Public attitude towards autonomous vehicles before and after crashes: A detailed analysis based on the demographic characteristics

Abstract

Autonomous vehicles (AVs) have the potential to offer a large number of benefits such as reducing the energy consumed and reducing the anxiety of the drivers. On the other side, the degree to which AVs will be adopted mainly depends on the public attitude and acceptance of this emerging technology. Over the last few years, AVs got involved in multiple accidents with different levels of severity. These accidents were widely covered in the media, creating a debate about the safety of this technology and discouraging people from adopting this new technology even if it offers a safer environment. In this study, a questionnaire survey was conducted to understand the impact of accidents involving AVs on the public perception of this technology for respondents with different demographic characteristics (age, gender, education, income, and prior knowledge about AVs). The results show the most negative shift in the attitude occurs for respondents who are older, female, and have no prior knowledge about AVs or their incidents. Additionally, the results shed light on the importance of educating the public about AVs in order to guarantee the highest level of acceptance. Finally, the findings of this paper can help AVs developer, policymakers, and transport planning agencies in understating the public attitude after accidents in order to react properly to avoid discouraging people from adopting AVs.

Keywords:

• Accidents
• autonomous vehicles
• public acceptance
• safety
• self-driving cars

PUBLIC INTEREST STATEMENT

Over the last few years, the public attitude towards AVs varied significantly from country to country and from year to year. While this variability has been observed, it has never been translated before. This paper shows that accidents involving AVs is one of the major factors that negatively influence the public attitude towards AVs. Thus, a questionnaire survey was conducted in order to understand the impact of these accidents on the public attitude towards AVs for different respondents with different demographic characteristics. The results show that people become more pessimistic towards AVs after accidents. However, respondents with a strong background about AVs showed a minor negative shift in their opinion of AVs. Indicating that people with stronger backgrounds about AVs are the least likely to change their opinion after AVs’ accidents.

1. Introduction and background

The idea of the automation of vehicles has started years ago. One of the early vehicle automation features, in the 1920s, that attracted research was the establishment of the vehicle-to-vehicle communication using radio waves (The Milwaukee Sentinel, 1926). This was followed by the research that focused on the invention of the electromagnetic guidance system in the 40s (The Victoria Advocate, 1957). Then, the University of Bundeswehr cooperated with Mercedes-Benz and this cooperation resulted in the invention of the first system for self-driving vehicles. (Davidson & Spinoulas, 2015), which was one of the main milestones that attracted huge investments from different disciplines with the objective of developing autonomous vehicles (AVs; Berrada & Leurent, 2017). In general, there are six levels of automation as per the National Highway and Transportation Safety Administration (from level zero to level five). Level zero refers to human-driving vehicles where all functions must be done by the human driver, while level one refers to vehicles with some driver assistance features. Level two refers to partial automation with the human driver control as the driver will be responsible for all driving tasks, while level three refers to conditional automation where the vehicle can travel autonomously in specific conditions and the driver should be ready to take control back at any point. Level four refers to vehicles with high automation where the vehicle can travel autonomously but the driver can take control back at any point, while level five refers to vehicles with full automation features that allow the vehicle to operate safely under all conditions (Greenblatt & Shaheen, 2015). In general, both researchers and manufacturers are putting tremendous effort into the development of AVs at a multidisciplinary level (Mallozzi et al., 2019). For example, Google, Uber, Apple, and Volvo have launched their separate projects for the development of AVs (Hartmans, 2016; O’Kane, 2019; Staff, 2019).

Over the last few years, a large number of studies investigated different aspects of AVs and their implications. In general, AVs have the potential to offer a large number of benefits such as reducing the energy consumption and emissions, while increasing the level of mobility, level of comfort, and productivity (Othman, 2022a; Bansal et al., 2016; Othman, 2021a; Antov et al., 2012; P. Wang et al., 2019; Massar et al., 2021). On the other side, AVs might pose risks such as increasing the level of congestion as a result of the empty trips that result in an increase in the vehicles kilometers traveled (VKT) (Othman, 2022b). While large attention has been paid to studying the implications of AVs, less attention has been paid to understanding the main factors that affect the public attitude towards AVs. However, it is generally known that the public attitude is a critical factor for the success of any new technology such as AVs and in many cases the non-technological issues represent a major barrier to the success of new technologies. For example, the IEEE report that investigates the future of AVs has shown that the main risk for the wide deployment of AVs will be the public attitude, not the technological realization (Newcomb, 2012).

In general, multiple studies have investigated the public attitude towards AVs such as the studies by Othman (2021b), Jing et al. (2020), Hilgarter and Granig (2020), S. Wang et al. (2020), Butler et al. (2021), Ahmed et al. (2021), and Ullah et al. (2019). These studies mainly focus on studying the public attitude towards AVs in general for respondents with different demographics such as the age, gender, income, and educational level. On the other side, these studies have shown different levels of perception of AVs technology across countries (Moody et al., 2020) and years without investigating the main factors that cause these variabilities. Thus, it can be stated that while the public attitude has been investigated in the literature, the factors affecting the public attitude towards AVs have been rarely discussed. One of the early studies that revealed this gap is the study by Šinko et al. (2017) which replicated the survey completed by Schoettle and Sivak (2014) that was conducted in 2014. The results of these two surveys show that although people became more familiar with AVs, they became less optimistic towards this technology as the level of interest in AVs dropped from 40% in 2014 to 10% in 2017. Additionally, a similar trend can be observed in the study by Panagiotopoulos and Dimitrakopoulos (2018) which showed that people became more pessimistic towards AVs over time. Although this negative shift was observed in the two studies, it was not explained until 2020 in the study by Abdelgawad and Othman (2020). Abdelgawad and Othman (2020) have shown that this negative shift in the public attitude towards AVs in the two studies by Šinko et al. (2017), and Panagiotopoulos and Dimitrakopoulos (2018) occurred because of the negative news about AVs that accompanied these two surveys as these two surveys were conducted during or shortly after the first fatal accident involved an AV in 2016. Thus, these results clearly show the negative impacts of accidents involving AVs on the public acceptance and perception of this emerging technology. Additionally, one of the studies that investigated the relationship between the public attitude towards AVs and AVs’ accidents is the study by Othman (2021b) that draw the relation between the number of fatal accidents of AVs per year and the percentage of respondents afraid of AVs on the same years for the data collected between 2016 to 2019. The results of Othman (2021b) study are shown in Figure 1 and it can be noted that there is a clear and direct relationship between the number of fatal accidents involving AVs and the percentage of people afraid of the technology. These results necessitate the need for a detailed and clear understanding of the impact of AVs’ accidents on the public attitude to prepare for the future.

