Acceptance of Navigate on Autopilot of New Energy Vehicles in China: An Extended Technology Acceptance Model

Abstract

This study investigated the factors influencing user acceptance of the Navigate on Autopilot (NOA) functionality in new energy vehicles in China. An extended Technology Acceptance Model (TAM) was developed, incorporating additional factors such as social influence, travel scenarios, price value, perceived trust and perceived risk. A questionnaire survey was conducted in Guangzhou, China, and 260 valid responses were obtained. Structural equation modeling (SEM) was used to analyze the relationships between the factors. The results indicated that perceived ease of use, perceived usefulness, travel scenarios, price value, and perceived trust had significant positive effects on attitudes towards NOA, whereas social influence and perceived risk did not. Attitude was the primary determinant of the behavioral intention to use NOA. The findings suggest that to enhance NOA acceptance, new energy vehicle companies should emphasize specific application scenarios, reduce technology costs, provide value-added services, and strengthen user trust in the technology. This study contributes to the understanding of NOA acceptance and provides practical insights into the promotion of driver assistance systems in the context of new energy vehicles in China.

1. Introduction

In recent years, with the emergence of the global energy crisis and related environmental issues, countries have transitioned to a low-carbon economy [1,2]. The automotive industry plays a key role in low-carbon transformation [3]. Compared to traditional fuel vehicles (TFVs), new energy vehicles (NEVs) are more suitable for sustainable development and have become an important means of reducing carbon emissions in the automotive industry [4]. In China, government subsidies have significantly boosted the sales of NEVs, but research shows that incentive policies cannot be sustained in the long term; technological progress is the key to the promotion and development of new energy vehicles [5]. Consequently, new energy vehicles are increasingly adopting Advanced Driver Assistance Systems (ADAS) as a key selling point and a means to improve safety and convenience [6]. ADAS has become a standard feature in many NEVs. Consumer demand for technology and the competitive landscape of the new energy vehicle market are driving this trend [7].

NEV companies worldwide are studying Tesla’s autonomous driving technology and integrating NOA (Navigate on Autopilot) functions into their cars as a key selling point, such as XPeng’s NGP (Navigation Guided Pilot), Huawei’s NCA (Navigate Connected Autopilot), and NIO’s NOP (Navigate on Pilot) [8]. Although each automaker has a different name for the NOA, they are all advanced ADAS. These systems help drivers operate their vehicles more easily on highways or under specific urban conditions, enabling functions such as lane changing and entering/exiting ramps without human intervention. To enhance their marketing, some car companies define their products as “Level 2+,” indicating an intermediate stage between Level 2 and Level 3 [9]. However, according to the classification standards of the Society of Automotive Engineers (SAE) in the United States, from Level 3 onward, when an accident occurs in the autonomous driving mode, the responsibility shifts from the driver to the vehicle manufacturer [10]. Therefore, Level 3 vehicles are suitable only for specific areas and uses. Legally speaking, new energy vehicles with NOA functionality still fall under Level 2. The initial reaction of drivers to the promotion of NOA has been positive [11,12]. However, if drivers cannot truly accept these features and use them in real traffic scenarios, the technology cannot realize its potential. In other words, the widespread adoption of technology depends not on the technology itself but on the public’s willingness to use it [13]. Therefore, studying drivers’ acceptance of NOA is of great significance for its commercialization and the future popularization of higher-level autonomous vehicles.

In the automotive field, the Technology Acceptance Model (TAM) is a widely recognized and utilized theory [14,15,16]. Since its introduction by Davis in 1989, the TAM has become a classic framework for explaining technology acceptance behavior. This model holds that two beliefs—Perceived Usefulness (PU) and Perceived Ease of Use (PEOU)—are the core variables influencing behavioral intention to use a technology, thereby affecting actual usage behavior [17]. However, the TAM has certain shortcomings when it comes to explaining complex technologies, such as NOA. First, the perceived risk and trust factors were not considered. Research has found that, for advanced levels of autonomous driving, perceived risk is a major barrier to individuals’ use of the technology [18], whereas trust affects the extent to which individuals are willing to use it [19]. Therefore, perceived trust and perceived risk are crucial factors in studying willingness to use complex technologies in the automotive domain [20]. Second, it overlooks the application scenarios. Not all scenarios are suitable for advanced assisted driving. Generally, people are more willing to use advanced assisted driving in relatively simple scenarios (such as highways) [21]. Thus, the application scenario of the technology is an important factor in individual usage. Third, it did not consider basic or value-added services. While satisfying basic services, the provision of value-added services can significantly enhance individual satisfaction with a particular technology [22]. Fourth, it does not consider social environment or media perception. Research indicates that mass media increases potential users’ self-efficacy regarding autonomous driving technology, whereas social media strengthens their subjective norms. Both the usefulness and risk of technology are perceived through mass media, whereas risk perception can be significantly reduced via social media [23]. Therefore, it is essential to study the factors influencing people’s decisions to use NOA and provide feasible recommendations for new energy vehicle companies.

NOA is an intermediate technology that facilitates the transition from L2 to L3 autonomous driving. It is designed to assist drivers in specific scenarios such as highway travel or urban commuting by utilizing multiple sensors and algorithms to reduce driving fatigue. NOA is currently a major feature promoted by China’s emerging new energy vehicle manufacturers such as NIO (Hefei, China), XPeng (Guangzhou, China), and Li Auto (Beijing, China). In the early stages, scholars employed the TAM to study the acceptance of ADAS [24,25,26], revealing that perceived usefulness and perceived ease of use were important predictors of users’ adoption of ADAS. Recently, researchers have also used TAM to investigate the acceptance of fully autonomous driving technologies [27,28,29,30], uncovering that factors such as ease of use, usefulness, attitude, social norms, and trust influence user adoption of autonomous driving technologies. However, research on NOA has not yet been conducted.

Therefore, the objectives of this study are: (1) to propose an extended TAM suitable for NOA analysis, (2) to empirically validate the model using SEM analysis, and (3) to provide management recommendations to new energy vehicle companies based on practical needs. This study provides guidance for new energy vehicle companies on which technical aspects of the NOA function should be emphasized or overcome in their promotion efforts to offer users more reliable and user-friendly services and improve the quality of mobility services. Additionally, from a theoretical perspective, this research not only verifies new factors in the existing literature regarding acceptance but also offers a new perspective on the application of technology acceptance theory. The chapters are arranged as follows: Section 2: Literature Review, Section 3: Hypothesis Development, Section 4: Method, Section 5: Results, Section 6: Discussion, Section 7: Conclusions.

2. Literature Review

2.1. ADAS and NOA

ADAS refers to a system in which onboard sensors are used during vehicle operation to perceive and identify information about the surrounding environment. The system then processes and analyzes the sensory information to provide an early warning of danger to the driver. Building on this system, various technologies and functions for fully autonomous vehicles can be developed. The concept of the ADAS originated in the 1980s when the European Union launched a (Dedicated Road Infrastructure for Vehicle Safety in Europe) project, aiming to reduce the risk of traffic accidents and enhance driver awareness [31]. With technological advancements, the number of vehicles equipped with ADAS continues to increase [32]. ADAS typically include Adaptive Cruise Control (ACC), Blind Spot Monitoring (BSM), Parking Assistance (PA), Automatic Emergency Braking (AEB), Driver Alertness Warning (DAW), Lane Departure Warning (LDW), and Pedestrian Detection System (PDS) [33]. Overall, the ADAS was designed to improve driving safety [34].

The NOA is an advanced driver-assistance feature developed by Tesla. It is part of Tesla’s broader assisted-driving system and is designed to reduce the driver’s workload by assisting with driving tasks. The NOA was specifically designed to enhance the highway driving experience. By using the vehicle’s multiple cameras, ultrasonic sensors, and radar to monitor road conditions and employing an onboard computer to process data for driving decisions, it automates complex highway operations such as lane changes and entering or exiting ramps. Drivers are still required to supervise at all times and to be ready to take control [35]. Emerging car manufacturers, such as XPeng Motors [36], Li Auto [37], and Xiaomi [38], influenced by Tesla, have focused their research on advanced technologies to improve ADAS functionality. However, these studies emphasize technological aspects more than user perspective, resulting in a limited understanding of consumer expectations regarding NOA. As NOA technology has become more widespread, it is necessary to further understand the consumer acceptance process. This study aimed to construct a comprehensive research model validated by municipalities, considering multiple factors that influence consumer acceptance.

2.2. Technology Acceptance Model

The Technology Acceptance Model is a fundamental behavioral model designed to predict individuals’ intentions to use new technologies [39]. This explains how people’s beliefs and attitudes influence their willingness to use certain technologies. The Technology Acceptance Model proposes two key beliefs: Perceived Ease of Use (PUE) and Perceived Usefulness (PU). The former is defined as the degree to which an individual believes that using a new technology is effortless, whereas the latter refers to the actual benefits that the technology can provide. These two beliefs influence attitudes toward using technology, which in turn affects behavioral intention. In most cases, the intention can be directly translated into the actual behavior [40]. The Technology Acceptance Model has been applied in research on various technologies, including studies on the acceptance of autonomous driving in the automotive field [41]. Moreover, TAM is one of the most commonly used theories for understanding users’ acceptance of automation technologies [20].

According to TAM, perceived ease of use positively influences perceived usefulness. Both perceived ease of use and perceived usefulness positively influenced attitudes toward technology use, and perceived usefulness acted as a positive driving factor influencing individuals’ behavioral intentions. In this study, perceived ease of use refers to the degree to which drivers believe that using NOA is effortless. Perceived usefulness refers to the extent to which individuals believe that NOA can help drivers reduce their driving load. This is consistent with TAM assumptions. Existing research has found that perceived ease of use has a positive impact on perceived usefulness, and perceived usefulness also positively affects behavioral intention [42,43,44,45]. However, the impact of perceived ease of use on behavioral intention has yielded mixed results. Sakuljao et al. [42] and Zhang et al. [43] reported positive effects, whereas Nordhoff et al. [44] and Bernhard et al. [45] did not reach this conclusion. This study follows the assumptions of the TAM, positing that individuals who perceive NOA as both easy to use and useful are more likely to accept this function.

2.3. Potential Users

To provide a more comprehensive explanation of the public’s willingness to use new technologies, some studies have investigated the characteristics of potential users of fully autonomous driving technology, including gender, age, and educational level. In terms of gender [46], men are generally more proactive than women in adopting autonomous driving technology. Regarding age [47], there are significant differences between the younger and older groups, with younger individuals being more likely to accept technology. In terms of education level [48], groups with higher educational backgrounds tend to be more open to new technologies. Some researchers believe that today’s young people, who have grown up in an environment surrounded by new technology, may have a higher level of acceptance. Some scholars consider adults aged 18–50 years to be young and have studied this group’s acceptance of autonomous vehicles. NOA technology is a prerequisite for the development of fully autonomous driving and is also one of the key technologies for the future advancement of new energy vehicles. Given that the younger generation is heavily influenced by the Internet and that their consumer views and needs differ significantly from those of other age groups, ensuring that the development of NOA technology meets the potential car purchasing needs of young users is an urgent issue that needs to be addressed.

