A Study of Young People’s Intention to Use Shared Autonomous Vehicles: A Quantitative Analysis Model Based on the Extended TPB-TAM

Abstract

Shared autonomous vehicles (SAVs) have the advantages of both autonomous driving technology and shared transportation, which is an important development direction for intelligent and green transportation in the future. However, a lack of trust and a high perceived risk have reduced the public’s willingness to use this mode of travel. To improve the public’s willingness to use it, many scholars have conducted research, but young people are still a neglected group. A structural equation model was used to test the models based on 316 survey samples. The results indicated that initial trust had a significant positive effect on the intention to use SAVs, while perceived security risk and perceived privacy risk had no significant effect on the intention to use, but perceived security risk can indirectly negatively affect the intention to use. In addition, attitude and face consciousness had a significant positive influence on intention to use, while subjective norms, perceived behavioral control, and perceived usefulness had a significant indirect positive influence on intention to use SAVs. The findings showed that the model used in this paper was reasonable and valid for explaining young people’s willingness to use SAVs. This will provide guidance for formulating more effective strategies for this group.

1. Introduction

With the development of the social economy, the number of private cars is increasing, and the problems of road traffic safety and congestion are becoming more and more serious. Autonomous vehicles (AVs), as a safer and more sustainable mode of transportation, have attracted the attention of many automobile manufacturers and scholars [1]. Some domestic and foreign studies predict that AVs are expected to become an alternative solution to many current traffic problems, which will bring revolutionary changes to transportation, automobile manufacturing, logistics, and other industries [2]. However, in the early days of autonomous driving, due to the high research and development costs of new technologies, AVs will be more expensive. It cannot meet the demand of the public, which wants to use AVs as private cars.

In this context, shared autonomous vehicles (SAVs) combined with the sharing economy are more easily accepted by the public [3]. Due to the sharing characteristics of SAVs, people’s willingness to buy private cars can be reduced to a certain extent, so as to alleviate traffic congestion and reduce accidents [4]. And because of its driverless feature, it can also avoid the phenomenon of no one taking orders and improve the efficiency of travel [5].

Although SAVs have the advantages mentioned above, these benefits will not be realized until they are available on a large scale. A survey by Woldeamanuel showed that while respondents acknowledged the benefits that autonomous driving technology can bring, they were also very concerned about the risks that can arise from using such technology [6]. Haboucha et al. found that even if SAVs were completely free, only 75% of respondents would use them for commuting [7]. And Zhang et al. pointed out that the biggest obstacle to the adoption of autonomous driving technology was not the technology itself but the public’s low willingness to use it [8]. Therefore, it is necessary to study the public’s willingness to use SAVs and determine the influencing factors. Then analyze the interactions among the factors. In addition, it is also important to find the underlying rule of SAVs use intention, which will provide a basis for subsequent relevant studies.

In order to improve the public’s willingness to use autonomous driving technology, researchers have been conducting relevant investigations in recent years. It has been found that perceived risk is the main obstacle for individuals to use AVs [9], and trust influences the extent to which people use autopilot systems [10]. Since Bauer introduced the concept of perceived risk from psychology to the field of consumer behavior in 1960 [11], it has been widely used in various fields. And in the field of intention to use AVs, there have been studies to enhance the explanatory ability of the model by introducing perceived risk [1,12]. Meanwhile, trust in autonomous driving has become a hot topic in recent years [13]. The perceived risk of the public was considered to be the mediating factor influencing their trust and acceptance of AVs [14]. Kenesei et al. pointed out that the research on trust and perceived risk is the key to quantifying the willingness to use AVs, and it is difficult to apply to any model without integrating them [15]. However, there is not enough evidence to verify the causal relationship between trust and perceived risk, and how these two factors affect the public when using SAVs.

It has been argued that young people influence future transportation patterns [16]. Guiding young people to use SAVs more often and reducing the use of private cars will create a more sustainable future. At the same time, a study found that young people’s willingness to use autonomous driving technology is lower than that of other age groups [17]. Therefore, it becomes important to study how to guide young people to use SAVs.

In this paper, in order to deeply explore the potential laws affecting the intention to use SAVs among young people, a model of the young people’s intention to use SAVs was used for the study. The model extended the theory of planned behavior and the technology acceptance model by introducing initial trust and perceived risk. In addition to this, the very Chinese characteristic of face consciousness was also considered. In addition, the structural equation model was used to analyze how trust and perceived risk affect young people’s intention to use SAVs, as well as the interaction of each influencing factor. The flow chart of this research is shown in Figure 1.

The findings of this study are expected to provide a deeper understanding of how psychological factors influence young people’s willingness to use SAVs. In addition, it will provide a basis for formulating policies to improve the use intention of SAVs among young groups and a reference for studying the use intention of SAVs among other groups.

This study is structured as follows: Section 1 is the introduction. Section 2 gives a review of the related literature. Section 3 describes the structure of the model and explains young people’s willingness to use SAVs. Section 4 presents the methodology, which includes data collection methodology and analysis methodology. Section 5 presents the results of the models. Finally, Section 6 summarizes the significance, limitations, and prospects of this study.