Figure 1. The relation between the number of fatal accidents involving AVs and the percentage of people afraid of AVs in the USA [adopted from (Othman (2021b)].

The studies that investigate the public attitude towards AVs after accidents are rare and these studies utilize social media data in order to conduct sentiment analysis for social media users before and after AVs’ accidents. One of the early studies that utilized sentiment analysis of social media to understand this impact is the study by Jefferson and McDonald (2019) which utilized Twitter data to understand the perception of the public towards AVs before and after the accident (for a time span of three days) that involved an AV on the 10^th of February 2019. The results of the sentiment analysis have shown a major shift in the public attitude as a result of this accident as the word “crash” became one of the most frequently used words used in the tweets after the accident, while it did not show up in the sentiment analysis before the accident. Additionally, there was a shift in the most frequent words used before and after the accident. Before the accident, the most frequent words focused on investments and startups, while the most frequent words used after the accident were “vehicle”, “crash”, and “autopilot”. Finally, the sentiment analysis shows a major drop in the positive tweets by 50%. These results indicate that this accident decreased the level of trust in AVs that most of the users who had positive opinions towards AVs prior to the accidents became more pessimistic towards this technology and their opinions were shifted in the negative direction. Similarly, the study by Penmetsa et al. (2021) investigated the public attitude towards AVs before and after two fatal accidents that involved AVs in March 2018 using sentiment analysis of Twitter data. The first accident was a pedestrian-fatal accident that involved a Volvo AV that was operated by Uber in Arizona on the 18^th of March 2018 (Levin, 2018; Levin & Carrie, 2018). The second accident involved a Tesla Model X vehicle on the 23^rd of March 2018. This accident was fatal, and the autopilot mode was on when the vehicle speeded up and steered to hit the road barrier (Levin & Woolf, 2016; Guardian staff, 2018). The sentiment analysis shows major drops in the positive attitudes towards AVs as the analysis showed 32% increase in the negative tweets moving from 14% before the accident to 46% after the accident. Additionally, the analysis shows 6% decrease in the compound sentiment score for tweets involving Tesla and Uber, while this percentage reached 11% for tweets involving self-driving cars. Thus, these results clearly show the impact of the accidents on the level of trust and public perception of AVs technology. Finally, the most recent study that explored the relation between accidents involving AVs and the public attitude is the study by Jing et al. (2022) that conducted a sentiment analysis for social media data before and after the NIO crash in August 2021 in China. In this study, the sentiment analysis was conducted on the data collected from Sina Weibo and Tik Tok, which are the main social media platforms used in China (CaiLianPress, 2021; Sina, 2020). The results show 18% increase in the negative sentiment after the accident (jumping from 29% before the accident to 47% after the accident). On the other side, the results show 10% decrease in the positive attitude (moving from 40% before the accident to 30% after the accident). One of the main outputs of this study is that the public perception of the different brands, that offer AVs, is negatively affected by any accident regardless of the brand of the AV that was involved in the accident. Although the analysis was conducted for an accident that involved a NIO car, the results show that the positive sentiment score for Tesla decreased by 9% after the accident moving from 32% before the accident to 23% after the accident.

As shown from the previous discussion, all the previous studies that investigate the relation between AVs’ accidents and the public attitude utilize social media data to conduct sentiment analysis of the available data. In general, the main benefit of utilizing social media data is that social media platforms are data-rich and offer a large amount of data. Thus, using this data for understanding the impacts of safety related factors of AVs on the public attitude can provide a lot of insights about this relation. On the other side, this analysis is subjected to multiple limitations that prevent it from providing a detailed analysis. For example, this analysis is general, and it cannot capture the impacts of these accidents on people with different demographic characteristics. Additionally, the results of the sentiment analysis for social media data might be biased as the characteristics of the sample collected from social media might deviate from the characteristics of the main population resulting in biased results. For example, Twitter data have shown that every one of three users in the USA earn more than 75.000 USD a year and have a college degree. Comparing these numbers with the overall population in the USA shows that Twitter users earn more and have more education than the average population in the USA (Alambeigi et al., 2021). From the gender perspective, the latest report published by Twitter shows that 62% of the user are males (Luo & He, 2021), which indicates gender unbalance in the samples taken from it. A similar trend can be observed for the most common social media platforms in China (Sina Weibo and Tik Tok) as the most recent reports show that 80% of the users have an age of 30–40 years (CaiLianPress, 2021). Thus, it can be concluded that social media samples might be biased towards a specific group, which necessitates the need for studies that are based on data that are representative of the general populations in order to offer generalized results that offer the developers of AVs and transportation planners insights about the impact of safety-related factors of AVs on the public attitude.

Similarly, previous studies that utilized sentiment analysis for analysing the public attitude before and after accidents have reported multiple deficiencies in the applications of this technique for AVs’ crash domain (Jefferson & McDonald, 2019; Penmetsa et al., 2021). Firstly, the analysis shed light on the mismatch between the score assigned to the different words in the dictionary and the score of the word in the Av domain as highlighted in the study by Penmetsa et al. (2021). This mismatch necessitates the need for the development of a separate dictionary that can be fairly used in the AV domain. Secondly, the analysis shows that the users of social media platforms use the concept of blaming the vehicle through the use of the words such as “blame” and “confused”, which suggests that Twitter users may be drawn to discussions of crashes involving compelling and concise narratives. The use of these new terms necessitates the need for a more advanced search method to avoid the limitations on the keywords search that occurs while finding the data on the social media platform. Additionally, previous studies highlighted that most of the tweets include a website link from a newspaper (McDonald et al., 2021), which makes it hard to get true sentiment scores for these tweets. Thus, these limitations necessitate the need for further research in order to appropriately understand the public perception of AVs before and after accidents. As a result, this study offers a detailed analysis of the public attitude towards AVs before and after accidents for a representative sample (the characteristics of the sample are similar to the characteristics of the population). Additionally, the analysis was conducted for the different groups with different demographic characteristics separately. In general, the main objectives of this study are:

• Quantify the shift in the levels of public trust, interest, and concern regarding AVs as a result of AVs accidents for respondents with different demographic characteristics (age, gender, income, education).

• Understand the impact of the level of awareness of AVs and previous accidents that involved AVs on the public perception of AVs technology before and after accidents.

The remained of the paper is organized as follows: section 2 offers the gaps in the previous studies and the main objectives of the paper, while section 3 summarizes the methodology followed in this study and how the survey was conducted. Section 4 summarizes the main results of the study and the public attitude towards AVs for the respondents with different demographic characteristics before and after the accidents. Finally, section 5 shows the main conclusions and results of this study.

