

Bayesian diagnosis is the reverse operation of Bayesian inference. Specifically, Bayesian diagnosis is to analyze the “causes” of the “effect”. It obtains the change in the independent variable through the state of the dependent variable. Figure 6 shows the results of Bayesian diagnosis for switching intention at high and low states. When switching intentions remained at high status, the probability of high status for all variables except face consciousness showed steady upward trends, with the probability of high status for attitude and self-efficacy increasing from 0.346 and 0.566 to 0.40 and 0.665, respectively, showing the largest increases relative to pre-diagnosis (increases of 17% and 16%, respectively). Similarly, while in the low state, the probability of low states increased for all variables. The attitude changed from 23.5% before diagnosis to 34.4%; the growth was 46%. The Bayesian diagnostic results indicate that personal factors have stronger direct effects on intention generation compared to environmental factors. Among them, attitude and self-efficacy had the strongest effect on intention.

Finally, key factors influencing people’s intention to switch to NEVs were identified through sensitivity analysis. Sensitivity analysis minimizes the prediction uncertainty of the target node by detecting specific influence nodes. Mutual info in

Table 8. Sensitivity to findings of switching intention.

Latent Variable	Mutual Info	Percent (%)
ATT	0.03163	2.14
SE	0.02474	1.68
EC	0.01932	1.31
IB	0.01832	1.24
PMB	0.01113	0.753
SN	0.00926	0.627
FC	0.00176	0.119

indicates the amount of information shared between two variables; larger mutual info between two variables indicates the stronger dependence between them. Table 8 shows that attitudes, self-efficacy, environmental consciousness, and infrastructural barriers have the greatest impact on switching intentions, whereas subjective norms and face consciousness were less effective.

5. Discussion

This study attempts to construct factor relation models of consumer vehicle SI from SCT in the context of record high fuel prices in China and the imminent removal of subsidy policy, and integrates structural equation modeling with Bayesian networks to quantify the effect degree and infer and diagnose the significant relationships between factors from the perspective of probability. The following section discusses the model results by dividing them into three sections.

For environmental factors, the SEM results show that subjective norms, environmental consciousness, and face consciousness have immediate significant effects on personal factors. They also showed that subjective norms and environmental consciousness exerted positive effects on consumer vehicle SI (β = 0.114, p < 0.05; β = 0.095, p < 0.05). Face consciousness has a significant positive effect on consumer attitudes but a negative effect on switching intentions, which is inconsistent with previous results. We propose that this is related that market development characteristics. According to the 2022 auto industry production and sales released by the China Association of Automobile Manufacturers, China’s passenger car market is characterized by “high-end fuel vehicles and comprehensive NEVs” [101]. The switch from fuel to NEVs does not provide consumers with a perceived improvement in image and status. Consistent with the conclusions of previous related SCT studies [19,20,21], environmental factors serve an essential function in explaining changes in individuals’ behavioral intentions, while having significant direct effects on behavioral intentions. The Bayesian inference and diagnosis results showed that when the subjective norm and environmental consciousness nodes varied from the low to high level, the high-state vehicle SI presented an increasing trend; conversely, when the node switching intention was in the high state, the prior probability of high-level subjective norm and environmental consciousness rose from 24.6% and 80.1% to 28.3% and 86.3%. The sensitivity analysis shows that environmental consciousness exerts greater sway over switching intentions. These conclusions show that when gasoline prices rise sharply to break through to record highs, the perceived external environment will have consequences for the psychological tendency of vehicle switching intentions. Firstly, as a technological innovation product, NEVs are novel to consumers who have never come into contact with them, and most people have a cautious wait-and-see attitude toward new things. At this point, the positive evaluation of NEVs by surrounding friends and news reports will definitely affect consumers’ propensity to switching. In addition, emotional approval from family members makes significant contributions to consumers’ behavioral decisions; if the surrounding family members and friends favor NEVs, the individual’s intention to switch to NEVs will rise. Secondly, the power source of new energy vehicles is relatively clean, they have low emission pollution, and they are quiet compared to fuel vehicles. When consumers perceive and realize that the adoption of NEVs will preserve the environment, protect ecology, relieve fossil fuel dependence, and raise our quality of travel and living environment, they will then subjectively evaluate the behavioral activities of switching to NEVs, which will generate bright visions and expectations of switching to NEVs, and ultimately they tend to select NEVs as their usual travel mode in the future.