2. Literature Review

As a new means of transportation, research on the public’s willingness to use AVs is gradually becoming the focus of academic and industrial circles [11]. Some researchers believe that AVs will account for more than 90% of the entire fleet by 2055, while autonomous driving technology may be operational by 2030 [12]. The combination of AVs and shared mobility technologies will be very helpful in achieving sustainable transportation [13]. SAVs combine AVs with traditional car sharing and taxi services, which can provide inexpensive on-demand mobile service and convenient last-mile solutions. However, there are not yet many AVs and SAVs on the road, so it is critical to identify the factors that influence the public’s willingness to use SAVs [14].

Before autonomous driving technology enters the market, the public’s willingness to use the technology largely determines the development and diffusion of the technology, and there are many factors affecting the public’s willingness to use the technology [15]. By consulting relevant literature, the existing research investigates the public’s willingness to use innovative technologies from three aspects: one is the attribute of innovative products themselves; One is the psychological factors of the public; and the other is the personal attributes of potential users.

2.1. Product Attributes

In studies related to autonomous driving technology, many consumers have expressed their concerns. A study by Woldeamanuel et al. [6] found that in terms of human-computer interaction, people are concerned about the amount of time it will take to learn to use this new technology. Safety is paramount, with concerns about equipment failure and system performance in bad weather. Next, data privacy is also a concern. Conversely, it was found that users with a higher value of travel time tend to choose the latter for both conventional and AVs [16]. In other words, whether users are willing to spend money in exchange for the value of travel time depends on the heterogeneity of the value of individual travel time. The symbolic value is the ability of consumers to highlight their social status when owning or using an innovative good. A study [17] pointed out that face consciousness has a wide impact on the psychology and behavior of Chinese consumers, and people with a strong sense of face are more likely to choose this new technology to “increase face”.

2.2. Public Psychology

In examining the psychological factors that influence the public’s willingness to use AVs, some studies have used variables from traditional models, including the Theory of Planned Behavior, the Technology Acceptance Model, and the Unified Theory of Acceptance and Use of Technology. As research continues, some researchers have optimized the traditional models to be more explicit about the key factors that influence the public’s willingness to use AVs. The optimization methods mainly include adding independent variables to the model or refining the dimensions of the model dependent variables while adding independent variables [18]. Choi et al. [19] built a model to explain the effects of different factors on intention to use AVs based on the Technology Acceptance Model and the Trust Theory, and the results showed that perceived usefulness and trust were the main important determinants of intention to use AVs. Ming et al. [20] used the Technology Acceptance Model as a basis, introduced two latent variables, the sense of gain and perceived trust, to investigate respondents’ acceptance of AVs, and demonstrated the mutual causality between the latent variables under different choices. Through scholars’ research, these theoretical models and other psychological latent variables well explain the public’s willingness to use AVs, and their influence effects are being continuously verified.

When exploring the psychological factors that negatively influence the public’s willingness to use AVs, perceived risk is a main barrier to individuals using AVs [21], and trust influences the extent to which people use autopilot systems [22]. Since Bauer introduced the concept of perceived risk from psychology to the field of consumer behavior in 1960 [23], it has been widely used in various fields. And in the field of intention to use AVs, there have been studies to enhance the explanatory ability of the model by introducing perceived risk [1,24]. Meanwhile, trust in autonomous driving has become a hot topic in recent years [25]. The perceived risk of the public was considered to be the mediating factor influencing their trust and acceptance of AVs [26]. Kenesei et al. [27] pointed out that the research on trust and perceived risk is the key to quantifying the willingness to use AVs, and it is difficult to apply to any model without integrating them. However, there is not enough evidence to verify the causal relationship between trust and perceived risk and how these two factors affect the public when using SAVs.

Finally, to deeply explore the potential laws affecting the willingness to use SAVs, the above traditional research framework and its improved model can only help researchers clarify the key factors affecting public acceptance, and it is difficult to give more precise analytical results [18]. Therefore, there is also a need to introduce more in-depth quantitative analysis tools. Similar to econometric models such as Mixed Logit Modeling, Structural Equation Modeling, etc.

2.3. Potential Users

To more fully explain the public’s willingness to use autonomous driving technologies, some studies have investigated the socio-demographic attributes of potential users of AVs, including sex, age, education, and income. Lingbo’s study found heterogeneity in the factors influencing the acceptance of driverless cabs across sex, age, and education cohorts. Zehua et al. [28] investigated whether individual differences affect the acceptance of self-driving buses among the population of Nanjing, China. According to the findings, both demographics and personality traits had a significant impact on the acceptance of self-driving buses. In terms of sex, there was a general difference between males and females in their willingness to use AVs or SAVs. Males had a more positive attitude toward AVs. Kyriakidis et al. [29] further confirmed that females were more concerned about issues related to AVs based on 5000 questionnaires from 109 countries. Herrenkind et al. [9] conducted a comparative study of the acceptance of driverless buses between younger and older age groups. Significant differences were found between younger and older people in terms of the effect of perceived usefulness on intention to use. Haboucha et al. [7] found that AVs were more likely to be used by people with higher levels of education. This conclusion was again validated by the findings of Gaojian et al. [30].

However, little is currently known about the potential users of SAVs [13]. Some researchers believe that the current young generation has grown up in an environment of new technology and realizes the importance of protecting the environment and paying more attention to sustainable development. And because of their low consumption level, they are easily ignored by the related tourism industry. However, they bring a lot of profits to the tourism industry in various places. Due to the changing demographics of today’s world, this generation will shape future travel needs [31]. Therefore, understanding these young travelers will be key to promoting SAVs.

