An Empirical Study on the Structural Assurance Mechanism for Trust Building in Autonomous Vehicles Based on the Trust-in-Automation Three-Factor Model

Abstract

The development of the Autonomous vehicle industry contributes to achieving the environmental, economic, and social sustainability goals. Autonomous vehicles (AVs) involve highly autonomous and complex intelligent driving technology, and their large-scale commercial application depends on the level of public trust in their safety and reliability. Therefore, how to establish and enhance public trust in AVs has become the key to the development of the AV industry. This study discusses the impact of technical structural assurance, social structural assurance, and individual cognitive factors on trust in AVs. This study uses a structural equation model to analyze a valid sample of 548 participants from China. The results show that autonomy has a negative impact on trust, and currently, personal cognitive factors exert a greater influence on trust compared to technical and social structural assurance factors in China. In theory, this study combines the trust-in-automation three-factor model with the concept of structural assurance to reveal subjective controllable factors that can promote public trust. In practice, this study reveals the important role of structural assurance factors in enhancing trust before fully automatic driving technology is officially launched.

1. Introduction

Autonomous vehicles (AVs) are emerging as a significant trend in the future of transportation owing to their potential to optimize traffic flow [1], enhance travel efficiency [2], reduce traffic accidents [3,4], improve urban space utilization [5], foster new industries [6], and increase public transportation equity [7,8]. The development of the AV industry can contribute positively to sustainable development. However, realizing these benefits depends on the widespread adoption of AVs.

The advancement of autonomous driving technology has transformed the relationship between humans and vehicles, shifting from traditional human control to intelligent driving system control. This transformation raises the issue of AV trust, focusing on whether individuals trust and delegate driving tasks to intelligent autonomous systems [9]. Despite significant technological advances, the actual usage of automated vehicle driving features is currently low, primarily due to public trust issues [10]. Therefore, it is crucial to thoroughly study the public’s trust in AVs and systematically construct a structural assurance mechanism for the establishment of public trust in AVs.

Trust is a core concept in the research of technology acceptance and application, and it has been identified as a crucial factor that influences the intention to adopt autonomous driving technology [11,12]. Research indicates that public trust in AVs is significantly influenced by a variety of factors, including the performance of the technology in terms of safety, reliability, transparency, and user experience [13,14,15]. Core variables in behavioral theory, such as perceived usefulness, perceived ease of use, and perceived risk, have been validated in numerous studies as key factors affecting trust [16,17,18]. Additionally, users’ attitudes and expectations toward autonomous driving technology, as well as individual characteristics (such as cultural background, age, and gender), have a substantial impact on their trust in and adoption of AVs [19,20,21]. To explore the influencing factors and mechanisms of public trust in AVs, Sun et al. [22] proposed a comprehensive model of trust in AV human–machine interaction, highlighting the critical role of the AV system’s operational model in fostering and establishing trust. Furthermore, Dirsehan and Can [23] utilized the Technology Acceptance Model (TAM) to confirm the substantial influence of trust on adopting autonomous technologies. Wu et al. [24] conducted a study on how levels of autonomy and anthropomorphic characteristics affect public acceptance and trust in Shared Autonomous Vehicles (SAVs) based on the Unified Theory of Acceptance and Use of Technology (UTAUT) Model. These groundbreaking studies shed light on the critical role of trust in shaping public perceptions and behavior towards AVs.

The factors influencing trust are complex and multidimensional. Most existing studies have only explored a single dimension, drawing factors from various theoretical models and sample data. To establish public trust in AVs, it is essential to develop a systematic model. Structural assurance, which refers to the infrastructure supporting technology use, has been found to have a positive impact on public trust development [25,26]. However, research on structural assurance in the context of AVs is limited compared to its exploration in e-commerce and mobile payments. This gap underscores the need for further investigation into the mechanisms of structural assurance for establishing trust in AVs.

Among the theories related to human trust formation, the trust-in-automation three-factor model proposed by Schaefer et al. [27] is more effective at explaining public trust-building in AVs [24]. Although the trust-in-automation three-factor model has been discussed and validated at a theoretical level [28,29], there is limited research on its real-world application. In particular, the practical challenges of trust-building during the deployment of various automation technologies have not been fully addressed. Existing studies mostly focus on simulated scenarios in laboratory environments. To fill this research gap, this study aims to thoroughly analyze the applicability of the model in real-world environments and verify its effectiveness under real conditions. Therefore, this study uses the model as the theoretical basis to construct a structural assurance model of trust in AVs across the three dimensions of technical, social, and personal cognition and systematically proposes a structural assurance mechanism for establishing trust in AVs. Unlike previous studies, this study uses structural equation modeling (SEM) to test these dimensions in a unified model and compare the strength of their influence, laying a theoretical foundation for mechanism construction.

This study conducted field research in major cities in the eastern coastal provinces of China, where consumers are expected to be the first users of AVs. Detailed data surveys targeting these consumers are of significant practical importance for the future promotion and application of AVs. The findings provide valuable references for policymakers and offer data support for enterprises in formulating market strategies, thereby more effectively promoting the adoption and application of AVs.

2. Literature Review

This study integrates the conceptual framework of structural assurance and the trust-in-automation three-factor model to explore the impact of human factors (i.e., individual cognitive factors), partner factors (i.e., technical structure assurance factors), and environmental factors (i.e., social structure assurance factors) on public trust in AVs.

2.1. Structural Assurance

2.1.1. Concept of Structural Assurance

Structural assurance is perceived as objective structural conditions based on institutions. Zucker [30] conceptualizes structural assurance as institutionalized structural constructs, encompassing mandatory legal contracts, commitments, and regulatory systems, which instill consumers with the belief in successful consumption. Shapiro [31] defines structural assurance as typically referring to objective structural conditions such as contracts, guarantees, regulations, commitments, and oversight. By fostering a secure and reliable environment, it engenders trust among participants in online transactional behavior.