3. Hypothesis Development

TAM is a widely used theoretical framework for studying technology adoption, and its advantages can quantify consumers’ intentions to accept technology. By integrating the TAM framework, we incorporated additional perspectives that concern consumers, such as “social influence,” “perceived risk,” and “perceived trust” perceived trust, into the model, thereby extending the foundational TAM to better accommodate users’ acceptance of NOA technology.

3.1. Attitude

Researchers, drawing on the original TAM and focusing on end users, conceptualize technology acceptance by constructing the dimensions of “attitude” and “behavioral intention.” Behavioral intention describes an individual’s intention to use a technology in the foreseeable future. This relationship has been confirmed in related studies, where attitude is regarded as an important determinant of behavioral intention [49]. Therefore, attitude serves as a conduit through which external influencing factors affect behavioral intention. This aligns with the model we intend to develop, leading us to propose the following hypothesis:

H1. 

The attitude to using NOA positively influences the intention to use.

3.2. Perceived Usefulness and Perceived Ease of Use

The TAM posits that perceived usefulness and perceived ease of use are decisive factors that determine attitude. Previous research in the automobile industry has also indicated that perceived usefulness and perceived ease of use play pivotal roles in the adoption of automotive technologies [50,51]. Additionally, perceived usefulness has a positive influence on behavioral intention and can increase with higher perceived ease of use. Hence, we propose the following hypotheses:

H2a. 

The perceived usefulness of NOA positively influences attitudes to use.

H2b. 

The perceived usefulness of NOA positively influences intention to use.

H3a. 

The perceived ease of use of NOA positively influences attitudes to use.

H3b. 

The perceived ease of use of NOA positively influences perceived usefulness.

3.3. Social Influence

Social influence is another crucial factor, as social norms and others’ behaviors can significantly affect individuals’ willingness to adopt new technologies [52]. Moreover, social networks and social media can have a major impact on personal behavior and decisions [53]. People live within social contexts; therefore, when studying the acceptance of new vehicle technologies, such as NOA, it is important to consider the influences of people around them or social media. Thus, we included social influence as a factor in the model by adding a new dimension to the TAM framework. Relevant research has shown that when respondents see friends or family using autonomous vehicles, their attitudes toward autonomous driving also become more positive [54]. This confirms the relevance of social influence and leads to the following hypothesis:

H4. 

Social influence positively influences attitudes to use.

3.4. Travel Scenarios

Travel scenarios are critical factors in the implementation of ADAS and autonomous driving technologies. Particularly on highways [55], where driving tasks are simpler, highways have become one of the most likely scenarios for the deployment of autonomous driving. Urban scenarios [56] represent an important target for the popularization and implementation of autonomous driving; however, their complex environments and traffic behaviors present significant challenges. From the perspective of practical NOA use, whether its technology can address complex urban scenarios is related to users’ attitudes and perceived usefulness of the technology. Therefore, the following hypothesis is proposed:

H5a. 

The complexity of travel scenarios positively influences attitude to use.

H5b. 

The complexity of travel scenarios positively influences the perceived usefulness.

H5c. 

The complexity of travel scenarios positively influences the perceived ease of use.

3.5. Price Value

As the TAM focuses on consumers’ perspectives, the cost of technology influences their acceptance of new technologies. That is, how consumers respond to changes in product pricing–that is, the degree to which they accept price increases in light of anticipated benefits [57]. In studies of autonomous driving technology acceptance, this construct is often referred to as “price value.” Some have defined “price value” as “the trade-off consumers perceive between expected benefits and costs of use” [58]. The higher the price value, the easier it is to perceive the technology as useful and easy to use, which is positively correlated with attitude. Additionally, the benefits that consumers perceive from technology or additional value can serve as an intrinsic motivation for personal choices [59]. Based on this, we propose the following hypothesis:

H6a. 

Price value positively influences attitude to use.

H6b. 

Price value positively influences the perceived usefulness.

H6c. 

Price value positively influences the perceived ease of use.

3.6. Trust

Some literature [60] defines “trust” as “holding a positive expectation or a favorable attitude toward a technology’s future performance even in vulnerable environments.” Because NOA technology minimizes human intervention during driving tasks, user trust in the system’s ability to perform driving tasks such as following and lane changes is critical [61]. Some studies regard safety as the primary consideration of trust, asserting that potential users must believe that autonomously controlled vehicles are safe to travel in [62]. Research shows a positive correlation between trust in technology and technology acceptance [63,64]. Therefore, this study proposes the following hypothesis:

H7. 

Perceived trust positively influences the attitude to use.

3.7. Perceived Risk

Perceived risk represents the uncertainty in specific situations [65]. In consumer studies, this concept is defined as the probability of incurring losses in uncertain situations. NOA technology uses sensors to perceive the environment and make operational decisions for the vehicle, making risk assessment during operation essential [66]. Many people remain concerned about the safety of this technology, which may hinder their willingness to use it. For example, after an accident involving Xiaomi vehicles in China, many potential users developed doubts about NOA [67]. Moreover, energy vehicles are all linked to users’ mobile apps, generating data such as location and travel patterns during trips and raising privacy concerns [68]. Research in the vehicle sector indicates that perceived risk is a negative factor influencing consumers’ willingness to use such vehicles [69,70]. Based on these findings, we propose the following hypothesis:

H8. 

Perceived risk negatively influences attitude to use.

Finally, based on the original TAM, this model incorporated five additional variables: perceived trust, perceived risk, social influence, travel scenarios, and price value. Summarizing the above hypotheses, the research model of this study is shown in Figure 1.

4. Method

This study consisted of three main phases: the questionnaire design, data collection, and model building and analysis.

Step 1: The questionnaire was designed based on TAM theory, taking into account the characteristics of NOA technology in new energy vehicles. This encompasses factors, such as social influence, travel scenarios, price value, perceived trust, and perceived risk. Before the official release, a pilot test was conducted, and the questionnaire was revised based on participant feedback to ensure its reliability.

Step 2: Survey data were collected online and offline. The target respondents were users who either planned to purchase a car or already owned one. The questionnaire was uploaded to the online platform WJX (the most popular survey platform in China, similar to SurveyMonkey). Volunteers were recruited from schools to send the questionnaire link or QR code to friends and family members with car purchase intentions. In addition, questionnaires were distributed in mall car showrooms, 4S dealerships, and auto repair shops. The respondents completed the questionnaire by scanning QR codes on their mobile phones.

Step 3: IBM SPSS 19 and AMOS 26 were used for data analysis and modeling to ensure accuracy and rigor of the results. Through an in-depth exploration and discussion of the research findings, targeted conclusions and recommendations were provided.

4.1. Questionnaire Design

Based on previous research, this study designed a multidimensional questionnaire by setting and optimizing measurement items. The questionnaire consisted of three parts: the first part introduced the NOA functionality, the second part investigated acceptance, and the third part gathered personal information.

In the first part, the purpose of the survey and the contents of the questionnaire were explained, with a commitment to safeguarding personal information. Pictures and text were used to help the respondents understand the characteristics and functions of the NOA. The questionnaire embedded images with accompanying text to illustrate its applications in scenarios such as lane changing, overtaking, and ramp navigation, allowing respondents to gain a clearer understanding of its functions.

In the second part, the acceptance survey covered nine variables: perceived ease of use, perceived usefulness, attitude, behavioral intention, social influence, travel scenarios, price value, perceived trust, and perceived risk. The questionnaire was drawn from reliable scales from related research and modified appropriately according to the features of NOA. After completing the initial draft, volunteers were invited to provide feedback and the language and related concepts were refined accordingly. The second part utilizes a five-point Likert scale, where 1 represents “strongly disagree” and 5 represents “strongly agree.”

Table 1. Variable measurement items.

Items	Item Code	Question Item Content	Source
Perceived Usefulness (PU)	PU1	The NOA function can significantly enhance my driving efficiency (such as reducing commuting time).	Jiang et al. [51] Müller et al. [71]
	PU2	The NOA function makes long-distance driving easier for me (such as reducing fatigue).
	PU3	The NOA function can help me deal with complex road conditions.
	PU4	I think the NOA function is a necessary technology for future travel.
Perceived Ease of Use (PUE)	PUE1	I think the operation interface of the NOA function is simple and intuitive, and easy to learn	Jiang et al. [51] Müller et al. [71]
	PUE2	I think the system can respond quickly to my instructions (such as changing lanes and following the vehicle).
	PUE3	I think vehicle prompts (such as voice and screen display) are clear and easy to understand
Attitude (AT)	AT1	I believe that NOA technology is the future development direction of intelligent driving.	Jiang et al. [51] Müller et al. [71]
	AT2	I support taking the NOA function as the core selling point of new energy vehicles.
	AT3	I tend to give priority to models equipped with the NOA function.
	AT-Op2	I don’t think NOA will be the core selling point of new energy vehicles.
Behavioral Intention (BI)	BI1	I am willing to use the NOA function for a long time in the future.	Jiang et al. [51] Müller et al. [71]
	BI2	I will recommend car models equipped with NOA function to my relatives and friends.
	BI3	I will upgrade to a more advanced version of the NOA function.
Social Influence (SI)	SI1	I think the positive coverage of NOA technology by the media will influence my attitude towards this feature.	Sharmeen et al. [53] Mir et al. [72]
	SI2	I think the usage evaluations of the NOA function by other users on social media will affect my usage.
	SI3	I think if my close friends and relatives around me choose to use NOA.
Travel Scenarios (TS)	TS1	I think I will use the NOA function in the highway scenario.	Sharmeen et al. [53] Mir et al. [72]
	TS2	I think it is necessary to be more cautious when using the NOA function in complex road conditions (work zone).
	TS3	I will use the NOA function during long-distance self-driving trips.
Price Value (PV)	PV1	I think the current pricing of the NOA function is reasonable.	Sharmeen et al. [53] Mir et al. [72]
	PV2	I think the additional services such as automatic parking that come with the NOA function would be better.
	PV3	I think regular OTA upgrades will enhance my satisfaction.
	PV4	I think the NOA function training is very important to me.
Trust (TR)	TR1	I believe that car manufacturers have fully verified the safety and reliability of the NOA system.	Kenesei et al. [66] Khan et al. [73]
	TR2	I trust the performance of the NOA system in extreme weather conditions such as heavy rain and thick fog.
	TR3	I think the emergency response measures taken by car manufacturers for NOA function failures make me feel at ease.
Perceived Risk (PR)	PR1	I’m worried that the NOA system won’t be able to respond promptly in case of emergencies, such as pedestrians intruding.	Kenesei et al. [66] Khan et al. [73]
	PR2	I’m worried that the system might misjudge and cause driving dangers (such as incorrect lane changes).
	PR3	I’m worried that the driving data collected by the NOA function might be misused or leaked.

lists the variables and the questions used in this study.

