H4: Risk tolerance in social culture moderates the strength of the relationship between technological ethical disputes and legal lag.
This study collects data using a questionnaire survey, which is aimed at the general public who have some knowledge of autonomous driving technology, and aimed to obtain data on the public’s attitudes toward the perception, trust, and legal adaptability of autonomous driving technology. The questionnaire design covers multiple dimensions, and strives to fully understand the public’s cognition of all aspects of autonomous driving technology, involving the advantages and risks of technology, the degree of perfection of laws and regulations, attitudes towards the popularization of technology, and concerns about ethical and safety issues.
This study takes China’s first-tier cities as a case study to deeply explore the differences in public perception of the adaptability of autonomous driving regulations. It combines the regional pace of technological advancement and regulatory response strategies to reveal the interactive relationship between regional legal evolution paths and social acceptance, providing empirical support and theoretical basis for international benchmark comparisons.
The participants in this study were adults aged 18 and above residing in various regions across China. A purposive sampling strategy was employed to ensure that respondents had basic awareness or exposure to autonomous driving technology. The questionnaire used the sample service of Wenjuanxing and was distributed through major online platforms in China to maximize geographic and demographic diversity. A total of 1061 valid responses were collected. While the survey was not restricted to a specific province, efforts were made to ensure regional balance in the sample.
In the regions covered by this study, policies and regulations related to autonomous vehicles are still in the initial exploration stage, the regulatory system is not yet sound, there is a lack of unified standards, and the regulatory mechanism is relatively backward. It is urgent to improve the legal norms in terms of technology access, road testing and liability identification to meet the needs of technological development.
The questionnaire survey of this study covers typical cities in North China, East China, Southwest China, and Southern China, including Shanghai, Nanjing, Hangzhou, Beijing, Chongqing, and Shenzhen. These cities were selected because they are highly representative in the research and development of autonomous driving technology and policy pilots. At the same time, they have diverse population structures and complex urban traffic conditions, which helps to fully reflect the public’s cognition and attitude towards autonomous vehicles.
The data was collected through an online questionnaire survey, which is published through various social platforms, online forums, and online channels of cooperative units. The questionnaire design fully considers the social impact of autonomous driving technology and its adaptability to laws and regulations, ensuring that the items cover the public’s trust, cognition, ethical risks, safety, technical functions, and laws and regulations on autonomous driving vehicles. The questionnaire is designed by autonomous driving technology experts and legal scholars to ensure the rigor and scientificity of its content. The respondents clearly inform their consent before filling out the questionnaire and confirm that their participation is voluntary. The questionnaire content involves the public’s personal information (such as gender, age, travel mode, etc.), understanding and attitude towards autonomous driving technology (such as views on the future development trend of autonomous driving, the advantages of technology, and ethical risks), and cognition of existing laws and regulations (such as the coverage of ethical issues by the law, the degree of perfection of autonomous driving regulations, etc.), as described in Table 1:
Table 1 indicates that 59.66% are male and 40.34% are female. The sample is mainly in the 18-30 age group, accounting for 47.5%. In terms of travel mode, 50.42% of the respondents say that they mainly drive; 35.25% use public transportation; the proportion of walking and non-motorized vehicles is relatively low, at 5.75% and 7.35%, respectively. This data reflects the potential differences in the cognition of autonomous driving technology among different groups, and provides a basis for analyzing the technology acceptance and legal adaptability.
The questionnaire is widely disseminated on major online platforms through Wenjuanxing, which is a popular online survey tool in China, and the exposure of the questionnaire is increased through Internet advertising and email push. After each participant completes the questionnaire, the system automatically records the time of answering the questionnaire and its source information (such as IP address, device type, browser, etc.). All participants must confirm that they have read and agreed to the informed consent form before they can continue to fill in the questionnaire. All questionnaires are equipped with multiple verification mechanisms to ensure the quality and accuracy of the data. Furthermore, if the participants do not complete the questionnaire or make obvious mistakes in the filling process (such as filling in too quickly or repeatedly selecting an option), their answers are excluded from the analysis to ensure the data validity.
The questionnaire survey is limited to the Wenjuanxing platform because they have high user activity and strong communication efficiency in the target area. However, the representativeness of inactive users is limited, and offline supplementary surveys will be considered in the future to increase coverage.
After the data collection is completed, the research team conducts strict data cleaning on the returned questionnaires. Firstly, invalid questionnaires (such as questionnaires with no answers or too short answer time) are eliminated. Secondly, possible extreme values or abnormal data are identified and processed to ensure the validity and representativeness of the data sample. After the data cleaning is completed, the effective sample size of the sample is 1061, covering a wider range of public groups.
