MASTER THESIS
NAME M.L. van Dorp | 12479217 Faculty of Economics & Business
DATE 29/06/2020 Executive Programme in Management Studies
VERSION Final Digital Business Track
SUPERVISORS drs. dr. ir. A.G.M. Pijpers MBA
Thesis on Autonomous Vehicles
A Quantitative Study on the Strength of Ownership on Consumers’ Usage Intentions
STATEMENT OF ORIGINALITY
This document is written by Margretha Lydia van Dorp who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Margretha Lydia van Dorp
TABLE OF CONTENTS
1 ABSTRACT 5
2 INTRODUCTION 6
2.1 DISRUPTION OF THE AUTOMOTIVE INDUSTRY 6
2.2 CHANGING MOBILITY DEMANDS 7
2.3 CONTRIBUTION 8
2.4 THESIS STRUCTURE 9
3 LITERATURE REVIEW 10
3.1 AUTONOMOUS VEHICLES 10
3.2 MOTIVATIONAL PATTERNS 11
3.3 TYPE OF OWNERSHIP 13
3.3.1 PRIVATELY OWNED 13
3.3.2 SHARED 14
3.4 USAGEINTENTIONS 16
3.5 TECH-SAVVY TARGET AUDIENCE 16
3.6 CONTROL VARIABLES 18
3.7 CONCEPTUAL MODEL 19
4 RESEARCH DESIGN 21
4.1 MEASUREMENTS OF VARIABLES 21
4.1.1 MOTIVATIONAL DRIVERS 21
4.1.2 MOTIVATIONAL BARRIERS 23
4.1.3 TYPE OF OWNERSHIP 23
4.1.4 USAGE INTENTIONS 24
4.1.5 PERSONALITY TRAITS |TECH-SAVVINESS 25
4.1.6 SOCIO-DEMOGRAPHIC VARIABLES 25
4.1.7 INTRODUCTION QUESTIONS 26
4.2 DATA COLLECTION 26
4.2.1 SURVEY CONTENT 27
4.2.2 ANALYSES 27
5 RESULTS 28
5.1 PREPARING THE DATA 28
5.2 NORMALITY OF THE DATA 28
5.2.1 OUTLIERS 28
5.3 RELIABILITY &VALIDITY 31
5.3.1 CORRELATION ANALYSIS 32
5.4 MULTIPLE REGRESSION 34
5.4.1 ASSUMPTIONS OF MULTIPLE LINEAR REGRESSION 36
5.4.2 LINEAR RELATIONSHIP BETWEEN INDEPENDENT AND DEPENDENT VARIABLES 37
5.4.3 NORMALLY DISTRIBUTED RESIDUALS 37
5.4.4 NO MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES 38
5.4.5 HOMOSCEDASTICITY 39
5.4.6 SUMMARY OF HYPOTHESIS TESTING 40
6 DISCUSSION AND CONCLUSION 41
6.1 DISCUSSIONS 41
6.1.1 THEORETICAL IMPLICATIONS 41
6.1.2 MANAGERIAL IMPLICATIONS 43
6.2 CONCLUSIONS 45
6.3 LIMITATIONS AND FURTHER RESEARCH 46
7 REFERENCES 48
8 APPENDICES 51
8.1 LIST OF TABLES 51
8.2 LIST OF FIGURES 51
8.3 SURVEY QUESTIONS 51
8.4 PRIOR RESEARCH 59
8.4.1 OVERVIEW OF STUDY BY MERFELD,WILHELMS,&HENKEL (2019) 59 8.4.2 OVERVIEW OF STUDY BY MERFELD,WILHELMS,HENKEL, ET AL.,(2019) 60
8.5 RESULTS & DESCRIPTIVE STATISTICS 60
8.5.1 TESTS FOR NORMALITY 60
8.5.2 OUTLIERS ANALYSIS BASED ON THREE TESTS 63
8.5.3 RELIABILITY &VALIDITY 63
8.5.4 OUTPUT LINEAR REGRESSION ANALYSIS 66
8.5.5 COEFFICIENT MATRIX 67
8.5.6 RESIDUALS 68
8.5.7 LINEAR RELATIONSHIP BETWEEN DEPENDENT AND INDEPENDENT VARIABLES 69
8.5.8 LEVENE’S TEST 69
8.5.9 INDEPENDENT SAMPLES T-TEST 70
1 ABSTRACT
Are autonomous vehicles going to disrupt the automotive industry? Before we can answer that, questions regarding safety, legislation, ethics, cyber, costs, technology, infrastructure, customer demands will have to be answered first. The aim of this study is to contribute to the knowledge on consumer behaviour towards autonomous driving. While different aspects of consumer intentions have previously been studied, the concept of ownership has been taken less into account. This thesis assesses the effect of the type of ownership on consumers’ usage intentions of autonomous vehicles.
This research was conducted by an online survey, resulting in 352 tech savvy respondents, followed by performing a four-stage hierarchical linear regression analysis. The results show that motivational drivers such as social responsibility, quality of life and social connections have a positive influence on consumers’
usage intentions. On the other hand, concerns regarding safety decrease their usage intentions.