2. Research methodology

In this study, a questionnaire survey was conducted with the objective of understanding the impacts of accidents involving AVs on the public perception of this technology in terms of the level of trust, interest, and concern towards AVs. Firstly, a pilot survey was designed and sent to five professors (from multiple universities across Canada and the USA), who have wide experience with public surveys, and 20 individuals from the general public in order to make sure that the survey is accurate, descriptive, and understandable. The results of the pilot survey were promising and showed high level of satisfaction of the survey from both the professors and the individuals from the public as all of them highlighted that they had no problem reading, underdoing, or responding to the different questions. The survey consisted of four main sections. Section one was a background section that provide the respondents with information about the survey and some background information about AVs technology. Section two focuses on understanding the public opinion and perception of AVs and in this section the respondents were asked to rank their level of interest, trust, and concern towards AVs technology on a Likert scale from 1 to 5. Then, section three focused on understanding the public attitude towards AVs after accidents by allowing the respondents to rank the same questions in section two but after introducing some accidents that involved AVs. Thus, at the beginning of section three, nine accidents that involved AVs were introduced to the respondents (before introducing the same questions asked in section two). These accidents had different levels of severity, from minor damage to fatal accidents, and show different malfunctions that might occur in AVs, including steering issues, spending issues, visualization issues, and battery explosions, as summarized in Figure 2. The main idea of introducing these accidents is to give the respondents information about the different AVs’ malfunctions and possible issues. Finally, section four focuses on collecting personal information of the respondents such as the age, gender, income, and educational level. For detailed information about the survey structure, please see a sample survey sheet in the appendix.

Figure 2. Details about the accidents introduced to the respondents during the survey (red text for fatal accidents and Orange text for minor injury accident.

The survey was then designed and published on SurveyMonkey platform and a total of 5880 responses from the USA were collected between February and June of 2022. The personal information of the respondents are summarized in Table 1 and the table shows that the survey sample is representative of the entire population as the representations of the different groups in the survey sample are consistent with the demographic characteristics of the entire population in the USA as per the latest United States Census Bureau (2022). The gender ratio was balanced as 48% of the respondents were males and the remaining 52% were females. These results comply with the most recent demographics in the USA according to the United States Census Bureau (2022) that shows that 49% of the USA population are males and 51% are females. Similarly, the different age groups are well represented in the survey sample as the different groups are almost equally represented in the sample which is representative of the USA population (United States Census Bureau, 2022). For the educational levels of three respondents, the analysis shows that the vast majority of the respondents have bachelor’s degrees or lower, while only 22% of the respondents have post-graduate degrees (master’s degrees or higher). Although the representation of the different groups with different educational levels seems unbalanced, the results are consistent with the educational attainment data published by the United States Census Bureau in February 2022 (United States Census Bureau, 2022). The survey sample shows that 32.8%, 45%, and 16% of the respondents have high school degrees or lower, bachelor’s degrees, and master’s degrees; while 36.8%, 48.9%, and 14.4% of the USA population have high school degree or lower, bachelor’s degrees, and master’s degrees as per the United States Census Bureau (2022). For the yearly household income, similar trends can be observed when the sample data are compared with the Census Bureau report (Shrider et al., 2021). The survey shows that 20.48%, 21.04%, 35.15%, and 20.17% of the respondents live in a household that has an income of 25 K$, 25–50 K$, 50–100 K$, and 100–200 K$, compared to 18.1%, 19.7%, 28.7%, and 23.3% of the USA population as per the Census Bureau report.

Table 1. Demographic information of the respondents participated in the survey

Demographic characteristics		Frequency (N)	Percentage (%)
Gender	Male	2805	47.70%
	Female	3075	52.30%
Age	18–29	1315	22.36%
	30–44	1814	30.85%
	45–60	1345	22.87%
	>60	1406	23.91%
Education	High School or lower	1930	32.82%
	Bachelor	2648	45.03%
	Master’s degree	943	16.04%
	Higher than master’s degree	359	6.11%
Household Income ($/year)	<25,000	1204	20.48%
	25,000–50,000	1237	21.04%
	50,000–100,000	2067	35.15%
	100,000–200,000	1186	20.17%
	>200,000	186	3.16%

While the previous discussion mainly focused on understanding the demographic characteristics of the respondents and the representation of the survey sample, it is commonly known that prior knowledge about the technology has a significant impact on the public attitude towards this technology. Thus, during the survey the respondents were asked whether they had prior knowledge about AVs technology or not in order to understand the impact of this knowledge on the perception of AVs technology before and after accidents. Additionally, a separate analysis will be conducted based on the respondents’ prior knowledge about accidents involving AVs in order to understand the impact of this knowledge on the perception of this technology before and after accidents. A summary of the respondents’ prior knowledge about AVs and AVs’ accidents is summarized in Table 2. The results show that 91.6% of the respondents have heard, seen or read about AVs before this survey, while only 8.4% have no prior knowledge about AVs at all. Additionally, out of these 91.5% of the respondents who have prior knowledge about AVs, only 24% have a strong background about the technology, while the remaining 67.6% know little about AV technology and its features. On the other side, 60.5% of the respondents highlights that they are aware of accidents that involved AVs, while 39.5% of the respondents had no previous knowledge about AVs’ accidents.

Table 2. Summary of the respondents’ previous knowledge about AVs and accidents of AVs

Question		Frequency (N)	Percentage (%)
Have you ever seen, heard, or read anything about self-driving cars before participating in this survey?	Yes. A lot	1410	23.98%
	Yes, a bit	3975	67.60%
	No	495	8.42%
Are you aware of any accident that involved a self-driving car?	Yes	3555	60.46%
	No	2325	39.54%

3. Results

This study focuses on understanding how the public attitude towards AVs after accidents for respondents with different demographic characteristics. Thus, a questionnaire survey was designed and a total of 5880 complete responses were collected from respondents from the USA. The average levels of inset, trust, and concern about AVs were calculated and compared in the next subsections for the different groups that have different demographic characteristics. Additionally, hypothesis testing was adopted in order to test whether the changes in the public attitude were significantly different after the accident compared to before the accidents for the respondents with different demographic characteristics. In this case, the t-test was performed, and the null and alternative hypotheses were:

$H_0:{\mu }_{\mathit{before}}={\mu }_{\mathit{after}}\left(\mathit{the}\mathit{two}\mathit{means}\mathit{are}\mathit{equal}\right)$

$H_1:{\mu }_{\mathit{before}}\ne {\mu }_{\mathit{after}}\left(\mathit{the}\mathit{two}\mathit{means}\mathit{are}\mathit{not}\mathit{equal}\right)$

Additionally, a 95% level of confidence was adopted in order to judge on this hypothesis. In this case, the p-value for every case was calculated and compared with (0.05). If the p-value was higher than 0.05, the null hypothesis cannot be rejected indicating that the two means are equal. In other words, there is no significant difference in the public attitude before and after the accidents. Otherwise, if the p-value was lower than 0.05, we can say that there is a significant difference in the public attitude towards AVs before and after the accidents. In the next subsections, a detailed analysis and discussion about the effects of AVs’ accidents on the public attitude will be presented.