For personal factors, the SEM outcomes indicated that perceived monetary benefits, self-efficacy, and attitudes all significantly and directly affected personal vehicle SI (β = 0.238, p < 0.01; β = 0.156, p < 0.05; β = 0.240, p < 0.001). The direct effect of personal factors on behavioral intention alteration is shared with the conclusions of previous studies of SCT [15,16,27]. The Bayesian inference and diagnosis showed that when perceived monetary benefit, self-efficacy, and attitude node switched from the low to high states, the high level of switching intention tended to rise. Moreover, the attitude and self-efficacy nodes were the sub-nodes that produced the maximum increase in switching intention. When switching intentions were in the low states, perceived monetary benefits, self-efficacy, and attitudes all showed an upward trend at low levels as well. The sensitivity analysis outcomes also reflect the significance of attitudes and self-efficacy. This indicates that personal factors play an invaluable role in consumer vehicle SI. As major adopters and purchasers of new energy vehicles, certain financial ability is the prerequisite, especially when factors such as rising gasoline prices, retreating subsidies, and vehicle price fluctuations affect the consumers’ financial capability related to purchasing and adoption. It is inevitable for consumers to evaluate whether they have sufficient financial capacity when making the behavioral switch to new energy vehicles for travel. The model also showed that infrastructure barriers positively promote consumer switching to NEVs (β = 0.140, p < 0.01), which is consistent with previous findings [62]. The availability and convenience of charging infrastructure affect consumers’ vehicle switching intentions to some degree. Despite the fact that the perceived risk is not significant in this study, we believe that firstly, vehicle manufacturers have enhanced the security of new energy vehicles, secondly, positive government and corporate advocacy have contributed to the low perceived risk of new energy vehicles, and finally, the massive accumulation of users for new energy vehicles has generated a positive reputation in China.

In the Bayesian network model, the Bayesian inference and diagnosis results indicate that the factors of attitude, self-efficacy, environmental consciousness, and infrastructure barriers are crucial predictors of consumer intentions toward switching to new energy vehicles. This conclusion was further corroborated by the results of the sensitivity analysis. Meanwhile, the results of structural equation modeling confirm the correlation relationship between personal factors, environmental factors, and consumers’ vehicle switching intentions. From the Bayesian network model and structural equation model results, both models exhibit consistent findings, indicating that the causal relationships between variables derived from the structural equation analysis are reliable, which is in line with previous research findings [102]. Bayesian networks quantify the uncertainty of parameters and structure among variables through probabilities, effectively identifying key determining factors that affect consumers’ intentions to switch to new energy vehicles, and providing more comprehensive and compelling evidence for the conclusions of structural equations. By combining SEM with Bayesian network models, we identified key impact factors on consumers’ vehicle switching intentions, leveraging the empirical evidence-based confirmatory power of SEM while highlighting the Bayesian network’s ability to handle nonlinear relationships and its predictive and diagnostic capabilities.

6. Conclusions and Policy Implications

The significant increase in gasoline prices has the potential to drive consumers toward the adoption of new energy vehicles. However, the reduction in government subsidies may impede consumer behavior toward switching, and the inadequate charging infrastructure in China may also act as a barrier to consumer adoption. The combination of these factors could potentially impact consumer decision making. This study is helpful in analyzing the mechanisms that may affect consumers’ switching to new energy vehicles and in understanding and predicting the changing trends of consumers’ switching intention to new energy vehicles under the impact of multiple factors, which may provide effective countermeasures for the future formulation of relevant policy measures to accelerate the development of new energy vehicles. The SCT modeling results showed that personal factors had a significant positive impact on consumers’ willingness to switch from owning no vehicle or owning a traditional gasoline vehicle to adopting a new energy vehicle. Some variables related to environmental factors had a negative impact on the switching intention. Additionally, when the price of gasoline rises, consumers’ perception of the cost savings associated with new energy vehicles becomes a significant driving factor, thus increasing their intention to switch to a new energy vehicle. Bayesian network analysis indicated that under the influence of multiple factors, approximately 51% of consumers had a high intention to switch to a new energy vehicle. Additionally, in terms of attitudes representing consumers’ intentions to switch new energy vehicles, self-efficacy, environmental consciousness, and perceived infrastructure readiness are significant predictors of individual vehicle switching intentions. The following policy implications are presented based on the research results.