Perceived ease of use, perceived usefulness, behavioral attitude, and behavioral intention refer to scales from Jiang et al. [51] and Müller et al. [71]; social influence, price value, and travel scenarios refer to scales from Sharmeen et al. [53] and Mir et al. [72]; perceived trust and perceived risk refer to scales from Kenesei et al. [66] and Khan et al. [73]. Each variable had 3 to 5 questions, with a total of 29 questions. To improve the validity of the questionnaire, a reverse question (AT-Op2) was designed. This question is the opposite of AT2, and is mainly intended to eliminate invalid questionnaires. This will not be counted again during statistical analysis. In addition, the questionnaire investigated payment methods including one-time buyouts, subscription fees, and free trials. It also asked respondents to select the reasons for choosing vehicles with NOA functionality.

The third part collected personal attributes including gender, age, annual household income, current type of household vehicle, and experience with NOA. The type of current household vehicle and whether the respondent has experienced NOA are likely to affect their NOA preferences.

Overall, this questionnaire provides a quantitative basis for researching the acceptance of NOA technology and offers important insights for new energy vehicle manufacturers in China as they improve and market such functions in the future.

4.2. Data Collection

To further expand the coverage of the sample and target potential users, survey volunteers were recruited, and questionnaires were distributed to friends and family members around us who had expressed an intention to purchase a car or already had a car in their household. At the same time, questionnaires were distributed in shopping malls, car dealerships (such as the Automobile Sales Service Shop 4S), car maintenance, and detailing shops. Figure 2 shows the QR code link for the questionnaire and offline survey locations (BYD 4S and Geely 4S). To alleviate concerns about privacy breaches and other risks among the participants, a clear statement about the confidentiality and anonymity of the data was provided at the beginning of the questionnaire.

The questionnaires were distributed between 17 February and 1 March 2025. All questionnaire responses were recorded using an online WJX platform. In total, 272 questionnaires were distributed. Of these, 12 were excluded due to incomplete information or repetitive responses. After the screening, 260 valid responses were obtained, yielding a response rate of 95.6%. Loehlin [74] suggested that the median sample size in the surveys should be 198. Barrett [75] suggested that the sample size should be greater than 200 but less than 500 because when the sample exceeded 500, the chi-square value could become severely inflated, leading to poor model fit. These findings further demonstrate the rationality of the sample size used in this study.

The survey adopted a combination of online and offline methods to ensure broader audience coverage, thereby enhancing data diversity and representativeness. Online data collection provides great convenience and efficiency, allowing users to complete the questionnaire anytime and anywhere, breaking the time and space limitations of traditional offline surveys. We trained volunteers to help them understand certain terms in the questionnaire so that they could introduce them to the respondents. Volunteers sent the Questionnaire Star link to friends and family interested in purchasing a car via WeChat. To ensure sample diversity, an offline questionnaire was designed to supplement the online channel. Offline surveys mainly involved visiting Automobile Sales Service Shop 4S frequented by groups intending to purchase new energy vehicles. Volunteers explained their intentions to 4S sales staff and customers and prepared keychains, stationery, and other items to increase their participation rate in the survey. Simultaneously, they printed prominent QR codes for respondents to quickly scan their phones and fill out the questionnaire. Combining both online and offline approaches ensured not only a sufficient sample size but also a certain degree of representativeness. The online channel helped us reach groups with a potential interest in buying energy vehicles, while the offline channel ensured direct contact with groups intended to make a purchase. The dual questionnaire method improved both the response rate and comprehensiveness of the data.

4.3. Structural Equation Modeling

This study used structural equation modeling (SEM) for quantitative data analysis. SEM is a confirmatory model suitable for the study and analysis of complex multivariate data. This method excels in handling latent variables and is mainly applied to the identification, parameter estimation, and verification of causal models, thereby making the research more scientific, objective, and accurate. The basic principle of SEM is to analyze the relationships among variables based on the covariance matrix calculated from covariance theory. The degree of fit between the hypothesized model and sample data was determined by comparing the degree of difference between the covariance matrix of the sample data and hypothesized model. The smaller the difference and the higher the goodness-of-fit, the more the hypothesized model is supported by the sample data, and vice versa.

A complete SEM consists of two basic models: measurement and structure. The measurement model describes how latent variables are conceptualized or measured by their corresponding observed indicators [76]. The measurement model consists of latent variables, observed variables, and measurement error terms that reflect the correspondence between the observed and latent variables. The relationship between the measurement model and the structural model is shown in Figure 3.

In SEM models, latent variables refer to variables that cannot be directly observed or measured, such as traits, or abstract concepts such as cognition, attitude, and intention. Latent variables are typically represented using elliptical or circular symbols. Observed variables are quantifiable variables that can be directly observed or measured and are usually used to represent observed variables that are often depicted with rectangular or square shapes. The higher the factor loading of an observed variable, the stronger its influence on the latent variable. Conversely, the lower the factor loading, the weaker the influence of the latent variable. The structural model describes the causal relationships among the latent variables and includes the residual portions of the “dependent variable,” which are not explained. Latent variables were further categorized into exogenous and endogenous latent variables. Exogenous latent variables are those in the model that are not influenced by any other variables but directly influence other variables. Endogenous latent variables refer to all other latent variables apart from exogenous ones. The mathematical expression of the structural model is the structural equation, as shown in Equation (1), and the vector composed of endogenous variables is shown in Equation (2).

$$\eta =B\eta +\Gamma \xi +\zeta$$

(1)

$$Y_i={ B}_0+B_1{X}_{i1}+{ B}_2{X}_{i2}+\cdots +B_p{X}_{ip}+{\epsilon }_i$$

(2)

where $B$ is regression coefficient matrix among endogenous variables; $\Gamma$ is regression coefficient matrix between exogenous variables and endogenous variables, $\zeta$ is endogenous variable residual term, ${\epsilon }_i$ is residual value of endogenous variable, $X$ and $Y$ are observed variables.

5. Results

5.1. Descriptive Statistics

5.1.1. Survey Respondents

The statistical characteristics of the samples are presented in

Table 2. Statistics of respondents’ basic information.

Factor	Levels	N	Percent (%)
Gender	Male	170	65.4%
	Female	90	34.6%
Age	18~25	68	26.2%
	26~40	138	53.1%
	41~50	47	18.1%
	>50	7	2.7%
Education	Below High School	15	5.8%
	High School Graduate	30	11.5%
	Bachelor’s degree	205	78.8%
	Master’s degree or higher	10	3.8%
Annual Household Income (CNY)	100,000–150,000 CNY	146	56.2%
	150,000–200,000 CNY	59	22.7%
	200,000–300,000 CNY	39	15.0%
	300,000–500,000 CNY	12	4.6%
	>500,000 CNY	4	1.5%
Vehicles Type Ownership	Traditional Fuel Vehicles	205	78.8%
	New Energy Vehicle (without NOA)	34	13.1%
	New Energy Vehicle (with NOA)	21	8.1%
Experience with NOA	Yes	66	25.4%
	No	194	74.6%

. Males accounted for 65.4% and females accounted for 34.6%, with more males than females. The age distribution was mainly concentrated under 50 years old, accounting for 97.3% of the population. According to the “2024 New Energy Vehicle Consumer Social Media Big Data Insight Report” by IPS Consulting, 64% of new energy vehicle users in China are male, while 36% are female [77]. The “2025 South China & Southwest China New Energy Vehicle Market Bulletin” by QuestAuto reveals that 92.3% of new energy vehicle users in South China are under the age of 50, with those aged 25–40 accounting for 54.1% [78]. Data from the Mob Research Institute show that 71.1% of new energy vehicle users have a bachelor’s degree or higher education. In terms of gender composition, age distribution, and educational background, the respondent group is largely consistent with the broader new energy vehicle user group, further demonstrating that the research sample is representative to a certain extent [79]. In addition, the annual household income of 78.9% is approximately 100,000–200,000 yuan, matching the mainstream price range of energy vehicles. Of the respondents, 78.8% still mainly used fuel vehicles, making them potential users for switching to new energy vehicles.

5.1.2. Acceptance Measure of NOA

Statistical survey data (Figure 4) show that men had a higher acceptance of NOA (overall average behavioral intention score of 3.86) than women (overall average behavioral intention score of 3.78); in other words, men were more likely to accept it. The younger group, around the age of 25 years showed the highest willingness to accept NOA (average score of 3.93), followed by the 25–50 age group (average score of 3.75), and those over 50 years scored 3.61, indicating that the younger the group, the stronger their acceptance of NOA. Respondents from households with gasoline cars (average score of 3.8) and those with electric vehicles (without NOA function) (average score of 3.82) showed almost no difference in acceptance of NOA, but families already owning energy vehicles with NOA functionality had an even higher acceptance of NOA (average score of 3.85). Regarding another question, respondents who had experienced the NOA function (average score 3.91) had a higher acceptance than those who had not (average score 3.8). This further indicates that experiencing a function has a significant effect on increasing acceptance.

We further analyzed the characteristics of the different age groups. We directly calculated the mean values of each variable for different age groups (18–25 years old, 26–50 years old, and over 50 years), as shown in

Table 3. Average scores of each variable across different age groups.

Age	18–25	26–50	>50
PU	3.8	3.84	3.78
PUE	3.75	3.88	3.76
PR	3.76	3.74	3.81
SI	3.68	3.67	3.71
TR	3.5	3.56	3.47
TS	3.7	3.55	3.29
PV	3.72	3.72	3.89
AT	3.88	3.95	4

. Although there are only 7 samples in the over-50 group, certain trends can still be observed. As shown in Table 3, in terms of perceived risk, the group over 50 years old is clearly higher than the other two groups, indicating that older adults are more concerned about the risks inherent in the technology itself. Similarly, in terms of perceived trust, the group aged over 50 years scored lower than the other two groups. On the value dimension, the over-50 group had the highest score, indicating that older users pay more attention to supporting services such as after-sales training. Among the young and middle-aged groups, the lowest-scoring variables were perceived trust (3.5 and 3.56, respectively) and travel scenarios (3.7 and 3.55, respectively). This suggests that in actual sales processes, it is important to focus on improving these two factors for young and middle-aged customers.

Table 4. Descriptive statistics for each item.