3.1. Impact of accidents involving AVs on the general survey sample (all respondents)

This subsection mainly focuses on quantifying the shift in the public attitude towards AVs after accidents for all the respondents who participated in the survey. Thus, the mean levels of interest, trust and concern towards AVs were calculated before and after introducing the accidents besides with the hypothesis testing in order to understand whether the changes in the means are statistically significant or not. In other words, the hypothesis testing will tell whether there is a significant shift in the public attitude towards AVs after the accidents. Table 3 summarizes the results of analysing the overall responses of the respondents before and after the accidents. The table shows the mean levels of trust, interest, and concern in AVs before and after the accidents besides with the percentages of change in the means after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant shift in the public attitude towards AVs after the accidents for the three factors tested as the p-values are lower than 0.05 (so the null hypothesis can be rejected, and the alternative hypothesis can be accepted). Additionally, this negative shift in the public attitude is major as both the average levels of interest and trust in AVs have witnessed 9.4 and 9.6% decrease after the accidents moving from 3.19 and 3.22 before the accidents to 2.89 and 2.91 after the accidents. Similarly, the average level of concern has witnessed 6.8% increase after the accidents jumping from an average value of 3.54 before the accidents to 3.78 after the accidents. In general, the results of the general analysis conducted in this subsection are consistent with the results of previous studies that analysed social media data and showed a negative shift in the public attitude after the accidents (Jefferson & McDonald, 2019; Levin, 2018; Levin & Carrie, 2018; Levin & Woolf, 2016; Penmetsa et al., 2021).

Table 3. Average levels of interest, trust and concern for all respondents before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Overall Average		%Change	P-value
	Before	After
Average level of interest for all respondents before and after introducing the accidents (where 5 indicates “very positive opinion” and 1 indicates “very negative opinion”)	3.19	2.89	9.404389	0.00345
Average level of trust for all respondents in AVs before and after introducing the accidents (where 5 indicates “strongly agree” and 1 indicates “strongly disagree”)	3.22	2.91	9.627329	0.00322
Average level of concern for all respondents about AVs before and after introducing the accidents where 5 indicates “highly concerned” and 1 indicates “not concerned at all”)	3.54	3.78	−6.77966	0.00587

3.2. Changes in the public attitude of the different age groups

In general, previous studies analysed the relation between the age of the respondents and the level of interest in AVs. The results of these studies show that younger people are more optimist and interested in AVs as concluded in the studies by Piao et al. (2016), Richardson and Davies (2018), Abraham et al. (2017), Lee et al. (2017), and Park et al. (2021). In this subsection, the public attitude towards AVs for the respondent within different age groups will be investigated in order to understand the changes after the accidents. The results are summarised in Table 4 which shows the average levels of interest, trust, and concern for the different groups before and after the accidents besides with the percentages of change in these levels after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for all the different age groups as all the p-values in the table are lower than 0.05, so the null hypothesis can be rejected. In this case, it can be stated that the shift in the public attitude towards AVs after the accidents is statistically significant for all the age groups. Additionally, the results show that the shift in the public attitude towards AVs varies according to the age of the respondent. Finally, the analysis shows that the respondents with an age of 60 years or more are the most pessimistic towards AVs before and after the accidents as this group had the lowest levels of interest, the lowest level of trust in AVs, and the highest level of concern towards AVs before and after the accidents. Additionally, the results show that the largest shift in the public attitude occurred for the same age group (>60) as the level of interest and trust dropped from 2.75 and 3 before the accidents to 2.35 and 2.53 after the accidents showing 14.55% and 15.67% reduction. Similarly, the respondents with age (>60) have shown the highest levels of concern before and after the accidents with average levels of concern jumping from 3.86 before the accidents to 4.17 after the accidents (showing 8% increase in the level of concern after the accidents with 8%). These results show that the older age group was not only the most pessimistic group towards AVs, but also the group with the largest negative shift in the public attitude towards AVs after the accidents.

Table 4. Average levels of interest, trust, and concern for the different age groups before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Age	Level of interest	Level of trust	Level of concern
Before	18–29	3.42	3.32	3.19
	30–44	3.49	3.42	3.54
	45–60	3.04	3.31	3.52
	>60	2.75	3	3.86
After	18–29	3.28	3.19	3.45
	30–44	3.06	3.1	3.74
	45–60	2.87	3.02	3.73
	>60	2.35	2.53	4.17
% Change	18–29	4.09	3.92	−8.15
	30–44	12.32	9.36	−5.65
	45–60	5.59	8.76	−5.97
	>60	14.55	15.67	−8.03
P-value	18–29	0.002873532	0.002866906	0.00427374
	30–44	2.15159E-06	5.53328E-06	0.004
	45–60	0.017361807	0.000134013	0.035733266
	>60	6.16332E-05	8.71371E-06	0.00367

3.3. Changes in the public attitude of male and female respondents

Previous studies analysed the relation between the gender of the respondents and the level of interest in AVs. These studies showed that male respondents are more positive, optimistic and interested in AVs than female respondents as concluded in the studies by Polydoropoulou et al. (2021), Pigeon et al. (2021), Schoettle and Sivak (2014), Piao et al. (2016), Richardson and Davies (2018), and Abraham et al. (2017). In this subsection, the public attitude towards AVs for the respondent with different genders will be investigated in order to understand the changes after the accidents. The results are summarised in Table 5 which shows the average levels of interest, trust, and concern for male and female respondents before and after the accidents besides with the percentages of change in these levels after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for male and female respondents as all the p-values in the table are lower than 0.05, so the null hypothesis can be rejected. Thus, it can be concluded that there is a significant change in the public attitude towards AVs after the accidents for both male and female respondents. Additionally, comparing the mean levels of interest, trust, and concern before and after the accidents for both male and female respondents shows that male respondents are always more positive towards AVs than female respondents (before and after the accidents). Additionally, the results show that the negative shift in the public attitude is larger for female respondents than male respondents. The mean levels of interest, trust, and concern towards AVs for male response changed from 3.39, 3.47, and 3.33 before the accidents to 3.23, 3.18 and 3.58 after the accidents (showing 4.72%, 8.36%, and −7.51% change). On the other hand, larger shifts in the levels of interest, trust, and concern can be observed for female drivers after the accidents as the levels of interest, trust, and concern changed from 3.03, 3, and 3.75 before the accidents to 2.7, 2.67, and 4.06 after the accidents (showing 10.89%, 11%, and −8.27% change). Thus, it can be concluded that female respondents are not only more pessimistic towards AVs than male respondents, but also have the largest negative shift in their attitude after the accidents.