Structural assurance is regarded as the secure measures that support the establishment of trust. Wang and Shang [32] defined structural assurance from the perspective of e-commerce as security policies that facilitate successful transactions, such as regulations, guarantees, commitments, and contracts. Lin, Lu, and Zhang [33] defined structural assurance from the standpoint of online investment as secure measures that enhance investors’ success rates, including legal resources, guarantees, rules, policies, commitments, or formal agreements. Sassi et al. [34], from the perspective of protecting the security of online transactions, posited that structural assurances encompass legal provisions (laws, guarantees, and regulations) provided by the institutional environment to safeguard transaction security.

In studies pertaining to the structural assurance of AVs, Koester and Salge [35] interpreted structural assurance as a range of legal, technical, and regulatory measures aimed at safeguarding the safety and reliability of vehicles. Li et al. [26], in their research on the public’s willingness to use AVs, interpreted structural assurance as the guarantee, regulations, commitments, or other procedures ensuring the successful use of AVs by the public. Xie et al. [36], from the perspective of selecting rides in autonomous taxis, regarded structural assurance as the organizational policies, norms, and regulations provided by ride-hailing service providers.

2.1.2. Structural Assurance and Trust

Structural assurance serves as the foundation of trust, facilitating the establishment of trust effectively. The public tends to trust others or entities more readily in a secure and reliable environment. Therefore, formal contracts, legal regulations, and online monitoring are crucial structural assurance measures for enhancing trust in online transactions, particularly in the initial stages of transactions [37]. Kim and Prabhakar [38] argue that legal regulations and online monitoring can reduce opportunistic behavior among the public, which is essential for bolstering trust in mobile services. Sporleder and Goldsmith [39] suggested that when buyers are uncertain about the legitimacy of online stores and the quality of goods, sellers can rely on authoritative third-party assurances such as government standards and quality control. Research by Shao and Yang [40] indicates that structural assurance effectively increases trust in electronic channels among various customer segments, thereby enhancing customers’ intention to use online banking. Shao et al. [41] affirmed that structural assurance is a significant antecedent facilitating trusting beliefs and willingness to transact online. Structural assurance contributes to enhancing perceived reliability. Hanif et al. [42] argued that expatriates who perceive a higher level of structural assurance are more likely to have a positive attitude towards making mobile purchases. This is because they believe that their personal information is secure, the transactions are reliable, and their privacy is protected. This sense of security leads to an increased willingness to engage in mobile commerce activities.

Liu et al. [43] found that structural assurance can help individuals develop confidence in AVs in the absence of direct experience. Koester and Salge [35] argued that structural assurance plays an important role in establishing users’ initial trust in AVs, and by establishing structural assurance, users and stakeholders can be provided with a trustworthy environment [44]. Therefore, the construction of a structural assurance mechanism will help enhance the public’s confidence and promote the establishment of public trust in AVs, especially the initial trust.

2.2. Trust-in-Automation Three-Factor Model

Schaefer et al. [27] utilized a meta-analysis approach to synthesize factors from past research on the development of automated trust [28], and attributed the development of trust systems to three core factors: human factors, partner factors (automation or robot), and environmental factors.

First, trust assessment is user-centered, and human factors are usually considered extremely important. The model categorizes human-related factors into four types: operator traits, operator states, cognitive factors, and emotional factors, and argues that these factors directly influence people’s trust in automated systems [27]. For example, the evaluation strategies for automated trust may differ among operators of different ages [45]. Positive emotions can significantly increase trust levels [46]. Experiential knowledge can contribute to the understanding of automation [47] and directly influence the establishment and development of trust in automation [48]. In addition, personality traits tend to dominate the early development of trust in automation [49].

Second, partner factors refer to the features and capabilities of the system. For example, a highly reliable automated system will promote an increase in the operator’s trust [50]. The quasi-human nature of the interface can make the operator show stronger trust elasticity [51,52]. Designing more transparent automated systems can better promote appropriate trust [53]. However, there may also be cases where the higher the degree of automation, the more difficult it is for operators to understand, which may lead to a decrease in trust [54].

Third, environmental factors encompass the influence of external conditions on the development of trust, including the context in which the interaction with the robot takes place, the socio–cultural context, and the technological environment [27,29]. For example, transportation infrastructure [55], risks inherent in the interaction process [56], and societal influences [57] can alter public trust in automation.

Trust in AVs is a specific manifestation of human–machine trust, representing the public’s confidence in autonomous driving technology. The trust-in-automation three-factor model provides an important reference for exploring the factors affecting trust in AVs and constructing the structural assurance mechanism for trust establishment in AVs.

3. Research Model and Hypotheses

3.1. Research Model

Although the trust-in-automation three-factor model explains the factors that affect trust in AVs, some of these factors are difficult to directly quantify and operationalize. Compared with the trust-in-automation three-factor model, the concept of structural assurance emphasizes the importance of subjective control measures. Structural assurance can be applied to technical and organizational environments [36], which can be divided into two categories: one is the technical structural assurance based on technical standards, data privacy protection, technology application security, etc., and the other is the external social structural assurance, such as laws and regulations, policies and measures, and cultural institutions. Combining the trust-in-automation three-factor model and the conceptual framework of structural assurance, this study aims to reveal the subjective and controllable factors that can promote public trust and effectively improve public trust in AVs through the establishment of structural assurance mechanisms. Based on this, this study constructs a structural assurance model of AV trust for the Chinese public based on the technical characteristics of AVs, the social environment, and individual cognitive factors, as shown in Figure 1.