Item	Item Code	M	SD	Skewness	Kurtosis	Item M	Item SD
PU	PU1	3.980	0.808	−0.453	−0.063	4.010	0.536
	PU2	4.090	0.812	−0.642	0.160
	PU3	3.950	0.841	−0.486	−0.124
	PU4	4.040	0.663	−0.590	0.125
PUE	PUE1	3.980	0.788	−0.538	0.276	3.974	0.717
	PUE2	3.930	0.800	−0.324	−0.433
	PUE3	4.020	0.786	−0.556	0.281
PR	PR1	3.930	0.860	−0.666	0.337	3.912	0.766
	PR2	3.970	0.831	−0.628	0.426
	PR3	3.840	0.884	−0.496	0.086
SI	SI1	3.980	0.771	−0.483	0.257	3.953	0.692
	SI2	3.970	0.788	−0.430	0.024
	SI3	3.900	0.864	−0.529	−0.106
TR	TR1	3.830	0.759	−0.406	−0.130	3.750	0.759
	TR2	3.710	0.921	−0.312	−0.434
	TR3	3.660	0.947	−0.184	−0.751
TS	TS1	3.710	0.953	−0.526	−0.037	3.730	0.740
	TS2	3.820	0.896	−0.456	−0.329
	TS3	3.660	0.974	−0.275	−0.729
PV	PV1	3.570	0.803	0.094	−0.494	3.890	0.446
	PV2	3.980	0.773	−0.472	−0.049
	PV3	3.990	0.751	−0.153	−0.778
	PV4	4.040	0.516	−0.309	−0.283
AT	AT1	4.070	0.548	−0.335	−0.448	3.960	0.666
	AT2	3.900	0.816	−0.336	−0.443
	AT3	3.910	0.836	−0.512	0.197
BI	BI1	3.750	0.867	−0.244	−0.435	3.830	0.607
	BI2	3.840	0.669	−0.249	−0.295
	BI3	3.920	0.848	−0.642	0.560

Note: PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

presents the descriptive statistics and normality test results for the factors used in this study. According to the results of the descriptive statistical analysis, the mean scores for each variable ranged from 3 to 4, with the scale scored from 1 to 5 in the positive direction. Therefore, it can be seen that the group of participants in this study had a moderate or above-average level of NOA acceptance, showing a positive attitude. The normality of each measurement item was tested using skewness and kurtosis. According to the criteria proposed in [80], if the absolute value of the skewness coefficient is within 3, and the absolute value of the kurtosis coefficient is within 8, the data can be considered to meet the requirements for an approximate overall distribution. Based on the results in the table below, the absolute values of skewness and kurtosis for all measurement items in this study were within standard limits. Therefore, it can be concluded that the data for all measurement items satisfied the requirements for an approximate normal distribution.

5.1.3. NOA Payment Models

Currently, there are three main NOA payment models in China: buyouts, subscriptions, and free trials. The buyout model refers to obtaining the NOA function by paying a one-time fee for the intelligent driving package (e.g., Huawei ADS, 36,000 RMB) (Huawei Technologies Co., Ltd., Shenzhen, China). The subscription model allows customers to use NOA monthly (e.g., Huawei ADS, 720 RMB per month). The free trial model provides access to the NOA function for a certain number of years free of charge. After the trial period ended, continued use required additional payments. This study surveyed potential users in China regarding their acceptance of different NOA payment models. The results are shown in Figure 5. Among the samples, 162 individuals selected the free trial option, 85 selected buyout option, and only 13 selected subscription option. All three payment methods had an acceptance rate of over 50%, indicating that current users have a high level of recognition of NOA features in new energy vehicles. Among these, the free trial method had the highest acceptance rate, exceeding 80%. This suggests that users are relatively price-sensitive and also reflects that users may not expect to own a new energy vehicle for a long period. The acceptance rate for the buyout model (60%) was higher than that for the subscription model (55%). Users may feel that a buyout offers unlimited use, whereas paying a similar amount in the subscription model would allow only continuous use for a few years. This indicates that users have not considered their actual scenarios or the frequency of NOA use, and future promotion of the subscription model could be further tailored to meet users’ actual needs.

5.2. Reliability Test

In this study, the main factors were measured using scales; therefore, verifying the quality of the measurement data is an important prerequisite for ensuring the significance of the subsequent analysis. First, the Cronbach’s alpha reliability test was used to analyze the internal consistency of each dimension. According to reference [81], a reliability coefficient below 0.6 is considered unreliable, indicating that the questionnaire needs to be redesigned or data needs to be collected again for analysis. Reliability between 0.6 and 0.7, considered reliable; 0.7 and 0.8, relatively reliable; 0.8 and 0.9, highly reliable; and 0.9 and 1.0, extremely reliable. The results of the reliability analysis are presented in

Table 5. Reliability analysis.

Item	Cronbach’s Alpha	n
PU	0.917	4
PUE	0.891	3
AT	0.893	3
BI	0.913	3
SI	0.821	3
TS	0.895	3
PV	0.870	4
TR	0.904	3
PR	0.871	3
All items	0.973	29

Note: PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

. The reliability coefficients for both the overall NOA acceptance scale and each dimension were within the range of 0.8–1.0. This indicates that the scales used in this study had very good internal consistency.

5.3. Validity Test

Validity testing was used to evaluate whether the proposed model accurately reflected the essential characteristics of the participants. Generally, Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) are used to assess the questionnaire validity. The scale used in this study was adapted from previous research and required an initial screening of measurement items using EFA, followed by verification of the proposed measurement model.

5.3.1. Exploratory Factor Analysis

When conducting exploratory factor analysis, two key indicators are widely used to determine whether a dataset is suitable for factor analysis: the Kaiser-Meyer-Olkin (KMO) value and Bartlett’s Test of Sphericity. The KMO value measures the relative magnitude of the simple and partial correlation coefficients among variables, ranging from 0 to 1. When the KMO value is below 0.5, the dataset is generally considered unsuitable for factor analysis. Conversely, the closer the KMO value is to 1, the more suitable the dataset is for factor analysis. The significance level of Bartlett’s Test of Sphericity is reflected in the sig. value; when the sig. was below 0.05, the correlation coefficients among the variables were considered significant, which also meant that the dataset met the prerequisite conditions for factor analysis. Therefore, by comprehensively considering the results of the KMO value and Bartlett’s Test of Sphericity, it was possible to effectively assess and determine whether the dataset should undergo factor analysis.

From

Table 6. KMO and Bartlett’s test.

KMO and Bartlett’s Test
KMO value		0.951
Bartlett’s test of sphericity	Approximate Chi-Square	7490.411
	df	406
	p-value	0

, it can be observed that the current questionnaire had a KMO value of 0.951. Simultaneously, the significance test statistics (Sig. value) is almost zero, which is far below the commonly set significance level of 0.05. This indicated that the questionnaire data had a significant correlation structure and were suitable for factor analysis.

To examine whether there was a common method bias issue in the questionnaire, we employed Harman’s single-factor test for analysis. This study mainly used SPSS Statistics 19 factor analysis to conduct an unrotated exploratory factor analysis of the questionnaire measurement scales. The results are presented in

Table 7. Variance explanation rate table.

Variance Explained
Factor Number	Characteristic Root			Rotational Front Variance Explained			Variance Explained After Rotation
	Characteristic Root	Variance Explained %	Accumulation%	Characteristic Root	Variance Explained %	Accumulation%	Characteristic Root	Variance Explained %	Accumulation%
1	13.403	46.217	46.217	13.403	46.217	46.217	2.844	9.806	9.806
2	2.719	9.376	55.593	2.719	9.376	55.593	2.633	9.08	18.886
3	1.52	5.243	60.836	1.52	5.243	60.836	2.613	9.011	27.898
4	1.367	4.714	65.55	1.367	4.714	65.55	2.591	8.934	36.832
5	1.054	3.634	69.184	1.054	3.634	69.184	2.558	8.82	45.652
6	0.877	3.023	72.206	0.877	3.023	72.206	2.512	8.661	54.313
7	0.791	2.729	74.936	0.791	2.729	74.936	2.501	8.623	62.936
8	0.74	2.551	77.486	0.74	2.551	77.486	2.454	8.461	71.396
9	0.687	2.369	79.855	0.687	2.369	79.855	2.453	8.459	79.855
10	0.571	1.97	81.825
11	0.545	1.881	83.706
12	0.472	1.626	85.332
13	0.434	1.495	86.827
14	0.416	1.433	88.26
15	0.382	1.316	89.576
16	0.353	1.217	90.793
17	0.319	1.102	91.894
18	0.306	1.057	92.951
19	0.276	0.95	93.901
20	0.258	0.89	94.791
21	0.229	0.791	95.582
22	0.221	0.761	96.343
23	0.194	0.669	97.012
24	0.193	0.667	97.679
25	0.172	0.595	98.274
26	0.141	0.486	98.76
27	0.131	0.45	99.21
28	0.118	0.405	99.615
29	0.112	0.385	100

. The findings indicate that five factors with eigenvalues greater than 1 explain 69.184% of the total variance. The variance explained by the first factor was 46.217%, which did not exceed half of the total variance explained [82]. Therefore, there was no serious common method bias in this study. Table 6 also presents the variance explained after the rotation. Factor analysis extracted a total of 9 factors, and after rotation, the variance explained by each was 9.806%, 9.080%, 9.011%, 8.934%, 8.820%, 8.661%, 8.623%, 8.461%, and 8.459%. The cumulative variance explained after rotation was 79.855%.

The data in this study were rotated using the maximum variance rotation method (varimax method) to determine the correspondence between each factor and research items.

Table 8. Table of factor-loading coefficients after rotation (all items).

Factor Loading Coefficients After Rotation
Name	Factor Loading Coefficient									Commonality
	PU	BI	PUE	TS	TR	SI	PV	AT	PR
PU1	0.728	0.126	0.245	0.223	0.151	0.101	0.185	0.209	0.015	0.767
PU2	0.713	0.183	0.287	0.21	0.06	0.228	0.274	0.062	0.034	0.804
PU3	0.733	0.115	0.169	0.173	0.279	−0.007	0.139	0.228	0.013	0.813
PU4	0.604	0.208	0.215	0.098	0.132	0.246	0.095	0.361	0.013	0.758
PUE1	0.376	0.247	0.625	0.389	0.137	0.11	0.247	0.032	0.067	0.805
PUE2	0.268	0.255	0.698	0.238	0.211	0.009	0.194	0.165	0.009	0.791
PUE3	0.221	−0.044	0.769	0.148	0.12	0.264	0.151	0.2	0.115	0.824
PR1	−0.015	0.043	0.084	0.039	−0.017	0.151	0.066	0.034	0.879	0.812
PR2	0.014	0.056	−0.026	−0.036	−0.052	0.194	0.053	0.131	0.855	0.797
PR3	0.028	−0.049	0.028	0.052	0.114	0.091	0.073	−0.01	0.704	0.679
SI1	−0.003	0.138	0.081	0.147	0.129	0.754	0.199	0.115	0.293	0.772
SI2	0.168	0.061	0.152	0.038	0.048	0.804	0.193	0.054	0.241	0.803
SI3	0.132	0.158	0.121	0.369	0.218	0.656	0.125	0.194	0.151	0.747
TR1	0.144	0.239	0.46	0.152	0.435	0.382	0.126	0.157	0.009	0.688
TR2	0.155	0.306	0.271	0.164	0.688	0.208	0.049	0.306	−0.075	0.836
TR3	0.085	0.333	0.465	0.182	0.558	0.204	0.088	0.291	−0.036	0.814
TS1	0.184	0.166	0.259	0.746	0.239	0.253	0.126	0.223	0.041	0.873
TS2	0.232	0.164	0.108	0.789	0.198	0.17	0.109	0.236	0.021	0.797
TS3	0.151	0.271	0.22	0.546	0.557	0.126	0.177	0.198	0.01	0.893
PV1	0.143	0.087	0.062	0.244	0.756	0.054	0.313	0.038	0.139	0.784
PV2	0.138	0.056	0.184	0.202	0.055	0.254	0.717	0.364	0.125	0.827
PV3	0.148	0.361	0.187	0.062	0.229	0.142	0.714	0.123	0.156	0.813
PV4	0.166	0.24	0.119	0.138	0.182	0.241	0.632	0.325	0.037	0.781
AT1	0.251	0.168	0.175	0.152	0.137	0.128	0.319	0.528	0.128	0.738
AT2	0.143	0.267	0.28	0.257	0.274	0.092	0.278	0.746	0.083	0.855
AT3	0.23	0.386	0.002	0.29	0.138	0.137	0.295	0.609	0.093	0.79
BI1	0.09	0.546	0.234	0.343	0.157	0.064	0.193	0.339	−0.015	0.778
BI2	0.14	0.744	0.177	0.11	0.279	0.196	0.168	0.303	−0.037	0.854
BI3	0.218	0.733	0.015	0.279	0.146	0.147	0.363	0.115	0.116	0.864