Table 5. Average levels of interest, trust, and concern for male and female respondents before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Gender	Level of interest	Level of trust	Level of concern
Before	Male	3.39	3.47	3.33
	Female	3.03	3	3.75
After	Male	3.23	3.18	3.58
	Female	2.7	2.67	4.06
% Change	Male	4.72	8.36	−7.51
	Female	10.89	11.00	−8.27
P-value	Male	0.003499939	1.13303E-05	0.000203071
	Female	2.05036E-08	1.24718E-07	4.70151E-07

3.4. Changes in the public attitude of respondents with different household incomes

While previous studies analysed the relation between the level of interest in AVs, and the gender and age of the respondents. Less attention has been paid to understanding the impact of the income levels on the public attitude towards AVs. The studies by Yuen, Chua, et al. (2020) and Yuen, Wong, et al. (2020) are one of the early studies that investigated this relation at a national level and for different countries. In this subsection, the public attitude towards AVs of the respondent at different household income levels will be investigated in order to understand the changes in the attitude of the different groups after the accidents. The results are summarised in Table 6 which shows the average levels of interest, trust, and concern for the different household income groups before and after the accidents besides with the percentages of change in these levels after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for all the different groups except for the level of trust for respondents with household income less than 25 USD thousand and > 200 USD thousands as the p-values for these levels of trust are higher than 0.05 indicating that the null hypothesis cannot be rejected. However, the levels of interest and concern towards AVs for these two income groups show significant changes as the p-values for these factors are lower than 0.05 indicating that the null hypothesis cannot be accepted and that there is a significant change in the levels of interest and concern for these two groups after the accidents. Additionally, the results show that respondents who live in a household that has an income of 25–50 USD thousand are the most pessimistic towards AVs before and after the accidents as this group showed the lowest levels of interest, lowest levels of trust, and the highest levels of concern before and after the accidents. On the other hand, respondents with a household income (>200 USD thousand) are the most optimistic towards AVs as they show the highest level of interest, the highest level of trust, and the lowest level of concern towards AVs before and after the accidents. Additionally, the mean values shown in Table 6 show that the levels of interest and trust in AVs showed some decrease with the increase in the household income until an income of 25–50 USD thousand is reached; then, the level of interest and trust increased with the increase in the household income. These results are consistent with the results of the survey by Lee et al. (2019) in the USA that showed that there is a polynomial relation between the level of interest in AVs and the level of income as the interest in AVs shows some decrease with the increase in the income levels, at low incomes; then, the level of interest increased with the increase in the income levels.

Table 6. Average levels of interest, trust, and concern for respondents with different household income levels before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Income	Level of interest	Level of trust	Level of concern
Before	<25	3.35	3.36	3.4
	25–50	2.87	2.87	3.75
	50–100	3.33	3.36	3.43
	100–200	3.31	3.37	3.42
	>200	4	3.63	2.9
After	<25	3.24	3.28	3.7
	25–50	2.56	2.64	4.04
	50–100	3.05	3.03	3.75
	100–200	3.05	2.94	3.73
	>200	3.81	3.54	3.25
% Change	<25	3.28358209	2.380952381	−8.823529412
	25–50	10.80139373	8.013937282	−7.733333333
	50–100	8.408408408	9.821428571	−9.329446064
	100–200	7.854984894	12.75964392	−9.064327485
	>200	4.75	2.479338843	−12.06896552
P-value	<25	0.046079838	0.069098763	0.0004539
	25–50	0.000909879	0.014245662	0.000283644
	50–100	0.000121758	7.7228E-05	2.01777E-05
	100–200	0.008494981	1.70903E-05	3.28985E-05
	>200	0.04405217	0.06951923	0.000839013

3.5. Changes in the public attitude of respondents with different educational levels

In general, previous studies analysed the relation between the educational levels of the respondents and the level of interest in AVs. The results of these studies show that people become more optimistic towards AVs with the increase in their educational levels as highlighted in the studies by Piao et al. (2016), Rezaei and Caulfield (2020), and Zhang et al. (2022). In this subsection, the public attitude towards AVs for the respondent at different educational levels will be investigated in order to understand the changes in the attitudes after accidents. The results are summarised in Table 7 which shows the average levels of interest, trust, and concern for respondents with different educational levels before and after the accidents besides with the percentages of change in these levels after the accidents and the p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for respondents with different educational levels as all the p-values in the table are lower than 0.05, so the null hypothesis can be rejected. In this case, it can be stated that the shift in the public attitude towards AVs after the accidents is statistically significant for respondents with different educational levels. Additionally, the results show that the shift in the public attitude towards AVs varies according to the educational level of the respondent. Before introducing the accidents, the results show that the respondents with higher educational levels are more positive towards AVs in terms of the level of interest, trust, and concern. On the other side, after introducing the accidents, the results show that the same trend can be observed as the respondent became more positive with the increase in their educational levels except for the respondents with educational levels higher than master’s degrees who showed a major shift in their opinions towards AVs after the accidents that this group became the group with the most pessimistic attitude towards AVs after the accidents, taking the place of respondents with an educational level lower than a bachelor’s degree who had the most pessimistic attitude towards AVs before introducing the accidents.

Table 7. Average levels of interest, trust, and concern for respondents with different household income levels before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Educational Level	Level of interest	Level of trust	Level of concern
Before	Lower than BSc	3.12	3.07	3.6
	BSc	3.31	3.2	3.54
	MSc	3.32	3.37	3.42
	Higher than MSc	3.25	3.12	3.625
After	Lower than BSc	2.75	2.86	3.79
	BSc	3	2.8	3.77
	MSc	3.13	2.93	3.77
	Higher than MSc	2.75	2.9	3.9
% Change	Lower than BSc	11.86	6.84	−5.28
	BSc	9.37	12.50	−6.50
	MSc	5.72	13.06	−10.23
	Higher than MSc	15.38	7.05	−7.59
P-value	Lower than BSc	0.004740762	0.002808327	0.005
	BSc	0.003426811	0.002812883	0.002398782
	MSc	0.001068537	0.000155005	0.002648162
	Higher than MSc	0.005177586	0.034226417	0.017530833