First, enterprises should dynamically adjust their production, marketing, and vehicle pricing strategies in response to changes in gasoline prices. Governments should also make appropriate adjustments to subsidy policies and taxes based on gasoline price fluctuations. Whether an individual family owns fuel cars or is without a car, they will gradually switch to NEVs when the technology and infrastructure of NEVs become more mature in the future. However, the development of NEVs is related to green growth strategy initiatives and the realization of automotive strong country goals. Their implementation process will become the center of attention of the government, vehicle enterprises, and individuals. When gasoline prices rise sharply, the higher travel expenses drive people to clean, efficient, and cost-effective modes of travel. To rapidly increase their market share and improve their brand effect and competitiveness, vehicle manufacturers can dynamically adjust their R&D, production plans, and marketing strategies for NEVs based on factors such as gasoline prices and policy guidance, and appropriately adjust the prices of NEVs. For different income groups, they can customize the launch of models within the same price range, but also offer the purchase of vehicles to drive electricity bundled sales, etc. For the government, the push–pull policy can be implemented based on the gasoline price situation. When gasoline prices rise, the government can appropriately reduce the subsidies for NEVs and, conversely, appropriately increase them when gasoline prices fall. In the process of regulating and promoting NEVs, the government should also strengthen the regulation of the market.

Second, with the promotion of NEVs, the market demand for NEVs will expand and the market share will be further enhanced. The universality and convenience of charging facilities will focus the attention of users. It is recommended that the government should first formulate charging specification standards and regulatory mechanisms, followed by market demand analysis to popularize and improve charging supporting facilities, and then finally categorize and launch the service methods of power exchange, super-fast charging, fast charging, general charging, and slow charging according to the user needs of various locations. Improving consumers’ perception of charging infrastructure will further enhance people’s willingness to switch vehicles.

Third, environmental issues are also a major concern of society, and promoting NEVs already serves as a crucial measure to mitigate emissions in the automotive industry under the “double carbon” target. Raising people’s environmental awareness will facilitate consumers’ switching to NEVs. It is recommended that the government and vehicle manufacturers enhance the environmental effects brought by NEVs through short videos, posters, and slogans. The government can also impose pollution charges on existing fuel vehicles with varying emission standards and displacements, which is one measure to improve consumers’ intention to switch vehicles. Additionally, to continue growing the adoption of NEVs and further mitigate the pollution problems caused by traffic, the government could appropriately exempt private NEVs from highway tolls. The above policy recommendations from an environmental perspective will encourage consumers to switch to NEVs.

While our study has yielded interesting findings, it is not without limitations, which provide directions and suggestions for future research. Firstly, the sample collected in this study may not represent the attitudes of most consumers towards new energy vehicles after an increase in oil prices, as the online data collection conducted may not include consumers who do not use the internet, and the representativeness of online platforms may not be uniform across the country. For example, young people and well-educated groups are overrepresented in our sample, which may be due to the skewed participation of users of Questionnaire Star. Future research should expand to offline consumers and increase the sample size to further reduce sampling bias. Secondly, our study only focused on the influence of psychological latent variables on consumers’ conversion intention, but consumer groups with different genders, ages, and incomes have significant heterogeneity, which may lead to different attitudes toward switching to new energy vehicles. In future research, objective variables should be combined with psychological latent variables to explore the impact of the heterogeneity of different groups on the conversion to new energy vehicles, comprehensively considering the factors that affect consumers’ conversion intention. Lastly, our study only focused on consumers who either owned traditional fuel vehicles or did not own a vehicle at all, but not those who already owned new energy vehicles. Exploring their intention to switch through face-to-face interviews, semi-structured in-depth interviews, and other survey forms is worthy of future research.