In previous studies on young people, Herrenkind et al. studied young people’s acceptance of driverless buses and defined young people as those under 35 years old [9]. Wan et al. considered adults aged 18–50 as young adults and surveyed their acceptance of AVs [20]. According to China’s Medium and Long Term Youth Development Plan (2016–2025), the age range of the youth group is defined as 14–35 years old, and due to legal provisions, the young group is defined as 18–35 years old.

Automated driving technology is the future development trend. Now that the younger generation is influenced by the Internet and its consumption concepts and needs are very different from other age groups of users, how to make the development of automated driving technology keep up with the times and the changes in user needs is the problem that should be solved at the moment.

3. SAVs Use Intention Model Construction

3.1. Theoretical Background

So far, researchers have developed many models to explain human behavior and their willingness to use new technologies.

The Theory of Planned Behavior (TPB) is the most fundamental conceptual model and theoretical basis in the field of individual behavior research and is mainly used to explain the decision-making process of individual behavior [32]. TPB was finally formed after three stages of development, from the initial multi-attribute attitude theory to the rational behavior theory and then to the Theory of Planned Behavior [33,34,35]. In 1985, Ajzen proposed the complete TPB framework. According to TPB, the most influential factor in an individual’s behavior is his or her intention to perform a behavior. And behavioral intention is interpreted as the result of attitudes, subjective norms, and perceived behavioral control. The TPB model has gained a lot of recognition for predicting and explaining human behavior and intention. And the theory has been used in a wide variety of research areas, including environmental science, healthcare, supply chain management, and transportation [36]. Therefore, TPB was selected as the theoretical support for the whole paper.

The technology acceptance model (TAM) was first proposed by Davis in 1985, based on rational behavior theory [37]. It is used to explain the extent to which users are willing to accept and use emerging technologies and is the most commonly used theoretical model in empirical studies of autonomous driving acceptance. According to this model, behavioral intention is influenced by both attitude (AT) and perceived usefulness (PU). Furthermore, perceived usefulness and perceived ease of use (PEU) are two main predictors, which jointly affect behavior and attitude [10]. At present, SAVs are in the early stages of new technology development and have not been widely used. Meanwhile, considering that most travelers have no experience using SAVs, this study mainly focused on PU and AT and did not include PEU.

To gain a deeper understanding of public attitudes and willingness to use AVs, several scholars have optimized traditional models for research. Zhang et al. introduced initial trust and perceived risk into TAM to explore users’ acceptance of L3-level AVs [8]. Peng et al. proposed and tested an extended TPB model by integrating the two variables of cognition and perceived risk [1]. These studies demonstrated the good applicability of TPB and TAM in the context of autonomous driving technology.

TPB and TAM, which are both derived from rational behavior theory, have two common variables: behavioral attitudes and behavioral intentions. Only the focus is slightly different, making the two models theoretically compatible and complementary. Some scholars combined TPB with TAM and proved that combining the two models could better explain people’s intention to use automatic driving [10]. Therefore, this paper combined these two models to study young people’s willingness to use SAVs.

3.2. SAVs Use Intention Model

This paper integrated face awareness, trust, and two perceived risks (i.e., perceived security risk and perceived privacy risk) to better explain the young group’s Intention to Use (IU) for SAVs. The extended TPB-TAM theoretical framework model was used to analyze the young group’s willingness to use SAVs. The framework Model is shown in Figure 2.

The model included nine components: intention to use, attitude, perceived usefulness, subjective norms, perceived behavioral control, initial trust, perceived security risk, perceived privacy risk, and face awareness. And based on the research background and the structure of the research model, 14 hypotheses were proposed and tested in this paper.

3.2.1. Attitude and Perceived Usefulness

It is generally believed in TAM that PU refers to an individual’s understanding of whether a new technology is conducive to study or work, and PUE refers to an individual’s understanding of whether a new technology is easy to learn or use.

AT and PU positively influence the intention to use new technologies and thus contribute to actual consumer behavior. Moreover, Panagiotopoulos et al. showed that perceived usefulness was the main driving force for the adoption of AVs [38]. Therefore, the following assumptions were made in this paper:

H1. 

Young people’s attitudes have a significant positive impact on their intention to use SAVs.

H2. 

Young people’s perceived usefulness of SAVs has a significant positive effect on their intention to use SAVs.

H3. 

Young people’s perceived usefulness of SAVs has a significant positive impact on attitudes.

3.2.2. Subjective Norm and Perceived Behavioral Control

Subjective norms (SN) refer to the perceived social opinion and pressure when an individual makes a certain behavioral decision. And his perception may be influenced by the people around him, such as family, friends, or colleagues.

Perceived behavioral control (PBC) refers to individuals’ perceptions of their abilities, opportunities, and resources. Further, the more resources and opportunities individuals perceive they have, the fewer barriers they anticipate and the greater perceived control they have over their behavior.

In TPB, SN and PBC have significant effects on IU. Moreover, it has been empirically proven that the public’s SN and PBC have a significant positive impact on their intention to use autonomous driving technology [1]. At the same time, the usefulness of new technology is also the focus of users, and an individual’s perception of the accessibility of new technology is related to the practicality and functionality of the technology [8]. Based on this, this paper makes the following assumptions:

H4. 

Young people’s subjective norms have a significant positive effect on their intention to use SAVs.

H5. 

Young people’s subjective norms have a significant positive effect on the perceived usefulness of SAVs.