3.2. Hypotheses Development

3.2.1. Technical Structural Assurance and Trust in AVs

Safety is the public’s confidence in the ability of AVs to adapt to complex traffic rules and mitigate unpredictable risks in a changing traffic environment. Safety stands out as the paramount factor in enhancing trust in AVs [11,58]. Compared to activities such as e-commerce and online payments, AVs entail greater physical risks [14]. Papadoulis et al. [59] and Vander and Sadabadi [60] find that AVs exhibit quicker response times and safer driving operations compared to human drivers. The intelligent sensor technology of AVs assists them in making prudent decisions in unforeseen road situations, thereby enhancing road safety [61]. If the public perceives that AVs can effectively safeguard their personal and information security, they are more likely to trust AVs [62]. Therefore, this study proposes the following hypothesis.

H1. 

Safety has a positive (+) effect on trust in AVs.

Autonomy in the context of AVs refers to their capacity to effectively perceive the surrounding environment, make decisions, and execute tasks without external control or supervision [63]. As a core functionality of AVs, autonomy not only embodies their ultimate developmental value but also significantly influences trust [35,64]. The higher the technical autonomy of AVs, the more accurately they can perceive the surrounding environment and the more rapidly they can make decisions, thus providing a safer and more efficient riding experience [24,65]. When the public becomes aware of the high level of technical autonomy in AVs, they are more inclined to trust the reliability and safety of these vehicles, thereby increasing their level of trust. Therefore, this study proposes the following hypothesis.

H2. 

Autonomy has a positive (+) effect on trust in AVs.

Human–machine interaction (HMI) refers to the system that aids drivers in understanding vehicle status, controlling vehicle operations, and participating in decision-making processes. HMI is responsible for providing passengers with vehicle-related information, including vehicle status, road conditions, navigation information, etc. The accuracy, clarity, and timeliness of information are crucial for fostering public trust [66]. HMI influences the establishment of public trust in AVs [67], and anthropomorphic interaction design facilitates enhancing public experience, comfort, operational convenience, and intuitiveness [51]. HMI design should consider public involvement and sense of control, making the public feel they can comprehend and participate in the decision-making process of AVs to reduce public anxiety and enhance trust in the system [68]. Therefore, this study proposes the following hypothesis.

H3. 

HMI technology has a positive (+) effect on trust in AVs.

Ubiquitous connectivity refers to the continuous and efficient communication capability between AVs and the external environment, other vehicles, and infrastructure. Through V2V (Vehicle-to-Vehicle) and V2I (Vehicle-to-Infrastructure) communication, AVs can engage in cooperative driving with other vehicles and infrastructure, assisting them in better identifying the positions, speeds, and intentions of other vehicles on the road, thereby reducing the risk of traffic accidents [4]. By connecting to external services and emergency response systems, AVs can promptly notify passengers and take appropriate measures to address issues, thus enhancing users’ trust in AVs through immediate response and handling [26]. Through real-time navigation, route planning, intelligent parking, and other functionalities, AVs provide passengers with a more convenient and efficient travel experience, thereby fostering greater trust in the reliability and practicality of AVs among the public [69]. Therefore, this study proposes the following hypothesis.

H4. 

Ubiquitous connectivity has a positive (+) effect on trust in AVs.

3.2.2. Social Structure Assurance and Trust in AVs

As the primary entity responsible for promoting the application of innovative technologies, the government wields significant influence over public trust, acceptance, and adoption of such innovations [70]. Clarifying accountability for AV accidents, protecting public personal privacy, and ensuring data security through legal means can help instill trust among the public [11,35]. There exists a close relationship between the development of relevant infrastructure and the establishment of trust in AVs. Well-maintained road facilities, functioning traffic signal lights, clear and visible road signs, and accurate and timely updates on traffic information and road conditions play a crucial role in fostering trust in AVs [55,71]. These factors drive users and passengers to trust the navigation and decision-making capabilities of AVs’ autonomous driving technology. Therefore, this study proposes the following hypothesis.

H5. 

Government support has a positive (+) effect on trust in AVs.

Media impact refers to the extent to which the public obtains objective information about AVs through the media. Media coverage serves as a conduit for delivering information and the latest developments regarding AV technology to the public [72]. The content, presentation, and tone of media coverage influence public trust in AVs [73,74]. Through the accurate and objective release of AV-related information, the media can assist the public in developing a proper understanding of this technology, thereby enhancing trust in AVs [75]. Positive media coverage can increase public trust and acceptance of AVs, while negative coverage may raise doubts and distrust among the public [76]. Excessive media coverage of AV accidents can lead to public concerns about the safety of AVs. Therefore, this study proposes the following hypothesis.

H6. 

Media impact has a positive (+) effect on trust in AVs.

3.2.3. Personal Cognitive Factors and Trust in AVs

Knowledge of AVs refers to the extent of the public’s knowledge regarding the performance attributes and advantages of AVs. Studies have shown that increasing understanding of autonomous driving systems through various means significantly increases trust towards these systems [77,78]. People’s trust in AVs increases with their level of understanding of the technology because as they gain more knowledge about the performance attributes and advantages of AVs, they may perceive themselves to possess perceptual control capabilities for operating and driving AVs [79]. Research by Liu et al. [80] found that the more the public understands Robotaxi, the higher their willingness to use the service, whereas a lack of understanding of Robotaxi may influence passengers’ emotional and situational trust. Therefore, this study proposes the following hypothesis.

H7. 

Knowledge of AVs has a positive (+) effect on trust in AVs.

Effort expectancy refers to the perceived ease of operating AVs by the public. Individual perceptions of the difficulty and effort involved in adopting new technology directly influence their level of trust in that technology [64]. Individuals with lower effort expectations are more likely to trust AVs because they are more inclined to believe that AVs can easily meet their needs without requiring excessive cognitive and psychological effort [70]. When public perceptions and behavioral efforts regarding AVs exceed the individual’s own cognitive and behavioral scope, doubts and uncertainties may arise, thereby affecting the establishment of trust in AVs among the public [81,82]. Therefore, this study proposes the following hypothesis.

H8. 

Effort expectancy has a positive (+) effect on trust in AVs.