Note: The bold numbers in the table indicate that the absolute value of the load factor was greater than 0.5. Rotation method: Varimax. PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

presents the information extraction of each factor corresponding to the research items, as well as the correspondence between the factors and research items. As shown in Table 8, the common variance values for all research items are above 0.4, indicating a strong correlation between the research items and the factors and that these factors are able to effectively extract information. At the same time, for most variables, the corresponding items were consistent with the expected divisions of the scale, and the respective factor loadings were greater than 0.5. However, in the TR variable, the factor loading of research item TR1 was 0.435, which was less than 0.5. TS3 had a factor loading of 0.557, which was greater than 0.5, and PV1 had a factor loading of 0.756, which was greater than 0.5. This indicates that the validity of the scale does not meet the requirements and can be improved by removing certain research items [83]. Therefore, items with poor internal consistency (TR1, TS3, and PV1) should be removed.

After removing items TR1, TS3, and PV1, the scale was again subjected to a rotated factor analysis, and the results are shown in

Table 9. Table of factor loading coefficients after rotation (deleted TR1, TS3 and PV1).

Factor Loading Coefficients After Rotation (Delete TR1, TS3 and PV1)
Name	Factor Loading Coefficient									Commonality
	PU	PR	SI	TS	PUE	BI	PV	TR	AT
PU1	0.744	0.019	0.085	0.254	0.21	0.086	0.238	0.224	0.139	0.803
PU2	0.677	0.018	0.247	0.183	0.378	0.236	0.21	−0.019	0.138	0.815
PU3	0.723	0.01	0.006	0.168	0.211	0.135	0.115	0.246	0.225	0.738
PU4	0.629	0.013	0.228	0.102	0.225	0.203	0.09	0.169	0.362	0.812
PUE1	0.34	0.059	0.122	0.383	0.638	0.26	0.2	0.13	0.07	0.816
PUE2	0.241	0.01	0.013	0.244	0.692	0.224	0.175	0.306	0.152	0.794
PUE3	0.202	0.116	0.254	0.132	0.744	−0.086	0.179	0.268	0.143	0.821
PR1	−0.034	0.874	0.169	0.017	0.128	0.074	0.042	−0.073	0.078	0.829
PR2	−0.022	0.846	0.22	−0.08	0.063	0.118	−0.022	−0.143	0.231	0.862
PR3	0.088	0.711	0.089	0.141	−0.117	−0.164	0.204	0.238	−0.16	0.858
SI1	−0.016	0.274	0.776	0.132	0.104	0.15	0.161	0.13	0.14	0.791
SI2	0.178	0.226	0.803	0.049	0.119	0.029	0.229	0.118	0.016	0.812
SI3	0.127	0.142	0.659	0.381	0.129	0.171	0.079	0.196	0.223	0.756
TR2	0.173	−0.063	0.198	0.19	0.212	0.272	0.1	0.751	0.186	0.836
TR3	0.086	−0.028	0.2	0.216	0.411	0.291	0.099	0.654	0.213	0.832
TS1	0.184	0.038	0.254	0.749	0.262	0.164	0.117	0.232	0.202	0.864
TS2	0.151	0.006	0.13	0.801	0.222	0.263	0.16	0.191	0.189	0.898
PV2	0.145	0.154	0.138	0.082	0.188	0.365	0.744	0.186	0.119	0.841
PV3	0.123	0.111	0.265	0.214	0.209	0.064	0.683	0.046	0.405	0.823
PV4	0.139	0.024	0.261	0.129	0.175	0.272	0.627	0.147	0.376	0.765
AT1	0.177	0.094	0.062	0.324	0.163	0.163	0.354	0.467	0.458	0.755
AT2	0.238	0.128	0.12	0.154	0.182	0.151	0.283	0.236	0.752	0.868
AT3	0.236	0.093	0.124	0.309	−0.014	0.355	0.294	0.224	0.589	0.784
BI1	0.077	−0.01	0.051	0.367	0.212	0.486	0.166	0.269	0.34	0.771
BI2	0.137	−0.027	0.175	0.122	0.141	0.699	0.194	0.418	0.248	0.847
BI3	0.213	0.117	0.141	0.29	0.019	0.72	0.379	0.179	0.102	0.868

Note: The bold numbers in the table indicate that the absolute value of the load factor was greater than 0.5. Rotation method: Varimax. PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

. As shown in Table 9, there was good internal consistency between the variables. However, the factor loadings for research items AT1 and BI1 were 0.458 and 0.486, respectively, both less than 0.5. Therefore, items AT1 and BI1, which had poor internal consistency, were removed.

After removing items AT1 and BI1, a rotated factor analysis was conducted on the scale again, and the results are shown in

Table 10. Table of factor loading coefficients after rotation (deleted TR1, TS3, PV1, AT1, and BI1).

Factor Loading Coefficients After Rotation (Deleted TR1, TS3, PV1, AT1, and BI1)
Name	Factor Loading Coefficient									Commonality
	PU	PR	BI	PUE	SI	TS	PV	TR	AT
PU1	0.727	0.013	0.087	0.251	0.241	0.192	0.176	0.162	0.165	0.798
PU2	0.646	0.004	0.243	0.394	0.191	0.171	0.276	−0.013	0.183	0.811
PU3	0.75	0.014	−0.007	0.16	0.176	0.187	0.077	0.301	0.194	0.763
PU4	0.624	0.003	0.228	0.245	0.104	0.077	0.222	0.155	0.4	0.817
PUE1	0.306	0.048	0.103	0.641	0.397	0.159	0.287	0.153	0.111	0.814
PUE2	0.255	0.004	0.002	0.648	0.252	0.218	0.17	0.373	0.118	0.796
PUE3	0.181	0.109	0.254	0.77	0.13	0.168	−0.034	0.246	0.152	0.825
PR1	−0.069	0.87	0.161	0.154	0.03	0.005	0.109	−0.083	0.131	0.826
PR2	−0.032	0.844	0.207	0.036	−0.056	0.02	0.058	−0.083	0.246	0.844
PR3	0.143	0.713	0.121	−0.102	0.113	0.213	−0.076	0.166	−0.253	0.824
SI1	−0.015	0.266	0.77	0.081	0.152	0.194	0.13	0.168	0.129	0.791
SI2	0.154	0.21	0.825	0.177	0.041	0.169	0.133	0.05	0.046	0.828
SI3	0.125	0.143	0.623	0.076	0.416	0.147	0.102	0.271	0.227	0.758
TR2	0.166	−0.05	0.152	0.171	0.214	0.119	0.246	0.786	0.209	0.881
TR3	0.092	−0.024	0.173	0.366	0.234	0.131	0.236	0.702	0.204	0.838
TS1	0.161	0.036	0.217	0.259	0.765	0.12	0.176	0.237	0.229	0.866
TS2	0.154	0.001	0.108	0.203	0.806	0.181	0.239	0.209	0.177	0.898
PV2	0.14	0.111	0.25	0.193	0.205	0.744	0.1	0.056	0.32	0.839
PV3	0.11	0.148	0.12	0.224	0.078	0.690	0.518	0.143	0.126	0.833
PV4	0.15	0.028	0.225	0.121	0.145	0.651	0.267	0.214	0.31	0.787
AT2	0.225	0.133	0.085	0.16	0.163	0.366	0.113	0.265	0.732	0.879
AT3	0.196	0.087	0.099	0.011	0.313	0.273	0.401	0.192	0.615	0.777
BI2	0.108	−0.041	0.163	0.134	0.137	0.132	0.692	0.44	0.29	0.859
BI3	0.175	0.102	0.128	0.036	0.296	0.274	0.769	0.177	0.163	0.868

Note: The bold numbers in the table indicate that the absolute value of the load factor was greater than 0.5. Rotation method: Varimax. PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

. The covariance values for all research items were higher than 0.4. The items corresponding to each variable were consistent with the expected divisions of the scale and the respective factor loadings were greater than 0.5. This indicates that the adjusted scale had good validity.

5.3.2. Confirmatory Factor Analysis

Confirmatory factor analysis was used to conduct an in-depth exploration and validation of the data with the help of a pre-constructed theoretical model. In this study, the initial model for confirmatory factor analysis of the factors influencing NOA acceptance was established using AMOS 26, as shown in the Figure 6. In the CFA model, there were only covariance relationships between the measurement models without causal relationships. The following section describes the parameter testing, convergent validity testing, and discriminant validity testing of the measurement model.

(1)

Measurement Model Parameter Testing

The following section examines whether the estimated parameters in the measurement model, such as factor loadings and goodness of fit, are reasonable. The results show that the standardized factor loadings of the observed indicators are greater than 0.5, indicating that the observed variables effectively measure their corresponding latent variables with a significant association between the two.

Further testing of the initial model’s fit in

Table 11. Goodness of fit measures for confirmatory factor analysis.

Fit Index	Judgment Criteria	CFA Model	Fitting Results
CMIN/DF	1–3 Excellent, 3–5 Good	2.545	Excellent
RMSEA	<0.05 Excellent, <0.08 Good	0.077	Good
GFI	>0.9 Excellent, >0.85 Good	0.855	Good
AGFI	>0.9 Excellent, >0.8 Good	0.799	Accept
IFI	>0.9 Excellent, >0.8 Good	0.926	Excellent
TLI	>0.9 Excellent, >0.8 Good	0.904	Excellent
CFI	>0.9 Excellent, >0.8 Good	0.925	Excellent
PNFI	>0.5 Good	0.692	Good
PCFI	>0.5 Good	0.724	Good
PGFI	>0.5 Good	0.616	Good

shows that according to the evaluation criteria [84], CMIN/DF (chi-square/degree of freedom ratio) = 2.545, which falls within the favorable range of 3–5. RMSEA = 0.077 (less than 0.08), GFI = 0.855 (less than 0.85), and AGFI = 0.799 (less than 0.8), all of which do not meet the fit standards. The IFI, TFI, and CFI were all greater than 0.8, and thus acceptable. The PNFI, PCFI, and PGFI were all greater than 0.5 and acceptable. The CFA model had a good goodness of fit.