H6. 

Perceived behavioral control among young people has a significant positive effect on their intention to use SAVs.

H7. 

Perceived behavioral control among the young group has a significant positive effect on the perceived usefulness of SAVs.

3.2.3. Initial Trust and Perceived Risk

Besides the original variables in TAM and TPB models that are widely used in the use of autonomous driving, trust is also an important prerequisite for the public to accept and intend to use autonomous driving technology.

Trust is considered to be a key determinant of human-computer interaction. And lack of trust in autonomous driving technology is the most common reason why the public does not accept it. Moreover, trust is expressed in a variety of ways in people’s minds, not only in the performance of the car itself but also in the car manufacturers and the institutions that set the relevant regulations.

For new technologies, trust is expressed by users having positive expectations of the technology. However, it is important to note that most consumers do not yet have experience interacting with autopilot, so trust is more accurately referred to as initial trust (IT).

In the early stages of the marketization of new technologies, consumers should have enough trust to overcome their perception of risks and form a positive attitude towards them [39]. Kenesei et al. conducted correlation research based on the public’s trust-risk-use intention on AVs and found that trust has a great impact on the use intention [27].

Therefore, the following assumptions were made in this paper:

H8. 

Young people’s initial trust in SAVs has a significant positive effect on their intention to use them.

H9. 

Young people’s initial trust in SAVs has a significant positive effect on attitudes.

Trust in automation is always relative to the functionality the consumer wants to implement. To build consumer trust, automation must prove itself useful to users. And the important role of perceived usefulness in trust can be understood as the convenience and benefits that travelers think this travel mode can bring to travel or society. Dikmen et al. found that the initial trust of users in Tesla’s autonomous driving system was positively related to perceived usefulness [40]. Thus, this paper makes the following hypotheses.

H10. 

Young people’s perceived usefulness to SAVs has a significant positive effect on initial trust.

Perceived Risk (PR) refers to an individual’s uncertainty about the prospect of using a new technology. And PR may be the main barrier for individuals to use SAVs, as individuals’ perceived risk and their willingness to use are inversely correlated: the higher the perceived risk of the purchased good or service, the lower the willingness to use it.

As a new product that has not entered the market yet, there are many uncertainties about its use risk. Consumers are also worried that their safety will be threatened if the automatic driving system fails [41]. And Bansal et al. pointed out that consumers worry about privacy leakage in addition to safety [42]. In this study, PR was defined according to the young people’s perceptions of the performance of SAVs and the degree of personal privacy protection when using SAVs. Hoff et al. pointed out that initial trust develops largely based on the public’s perceived benefits and risks associated with AVs [43]. In other words, a certain amount of risk is necessary for trust to function.

Thus, this paper introduced perceived safety risk (PSR) and perceived privacy risk (PPR) and made the following hypotheses:

H11. 

Young people’s perceived safety risk of SAVs has a significant negative impact on initial trust.

H12. 

Young people’s perceived privacy risk to SAVs has a significant negative effect on initial trust.

3.2.4. Face Consciousness

Face consciousness (FC) refers to an individual’s desire to obtain a sense of self-worth from the attitude or behavior of others. It emphasizes that people have more social needs than personal needs, and consumers with strong face consciousness have higher social needs in consumption.

Chinese believe that their face not only represents their reputation but also the reputation of their family, relatives, and friends. So Chinese people tend to have a strong sense of face [44].

Jinpeng et al. empirically demonstrated that consumers’ psychological pursuit of social recognition partially influences the role of face consciousness in consumption behavior [45]. And Huang et al. pointed out that in the future, AV will be an important representative of fashionable technology products, and consumers with a strong sense of face are more likely to “increase face” by actively using this product [17]. Therefore, it was hypothesized here that:

H13. 

Young people’s face consciousness has a significant positive effect on their intention to use SAVs.

H14. 

Young people’s face consciousness has a significant positive impact on subjective norms.

4. Method

The main purpose of this study was to explore the complex relationship between each factor and the intention to use. Although there have been some studies on these factors, the relationship between these structures is not clear in the context of using SAVs. Therefore, based on the previous research, this paper put forward the relationship between the research model and its structure and validated it by designing experiments and obtaining data.

4.1. Development of Research Instruments

Considering that SAVs are a travel mode in the future, in order to avoid influencing respondents’ intention to use SAVs due to their ignorance of SAVs, the concept of SAVs was introduced at the beginning of the questionnaire. In addition, it was explained to the respondents that this questionnaire was anonymous and their information would not be disclosed, so that they could fully understand the purpose of this study (see Appendix A).

The questionnaire consisted of three parts. The first part was a description of the research objectives of this project and the concept of SAVs. The second part consisted of a number of questions related to psychological factors that influence respondents’ intentions to use SAVs. And the third part was demographic information, including sex, age, occupation, etc. The items related to willingness to use SAVs were designed on the theoretical basis of TPB and TAM. These items of psychological latent variables were all adopted on the five-point Likert scale (from 1 “strongly disagree” to 5 “strongly agree”), and the respondents chose the items according to their own opinions and attitudes.

The items involved in this paper (see Appendix B) were all from the maturity scale and have been verified in existing studies. The subjective norms, perceived behavioral control, and use intention of TPB referred to the scale developed by Yuexia et al. and Yuen et al. [46,47]. The attitude and perceived usefulness of TAM referred to the scale designed by Jing Peng et al. [37] and Sun Lingbo et al. [48]. The introduced variables, initial trust, perceived security risk, and perceived privacy risk, were mainly adopted from the questions designed by Zhang et al. [8]. Face awareness was measured using the scale developed by Jing-Peng et al. [37].