4. Research Methodology

4.1. Instrument Design

This study utilized the existing literature to develop the instrument, with all items being assessed using a 5-point Likert scale. In order to better align with the research context of AVs, several items were revised. Prior to the main data collection phase, a preliminary pilot study was conducted. A sample survey of 50 undergraduate students was administered to assess the reliability and validity of the scale. Two items with factor loadings lower than 0.7 were deleted in order to improve the validity of the constructs [83]. The finalized instrument is presented in

Table 1. Measurement scale.

Constructs	Description	References
Safety	AVs can reduce the occurrence of accidents. AVs has the ability to cope with unexpected situations. AVs can drive safely even at night or in poor weather. AVs can ensure personal and private data security.	[11,61]
Autonomy	AVs can independently provide me with an action plan. AVs can independently complete travel tasks. AVs can ensure autonomous travel through advanced technology. AVs can optimize routes in real-time to ensure safe and efficient travel.	[24,65]
Human –Machine Interaction	AVs can provide comprehensive and timely feedback on vehicle travel information. AVs can timely and accurate reception and execution of travel instructions. AVs’ human–machine interaction can be highly anthropomorphic and easy to understand and operate. AVs can quickly hand over control of the vehicle to me in case of emergency.	[51,66]
Ubiquitous Connectivity	Ubiquitous connectivity can obtain vehicle information and monitor vehicle status at any time. Ubiquitous connectivity can help the vehicle arrive at the commanded location in time and accurately. Ubiquitous connectivity can accurately maintain the distance between vehicles to ensure safe travel. Ubiquitous connectivity can capture real-time road traffic conditions in a timely manner.	[4,26]
Government Support	I will use AVs if the government provides some supportive cost subsidies. I will use AVs if the government develops a sound legal system. I will use AVs if the government can develop a comprehensive insurance system. I will use AVs if the government invests in relevant infrastructure (road facilities, information systems, signals).	[41,79]
Media Impact	I frequently see reports on AVs in various media. I frequently see reports on the intelligent travel of AVs in various media. I frequently see reports on the safe travel of AVs in various media. I frequently see reports on the current state of development of AVs in various media	[73,75]
Knowledge of AVs	I know the current status of AV technology development. I know the advantages of AVs such as improving travel safety, reduced traffic accidents, increased mobility, energy savings, etc. I know the AV Robotaxi trial run information. I know the current state of the development of AVs.	[78,79]
Effort Expectancy	I think the AV driving environment is easy to adapt to. I think AVs are easy to use. I think AVs are easy to interact with.	[64,70]
Trust	I trust that AVs can drive on their own without my assistance. I trust AVs can be safe and reliable in bad weather conditions. I trust AVs’ driving skills more than my own. I trust AVs can make my trip go smoothly.	[24,43]

.

4.2. Data Collection

The survey areas for this study were selected as the eastern coastal provinces of China. To facilitate data collection, both questionnaire design and distribution were conducted through the website Questionnaire Star. The questionnaire was distributed from 25 November 2023 to 15 December 2023, lasting for approximately three weeks. In order to maintain questionnaire integrity, 65 responses deemed invalid were excluded due to criteria including overtly illogical answers or completion times below 120 s. Finally, a total of 548 valid questionnaires were obtained and the demographic features of the respondents are summarized in

Table 2. Demographic features.

Items	Types	Numbers	Percentage (%)
Gender	Male	247	45.1%
	Female	301	54.9%
Age	18–25 years old	138	25.2%
	26–30 years old	193	35.2%
	31–35 years old	92	16.8%
	36–40 years old	78	14.2%
	41–50 years old	28	5.1%
	Above 50 years old	19	3.5%
Education level	High school graduates	46	8.4%
	Junior colleges	149	27.2%
	Four-year colleges	281	51.3%
	Graduate schools and above	72	13.1%
Average monthly income	Below USD 1000	98	17.9%
	USD 1000~USD 1500	211	38.5%
	USD 1500~USD 2000	134	24.5%
	USD 2000~USD 2500	83	15.1%
	More than USD 2500	22	4%

.

According to the gender distribution, out of 548 respondents, 45.1% were male and 54.9% were female. A total of 90% of the respondents fell within the age range of 25 to 40, which is consistent with the characteristics of younger age groups seen in new energy vehicle consumer demographics. In terms of academic background, more than half of the respondents possessed a bachelor’s degree or above. Additionally, 43.6% of respondents had an average monthly income greater than USD 1500, reflecting the high level of economic development in eastern coastal areas as well. Among the respondents, 384 individuals reported having driving experience, while 164 did not. Additionally, 247 respondents had experience using assisted driving features, whereas 301 did not. Among these respondents, 72% opted for individual purchase, while 28% chose shared usage. This indicates a greater inclination among respondents toward individual ownership of autonomous vehicles rather than opting for shared usage. It is also evident that, presently, shared mobility in the AVs domain has not yet gained widespread acceptance or popularity.

4.3. Structural Equation Modeling (SEM) Analysis

4.3.1. Measurement Modeling

This study conducts a confirmatory factor analysis to evaluate the reliability and validity of the measures, as it focuses on latent variables. The findings are presented in

Table 3. Confirmatory factor analysis results.