(2)

Convergent Validity Test

Convergent validity refers to the degree of consistency and correlation between all the observed indicators that measure the same factor. Two indicators were used to assess convergent validity: composite reliability (CR) and average variance extracted (AVE). The formulas for calculating CR and AVE are shown in Equations (3) and (4), respectively.

$$\mathrm{C}\mathrm{R}={\displaystyle \frac{{\left(\sum {\lambda }_i\right)}^2}{{\left(\sum {\lambda }_i\right)}^2+\sum Var\left({\epsilon }_i\right)}}$$

(3)

$$\mathrm{A}\mathrm{V}\mathrm{E}={\displaystyle \frac{{\left(\sum {\lambda }_i\right)}^2}{\sum {\lambda }_i^2+\sum Var\left({\epsilon }_i\right)}}={\displaystyle \frac{{\sum }_{i=1}^n{L}_i^2}{n}}$$

(4)

where ${\lambda }_{\mathrm{i}}$ is standardized factor loading, $Var\left({\epsilon }_{\mathrm{i}}\right)$ is standardized residual, $L_i^2$ is the square of the standardized factor loading, $n$ is number of items per dimension. CR > 0.7 is considered to meet the standard, and AVE > 0.5 is considered to meet the standard [85]. The calculation results of the convergent validity indicators for the optimized model are listed in

Table 12. Composite reliability and average of variance extracted tests.

Latent Variables	Explicit Variables	Coef.	SE	t	p	Factor Loading	SMC	AVE	CR
PU	PU1	1.000	-	-	-	0.768	0.683	0.671	0.890
	PU2	1.079	0.076	14.198	0.000	0.840	0.705
	PU3	1.050	0.075	13.938	0.000	0.826	0.590
	PU4	1.053	0.074	14.181	0.000	0.839	0.704
PUE	PUE1	1.000	-	-	-	0.744	0.761	0.666	0.857
	PUE2	1.122	0.085	13.246	0.000	0.828	0.685
	PUE3	1.181	0.085	13.939	0.000	0.873	0.553
AT	AT2	1.000	-	-	-	0.826	0.671	0.740	0.850
	AT3	1.010	0.067	15.053	0.000	0.819	0.682
BI	BI2	1.000	-	-	-	0.819	0.777	0.723	0.839
	BI3	1.077	0.069	15.582	0.000	0.881	0.671
SI	SI1	1.000	-	-	-	0.794	0.671	0.627	0.834
	SI2	0.892	0.072	12.377	0.000	0.761	0.579
	SI3	0.966	0.073	13.321	0.000	0.819	0.631
TS	TS1	1.000	-	-	-	0.886	0.852	0.819	0.901
	TS2	1.042	0.053	19.589	0.000	0.923	0.786
PV	PV2	1.000	-	-	-	0.802	0.723	0.678	0.863
	PV3	0.969	0.068	14.290	0.000	0.817	0.667
	PV4	1.037	0.069	14.984	0.000	0.850	0.643
TR	TR2	1.000	-	-	-	0.901	0.747	0.779	0.876
	TR3	1.018	0.058	17.489	0.000	0.864	0.812
PR	PR1	1.000	-	-	-	0.662	0.743	0.640	0.840
	PR2	1.229	0.111	11.103	0.000	0.859	0.738
	PR3	1.280	0.115	11.109	0.000	0.862	0.439

Note: PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

.

As shown in Table 12, the composite reliability (CR) values of the nine latent variables in the measurement model ranged from 0.834 to 0.901, all exceeding the critical value of 0.7, indicating high internal consistency and reliability among the latent variables. The corresponding sets of observed indicators for each latent variable also demonstrated good convergent validity. In addition, the AVE values of all latent variables ranged from 0.640 to 0.819, all above the reference critical value of 0.5, indicating that the latent variables had strong explanatory power for the observed variables and further illustrating that the observed indicators under the same factor had good convergent validity.

(3)

Discriminant validity test

Discriminant validity refers to the differences between the latent variables.

Table 13. Results of discriminant validity test.

	PU	PUE	AT	BI	SI	TS	PV	TR	PR
PU	0.819
PUE	0.801	0.816
AT	0.713	0.635	0.860
BI	0.638	0.643	0.794	0.850
SI	0.516	0.551	0.576	0.568	0.792
TS	0.652	0.752	0.684	0.673	0.602	0.905
PV	0.662	0.664	0.817	0.765	0.664	0.618	0.823
TR	0.631	0.745	0.658	0.758	0.556	0.717	0.626	0.883
PR	0.124	0.159	0.255	0.150*	0.505	0.140	0.298	0.057	0.800

Note: The diagonal numbers are the square root values of AVE. PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

presents the indicators and correlation coefficients for the discriminant validity of the acceptance factor measurement model. According to the analysis in Table 13, most of the standardized correlation coefficients between the dimensions in this discriminant validity test were smaller than the square root of the corresponding AVE values for each dimension, indicating good discriminant validity between the dimensions.

Another method to assess discriminant validity is the heterotrait–monotrait ratio (HTMT). Kline [76] suggested that all HTMT values should be less than 0.85. All HTML values in this study were smaller than 0.85 (

Table 14. HTMT values.

	PU	PUE	AT	BI	SI	TS	PV	TR	PR
PU
PUE	0.795
AT	0.716	0.638
BI	0.64	0.619	0.788
SI	0.514	0.571	0.576	0.57
TS	0.651	0.738	0.682	0.688	0.592
PV	0.66	0.666	0.822	0.778	0.663	0.623
TR	0.635	0.747	0.665	0.752	0.552	0.717	0.624
PR	0.126	0.182	0.255	0.16	0.518	0.145	0.311	0.059

Note: PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

); therefore, all constructs had discriminant validity.

5.4. Correlation Analysis

An exploratory analysis of the relationships between the variables was conducted using the Pearson correlation method, and the results are presented in

Table 15. Pearson correlation analysis results among different dimensions.

	PU	PUE	PR	SI	TR	TS	PV	AT	BI
PU	1
PUE	0.695 **	1
PR	0.119 *	0.154 *	1
SI	0.444 **	0.484 **	0.430 **	1
TR	0.560 **	0.645 **	0.150 *	0.474 **	1
TS	0.583 **	0.648 **	0.126 *	0.517 **	0.636 **	1
PV	0.578 **	0.572 **	0.263 **	0.563 **	0.543 **	0.549 **	1
AT	0.622 **	0.545 **	0.214 **	0.488 **	0.573 **	0.597 **	0.703 **	1
BI	0.553 **	0.525 **	0.133 *	0.479 **	0.644 **	0.597 **	0.663 **	0.665 **	1

Note: **. Significantly correlated at the 0.01 level (two-tailed). *. Significantly correlated at the 0.05 level (two-tailed). PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

. Specifically, there were significant correlations between the variables in this analysis at the 99% significance level. According to the results of the correlation analysis, the correlation coefficients (R) between variables were greater than 0. Therefore, it can be concluded that there were significant positive correlations among all the variables in this study.

5.5. Structural Equation Model Analysis

Structural equation modeling (SEM) allows for a comprehensive understanding of the relationships between actual variables. This is especially valuable in situations where it is not possible to explore causal relationships between latent variables solely through regression analysis. SEM offers a more significant depiction of the influence relationships among latent variables. The SEM structure after the optimization of the measurement model is shown in Figure 7.

5.5.1. Multicollinearity Diagnostics

This study used the variance inflation factor (VIF) to assess collinearity issues among the model structures. The linear regression method in SPSS Statistics 19 was used to analyze different variables. The VIF of each construct in this study is presented in

Table 16. Collinearity diagnosis (VIF values).

	PR	SI	TR	TS	PV	PU	PUE	AT	BI
PR								1.313
SI								1.968
TR								2.856
TS						2.068	1.681	2.551
PV						1.884	1.681	2.371
PU								2.238	2.297
PUE						1.946		2.627	2.089
AT									1.854
BI

Note: PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

, and the VIF values are less than 3 [86], which can confirm that no collinearity exists in the regression model.

5.5.2. Model Fit Evaluation

A preliminary analysis of the CB-SEM (Covariance-Based Structural Equation Modeling) was conducted using AMOS 26. The fitting parameter results for the structural model are presented in

Table 17. Goodness of fit measures of SEM.

Fit Index	Judgment Criteria	Fitted Value	Fitted Result
CMIN/DF	1–3 Excellent, 3–5 Good	2.437	Excellent
RMSEA	<0.05 Excellent, <0.08 Good	0.074	Good
GFI	>0.9 Excellent, >0.85 Good	0.852	Good
AGFI	>0.9 Excellent, >0.8 Good	0.805	Good
IFI	>0.9 Excellent, >0.8 Good	0.926	Excellent
TLI	>0.9 Excellent, >0.8 Good	0.909	Excellent
CFI	>0.9 Excellent, >0.8 Good	0.925	Excellent
PNFI	>0.5 Good	0.727	Good
PCFI	>0.5 Good	0.764	Good
PGFI	>0.5 Good	0.647	Good

. Compared with the optimized measurement model, all indices improved, indicating that the established structural model better reflected the potential causal relationships among the factors. According to the evaluation index results, the chi-square to degrees of freedom ratio (CMIN/DF) was 2.437, which fell within the excellent range of 1–3. RMSEA = 0.074 (less than 0.08), GFI = 0.852 (greater than 0.85), and AGFI = 0.805 (greater than 0.8), all of which reached acceptable levels. The IFI, TFI, and CFI values were all greater than 0.9, indicating excellent reliability. The PNFI, PCFI, and PGFI values were all >0.5, indicating a good level. Taking all of these fit indices into comprehensive consideration, the theoretical model has reached the standard required for fit with the actual collected data, and the degree of congruence between the two is high.

5.5.3. Path Analysis

To verify whether the research hypotheses of the NOA acceptance mechanism model held, AMOS 26 was used to conduct a path analysis of the NOA acceptance structure model to examine the path relationships between the latent variables in the model. Table 11 presents the parameter estimation results for model paths.

Table 18. SEM Path Parameters.

Path			Std. (β)	S.E.	C.R.	p	R2
PUE	←	TS	0.608	0.066	7.199	***	0.655
PUE	←	PV	0.27	0.076	3.52	***
PU	←	PUE	0.56	0.118	5.384	***	0.676
PU	←	PV	0.137	0.084	1.442	0.149
PU	←	TS	0.199	0.082	2.714	**
AT	←	PUE	0.145	0.091	1.978	*	0.861
AT	←	PU	0.155	0.088	2.063	*
AT	←	TR	0.3	0.094	2.877	**
AT	←	PV	0.477	0.091	5.895	***
AT	←	TS	0.256	0.118	2.214	*
AT	←	SI	0.096	0.079	1.212	0.225
AT	←	PR	−0.013	0.051	−0.236	0.813
BI	←	AT	0.995	0.103	9.374	***	0.855
BI	←	PU	−0.098	0.087	−1.1	0.271

Note: * p < 0.050 ** p < 0.010 *** p < 0.001. PU (Perceived Usefulness), PUE (Perceived Ease of Use), PR (Perceived Risk), SI (Social Influence), TR (Trust), TS (Travel Scenarios), PV (Price Value), AT (Attitude), BI (Behavioral Intention).

shows the causal relationships between the latent variables in the model, including the significance (p value) of the unstandardized path coefficients, the absolute value of the critical ratio (C.R.), and the standardized path coefficient β value.