Simultaneously social responsibility is influenced by the type of ownership and that relationship in fact has the strongest impact. This study extends the existing literature by providing insight into the strength of the motivational patterns on usage intentions of autonomous vehicles. Limitations of the study and
suggestions for future research are provided for.
Key words: autonomous vehicles (AV), autonomous driving, ownership, self-driving vehicles
2 INTRODUCTION
Being stuck in traffic on your way to work, having trouble finding parking space, getting tired after long commutes. Wouldn’t it be nice if there was a vehicle so smart, that it would bring you safely to your destination? Without thinking, any fuss or stress? Although this may sound like a vague dream, many people agree this dream will become reality. Some say even sooner than you think. Researchers are looking into the possibilities and challenges, car manufacturers are already testing different levels of automation, the government stimulates pilots and municipalities already reorganizing their infrastructure.
Can autonomous driving contribute to a sustainable future? Is it the coming disruption that the automotive industry needs and fears? Who would want to buy this vehicle? What hurdles are there to overcome when it comes down to topics like safety, legislation, ethics, cyber, costs, technology, infrastructure, customer demands? That last element is the focus of this thesis: to get better insight into consumers’ intentions and their thoughts on owning such a vehicle. Personally, I am very passionate about the opportunity
autonomous driving potentially brings to our way of living. That is why I have chosen to dedicate my time to contribute to research that will help implement this new technology and hopefully drive automation.
2.1 DISRUPTION OF THE AUTOMOTIVE INDUSTRY
According to recent studies, the automotive industry is most likely going to be disrupted by four mayor trends, referred to by the acronym CASE: Connectivity, Autonomous driving, Shared & Services and Electric mobility (Capgemini et al., 2018). Electric vehicles have started this disruption (Bohnsack et al., 2014) and will enforce business model innovation via a non-ownership model from products to services, also referred to as Mobility as a Service (MaaS) (Monios & Bergqvist, 2020). It is expected that this will trigger a deep and transformative sociotechnical transition (Skeete, 2018). One example of business model innovation is the high margin collision repair business of dealers and repair centres will have to reinvent themselves since self-driving vehicles are expected to enhance safety (KPMG, 2019). The technology will also converge existing taxi, car-rental and car-share business models to Mobility as a Service (MaaS) or Transportation as a Service (TaaS) (Kirk et al., 2015).
Although these trends influence the vehicle-dominated ways of both personal- and freight transport, this study will focus on connected autonomous vehicles for personal transportation.
2.2 CHANGING MOBILITY DEMANDS
Urbanisation is an important trend that will contribute to changing the automotive industry. People without access to vehicles are often forced to go on foot or by cycle to get to their destinations, leaving them in an unattractive and often unsafe environment. Though there is a correlation between traffic growth (car ownership) and economic growth, road congestion caused by increased vehicle use and speeds also costs economies (Douglas et al., 2011). Mobility demands and costs are being affected by an aging population, more people working from home, reduced youth drivers’ licence and user preferences regarding urban living, walking and cycling (Litman & Litman, 2015). Their study also suggests that the way people use personal vehicles is changing. It expects that by 2030-2040 more households will rely on convenient and inexpensive taxi- and carsharing services made possible by autonomous vehicles, which would lead to reduced vehicle ownership.
Municipalities are already anticipating, e.g. with the city of Amsterdam eliminating 11.000 parking spaces by 2025 and replacing them with trees, bicycle sheds and wider pavements (Amsterdam, 2019). In addition, the city of Helsinki plans on transforming its public transport network into a “mobility on demand”
system by 2025. It’s goals is to provide cheap, flexible and well-coordinated transportation in order to become competitive with private car ownership due to convenience and ease of use (Greenfield, 2014).
Potentially, self-driving vehicles present a beneficial change with regard to safety, congestion, an alternative to ownership, lack of parking spaces and travel behaviour for target groups that depend on (public) transportation such as elderly and/or physically less able people, that will drive cost efficiency and sustainability (Fagnant & Kockelman, 2015; Kirk et al., 2015). However, policymakers and researchers identify the general public is not yet familiar with the new opportunities or challenges of autonomous vehicles, such as sharing (Zmud et al., 2016).
2.3 CONTRIBUTION
This thesis contributes to research on consumer behaviour towards autonomous driving. Since 2012 research has focussed on capturing the behavioural characteristics of individuals and perceptions regarding an individuals’ willingness to pay for / use of an autonomous vehicle. An extensive review of studies about autonomous vehicles identifies the following factors affecting opinions and attitudes towards paying for and/or usage of an autonomous vehicle: level of awareness of autonomous vehicles; consumer innovativeness (comfort with innovation); safety; trust of strangers (shared mobility); environmental concerns; relative advantage, compatibility and complexity (lifestyle related) subjective norms (influence);
self-efficacy (control); and driving- related seeking scale (driver or being driven) (Gkartzonikas & Gkritza, 2019).