Construct	Items	λ	CR	AVE	α
Safety (SAF)	SAF1	0.840	0.848	0.583	0.846
	SAF2	0.691
	SAF3	0.747
	SAF4	0.768
Autonomy (AUTO)	AUTO1	0.737	0.804	0.506	0.802
	AUTO2	0.705
	AUTO3	0.678
	AUTO4	0.724
HMI	HMI1	0.784	0.841	0.57	0.841
	HMI2	0.708
	HMI3	0.746
	HMI4	0.779
Ubiquitous Connectivity (UC)	UC1	0.855	0.879	0.645	0.875
	UC2	0.800
	UC3	0.709
	UC4	0.841
Government Support (GS)	GS1	0.833	0.883	0.654	0.882
	GS2	0.780
	GS3	0.776
	GS4	0.844
Media Impact (MI)	MI1	0.813	0.818	0.531	0.812
	MI2	0.739
	MI3	0.634
	MI4	0.716
Knowledge of AVs (KN)	KN1	0.823	0.864	0.614	0.863
	KN2	0.738
	KN3	0.758
	KN4	0.811
Effort Expectancy (EE)	EE1	0.828	0.830	0.621	0.826
	EE2	0.720
	EE3	0.811
Trust (TRU)	TRU1	0.786	0.844	0.576	0.844
	TRU2	0.769
	TRU3	0.772
	TRU4	0.705

Note: Model fit statistics: χ² = 806.978, df = 657, χ²/df = 1.228, CFI = 0.986, GFI = 0.931, AGFI = 0.919, IFI = 0.987, NFI = 0.932, RMR = 0.025, and RMSEA = 0.020.

and

Table 4. Discriminant validity of the constructs.

	SAF	AUTO	HMI	UC	GS	MI	KN	EE	TRU
SAF	0.764
AUTO	−0.287	0.711
HMI	0.298	−0.265	0.755
UC	0.430	−0.322	0.341	0.803
GS	0.276	−0.235	0.215	0.385	0.809
MI	0.340	−0.282	0.300	0.387	0.338	0.729
KN	0.396	−0.198	0.292	0.422	0.393	0.268	0.784
EE	0.284	−0.202	0.288	0.368	0.329	0.395	0.223	0.788
TRU	0.497	−0.341	0.406	0.533	0.475	0.462	0.540	0.450	0.759

Note: In bold on diagonals is the square root of AVE. Off diagonals are Pearson correlation of constructs.

. Table 3 provides information on model fit indices, standardized factor loadings (λ), Cronbach’s α, composite reliability (CR), and average variance extracted (AVE) for each measure. In addition, Table 4 examines the discriminant validity of the questionnaire by comparing correlation coefficients among the five latent variables and the square roots of the AVE.

As indicated in Table 3, the fitness measures of the CFA model all meet the specified value standards. Specifically, χ²/df is 1.228, which is below 3; CFI is 0.986, exceeding 0.95; and RMSEA is 0.020, lower than 0.05. Furthermore, the standardized factor loadings are all above 0.6, while both Cronbach’s α and composite reliability (CR) are above 0.8, and average variance extracted (AVE) exceeds 0.5, thus indicating that the measurement shows good convergent validity. According to the discriminant validity analysis results presented in Table 4, it is evident that the square root of the AVE for each construct significantly exceeds its correlation with other constructs. These findings indicate strong discriminant validity among the constructs.

4.3.2. Path Analysis

The path significance is tested using structural equation modeling, and the results are summarized in

Table 5. Significance test of the path coefficients.

Path				β	S.E.	C.R.	p-Value	Hypothesis
H1	Trust	<---	SAF	0.202	0.041	4.68	***	Accepted
H2	Trust	<---	AUTO	−0.079	0.045	−2.012	0.044	Rejected
H3	Trust	<---	HMI	0.099	0.042	2.522	0.012	Accepted
H4	Trust	<---	UC	0.113	0.047	2.525	0.012	Accepted
H5	Trust	<---	GS	0.135	0.04	3.323	***	Accepted
H6	Trust	<---	MI	0.144	0.05	3.281	0.001	Accepted
H7	Trust	<---	KN	0.289	0.041	6.542	***	Accepted
H8	Trust	<---	EE	0.182	0.044	4.315	***	Accepted

Note: *** p < 0.001.

and Figure 2. The results of the goodness-of-fit evaluation of the structural model are measured as CMIN/DF = 1.244, RMSEA = 0.021, GFI = 0.93, AGFI = 0.917, TLI = 0.984, and CFI = 0.985.

Based on the empirical results, it is evident that hypothesis H1 is supported, indicating a significantly positive effect of safety (β = 0.202, p < 0.001) on trust in AVs. This finding aligns with similar conclusions drawn in the existing research [11], suggesting that the safety of AV technology forms the foundation for building user trust. Although autonomy (β = −0.079, p < 0.05) has a significant effect on trust in AVs, the path coefficient is negative, which differs from the conclusions of the previous literature; thus, H2 is not supported. In contrast to the findings of Verberne et al. [65] and Wu et al. [24], the result of this study shows that as the level of autonomy in AVs increases, public trust in AVs decreases. Moreover, H3 is supported by the empirical results, indicating a significant positive effect of HMI (β = 0.099, p < 0.05) on trust in AVs. This finding aligns with similar findings in the existing research [27,67]. Through HMI, the interaction between the public and automated vehicle systems can be facilitated, enabling enhanced public engagement with AV technology. H4 is supported by the empirical results, suggesting a significant positive effect of ubiquitous connectivity (β = 0.113, p < 0.05) on trust in AVS. This finding is consistent with similar conclusions drawn in existing studies [26].

H5 is supported by the empirical results, indicating a significant positive effect of government support (β = 0.135, p < 0.001) on trust in AVs. This is consistent with similar conclusions drawn in existing studies [80]. Governments can ensure the safe operation of AVs on roads and protect the interests of the public by enacting relevant laws, regulations, and standards. Based on the empirical results, H6 is supported, indicating that media impact (β = 0.144, p < 0.01) has a significant positive effect on trust in AVs. This finding is consistent with similar conclusions drawn in the existing research [75]. Positive and constructive media coverage and commentary can help build public trust and acceptance of AVs, whereas negative coverage may raise doubts and concerns among the public.

Furthermore, from the empirical results, H7 is also supported, indicating that knowledge of AVs (β = 0.289, p < 0.001) has a significant positive effect on trust in AVs. This finding aligns with the existing research [11,26]. Acquiring knowledge about AVs, including relevant information, current developments, advantages, and disadvantages, helps the public better understand AVs. In addition to this, evidence from the empirical results supports H8 as well, which indicates that effort expectancy (β = 0.182, p < 0.001) has a significant positive effect on trust in AVs. This result is consistent with the findings of previous studies [24,64].