According to the analysis of Table 18, of the 14 path hypotheses proposed in the NOA acceptance theory model, four path coefficients were not significant: price value (β = 0.137, p > 0.05) had no significant effect on perceived usefulness, social influence (β = 0.096, p > 0.05), and perceived risk (β = −0.013, p > 0.05) had no significant effect on attitude, and perceived usefulness (β = −0.098, p > 0.05) had no significant effect on behavioral intention. The remaining 11 path coefficients were significant (|C.R.| > 1.96, p < 0.05).

H1: Path coefficient between behavioral intention and attitude (|C.R.| = 9.374 > 1.96, p < 0.001) reached a significant level and the standardized path coefficient was positive, indicating that individuals’ acceptance intention for NOA was significantly positively correlated with their attitude. Therefore, H1 is supported.

H2a and H2b: Path coefficients between attitude and perceived usefulness (|C.R.| = 2.063 > 1.96, p < 0.05) reached a significant level, whereas the path coefficient between behavioral intention and perceived usefulness (|C.R.| = 0.271 < 1.96, p > 0.05) did not reach significance, although both standardized path coefficients were positive. This indicates that Hypothesis H2a is significantly supported, whereas Hypothesis H2b is not supported by the data and should be rejected.

Similarly, hypotheses H3a, H3b, H5a, H5c, H6a, H6c, and H7 were supported. Thus, Hypotheses H4, H6b, and H8 were rejected. The hypothesis testing results are summarized in

Table 19. Hypothesis model path relationship test.

Hypothesis		Decision
H1	AT has a significant positive impact on BI.	Supported
H2a	PU has a significant positive effect on AT.	Supported
H2b	PU has a significant positive effect on BI.	Not supported
H3a	PUE has a significant positive effect on BI.	Supported
H3b	PUE has a significant positive effect on PU.	Supported
H4	SI has a significant positive effect on AT.	Not supported
H5a	TS has a significant positive effect on AT.	Supported
H5b	TS has a significant positive effect on PU.	Supported
H5c	TS has a significant positive effect on PUE.	Supported
H6a	PV has a significant positive effect on AT.	Supported
H6b	PV has a significant positive effect on PU.	Not Supported
H6c	PV has a significant positive effect on PUE.	Supported
H7	TR has a significant positive effect on AT.	Supported
H8	PR has a significant negative effect on AT.	Not supported

.

5.5.4. Robustness Test

The main methods of robustness testing include altering research variables and excluding specific samples. If, after these adjustments, the signs and significance of the model coefficients remain largely unchanged, the results are considered robust. In this study, robustness was tested by modifying both variables and samples, and the model (BI, AT, PU, PUE) was validated using multiple linear regression. First, regression was performed using the original samples and variables to obtain Model 1. Then, samples of participants over the age of 50 were excluded, and regression was performed again to obtain Model 2. Finally, all samples were retained, and an age variable was added, followed by another regression to obtain Model 3. As shown in

Table 20. Robustness test of BI.

	BI: Model 1		BI: Model 2		BI: Model 3
	B	t	B	t	B	t
Constant	0.584 *	3.151	0.62 *	3.339	0.776 *	3.912
AT	0.604 *	11.174	0.604 *	11.111	0.606 *	11.329
PU	0.184 *	3.2	0.174 *	3.01	0.183 *	3.217
Age					−0.1 *	−2.539
R-squared	0.547		0.549		0.558
Adjusted R-squared	0.543		0.545		0.553
F-value	154.977		152.115		107.654

Note: * p < 0.05; Model 1 is the base model; Model 2 excludes samples with age over 50; Model 3 adds the age variable.

,

Table 21. Robustness test of AT.

	AT: Model 1		AT: Model 2		AT: Model 3
	B	t	B	t	B	t
Constant	0.01	0.047	−0.067	−0.309	−0.052	−0.232
PU	0.23 *	3.878	0.235 *	3.918	0.23 *	3.884
PUE	0.112 *	2.164	0.105 *	2.076	0.119 *	2.286
PR	−0.052	−0.356	−0.058	−0.278	−0.052	−0.332
SI	0.02	1.162	0.016	1.302	0.019	1.167
TR	0.105	1.7	0.094	1.495	0.103	1.665
TS	0.184 *	2.7	0.184 *	3.213	0.194 *	3.41
PV	0.522 *	3.7	0.532 *	7.386	0.519 *	7.296
Age					0.034	0.924
R-squared	0.649		0.658		0.65
Adjusted R-squared	0.639		0.648		0.639
F-value	107.654		67.206		58.326

Note: * p < 0.05; Model 1 is the base model; Model 2 excludes samples with age over 50; Model 3 adds the age variable.

,

Table 22. Robustness test of PU.

	PU: Model 1		PU: Model 2		PU: Model 3
	B	t	B	t	B	t
Constant	0.601 *	3.07	0.595 *	3.005	0.609 *	2.926
TS	0.172 *	3.252	0.188 *	3.443	0.171 *	3.156
PV	0.24 *	3.572	0.225 *	3.276	0.24 *	3.566
PUE	0.448 *	7.59	0.448 *	7.511	0.449 *	7.493
Age					−0.004	−0.105
R-squared	0.551		0.554		0.551
Adjusted R-squared	0.546		0.548		0.544
F-value	104.876		103.017		78.356

Note: * p < 0.05; Model 1 is the base model; Model 2 excludes samples with age over 50; Model 3 adds the age variable.

and

Table 23. Robustness test of PUE.

	PU: Model 1		PU: Model 2		PU: Model 3
	B	t	B	t	B	t
Constant	1.044 *	5.307	1.044 *	5.236	0.831 *	3.943
TS	0.388 *	7.692	0.384 *	7.278	0.407 *	8.081
PV	0.373 *	5.564	0.376 *	5.458	0.355 *	5.337
Age					0.105 *	2.611
R-squared	0.486		0.481		0.499
Adjusted R-squared	0.482		0.477		0.494
F-value	121.542		116.041		85.133

Note: * p < 0.05; Model 1 is the base model; Model 2 excludes samples with age over 50; Model 3 adds the age variable.

, after excluding samples and modifying variables, the signs and significance of the main research variables in the model remain consistent, confirming the robustness of the empirical results of this study.

6. Discussion

This study aims to explore the factors that influence the user acceptance of NOA functions in new energy vehicles in China, as well as their mechanisms of influence, to enhance user acceptance of NOA. By integrating factors including social influence, travel scenarios, price value, perceived trust, and perceived risk, this study established an extended TAM. Using structural equation modeling, this study investigated the impact of each factor on NOA acceptance. The results showed that 10 of the 14 hypotheses were supported by the data. Although the four hypotheses were not supported, they still had a certain degree of influence.

Further discussion of the non-significant paths reveals nuanced insights into the acceptance mechanisms of NOA in China’s unique socio-technical context.

(1)

Non-significant effect of PU on BI (H2b: PU→BI)

This result aligns with classical TAM literature wherein attitude often fully mediates the effect of PU on BI. In the context of NOA, usefulness may be a necessary but insufficient condition for adoption—shaping one’s attitude rather than directly motivating intention. This implies that users may recognize the functional benefits of NOA (e.g., reduced fatigue), but their decision to use it depends on an overall favorable attitude influenced by multiple factors such as trust, ease of use, and scenario applicability.

(2)

Non-significant effect of SI on AT (H4: SI→AT)

The lack of a significant effect of social influence on attitude may be deeply rooted in China’s automotive consumption culture and digital media environment. Consumers, especially the young and well-educated group that dominates our sample, increasingly rely on accurate and objective data rather than social recommendations when evaluating smart technologies such as NOA. This is consistent with the findings of Liang et al. [87], who pointed out that Chinese electric vehicle consumers’ choices are influenced by quantifiable factors such as range, power, space, design, and comfort. Furthermore, the widespread use of influencer marketing and the occasional presence of exaggerated claims on social platforms may have contributed to a certain level of consumer skepticism, thereby diminishing the impact of social recommendations [88]. Consequently, social influence in its current manifestation may serve more as a source of initial awareness rather than a decisive factor in shaping attitudes toward NOA.

(3)

Non-significant effect of PV on PU (H6b: PV→PU)

The non-significance of this path indicates that respondents dissociated monetary value from perceived utility. That is, while users acknowledge that additional services (e.g., OTA upgrades and training) enhance usability and justify cost, these do not directly amplify their perception of the system’s usefulness. Usefulness is likely derived more from practical performance and reliability in specific driving scenarios, aspects that are evaluated independently of the pricing structure.

(4)

Non-significant effect of PR on AT (H8: PR→AT)

The non-significant relationship between perceived risk and attitude may be attributed to limitations in the conceptualization and measurement of risk. Our scale primarily captured functional and privacy risks, for example, “PR1 and PR2: system misjudgment” and “PR3: data leakage.” However, it omitted other critical dimensions salient, such as legal liability (e.g., “Who is responsible in case of an accident?”) [89], ethical risk (e.g., decision-making in dilemmas) [90], and long-term dependency risk (e.g., degradation of driving skills) [91]. Future studies should adopt a multi-dimensional risk scale incorporating legal, ethical, and performance liability aspects based on frameworks such as that of Tan et al. [89], who validated perceived risk in automated vehicles to include accountability. Revising the risk construct to reflect these facets could yield different and more nuanced effects on user attitude.

We recommend that future research incorporates broader social trust variables and a refined, multi-dimensional risk construct to better capture the underlying factors influencing consumer attitudes. Next, we will further analyze how these factors influence the following four variables: perceived ease of use, perceived usefulness, attitude, and behavioral intention.

6.1. Analysis and Discussion of PUE

The empirical data indicate that travel scenarios and price values have a direct and significant impact on users’ perceived ease of use, together explaining 65.5% of the variance. Further analysis of the path coefficients showed that the impact of travel scenarios on perceived ease of use was greater than that of price value. The path coefficient for travel scenarios reached 0.608, meaning that if travel scenarios increased by one unit, the perceived ease of use increased by 0.608 units. This suggests that integrating NOA use with travel scenarios makes it easier for users to understand the practical functions of NOA. These findings can guide real-world NOA promotion practices. To help users more easily understand how to use NOA, their functions should be explained in different travel scenarios. Additionally, price value had a significant impact on perceived ease of use, with a path coefficient of 0.270. In other words, providing certain value-added services enhances users’ perceived ease of use. This suggests that NOA training and OTA upgradation services can make it easier for users to understand the functional characteristics of NOA. Overall, to help users better understand the functions of NOA and make them feel that it is easy to use, it is necessary to provide user training and demonstrate NOA in users’ real travel scenarios.