4.2. Sampling and Data Collection

The survey began on 8 October 2022. The questionnaire data was collected mainly through the Internet, and the questionnaire was distributed through “questionnaire star”. The questionnaires were distributed from 4 November 2022 to 25 November 2022. Respondents in this study were fully informed about the purpose of this study before completing the questionnaire. Respondents in this study were fully informed of the purpose of this study before completing the questionnaire. Respondents were consulted, and their consent was obtained to conduct the survey. Researchers had no access to information that could identify individual participants during or after data collection.

A total of 339 questionnaires were returned, and after eliminating non-conforming and illogical questionnaires, a total of 316 valid questionnaires were obtained, with an effective rate of 93.22%. The demographic information of respondents is shown in

Table 1. Summary of demographics.

Demographic Variable	Value Set	Frequency	Proportion (%)
Sex	Male	133	42.09
	Female	183	57.91
Age	18–22	111	35.13
	23–35	205	64.87
Occupation	Student	217	68.67
	Worker	83	26.27
	Freelancer	7	2.22
	Other	9	2.85
Education level	Associate’s degree below	4	1.27
	Associate’s degree	20	6.33
	Bachelor’s degree	181	57.28
	Postgraduate degree	111	35.13
Daily commuting mode	Private Car	77	24.37
	Public transportation	146	46.20
	Walking or cycling	85	26.90
	Other	8	2.53

.

As can be seen from Table 1, 42.09% of the respondents were male and 57.91% were female. In terms of occupation, the student group was more numerous, accounting for 68.67%. In respect of academic qualifications, young people were highly educated, mainly undergraduate, master’s, and above. In addition, most of the respondents commuted by public transport, accounting for 46.20%. Detailed results are in Supplementary Dataset S1.

4.3. Research Methodology Used

This study used Structural Equation Modeling (SEM) to quantitatively analyze the survey data. SEM is a statistical analysis tool for data that is formed by using multiple regression analysis, path analysis, and confirmatory factor analysis. It includes a measurement model and a structural model, which can examine the relationship between the observed variable and latent variable and each latent variable.

There are often errors when measuring variables, and the variables are reflected by multiple indicators. SEM can deal with errors well and assume the relationship between independent variables and multiple dependent variables. Therefore, this study used SEM to analyze the willingness to use SAVs and the factors influencing it among young people.

The measurement model is a description of the relationship between each observed indicator and its corresponding latent variable. The specific form is as follows:

$$x={\Lambda }_x\xi +\delta$$

(1)

$$y={\Lambda }_y\eta +\epsilon$$

(2)

where:

$x$ is the exogenous indicator;

${\Lambda }_x$ is the factor loading matrix of the exogenous indicator on the exogenous latent variable;

$\xi$ is the exogenous latent variable;

$\delta$ is the measurement error of $x$;

$y$ is the endogenous indicator;

${\Lambda }_y$ is the factor loading matrix of the endogenous indicator on the endogenous latent variable;

$\eta$ is the endogenous latent variable;

$\epsilon$ is the measurement error of $y$.

The structural model reflects the relationship between latent variables that are not directly measurable, in the following form:

$$\eta =B_{\eta }+\Gamma \xi +\zeta$$

(3)

where:

$B_{\eta }$ is the relationship between the endogenous latent variables;

$\Gamma$ is the effect of exogenous latent variables on endogenous latent variables;

$\zeta$ is the error term.

The identification and estimation of the model provide the basis for the feasibility analysis of the model. In this process, it is necessary to estimate the parameters of the model to check whether the designed model conforms to the actual situation. Moreover, the maximum likelihood estimation (ML) and partial least squares method (PLS) are commonly used to estimate the model parameters. While ML requires a large sample size, PLS is suitable for small sample models, and PLS-SEM parameter estimation results are relatively stable, which is suitable for more complex models. Based on the above characteristics, this paper used SmartPLS 3.0 for parameter estimation based on PLS.

5. Results

5.1. Data Analysis of Measurement Models

Before testing the structural model, it is usually necessary to assess the internal consistency and reliability, convergent validity, and discriminant validity of the measurement model.

Internal consistency is tested by Cronbach’s alpha (α) and Composite reliability (CR). It is usually recommended that α > 0.7 and CR > 0.7 when the measurement model is considered to have internal consistency and reliability [49].

Factor loadings (FL) and average variance extracted (AVE) can test for convergent validity and discriminant validity. It is generally suggested that FL > 0.6 and AVE > 0.5. At this time, it is considered that the measurement model has good convergence validity [49].

Discriminant validity is evaluated by comparing the square root of the AVE of each structure with other structural correlation coefficients. When all the square roots of AVE are larger than other structural correlation coefficients, the measurement model is considered to have good discriminative validity.

As shown in

Table 2. The results of statistical analysis and confirmatory factor analysis.