4.4. Moderating Effect Tests

To examine whether there are differences in the influence of variables such as gender, age, education, and income of the research subjects as shown in

Table 6. Subgroups with different backgrounds.

Group	Subgroup	Group Size
Gender	Male	301
	Female	247
Age	less than 30 years old	331
	more than 30 years old	217
Education Level	Less than a bachelor’s degree	195
	Bachelor’s degree or above	353
Monthly income	less than USD 1500	309
	USD 1500 or more	239

, Multiple Group Analysis was conducted. Pairwise Parameter Comparison was used to confirm differences between individual groups, and the presence of differences between groups was determined using Critical Ratios for differences between parameters. When the absolute value of CR (critical ratios) is greater than 1.96, the multigroup structural path is incomplete and the moderating effect holds. The results of the moderating effect tests are shown in

Table 7. Results of roderating effect tests.

Group	Relationship			β0	t	β1	t	CR
Male (0) vs. Female (1)	SAF	→	Trust	0.253	4.132 ***	0.26	4.596 ***	−0.389
	AUTO	→	Trust	0.041	0.824	−0.326	−4.877 ***	−4.387
	HMI	→	Trust	0.081	1.429	0.114	2.43 *	−0.07
	UC	→	Trust	0.114	1.667	0.151	2.993 *	0.168
	GS	→	Trust	0.117	2.155 *	0.218	3.919 ***	0.842
	MI	→	Trust	0.14	2.441 *	0.103	1.747	−0.87
	KN	→	Trust	0.249	4.199 ***	0.209	3.579 ***	−0.884
	EE	→	Trust	0.263	4.665 ***	0.114	2.113 *	−2.449
Less than 30 years old (0) vs. More than 30 years old (1)	SAF	→	Trust	0.199	3.471 ***	0.209	3.388 ***	0.092
	AUTO	→	Trust	0.036	0.613	−0.112	−2.057 *	−1.813
	HMI	→	Trust	0.143	2.883 **	0.06	0.955	−0.714
	UC	→	Trust	0.228	4.032 ***	0.058	0.796	−1.505
	GS	→	Trust	0.16	3.141 **	0.09	1.436	−0.642
	MI	→	Trust	0.102	1.627	0.196	3.225 **	−1.493
	KN	→	Trust	0.243	4.505 ***	0.368	5.259 ***	2.126
	EE	→	Trust	0.28	5.184 ***	0.134	1.989 *	−0.951
Less than a bachelor’s degree (0) vs. Bachelor’s degree or above (1)	SAF	→	Trust	0.224	4.355 ***	0.088	1.083	−1.565
	AUTO	→	Trust	−0.07	−1.705	−0.444	−2.428 *	−2.106
	HMI	→	Trust	0.108	2.456 *	0.046	0.577	−0.779
	UC	→	Trust	0.063	1.254	0.157	1.518	−0.868
	GS	→	Trust	0.163	3.557 ***	0.103	1.072	−0.701
	MI	→	Trust	0.114	2.432 *	0.181	1.445	−0.491
	KN	→	Trust	0.302	5.804 ***	0.062	0.662	−2.216
	EE	→	Trust	0.258	5.399 ***	−0.075	−0.724	−2.937
Less than USD 1500 (0) vs. USD 1500 or More (1)	SAF	→	Trust	0.148	3.213 ***	0.268	3.64 ***	1.314
	AUTO	→	Trust	0.081	1.684	−0.117	−1.892 *	−2.515
	HMI	→	Trust	0.129	3.079 **	0.08	1.136	−0.412
	UC	→	Trust	0.253	5.098 ***	0.022	0.293	−2.568
	GS	→	Trust	0.095	1.956	0.155	2.366 *	0.733
	MI	→	Trust	0.237	3.48 ***	0.11	1.736	−1.237
	KN	→	Trust	0.366	7.719 ***	0.227	2.79 **	−0.65
	EE	→	Trust	0.282	5.387 ***	0.117	1.789	−1.445

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

.

The moderating effect of gender on the relationship between autonomy (CR = −4.387), effort expectancy (CR = −2.449), and trust is found to be statistically significant. In the pathway of the impact of autonomy on trust, the female group (β₁ = −0.326, p < 0.001) exhibits a significant effect, while the male group (β₀ = 0.041, p > 0.05) shows no significant effect. In the pathway of the impact of effort expectancy on trust, both the male group (β₀ = 0.263, p < 0.001) and the female group (β₁ = 0.114, p < 0.05) exhibit significant effects but the coefficient of influence for the male group is larger than that of the female group.

The moderating effect of age on the relationship between the knowledge of AVs (CR = 2.126) and trust is found to be statistically significant. On the positive effect of knowledge of AVs on trust, both the group less than 30 years old (β₀ = 0.243, p < 0.001) and the group more than 30 years old (β₁ = 0.368, p < 0.001) have a significant effect.

The moderating effect of education on the relationship between autonomy (CR = −2.106), knowledge of AVs (CR = −2.216), effort expectancy (CR = −2.937), and trust is found to be statistically significant. On the effect of autonomy on trust, there is a significant negative effect on the bachelor’s degree or above group (β₁ = −0.444, p < 0.05), but no significant effect on the less than a bachelor’s degree group (β₀ = −0.07, p > 0.05). In terms of personal perceptions, the level of education shows a significant difference. For the less than a bachelor’s degree group, knowledge of AVs (β₀ = 0.302, p < 0.001) and effort expectancy (β₀ = 0.258, p < 0.001) have a significant effect on trust, but not for the bachelor’s degree and above group.