6.2. Discussion and Analysis of PU

The study found that perceived ease of use (β = 0.560, p < 0.001) and travel scenarios (β = 0.199, p < 0.05) had significant effects on perceived usefulness, highlighting the role of perceived ease of use and travel scenarios in shaping perceived usefulness. However, price value (β = 0.137, p > 0.05) did not have a significant effect on perceived usefulness, revealing that the value provided by the service does not clearly indicate the usefulness of NOA. Together, these three variables explained 67.6% of the variance in perceived usefulness. Further analysis of the path coefficients indicated that perceived ease of use had the greatest impact on perceived usefulness. This suggests that users consider NOA to be useful only when they are easy to use. This finding aligns with the basic TAM theoretical framework, which posits that, when a product is too complex and difficult to use, users’ perceptions of its usefulness diminish. At the same time, travel scenarios are also important: introducing NOA functions in the context of specific travel scenarios not only makes users feel that they are easy to use but also increases their perceived usefulness. Therefore, by lowering the operating difficulty of NOA functions and demonstrating them in specific travel scenarios, user recognition of NOA’s usefulness can be further enhanced, thereby increasing the acceptance of NOA features.

6.3. Discussion and Analysis of AT

Empirical data show that perceived ease of use (β = 0.145, p < 0.05), perceived usefulness (β = 0.155, p < 0.05), perceived trust (β = 0.300, p < 0.01), price value (β = 0.477, p < 0.001), and travel scenarios (β = 0.256, p < 0.05) all have significant impacts on attitude. Among these, the effect of price value was the most significant, followed by travel scenarios, perceived trust, perceived ease of use, and perceived usefulness. All path coefficients were positive, indicating that as these factors improved, users’ acceptance attitudes toward NOA also increased. Price value is the primary factor influencing attitudes, meaning that people prioritize price value when deciding whether to use an NOA. This further confirms that during the promotion of new technology, users are most concerned about reasonable pricing that delivers corresponding services. Therefore, during the promotion of NOA technology, its cost should be further reduced, and users’ sense of value should be improved.

Travel scenarios also significantly affect attitude, implying that users prefer NOA features to have a clear scope and scenario for application. Thus, during promotions, the applicable scope should not be exaggerated, and clarifying NOA boundaries helps significantly improve user attitudes towards its use. Trust also demonstrates a certain degree of influence in this model, with a standardized path coefficient of 0.256, indicating that, as users’ trust in NOA increases, they become more willing to choose or accept it. Thus, during the promotion of NOA, users’ trust should be enhanced through authoritative certification, test drives, and other measures. Likewise, perceived ease of use and perceived usefulness have a certain impact on attitudes.

Additionally, social influence (β = 0.096, p > 0.05) and perceived risk (β = −0.013, p > 0.05) did not have a significant impact on attitude. This indicates that the attitudes of friends and social policy environment do not significantly influence people’s attitudes towards NOA; individuals prefer to make decisions based on their own experiences. However, this does not imply that social influence is unimportant; attention should be paid to the positive coverage of NOA on social media and elsewhere. Perceived risk negatively affected attitudes but not significantly. This may be related to the fact that NOA has already been used in some energy vehicles, and users recognize the technical competence of existing brands, so perceived risk is not prominent. However, in actual technological applications, potential risks may negatively affect acceptance. Therefore, users’ perceived risk of NOA should be minimized.

In summary, the seven variables influencing attitude together explained 86.1% of the variance in attitude, indicating the rationality of the theoretical model structure.

6.4. Discussion and Analysis of BI

Empirical data showed that attitude (β = 0.995, p < 0.05) had a significant impact on behavioral intention, whereas perceived usefulness (β = −0.098, p > 0.05) did not have a significant impact. Together, these two variables explained 85.5% of the variance in behavioral intention. This is consistent with the original TAM, in which attitude directly determines behavioral intentions. Therefore, to increase the acceptance of NOA technology during its promotion, attention should be paid to factors that affect user attitudes and enhance these influencing factors.

Based on the above analysis and the current development of NOA technology in new energy vehicles, the following three comprehensive strategies are proposed to improve public acceptance of NOA.

(1)

Clarify the applicable scenarios for NOA functions.

It is recommended that new energy vehicle companies classify different scenarios and specify the applicable scope for highways, urban roads, underground garages, and other environments, so that users can understand in which scenarios NOA features can be highly trusted and in which situations users need to operate the vehicle themselves. This further enhances users’ trust in NOA features and increases their public acceptance.

(2)

Reduce hardware costs and enhance service value.

Further reduction in vehicle hardware costs and compensation for hardware limitations by improving algorithm capabilities, as exemplified by Tesla’s vision-based assisted driving approach. Meanwhile, ensuring the value of the services provided through regular OTA updates and value-added services increases user stickiness. Additionally, we introduced incentive policies tailored to younger users, such as safety-driving reward points and redemption schemes.

(3)

Improve system transparency and strengthen user trust.

Based on technological capabilities, system capability boundaries are communicated to users to enhance transparency. Simultaneously, human–machine interface technologies can be leveraged to translate complex functions into easily understood voice or text prompts for users, thereby improving decision-making efficiency. Moreover, in the event of accidents, promptly disclosing the cause to the public and investigating potential issues in the NOA system process further strengthens user trust and increases public acceptance.

(4)

Advanced technical capability and increased perceived value.

Through ongoing technological iteration, the safety and convenience of NOA are continuously enhanced, allowing users to tangibly experience their usefulness and ease of use, thereby increasing their acceptance. Meanwhile, promoting and popularizing the technology, highlighting its features and advantages, and guiding the public toward developing correct consumer perceptions of NOA.

7. Conclusions

This study explored the factors and mechanisms influencing the acceptance of NOA functionality in new energy vehicles in China. By establishing a structural equation model based on acceptance theory, testing its goodness of fit, and conducting a path analysis, we verified whether the hypotheses in the model were true. This study reveals the impact of perceived ease of use, perceived usefulness, travel scenarios, price value, perceived trust, and perceived risk on attitudes as well as the mechanism by which attitude affects final behavioral intention. According to the survey results, respondents held a positive attitude toward NOA technology and were willing to use energy vehicles equipped with NOA functionality. Among the many influencing factors, we found that travel scenarios, price value, and perceived trust were the key factors affecting users’ perceived value and attitude, whereas social influence and perceived risk were not as significant.

Further analysis revealed that the mechanism by which travel scenarios influenced NOA acceptance was TS→PUE/PU→AT→BI and TS→AT→BI. Price value influenced NOA acceptance through PV→AT→BI, and PV→PUE→AT→BI. The mechanism of perceived trust was TR→AT→BI.

To further improve the acceptance rate of the NOA functionality, its promotion should highlight specific scenarios for NOA applications, making it easier and more convenient for users to adopt the technology. It is also necessary to continue lowering the price of NOA technology, provide substantial value-added services, and employ various measures to enhance users’ trust in NOA functionalities. These findings contribute significantly to the promotion of NOA technology.

7.1. Theoretical Contributions and Practical Impact

This study was based on the TAM. By collecting data through both online and offline questionnaires, it provides a comprehensive analysis of the acceptance of NOA functions in new energy vehicles. Using statistical methods and path analysis techniques, we investigated the influence of perceived ease of use, perceived usefulness, social influence, perceived risk, travel scenarios, price value, and perceived trust on attitudes and willingness to accept NOA. The sample mainly reflects the acceptance of NOA among young and middle-aged groups with higher levels of education. This research enriches the application of the TAM in the NOA field and offers theoretical guidance and empirical evidence for the development of NOA. The significance of this study lies in providing a solid theoretical foundation and guidance for new energy vehicle companies in promoting and improving NOA functions. Specifically: (1) When promoting NOA features, it is important to clearly state the applicable scenarios and limitations of NOA. (2) When promoting to different age groups, special attention should be paid, for the elderly, emphasize low risk and comprehensive after-sales service; for young and middle-aged people, focus on increasing their trust in the technology and highlight how the technology is applied in various scenarios. (3) Reduce the technical costs of NOA to make them available in more entry-level products. (4) Enhance service value through regular OTA updates to maintain user engagement. (5) Improve system transparency by promptly disclosing the causes of incidents when accidents occur. (6) Continuously iterate technology and promote updates to users, thereby increasing perceived user value.

7.2. Limitations and Future Research

It is important to note that this study has certain limitations. First, we could not verify whether the survey sample represented the overall target population. Due to the location-specific nature of the survey, the sample was concentrated. The samples in this study were mainly from Guangzhou, with a focus on young and middle-aged groups, which to some extent limits the generalizability of the conclusions nationwide. Previous research on ADAS and autonomous driving has found that young and elderly drivers have different levels of acceptance toward ADAS [92], and the acceptance of autonomous driving technology also varies across countries and regions [71]. Therefore, awareness, acceptance, and purchasing power for NOA may differ depending on the consumer’s age group or the level of the city they live in. For example, consumers in first-tier cities may place greater emphasis on technological experiences and brand value, whereas those in lower-tier cities may be more concerned with practicality and cost-effectiveness. Future research should sample across multiple regions and age groups nationwide to further validate the universality of the model. In addition, we recommend that subsequent studies conduct subgroup analyses (such as grouping by city tier or age) to test the model’s stability. Other background variables, such as family size and whether the participant holds a driver’s license, should also be considered in future analyses of influencing factors.

This study only expands on the TAM by adding variables such as travel scenarios, price value, social influence, perceived trust and perceived risk. The research framework did not cover all possible influencing factors, and further exploration and validation were required. In particular, the construct of payment methods (free trial, one-time purchase, subscription) could not be included as a moderating variable in the model because of sample limitations. We only conducted descriptive statistical analysis. Although payment methods were not included as a moderating variable in this model, their descriptive results still offer valuable insights. For example, the high acceptance of the free trial model suggests that car companies can lower users’ initial barriers to adoption through experiential marketing. However, such research is insufficient. In the future, we hope to evaluate willingness to adopt in different groups when the sample size is sufficient to utilize the moderating role of payment methods. This will further deepen our understanding of “price value” and provide insights into consumer attitudes and intentions under different payment models. This can directly provide evidence to support automakers in formulating flexible business models and pricing strategies. For example, it can help determine which types of customers (such as free trial, one-time purchase, subscription) are better suited for the promotion of subscription or one-time purchase models. At the same time, more practically valuable influencing factors should be identified, and more acceptance-related models should be established to deepen the understanding and guidance regarding NOA acceptance. In addition, this study did not consider the endogeneity of observed variables when introducing variables. Future research should further verify these causal relationships using experimental designs. This will further enhance public recognition and acceptance of NOA and lay a foundation for the future promotion and application of autonomous driving technology, thereby facilitating the healthy and rapid development of the new energy vehicle industry as a whole.