Constructs	Item	Mean	SD	Factor Loadings	Cronbach’s Alpha	Composite Reliability	Average Variance Extracted.
Attitude (AT)	AT1	3.684	0.861	0.874	0.881	0.926	0.808
	AT2	3.756	0.89	0.919
	AT3	3.646	0.918	0.902
Perceived Usefulness (PU)	PU1	3.769	0.981	0.833	0.779	0.857	0.602
	PU2	3.633	0.944	0.841
	PU3	3.399	1.034	0.718
	PU4	3.082	1.049	0.708
Subjective Norm (SN)	SN1	3.535	1.029	0.816	0.836	0.901	0.753
	SN2	3.649	0.917	0.902
	SN3	3.687	0.917	0.883
Perceived Behavioral Control (PBC)	PBC1	3.234	1.026	0.833	0.737	0.849	0.652
	PBC2	3.608	0.947	0.835
	PBC3	3.696	0.995	0.752
Initial Trust (IT)	IT1	3.25	0.902	0.917	0.904	0.94	0.84
	IT2	3.253	0.914	0.929
	IT3	3.304	0.915	0.903
Perceived Safety Risk (PSR)	PSR1	3.867	0.911	0.943	0.813	0.912	0.839
	PSR2	4.041	0.932	0.888
Perceived Privacy Risk (PPR)	PPR1	3.842	0.951	0.945	0.937	0.96	0.889
	PPR2	3.962	0.892	0.914
	PPR3	3.88	0.93	0.97
Face Consciousness (FC)	FC1	2.965	1.047	0.908	0.918	0.948	0.859
	FC2	3.028	1.029	0.94
	FC3	2.87	1.067	0.932
Intention to Use (IU)	IU1	3.633	0.79	0.861	0.844	0.906	0.762
	IU2	3.326	0.923	0.875
	IU3	3.598	0.838	0.882

, all values of α were above the recommended minimum value of 0.7. This indicated that the sample data had good internal consistency and were relatively realistic and reliable.

All Factor loading values were greater than 0.6, which means that the selected study variables were reasonable.

The CR value and AVE value of the latent variables were all greater than 0.7 and 0.5, respectively. It proved that the factors were strongly interpretive, the measurement errors were reasonable, and the convergence was good.

As shown in

Table 3. Results of discriminant validity test.

	AT	PU	SN	PBC	IT	PSR	PPR	FC	IU
AT	0.899
PU	0.76	0.776
SN	0.49	0.501	0.868
PBC	0.16	0.566	0.385	0.808
IT	0.673	0.719	0.482	0.512	0.916
PSR	0.053	0.063	0.181	0.126	−0.077	0.916
PPR	0.16	0.209	0.255	0.236	0.115	0.602	0.943
FC	0.391	0.49	0.375	0.309	0.535	0.004	0.109	0.927
IU	0.667	0.626	0.467	0.488	0.706	0.036	0.168	0.587	0.873

AT = attitude; PU = perceived usefulness; SN = subjective norms; PBC = perceived behavioral control; IT = initial trust; PSR = perceived safety risk; PPR = perceived privacy risk; FC = face consciousness; IU = intention to use.

, all AVE square roots were greater than the other structural correlation coefficients, which can reflect good discriminant validity. In general, the valid sample data was suitable for further research on model fitting.

5.2. Data Analysis of the Structural Model

Multiple Coefficient of Determination (R²) indicates the degree to which the variance of a latent variable can be explained by other latent variables. The coefficient was introduced here to analyze the fitting effect of the model.

According to available studies [50]:

Recommended minimum R² > 0.10;

When R²= 0.1~0.5, the explanatory power is weak;

When R²= 0.5~0.75, the explanatory power is strong;

And the larger the R², the stronger the explanatory power.

The results of the model explanatory power and significance analysis showed that IU: R² = 0.631, IT: R² = 0.535, AT: R² = 0.611, PU: R² = 0.414, and SN: 0.140, which proved that the model had good explanatory power. The results are shown in Figure 3.

Using the bootstrapping algorithm with 5000 subsamples for sampling tests, the path coefficients and t-values (T) between the structures can be obtained. Whether the hypothesis is supported can be determined from the path coefficients and T.

When the t-test value of the path coefficient is greater than 1.96, it is considered to pass the significance test. As shown in

Table 4. Results of hypothesis testing.

Hypotheses	Path	Path Coefficients	T-Values(T)	Supported?
H1	AT→IU	0.336 ***	5.274	Yes
H2	PU→IU	−0.063	1.039	No
H3	PU→AT	0.572 ***	11.415	Yes
H4	SN→IU	0.069	1.252	No
H5	SN→PU	0.332 ***	5.87	Yes
H6	SN→IU	0.099	1.83	No
H7	PBC→ PU	0.438 ***	9.857	Yes
H8	IT→IU	0.304 ***	5.065	Yes
H9	IT→AT	0.261 ***	4.725	Yes
H10	PU→IT	0.716 ***	20.02	Yes
H11	PSR→IT	−0.16 *	2.447	Yes
H12	PPR→IT	0.062	1.089	No
H13	FC→IU	0.272 ***	4.267	Yes
H14	FC→SN	0.375 ***	6.542	Yes

AT = attitude; PU = perceived usefulness; SN = subjective norms; PBC = perceived behavioral control; IT = initial trust; PSR = perceived safety risk; PPR = perceived privacy risk; FC = face consciousness; IU = intention to use. * p < 0.05, *** p < 0.001.

, except for H2, H4, H6, and H12, the other hypotheses (H1, H3, H6, H7, H8, H9, H10, H11, H13, and H14) were all supported by the data, and all of them were highly significant.