The questionnaire after the preliminary design is reviewed by experts and verified by a small range of trial distribution to ensure the validity of the measurement structure and items of the questionnaire. Before the formal survey, all items are subjected to reliability analysis, and the Cronbach’s α value reaches above 0.85, indicating that the questionnaire has a high internal consistency and can accurately reflect the attitudes and cognition of the respondents. The validity of the data is also guaranteed. The questionnaire questions are all expressed in clear and concise terms, avoiding leading and ambiguous questions to ensure that the respondents’ answers can truly reflect their attitudes and cognition. In addition, the respondents’ answers are repeatedly verified during the survey to eliminate possible answer bias.
In this study, descriptive statistical analysis is utilized to show the basic situation of the sample and provide necessary background information for subsequent hypothesis testing. After data collection, all questionnaire data are preliminarily cleaned and sorted to eliminate invalid data and missing values to ensure analysis results’ accuracy and validity.
In the descriptive statistical analysis part, the frequency distribution of all variables is calculated to determine the frequency and proportion of each variable. Group statistics and comparative analysis are implemented for cognitive differences among different groups (such as gender, age group, travel mode, etc.). A further research is conducted on the respondents’ attitudes towards trust, ethics, and safety concerns about autonomous driving technology, and their support for the promotion of autonomous driving technology and related views on policy recommendations are evaluated.
Sample subtotal for the question: Do you know anything about self-driving vehicles? The distribution of questions is shown in Fig. 1.
Sample subtotal for the question: I am worried about a technical failure of the self-driving system. The distribution of questions is shown in Fig. 2.
Sample subtotal for the question: Do I think there are ethical issues with self-driving vehicles? The distribution of questions is shown in Fig. 3.
Sample proportion for the question: Do you know anything about self-driving vehicles? The distribution of questions is shown in Fig. 4.
Sample proportion for the question: I am worried about a technical failure of the self-driving system. The distribution of questions is shown in Fig. 5.
Sample proportion for the question: Do I think there are ethical issues with self-driving vehicles? The distribution of questions is shown in Fig. 6.
The data in Fig. 1 shows that there are significant differences in the respondents’ understanding of autonomous driving technology. Most respondents only have a basic level of cognition and a limited degree of in-depth understanding. In terms of trust, the public expresses obvious concerns about the technical failures that autonomous driving systems may encounter, which reflects the public’s concern about the reliability and safety of the technology. Regarding ethical issues, most respondents are negative or indifferent to the ethical dilemmas that autonomous driving may cause, indicating that the public has a low level of concern about technological ethical issues. Overall, there are certain differences in the public’s cognition of autonomous driving technology, and there are still major concerns about trust and ethical issues, which may pose certain obstacles to the widespread acceptance and promotion of the technology.
When analyzing the cognition and trust in autonomous driving technology, the respondents’ views on the popularization trend of autonomous driving, technical functions, ethical safety issues, and social impact are focused on.
Sample statistical chart of the question: I think autonomous driving affects drivers’ employment and reduce the employment rate. The statistics are shown in Fig. 7.
Sample statistical chart of the question: I think autonomous driving reduces road traffic accidents. The statistics are shown in Fig. 8.
There are some differences in the respondents’ views on the social impact of autonomous driving technology. Regarding employment, 79.18% of the respondents believe that autonomous driving affects the employment of drivers, resulting in a decline in the employment rate. In terms of traffic safety, 89.16% of the respondents agree that autonomous driving helps reduce road traffic accidents, and a high proportion of respondents believe that it brings safety benefits.
Through the results of descriptive statistics, it is possible to identify significant differences in the perception and acceptance of autonomous driving technology among different groups. Overall, the descriptive statistical analysis of this study not only helps to understand the basic public perception of autonomous driving technology, but also lays the foundation for further exploring the relationship between public perception and legal framework adaptability.
In this study, SEM is utilized to test the hypotheses of H1-H3, focusing on the relationship between public cognition, legal system perfection, and social acceptance, and how the government policy communication strength affects the willingness to apply technology through technology cognition and legal cognition. SEM is a multivariate statistical analysis method utilized to analyze complex relationships between variables (Amini and Alimohammadlou, 2021; Igolkina and Meshcheryakov, 2020). It can handle multiple dependencies at the same time and reveal the direct and indirect effects between different variables (Herwin et al., 2022; Mulang, 2021). In this paper, SEM is used to test the chain mediation effect and moderating effect in hypothesis H1-H3. AMOS software was used for SEM analysis.
The H1 hypothesis examines how the public’s confidence in autonomous driving technology, the coverage of the existing legal system, and moral safety affect social acceptance. It also examines the beneficial moderating effect of the degree of perfection of the legal system. In the model construction, the public’s trust in autonomous driving technology and the degree of perfection of the legal system are regarded as independent variables; social acceptance is the dependent variable; the degree of perfection of the legal system is used as a moderating variable to test the independent and interactive effects of trust and legal system perfection on social acceptance. Through structural equation model analysis, the relationship between each variable and the dependent variable is evaluated, and the moderating role of the legal system in the relationship between the two is explored.