3.6. Impact of AVs’ accidents on respondents with different levels of awareness about AVs

Previous studies show that previous experience or awareness about AVs technology and features is one of the main factors affecting the public attitude towards AVs. In general, respondents with previous experience are more optimistic towards AVs and concluded in multiple studies (Charness et al., 2018; Cunningham et al., 2019; Nordhoff et al., 2019; Piao et al., 2016). Thus, this subsection mainly focuses on understanding the public attitude of the respondents with and without prior knowledge about AVs technology before and after introducing the accidents. The results are summarised in Table 8 which shows the average levels of interest, trust, and concern for the respondents with different levels of awareness before and after the accidents besides with the percentages of change in these levels after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for respondents with no or little background about AVs as all the p-values (for these two groups in the table) are lower than 0.05, so the null hypothesis can be rejected. In this case, it can be stated that the shift in the public attitude towards AVs after the accidents is statistically significant for respondents with no or little knowledge about AVs. On the other side, the results of the hypothesis testing for respondents who have a strong background about AVs show that there is no significant difference in their level of interest and trust in AVs after the accidents as they have p-values higher than 0.05. Thus, it can be concluded that there is no significant change in the level of interest and trust in AVs after accidents for respondents with a strong background about AVs. For the general option towards AVs before introducing the accidents, the results show that the positive opinion towards AVs significantly increase with the increase in the level of awareness as the average positive opinions about AVs were 3.77, 3.03 (19.6% decrease), and 2.85 (24.4% decrease) for respondents with extensive experience about AVs, respondents who have limited information about AVs, and respondents who have never heard of AVs. After introducing the accidents, the positive attitude towards AVs decreased for the three groups but at different levels. The positive opinion towards AVs decreased from 3.77, 3.03, and 3.85 to 3.63 (3.7% decrease), 2.75 (9.2% decrease), and 2.49 (9.5% decrease) for respondents with strong background about AVs, respondents with limited background about AVs, and respondents who did not know about AVs before the survey. Thus, respondents with no background about AVs do not only have the most pessimistic opinion about AVs, but also the group with the biggest decrease in their positive opinion when they knew about the accidents. Thus, educating the public about AVs is a major step in order to improve the public attitude towards AVs.

Table 8. Average levels of interest, trust, and concern for respondents with different levels of knowledge about AVs before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Prior Knowledge about AVs	Level of interest	Level of trust	Level of concern
Before	Yes, a lot	3.76	3.77	3.32
	Yes, a bit	3.03	3.07	3.52
	No	2.84	2.84	3.86
After	Yes, a lot	3.62	3.62	3.57
	Yes, a bit	2.75	2.73	3.72
	No	2.49	2.31	4.18
% Change	Yes, a lot	3.72	3.98	−7.53
	Yes, a bit	9.24	11.07	−5.68
	No	12.32	18.66	−8.29
P-value	Yes, a lot	0.050865155	0.061347156	0.001564159
	Yes, a bit	3.89083E-09	3.62963E-10	0.000654467
	No	0.006691552	0.000173787	0.001719688

Secondly, the results show that the level of trust in AVs increases with the increase in the level of awareness about AVs before and after introducing the accidents. Before introducing the accidents, the levels of trust in AVs were 3.78, 3.07 (18.7% decrease), and 2.85 (24.6% decrease) for respondents with extensive experience about AVs, respondents who have limited information about AVs, and respondents who have never heard of AVs. Then, after introducing the accidents to the respondent, the level of trust in AVs decreased for all the respondents within the three groups but with different levels. The level of trust in AVs decreased from 3.78, 3.07, and 2.85 to 3.63 (4% decrease), 2.72 (11.4% decrease), and 2.31 (15.1% decrease) for respondents with strong background about AVs, respondents with limited background about AVs, and respondents who did not know about AVs before the survey. Thus, the results clearly show that respondents with no background about AVs have the lowest level of trust in AVs and their level of trust significantly decreased after giving them information about AVs’ accidents. These results are consistent with the results presented in the previous paragraph and show the importance of educating the public about AVs in order to increase their level of trust in this emerging technology.

Finally, for the level of concern about AVs, respondents with higher levels of awareness about AVs are less concerned about this technology for the two cases before and after introducing the accidents. Before introducing the accidents, the levels of concern about AVs were 3.32, 3.53 (6.3% increase), and 3.86 (16.3% increase) for respondents with extensive experience about AVs, respondents who have limited information about AVs, and respondents who have never heard of AVs. Then, after introducing the accidents to the respondent, the levels of concern about AVs increased for all the respondents within the three groups but with different levels. The levels of concern about AVs increased from 3.32, 3.53, and 3.86 to 3.58 (7.8% increase), 3.72 (5.4% increase), and 4.18 (16.8% increase) for respondents with strong background about AVs, respondents with limited background about AVs, and respondents who did not know about AVs before the survey. Thus, the results clearly show that people with no background about AVs are highly concerned about the technology and this level of concern significantly increase when these respondents received information about issues and accidents of AVs. These results show the significant impact of the previous experience on the public attitude towards AVs before and after accidents.

3.7. Impact of AVs’ accidents on respondents with different levels of awareness about accidents involving AVs

While previous studies investigated the impact of prior experience with AVs features on the public attitude towards AVs, none of these studies investigated the public attitude of people with different levels of awareness about accidents involved AVs. Thus, in this subsection, the opinion, level of trust, and concern of the respondents who have and have no prior knowledge about AVs’ accidents will be investigated before and after introducing the accidents. The results are summarised in Table 9 which shows the average levels of interest, trust, and concern for the different groups before and after the accidents besides with the percentages of change in these levels after the accidents and p-values resulting from the hypothesis testing. The results of the hypothesis testing show that there is a significant change in the levels of trust, interest, and concern towards AVs after the accidents for respondents with and without prior knowledge about accidents involving AVs as all the p-values in the table are lower than 0.05, so the null hypothesis can be rejected. In this case, it can be stated that the shift in the public attitude towards AVs after the accidents is statistically significant for respondents with and without prior knowledge about accidents involving AVs. Additionally, for the general opinion towards AVs before introducing the accidents, the results show that respondents with prior knowledge about AVs’ accidents are more positive towards AVs as the average positive opinions were 3.22 and 3.15 (2.5% decrease) for respondents with and without prior knowledge about accidents. This suggests that respondents with prior knowledge about some accidents accept the idea that AVs will not be perfect in decision making and at some points AVs will make faulty decisions similar to a traditional human-driver. Additionally, a second factor that contributes to this negative shift in the opinion from respondents with and without previous experiences is the level of awareness about AVs technology as most of the respondents who highlighted that they have prior knowledge about accidents involved AVs have already indicated that they have either a strong or little knowledge about AVs technology. Thus, these respondents are aware of the general benefits of AVs and the current state of technology. On the other side, all the respondents who had no prior knowledge about AVs’ accidents indicated that either they have no prior experience about AVs technology, or they have limited knowledge about the technology. Thus, this lack of knowledge about the benefits of AVs might affect the opinions of this group. After introducing the accidents to the respondents, the general opinion towards AVs shifted in the negative direction for the two groups but with a larger impact on the respondents with no prior experience about AVs’ accidents. The average general opinions towards AVs moved from 3.22 to 3.03 (6.2% decrease) for respondents with prior knowledge about AVs, and from 3.15 to 2.82 (10.5% decrease) for respondents without prior knowledge. It might be expected that respondents with prior knowledge about AVs’ accidents should not show a decrease in their general opinion; however, introducing these accidents with different levels of severity and multifunction in AVs might have provided the respondents with more information about the issues of AVs that they had no prior knowledge about. For example, some respondents might be aware of two of the nine accidents presented in the survey prior to the survey, which might decrease the positive attitude towards AVs. Additionally, introducing the accidents to the respondents should have recalled some negative sentiment towards AVs, which in turn decreases the level of interest in AVs.