To determine the key factors and influencing paths of the intention to use SAVs, the Bootstrapping algorithm was used to calculate and analyze the total, direct, and indirect influences of all latent variables on the intention to use SAVs (

Table 5. Analysis of the effect of each variable on the intention to use SAVs.

Influencing Relationships	Direct Impact	Indirect Impact	Total IMPACTS	Pathways to Significant Indirect Effects
AT-IU	0.325 ***	—	0.325 ***	—
PU-IU	−0.063	0.464 ***	0.401 ***	PU→IT→IU; PU→AT→IU; PU→IT→AT→IU.
SN-IU	0.069	0.133 ***	0.202 ***	SN→PU→IT→AT→IU; SN→PU→AT→IU; SN→PU→T→IU.
PBC-IU	0.099	0.176 ***	0.275 ***	PBC→PU→IT→AT→IU; PBC→PU→AT→IU; PBC→PU→IT→IU.
IT-IU	0.304 ***	0.085 ***	0.389 ***	IT→AT→IU.
PSR-IU	—	−0.062 *	−0.062 *	PSR→IT→IU.
PPR-IU	—	0.024	0.024	—
FC-IU	0.272 ***	0.076 **	0.348 ***	FC→SN→PU→IT→AT→IU; FC→SN→PU→AT→IU; FC→SN→PU→UT→IU.

AT = attitude; PU = perceived usefulness; SN = subjective norms; PBC = perceived behavioral control; IT = initial trust; PSR = perceived safety risk; PPR = perceived privacy risk; FC = face consciousness; IU = intention to use. * p < 0.05; ** p < 0.01; *** p < 0.001.

).

The results showed that attitude, initial trust, and face awareness had a significant direct effect on the intention to use SAVs, while other variables except attitude and perceived privacy risk significantly and indirectly influenced intention to use. And although perceived security risks could not directly affect the intention to use SAVs, they could also negatively affect the intention to use SAVs through initial trust.

Previous studies on the intention to use automated driving have shown that perceived behavioral control and subjective norms in TPB and perceived usefulness performance in TAM directly and significantly affect behavioral intention [48]. In contrast, the results of this study showed that these three variables have no significant direct effect on the intention to use but can significantly affect the intention to use indirectly.

5.3. Discussion

The purpose of this study was to investigate the influence mechanisms of initial trust and perceived risk on the willingness to use SAVs among young Chinese to improve their trust and acceptance of SAVs. By integrating initial trust, perceived security risk, perceived privacy risk, and face awareness, an extended TPB-TAM model was applied in this paper.

This study used partial least squares to clarify the effect of each variable on the intention to use SAVs and tested the model. The results showed that ten of the fourteen hypotheses (H1, H3, H6, H7, H8, H9, H10, H11, H13, and H14) were supported by the data, while four hypotheses (H2, H4, H6, and H12) were not supported by the data. The results indicated that although perceived usefulness, subjective norms, and perceived behavioral control do not have a direct and significant effect on the intention to use SAVs, they influence use intentions to a large extent indirectly through other variables.

Hypotheses H1, H8, and H13 were highly significant, indicating that the three latent variables of attitude, initial trust, and face awareness had a significant positive effect on the intention to use SAVs. It also confirmed that the more positive attitude young people have towards SAVs and the stronger their sense of face, the more likely they are to use SAVs in the future. Hypothesis H8 was valid and indicated that young people’s choice of SAVs was largely based on their initial trust level in SAVs. Of the three hypotheses that had a direct impact on initial trust, perceived usefulness and perceived safety risk had a significant impact on initial trust, while perceived privacy risk did not.

The study found that the young people’s concern about privacy risks did not significantly affect their initial trust in SAVs, which may be because contemporary young people are in the era of big data and are less sensitive to data privacy. In contrast, their perceived security risk harmed their willingness to use SAVs, and perceived security risk had a significant negative relationship with initial trust. Therefore, it can be concluded that reducing young people’s perceived security risk helps increase initial trust. Also, reducing respondents’ concern about the risk of use will increase their initial trust, thus improving their willingness to use SAVs. Increasing perceived usefulness can also indirectly increase willingness to use SAVs.

Based on the above findings, it is suggested that automobile manufacturers pay more attention to improving the safety of autonomous driving technology. And vehicle tests can be used to establish a trusting relationship between young people and SAVs. At the same time, the government can strengthen the propaganda on the role of SAVs, such as reducing traffic congestion and pollution, in order to improve young people’s perception of the positive impact of SAVs on society.

In the future, when SAVs are introduced to the market early on, they are likely to be used only by a small group of people who are receptive to new things. Because the benefits that SAVs bring are not immediately realized. In this case, advocacy efforts need to identify people who are more likely to be early adopters of SAVs and target specific marketing strategies accordingly.

For example, this paper found that face consciousness had a significant direct and indirect impact on young people. However, subjective norms and perceived behavioral control had no significant direct effect on the use intention, but they indirectly and positively affected the use intention through perceived usefulness. In response, this study has the following recommendations: (1) SAVs publicity can be carried out for consumer groups with strong facial awareness. As they are more likely to use this new technology to gain the approval of others, it can also influence the decision-making of young people through social media communication. (2) More pilot projects on SAVs can be promoted in the future to reduce the expected psychological barriers to the use of SAVs among young people. (3) In addition, when improving relevant laws and regulations, relevant departments can also put forward some incentive policies for using SAVs according to the needs of young people, such as subsidies, sharing bonuses, and low-carbon points.