The moderating effect of income on the relationship between autonomy (CR = −2.515), ubiquitous connectivity (CR = −2.568), and trust is found to be statistically significant. In the pathway of autonomy’s impact on trust, it significantly affects the group with a monthly income of USD 1500 or more (β₁ = −0.117, p < 0.05), but has no significant impact on the group with a monthly income less than USD 1500 (β₀ = 0.081, p > 0.05). The positive impact of ubiquitous connectivity on trust significantly affects the group with a monthly income of USD 1500 or less (β₀ = 0.253, p < 0.001) but has no significant effect on the group with a monthly income above USD 1500 (β₁ = 0.022, p > 0.05).

5. Discussion

In order to enhance public trust in AVs, this study investigates the factors influencing the establishment of trust in AVs from a subjective and controllable perspective by integrating the concept of structural assurance and the trust-in-automation three-factor model. This study also introduces an innovative structural assurance model for the trust establishment of AVs (See Figure 1). The development of the model addresses gaps in existing theoretical research and enriches the knowledge system within related fields in China. The analysis of field survey data contributes to the scientific construction of an effective structural assurance mechanism for AVs, which in turn enhances trust in AVs and supports the sustainability of urban transportation systems.

5.1. Reflections on the Model

Firstly, in terms of the technical structural assurance factors, safety, HMI, and ubiquitous connectivity have direct positive impacts on trust in AVs, while autonomy has a negative impact. This indicates that safety remains a crucial factor in influencing public trust in AVs. Enhancing public awareness of the safety of AV technology is critical for gaining public trust and acceptance. HMI technology facilitates interactive information exchange between the public and intelligent AV driving systems, fosters greater public engagement with AVs, and increases perceived control over AVs, thereby fostering trust in AVs. Ubiquitous connectivity technology enhances the safety performance of AVs and fosters public trust in autonomous driving systems by providing data support, real-time updates, collaborative driving, and other functionalities. However, contrary to prior research findings, autonomy exhibits a negative impact on trust in AVs. Although China’s autonomous driving technology has developed rapidly, it started relatively late. Relevant laws, regulations, and infrastructure are not yet fully developed, and public understanding of autonomous driving technology is relatively limited. Consequently, as the autonomy of AV technology increases, public trust in AVs in China tends to decline.

Secondly, in terms of the social structural assurance factors, government support and media impact have direct positive impacts on trust in AVs. This indicates that government efforts, such as enacting relevant laws and regulations to protect public interests and promoting infrastructure development and upgrades to ensure AV safety, can increase public trust. Moreover, with the increasing influence of social media on public life, positive and constructive coverage and comments play a pivotal role in fostering public trust in AVs, whereas negative coverage may evoke skepticism and concern among the public. Objective and comprehensive coverage of information related to AVs, especially the subsequent accident investigation and handling, has a significant impact on the public’s correct understanding of AV technology and is essential for building trust in AVs.

Thirdly, in terms of individual cognitive factors, knowledge of AVs and effort expectancy exert a greater influence on trust compared to technical and social structural assurance factors. This underscores that a more comprehensive understanding of AVs among the public facilitates the establishment of trust in them. As the level of AVs’ intelligence increases, their operation will be more humanized, and the public will think AV functions simpler and more user-friendly, which may thereby aid in the establishment of trust.

Fourthly, the results of the multiple-group analysis indicate significant differences in trust in AVs among different gender, age, education, and income groups.

(1) Gender differences: Males tend to prioritize technological autonomy enhancement, while females place greater emphasis on the safety and reliability associated with technological autonomy improvements. Male respondents tend to associate the use of technology with positive outcomes, making them more likely to believe that advancements in AVs will lead to positive travel experiences. In contrast, female respondents may exhibit more cautious expectations regarding the use of AVs.

(2) Age differences: The group over 30 years old places a greater emphasis on understanding AVs, and is more concerned about the level of knowledge about AVs. The group under 30 years old demonstrates a more open attitude towards trusting AVs, showing quicker acceptance of new things.

(3) Educational differences: The group with a bachelor’s degree or above has higher expectations for technology and is more aware of the limitations and potential safety hazards of autonomous driving technology, leading to skepticism towards the autonomy of AVs. For the group with an educational background below a bachelor’s level, trust in AVs is contingent upon understanding the current development status, advantages, and user-friendly operation.

(4) Income differences: The group with a monthly income of USD 1500 or more may be more concerned about the technological complexity and potential risks associated with enhancing of autonomy, thus holding a negative attitude towards autonomy. Conversely, the group with a monthly income of USD 1500 or less may prioritize the practicality and convenience brought by AVs.

5.2. Theoretical Implications

Firstly, this study conducts a comprehensive field study in major cities in China’s eastern coastal provinces. The consumers in these urban areas are anticipated to be the initial adopters of AVs. The detailed data collected from these consumers have significant practical value for the future promotion and application of AVs. This study not only offers valuable insights for policymakers but also provides critical data support for enterprises, enabling them to develop more effective market strategies to facilitate the widespread adoption and application of autonomous driving technology.

Unlike the more theoretical studies in the existing literature, this study validates the effectiveness of the trust-in-automation three-factor model in practical applications through empirical analysis and provides empirical support for the theoretical exploration of trust establishment. By integrating the concept of structural assurance, this study innovatively develops a structural assurance model for establishing trust in AVs. This approach fills gaps in existing basic theoretical research and enhances the body of knowledge in related fields in China.

Although technical factors are often regarded as the core element of trust in laboratory environments, the empirical results of this study show that personal cognition has emerged as the most significant influencing factor. The findings indicate that, at the current development stage of automated driving technology in China, personal factors have a greater impact on trust compared to technical and social factors. Among these factors, the understanding of AVs, surpassing system reliability and capability, emerges as the most significant determinant of trust. Individual experience and knowledge significantly enhance the understanding of AVs and play a crucial role in the establishment of trust.