Table 9. Average levels of interest, trust, and concern for respondents with different levels of knowledge about accidents involving AVs before and after introducing the accidents besides with the percentages of change in the means and p-values resulting from the hypothesis testing

	Prior Knowledge about AVs’ accidents	Level of interest	Level of trust	Level of concern
Before	Yes	3.22	3.22	3.45
	No	3.14	3.22	3.69
After	Yes	3.03	2.98	3.66
	No	2.82	2.8	3.94
% Change	Yes	5.90	7.45	−6.09
	No	10.19	13.04	−6.78
P-value	Yes	0.000153907	4.85191E-06	0.00025351
	No	4.28761E-07	2.76687E-07	0.00043273

Secondly, for the level of trust in AVs before introducing the accidents, both respondents with and without prior knowledge about AVs showed the same level of trust in AVs with an average value of 3.22. On the other side, after introducing the accidents, the level of trust for the two groups decreased but with a higher decrease for the respondents without prior knowledge about AVs’ accidents. The level of trust in AVs decreased from 3.22 to 3.03 (5.9% decrease) for respondents with prior experience about AVs’ accidents, and from 3.22 to 2.81 (12.7% decrease) for respondents without prior experience about AVs’ accidents. Thus, it can be observed that the respondents with no prior knowledge about AVs’ accidents do not only have the lowest level of trust in AVs, but also the group that their level of trust is significantly affected by the accidents introduced to them. Again, this can be explained as the limited knowledge about the features and benefits of this new technology made this group skeptical about AVs as explained earlier.

Finally, for the levels of concern about AVs before introducing the accidents, the results show that respondents with prior knowledge about AVs’ accidents are less concerned about AVs before and after introducing the accidents. Before introducing the accidents, the levels of concern about AVs had average values of 3.46 and 3.69 (6.6% increase) for respondents with and without prior knowledge about AVs’ accidents. On the other side after introducing the accidents, the levels of concern regarding AVs increased for the two groups. The levels of concern increased from 3.46 to 3.66 (4.8% increase) for respondents with prior knowledge about AVs’ accidents, and from 3.69 to 3.94 (6.8% increase) for respondents without prior knowledge about AVs’ accidents. Thus, the results show that respondents without prior knowledge about AVs’ accidents have higher levels of concern and are more negatively affected by the accidents when compared with respondents with prior knowledge about AVs’ accidents. As a result, the previous discussion does not only shed light on the importance of educating the public about AVs and the benefits of this new technology, but also showed the importance of educating the public about AVs’ accidents and the state of technology as AVs in order to guarantee the highest level of trust and lowest level of concerns regarding AVs.

4. Conclusion

Over the last few years, autonomous vehicles have attracted research and manufacturers from different disciplines in order to introduce AVs. In general, AVs have the potential to offer a lot of benefits; however, the degree to which AVs can achieve these benefits depends on the public attitude towards this technology. While research on AVs mainly focuses on studying the implications, benefits, and technological realization, little attention has been paid to the factors that affect the public attitude towards this technology. However, previous studies show that accidents involving AVs have a major impact on the public perception of this technology. Thus, this study focuses on understanding and quantifying how the public attitude towards AVs changes after AVs get involved in accidents. For this aim, a questionnaire survey was designed and conducted between February and June of 2022 and a total of 5880 complete responses were received from respondents who live in the USA. The survey was designed to capture the respondents’ opinions on AVs before and after accidents using a Likert scale to allow the respondents to rank their level of interest, trust, and concern about AVs technology before and after accidents. Then, the data were analysed to show how the opinions of the respondents with different demographic characteristics (age, gender, educational level, household income, prior knowledge about AVs, and prior knowledge about AVs’ accidents) changes after the accidents. The results of this study can be summarized as follows:

• The results show that the opinions of the respondents move in the negative direction towards AVs after introducing the accidents as the levels of interest and trust in AVs decrease after introducing the accidents, while the level of concern about AVs increases.

• The analysis shows that there is an inverse relationship between the age of the respondent and the level of interest in AVs as younger respondents showed higher levels of interest, and trust in AVs before and after the accident. Additionally, the negative shift in the public opinion towards AVs increased with the increase in the age of the respondent as older respondents showed the largest reduction in their level of interest and trust in AVs, and the largest increase in the level of concern towards AVs after accidents.

• The results show that male respondents are more positive towards AVs than female respondents. Additionally, comparing the negative shift in the public attitude towards AVs show that the shift in the level of interest, trust, and concern about AVs is much larger for female respondents than male respondents.

• The results show that respondents who live in households that have higher household income are more optimistic towards AVs that respondents who live in a household with an income of (>200 USD thousand) have the lowest levels of concern about AVs and the highest level of interest and trust in AVs before and after the accidents. On the other hand, respondents with a household income between 25 USD thousand and 50 USD thousand have the most pessimistic opinions towards AVs before and after the accident. Additionally, the results show that respondents from this group have the largest drop in their interest in AVs after accidents.

• The results show that there is a direct relation between the level of education attained and the level of interest in AVs, before introducing the accidents, as respondents with higher education levels show more positive attitudes than respondents with lower educational levels. After introducing the accidents, a similar trend can be observed, as the level of interest increases with the increase in the educational levels attained, except for respondents who have educational levels higher than master’s degrees who had the largest shift in their opinions towards AVs after the accidents making this group the most pessimistic group towards AVs after the accidents taking the place of respondents with an educational level lower than a bachelor’s degree.

• Investigating the impact of prior knowledge about AVs on the public attitude shows that respondents with prior knowledge about AVs are more optimistic about this emerging technology for both cases before and after introducing the accidents. Additionally, after introducing the accidents, the negative shift in the attitude decreases with the increase in the level of prior knowledge of the respondents about AVs. This analysis sheds light on the importance of educating the public about AVs and their benefits in order to achieve the highest level of acceptance of this emerging technology.