The H2 hypothesis mainly examines how the public’s cognition of the risk type of autonomous driving affects the choice of responsibility preference through risk attribution tendency. H2 requires the establishment of a path model covering risk type perception, attribution tendency, and responsibility allocation preference. Specifically, taking risk type perception as an independent variable, it affects the public’s attribution tendency, and then affects the public’s responsibility allocation preference. The mediating role of attribution tendency is tested through path analysis. In the structural equation model, the path coefficient significance test reveals how perceived risk affects the sense of responsibility through the individual’s attribution mechanism, thereby affecting the public’s attitude towards responsibility allocation. This analysis reveals how differences in risk perception lead to public awareness of the legal liability of autonomous driving technology.
H3 hypothesis explores how the effectiveness of government policy communication affects the public’s willingness to apply autonomous driving technology on a large scale through the chain effect of the level of technical cognition and the clarity of legal cognition. In the SEM of this hypothesis, the intensity of government policy communication is used as the independent variable; technical cognition and legal cognition clarity are used as mediating variables; the willingness to implement on a large scale is used as the dependent variable. Path analysis is used to test the indirect effect of policy communication on technical cognition and legal cognition, and to measure its impact on the willingness to use technology. The test of the chain mediation effect reveals the key role of government policies in promoting technology acceptance and promoting the widespread application of technology, especially by improving the public’s technical and legal cognition, thereby indirectly enhancing the public’s willingness to apply technology.
In the process of constructing the SEM model, the normality test and multicollinearity test of the data are the prerequisites for ensuring the validity of the model estimation results. For the fit of the model, multiple fit indicators are used for evaluation. The test results of goodness of fit help evaluate the degree of match between the theoretical model and the actual data, and adjust the model.
In the process of data processing, the collected sample data is first cleaned to remove missing values and outliers to ensure data quality. Next, the maximum likelihood estimation (MLE) method is used for parameter estimation to obtain the model path coefficient and its standard error (Huang, 2022; Yin et al., 2021). The significance of the model path coefficient is tested by t-value and p-value to determine whether each hypothesized path is established.
The test of the moderating effect is carried out by establishing a model containing interaction terms. The significance of the interaction term reveals the strength of the moderating effect. In particular, the moderating effect of the perfection of the legal framework on the relationship between public trust and social acceptance, the chain mediation effect of the government policy communication intensity on public technology cognition and legal cognition, etc., are analyzed by path analysis of the interaction effect and path test of the mediation effect.
Hierarchical regression analysis is utilized to explore the moderating effect of risk tolerance in social culture on the relationship between technological ethical disputes and legal lag. The H4 hypothesis holds that risk tolerance plays a moderating role in this relationship. Analyzing the boundary conditions of this moderating effect helps to deeply understand the impact of social and cultural factors on the promotion of autonomous driving technology. Hierarchical regression analysis conducts regression analysis by gradually applying independent variables and interaction terms. This method can effectively evaluate the moderating effect between variables (Dodanwala et al., 2021; Fadare et al., 2022). The analysis process mainly involves four steps: the gradual construction of the model, the application of interaction terms, the testing of the moderating effect, and the assessment of the goodness of fit of the model. All graphs and tables in this study were drawn using STATA version 17.0, using statistical techniques such as structural equation modeling (SEM) and hierarchical regression analysis to ensure the scientificity and reproducibility of the results.
Before conducting hierarchical regression analysis, it is necessary to construct a basic model and an extended model containing moderating effects. The basic model includes the predictive relationship between technological ethical disputes (independent variables) and legal lag (control variables) on social acceptance. On this basis, the extended model applies the main effect of risk tolerance and the interaction term between technological ethics disputes and legal lag to test the moderating effect of risk tolerance.
First, regression analysis is performed on technological ethics disputes and legal lag to explore their independent effects on social acceptance. At this time, risk tolerance is not considered, aiming to identify the direct impact of technological ethics disputes and legal lag on social acceptance. On the basis of the basic model, risk tolerance is added as an independent variable, and the interaction term between technological ethics disputes and legal lag is added.
The test of interaction effect is the core of hierarchical regression analysis. In this study, the step of applying interaction terms is crucial. The interaction term reflects the influence of risk tolerance on the relationship between technological ethics disputes and legal lag. In the extended model, a new regression equation is obtained by calculating the product of the interaction term and risk tolerance. By analyzing the regression coefficient of the interaction term, it is revealed how risk tolerance adjusts the impact of technological ethical disputes and legal lag on social acceptance at different levels.