6. Conclusions

When SAVs are not yet widespread, people’s intentions to use them are a key factor in determining their future development. Therefore, it is necessary to use models to investigate people’s intentions to adopt SAVs.

This paper argued that the younger group’s intention to use SAVs is largely influenced by their initial trust in SAVs, which is built on perceived risk and perceived usefulness, and that these latent variables are also influenced by other factors. Therefore, TPB-TAM was expanded by introducing initial trust, perceived risk, and face awareness, and young people were taken as the main survey objects for analysis. The results showed that the overall effect of the SAVs use intention model used in this paper was good.

The questionnaire used in this study was based on the research hypothesis model. After obtaining the questionnaire data, the structural equation model and path analysis were used to clarify the effect of each variable on the intention to use SAVs.

Subjective norms and perceived usefulness in the original TPB-TAM model had no significant direct effect on the intention to use but had an indirect positive effect on the intention to use. And the four newly introduced variables, except for privacy risk, all significantly affected intent to use.

According to the path results of the model, the influence mechanism of psychological latent variables on SAVs use intention was analyzed, and the effect of the TPB-TAM classical framework on the influence of young people on SAVs use intention was verified.

The results showed that the structural model used in this study can quantitatively evaluate user data and filter out the factors affecting the intention to use SAVs, which is conducive to improving people’s trust and acceptance of SAVs. Our study has the following theoretical and practical implications.

6.1. Theoretical Contributions

The SAVs use intention model constructed for the young people targeted in this study provides a theoretical basis for studying consumers’ attitudes and preferences towards SAVs in the future and provides a theoretical reference for research on the use intentions of other age groups.

In addition, this study can be extended to investigate consumers’ intentions to accept and use smart products. Regarding smart products, some scholars believe that investigating consumers’ perceptions of products before purchase is a major direction for future research [51]. If researchers collect the attribute data of new technology before individuals decide to use it, so that consumers can feel or experience it, then it will be more valuable to predict the utilization rate of new technology [52]. The study collected respondents’ assessments of the perceived characteristics of SAVs prior to their commercialization. The study may provide theoretical support for future promotion strategies for SAVs. Future research may explore the key factors that affect consumers’ acceptance of SAVs after their commercialization so as to study the changes in consumers’ acceptance intentions before and after their commercialization.

This paper enriches the literature on the willingness to use SAVs by investigating the willingness to use SAVs in China, the world’s largest automobile market. The influence of Chinese cultural values (i.e., face consciousness) was analyzed. And this influence plays an important role in shaping consumer behavior in the Chinese context. This study proves the significant influence of face awareness on consumers’ willingness to use it through empirical research, which has contributed to the marketing literature in China.

6.2. Practical Implications

The emergence of autonomous driving technology is a revolution in the automotive industry. Therefore, it is important for industry managers to understand the factors that drive consumer acceptance of this technology and products so that they can develop appropriate products to meet consumer expectations and needs. Developing appropriate products without understanding these influencing factors may lead to the decline of the entire product or the industry as a whole [53]. In other words, in the automotive industry, users’ willingness to use a new technology may depend on the consumer’s comprehensive perception and thorough understanding of the new technology. The emergence of autonomous driving technology is a revolution in the automotive industry; therefore, it is important for industry managers to understand the factors that drive consumer acceptance of this technology and products so that they can develop appropriate products to meet consumer expectations and needs. Developing appropriate products without understanding these influencing factors may lead to the decline of the entire product or the industry as a whole. In other words, in the automotive industry, users’ willingness to use a new technology may depend on the consumer’s comprehensive perception and thorough understanding of the new technology. Therefore, by expanding on the influencing factors behind consumers’ willingness to use SAVs, this study provides recommendations for automotive developers as well as marketing managers to increase consumers’ willingness to use SAVs.

At present, countries around the world strongly support the research and development of automated driving technology and the layout of an automated driving development strategy, which is in a critical phase of development. And a higher willingness to use is the key to new products entering the market. Increasing consumers’ trust in autonomous driving technology will help guide the commercialization of the technology in the automotive market and accelerate its development. Eight states in the U.S. have deployed pilot programs for SAVs [14]. Understanding the underlying mechanisms behind the acceptance and use of SAVs is particularly important for improving service performance indicators, efficiency, and technological productivity when implementing pilot programs, which is one of the practical implications of this study.

6.3. Limitations and Research Recommendations

Due to the following deficiencies in this study: (1) Respondents have not experienced the automatic driving system and have little understanding of SAVs. (2) Secondly, the questionnaire designed in this paper only allowed respondents to understand and perceive SAVs through words, pictures, and oral descriptions. Respondent’s reflections are not realistic enough. (3) Finally, the sample size was limited.

These limitations may have caused some bias in the questionnaire results. If future research can recruit a large number of respondents to experience real SAVs, conduct a simulation test, and then interview the respondents’ feelings, more realistic and targeted suggestions can be put forward for the development of SAVs.

Also, this study used quantitative research methods. In the future, qualitative research methods and a combination of qualitative and quantitative research can be used for in-depth research to verify the results of this study.

Secondly, under the current research situation, SAVs have not been popularized, and the research object is only the potential users of SAVs. However, with the development of technology, SAVs will gradually enter people’s lives, and the factors affecting their willingness to use them will also change, which will require further research in the future.