• A further analysis has been conducted in order to understand the impact of the respondents’ prior knowledge about AVs’ accidents on their attitude towards AVs. The results show that respondents with prior knowledge about AVs’ accidents are more optimistic towards AVs for the two cases studied: before and after introducing the accidents. This optimistic attitude occurs because most of the respondents with knowledge about AVs’ accidents highlighted that they have a strong background about the technology, its benefits, and its current state of technology so these respondents accept the idea that AVs will not be always able to take the correct decisions at this early level of adoption. Additionally, the results show that the negative shift in the attitude is much larger for the respondents with no prior experience about AVs’ accidents than the respondent with prior knowledge about AVs’ accidents. It might be expected that respondents with prior knowledge about AVs’ accidents should not show a decrease in their general attitude towards AVs; however, introducing these accidents with different levels of severity and multifunction in AVs might have provided the respondents with more information about the issues of AVs that they had no prior knowledge about or recalled some the negative sentiments towards AVs that negatively affect the opinion of these respondents. These results shed light on the importance of educating the public about AVs’ accidents and the state of technology in order to achieve the highest level of acceptance of this emerging technology.

• Finally, the previous analysis does not only shed light on the importance of educating the public about AVs and the benefits of this new technology, but also showed the importance of educating the public about AVs’ accidents and the state of technology of AVs in order to guarantee the highest level of trust and the lowest level of concerns regarding AVs.

The findings of this paper can help AVs developer, policymakers, and transport planning agencies in understanding the public attitude after accidents in order to react properly to avoid discouraging people from adopting AVs. Additionally, the results presented in this study show how the opinions of every group of people change after the accidents so that the appropriate reaction plans can be designed in order to react properly with every group of people through the channels or platforms that every group of respondents frequently use in a daily basis. Additionally, the results of this study show the importance of educating the public about the benefits and current state of technology of AVs in order to minimize the impact of these accidents on the public attitude as the results show that there is no significant change in the public attitude of respondents with strong background about AVs. Additionally, the analysis shows that respondents with strong background about AVs have the most positive attitude towards AVs with the lowest level of concern. Thus, developers of AVs need to work on familiarizing people more about their cars in order to guarantee the highest level of acceptance of this emerging technology. Finally, the results show that respondents who previously knew about AVs’ accidents had the lowest negative shift in their attitude after the accidents. Thus, educating the public about the possible deficiencies of AVs increase their level of acceptance of the technology. Thus, the developers of AVs should not market their vehicles as the perfect vehicles that do not make mistakes as this makes people more scared in case an accident occurred, and might discourage people from adopting this technology.

5. Study limitations and recommendations for future research

This study offers a large number of insights about the relation between accidents involving AVs and the public attitude towards AVs and the. However, this study has its own limitations. The study was conducted for respondents from the USA so it is recommended to replicate the same analysis for other countries in order to understand how respondents from different countries react to AVs’ accidents. Additionally, while the public attitude towards AVs is dynamic and changes with time so the public knowledge about AVs and their accidents should change over time so it is recommended to replicate the same survey and analysis periodically in order to capture these dynamics and understand the impact of this knowledge on the public attitude after accidents.

List of acronyms

AV, Autonomous Vehicles; VKT, Vehicles kilometers travelled

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author received no direct funding for this research.

Notes on contributors

Kareem Othman

Kareem Othman received his B.S and M.Sc degrees from the civil engineering department, faculty of engineering, Cairo University. His graduation project received the “Best Transportation Engineering Graduation Project” award in Egypt (Engineer Syndicate Competition) by the Egyptian Society of Engineering, The Egyptian Engineers Syndicate in 2016. In 2017, He received the “Excellence Award” from The Egyptian Engineers Syndicate. He did his Ph.D. at the University of Toronto, Canada, during which he received multiple distinguished awards such as the CAA award. His main research interests include public transit, autonomous vehicles, multimodal arterial control, and asphalt pavement design. This study is part of a bigger research project that focuses on understanding the public attitude towards autonomous vehicles and the factors that influence the interest of the public in autonomous vehicles.

Appendix:

Sample Survey Sheet

Section 1:

We are conducting this academic survey to understand the public opinion about self-driving cars.

This survey focuses only on fully self-driving cars. In this case, no driver is required behind the wheel at all, and the car might not even have a steering wheel or gas/brake pedals as shown in the following images.

Self-driving vehicle with no steering wheel:

Self-driving vehicle with a steering wheel:

Please contact me if you have any questions.

Thank you very much for participating in the survey.

Kareem Othman

Civil Engineering Department

University of Toronto

Section 2:

Have you ever seen, heard, or read anything about self-driving cars before participating in this survey?

• Yes, a lot

• Yes, a bit

• No

What is your general opinion about self-driving cars? Even if you have never heard of self-driving cars before participating in the survey, please give us your opinion based on the description you read at the beginning of the survey.

• 5 (Very positive)

• 4 (Somewhat Positive)

• 3 (Neutral)

• 2 (Somewhat Negative)

• 1 (Very negative)

“Self-driving cars can improve the level of safety when compared to human-driven vehicles?” Do you agree with this statement

• 5 (Strongly agree)

• 4 (Somewhat agree)

• 3 (Neutral)

• 2 (Somewhat disagree)

• 1 (Strongly disagree)

How concerned would you be about traveling in a self-driving car? where 1 indicates not concerned at all and 5 indicate highly concerned

• 1

• 2

• 3

• 4

• 5

Are you aware of any accident that involved a self-driving car?

• Yes

• No

Section 3:

Given your kentledge of these accidents, please answer the following questions:

What is your general opinion about self-driving cars? Even if you have never heard of self-driving cars before participating in the survey, please give us your opinion based on the description you read at the beginning of the survey.

• 5 (Very positive)

• 4 (Somewhat Positive)

• 3 (Neutral)

• 2 (Somewhat Negative)

• 1 (Very negative)

“Self-driving cars can improve the level of safety when compared to human-driven vehicles?” Do you agree with this statement

• 5 (Strongly agree)

• 4 (Somewhat agree)

• 3 (Neutral)

• 2 (Somewhat disagree)

• 1 (Strongly disagree)

How concerned would you be about traveling in a self-driving car? where 1 indicates not concerned at all and 5 indicate highly concerned

• 1

• 2

• 3

• 4

• 5

Section 4:

What is your gender?

• Male

• female

• Prefer not to say

• Other (please specify)

How old are you?

• Under 18

• 18–29

• 30-44

• 45–60

• 60+

What is the highest level of education you have completed?

• Lower than a bachelor’s degree

• Bachelor’s degree

• Master’s degree

• Higher than master’s degree

What is your overall yearly household Income?

• Less than 25,000 $

• 25,000 to 50,000 $

• 50,000 to 100,000 $

• 100,000 to 200,000 $

• More than 200,000 $