Compared to the findings of Wu et al. [24] in Singapore, this study reveals that, although the three-factor trust model has broad applicability, the roles and relative importance of each factor vary significantly across different geographical and cultural contexts. Specifically, in Singapore, technological autonomy has a significant positive impact on public trust. However, in China, as the level of technological autonomy in autonomous driving increases, public trust in autonomous vehicles tends to decrease.

5.3. Practical Implications

The establishment of a structural assurance mechanism can provide the public with a stable and trustworthy environment to ensure the reliability and security of AVs. Therefore, based on theoretical research findings and empirical survey results, this study primarily proposes strategic recommendations from the perspectives of technical structural assurance and social structural assurance.

(1) Automotive companies and research institutions have to provide strong technical structural assurance for AV users.

Firstly, safety is the foundation for ensuring the safe operation of AVs under various conditions. Research institutions should conduct continuous safety testing and verification to ensure that AVs can operate safely under diverse road conditions and traffic scenarios. Furthermore, it is imperative to transparently demonstrate the safety performance of AVs to the public. This can be achieved through the public release of AV safety data and reports, including accident rates, failure rates, and the effectiveness of safety features.

Secondly, designing intuitive and comprehensible user interfaces is essential. These interfaces should provide clear status and feedback information, deliver real-time safety prompts and warnings, and enable the public to understand the safety performance of AVs and take appropriate actions when necessary.

Thirdly, automotive companies and research institutions should offer test driving experiences to the public, allowing them to personally experience the advantages and convenience of AV technology. Increasing public understanding and awareness of AVs through real-world tests and test drives helps establish trust between the public and the vehicles. Automotive companies can also cooperate with insurance companies to introduce insurance products suitable for AVs, providing appropriate economic and legal protection for users.

(2) The government and the media should increase public trust in AVs by providing multifaceted social structural assurance.

Firstly, governments should enact relevant laws and standards to strengthen safety regulations and certification for AVs. They should also promote the development of intelligent road infrastructure to ensure the safe operation of AVs on roads. Establishing scientific and clear legal regulations regarding accident liability protects the public’s interests. Governments can further enhance public trust in AVs by implementing subsidy policies and measures related to AVs to guide their development. The government can set up third-party testing and certification to review and accredit AVs to increase public trust in the technology.

Secondly, the media should objectively report factual situations. Governments, automotive companies, research institutions, and other relevant entities can use the media to convey information about AVs, such as their potential benefits, safety, and reliability, thereby influencing public attitudes and perceptions of AVs. Automotive companies and research institutions can utilize various social media platforms such as WeChat Moments, Xiaohongshu, Weibo, and TikTok to share AV-related videos and experiences, increasing exposure to AVs. Additionally, in the face of negative news, governments, automotive companies, and research institutions should provide timely, objective, and accurate responses. Media should follow up on the results of subsequent accident investigations and resolutions to minimize the negative impact of adverse news.

(3) The development of a structural assurance mechanism should consider the individual characteristics of consumers, such as age, gender, education level, and income.

The marketing promotion of AVs should initially target the younger demographic, as they are more inclined to trust AVs and are more likely to share their positive experiences with them through new media platforms like WeChat and TikTok. Furthermore, there should be a focus on educating male consumers about the technological advancements and autonomy of AVs smart self-driving technology, while female consumers should be informed about the safety and reliability aspects. Additionally, it is important to present comprehensive and transparent information about AVs’ intelligent autonomous driving technology to individuals with higher levels of education and income in order to build trust effectively.

5.4. Limitations and Future Research

This study has several limitations. Firstly, the sample mainly consists of residents from economically developed regions of China, which may introduce regional bias. Although this study has validated the effectiveness of the three-factor trust model within the context of Chinese society, its universality still needs to be further tested across a broader range of geographical and cultural contexts.

Given the aforementioned limitations, in future research, the geographic scope of the survey should be expanded. This will help to more comprehensively analyze the mechanisms through which structural assurance factors influence the establishment of trust in AV technology among the Chinese public. Although this study focuses solely on the Chinese public, future research might apply the proposed model to conduct comparative analyses across different countries, facilitating cross-cultural comparisons of public perceptions and attitudes towards AVs. Cross-national comparative research will help understand public acceptance and trust mechanisms for AV technology in various cultural and technological contexts, providing valuable references for the global promotion of AVs.

6. Conclusions

This study combines the concept of structural assurance with the trust-in-automation three-factor model to systematically classify the influencing factors of trust in AVs into three categories: technical structural assurance factors, social structural assurance factors, and public personal cognitive factors. The research findings indicate that among the technical structural assurance factors, safety, human–machine interaction (HMI), and ubiquitous connectivity have direct positive impacts on trust in AVs, while autonomy has a negative impact. Among the social assurance factors, government support and media impact have direct positive impacts on trust in AVs. Individual cognitive factors, such as knowledge of AVs and effort expectancy, have direct positive impacts on trust. The results of multiple group analysis indicate significant differences in trust in AVs among the gender, age, education, and income groups. Comparative analysis with previous studies reveals that prior to the formal rollout of AV technology, structural assurance, such as robust policy support, leading technological innovations, high-quality infrastructure, and proactive public education, plays a crucial role in enhancing trust.

This study was conducted within a specific geographical context, and the factors constituting the trust model may vary because of differences in cultural or technological backgrounds. For instance, in Singapore, the government has systematically developed policies, laws, intelligent transportation infrastructure, and public awareness initiatives, leading to relatively high acceptance of new technologies [84]. In this context, autonomous driving technology becomes a key factor influencing public trust. In contrast, in China, despite the rapid development of autonomous driving technology, its progress started later compared to Singapore, and related laws, regulations, and infrastructure remain underdeveloped. As a result, public awareness of the technology is relatively limited. Therefore, in China, personal cognition factors and social structural assurances play a more critical role in fostering public trust in autonomous vehicles. Based on these findings, for countries and regions aiming to introduce and promote autonomous driving technology, it is of significant reference value to thoroughly consider the local technological context and design targeted strategies to enhance public trust.