In the papers reviewed, researchers have attempted to predict the market penetration rate and people’s willingness to pay for / use of an autonomous vehicle. Lee et al. (2019) researched the concept of ownership from a technology acceptance perspective. They suggest that while psychological ownership affects the intention to use, it does not affect perceived usefulness. This implies that if autonomous vehicles can provide personalized mobility services, people will feel like it is their own. This way a psychological bond is encouraged (i.e., psychological ownership) with an autonomous vehicle and this proves to be an effective strategy for promoting the use of autonomous vehicles.
Research by Merfeld et al. (2019) links up with Gkartzonikas & Gkritza (2019) and calls for future assessments since usage intentions can differ in purchase intention for privately owned autonomous vehicles and participation motives in shared autonomous vehicles. It remains unclear if the ownership model (privately-owned autonomous vehicles or participation in shared mobility adoption of autonomous driving) influences consumers’ motives. Another study by Merfeld on carsharing provides a holistic framework, but lacks the quantification of those relationships and calls for future research to provide a more comprehensive perspective (Merfeld, Wilhelms, Henkel, et al., 2019).
Accordingly this thesis aims to answer the following research question:
What is the effect of ownership (privately-owned or shared) on consumers’ autonomous vehicle usage intentions?
2.4 THESIS STRUCTURE
The literature review in chapter two will explain the theoretical background of autonomous vehicles, industry experts and consumers’ opinions towards usage of autonomous vehicles. Differences between private autonomous vehicles (PAV) and shared autonomous vehicles (SAV) are mentioned as well.
Furthermore, it will elaborate on the target group of this study as well as the conceptual model used to answer the research question.
Chapter three will elaborate on the collected data and the research design. By describing the variables used in the study as well as the subsequent analyses. The results of the study are being discussed in chapter four. First by describing how the data was prepared and checked with different types of tests to continue with conducting a Hierarchical Multiple Linear Regression Analysis. Final chapter five takes a closer look at the conclusion and discussion that can be drawn from the study results. Also, the study’s limitations and points of interest for future research are laid out.
3 LITERATURE REVIEW
This chapter contains the literature review and will discuss autonomous vehicles, which drivers and barriers of adapting this new technology are previously researched by scholars, how differences between private autonomous vehicles (PAV) and shared autonomous vehicles (SAV) might influence the adoption process and what light previous research can shine on consumers in particular.
3.1 AUTONOMOUS VEHICLES
Autonomous driving has been described in different ways and in different terminologies, such as
autonomous vehicles (AV) / cars, self-driving vehicles / cars, shared autonomous vehicles (SAV), Electric Autonomous Vehicles (EAV), Connected Autonomous Vehicles (CAV), driverless vehicles, automated vehicles, robotic vehicles. This thesis focusses on the definition of a ‘connected and autonomous’ vehicle, that integrates distinct but yet related technologies (GOV.UK, 2015). Six levels of autonomy have been defined (Embark, 2018; Skeete, 2018):
0. Driver only – Driver continuously in control (no system)
1. Assisted – Minor driving tasks performed by the system (park assist)
2. Partial automation – Driver must monitor dynamic driving tasks (traffic jam assist)
3. Conditional automation – Driver does not need to monitor driving tasks, must be able to resume control (highway pilot)
4. High automation – Driver not required during defined use case (urban automated driving) 5. Full automation – No driver required (full end-to-end journey)
A recent study by Kyriakidis et al. (2015) indicates that 69% of the respondents believe automatic vehicles could gain 50% market share by 2050. While most people are willing to pay for a fully automated vehicle, they are unwilling to pay extra for a partially automated vehicle. Additionally, 20% of respondents stated that they would be willing to pay $7,000 more for a Level 5 fully automatic vehicle.
3.2 MOTIVATIONAL PATTERNS
Two different studies Merfeld, Wilhelms, Henkel, et al., (2019) and Merfeld, Wilhelms, & Henkel, (2019) researched the drivers and barriers of being driven autonomously as well as carsharing with shared autonomous vehicles.
Motivational patterns (self-fulfilment, security and responsibility) for being driven are qualitatively researched by Merfeld, Wilhelms, & Henkel, (2019) and they point out that consumers want to utilize autonomous vehicles differently than they use their conventional vehicles.
According to their study “adoption of autonomous driving is essential to exploit the technology’s benefits for society, and motivational structures can serve as guidance for policies to encourage adoption.” Their study researched the overarching motives and made an implication matrix (Appendix: Prior research), depicting the frequency of association between elements. The study drilled down from a theme (e.g. social connections) down to an attribute (e.g. being driven) into a functional consequence (e.g. time for private matters) leading to a psychosocial consequence (e.g. nurturing social bonds). Five out of seven themes are driving the adoption process according to their research: social connections, career success, quality of life and social responsibility and personal integrity.
When it comes to barriers of the adoption process, accountability and safety are highlighted out of the five themes. Since their study is a qualitative study, this study will focus on the five most dominant motivational patterns, by using the ones with over 17 direct (left) and indirect (right) associations:
This matrix (Overview of study by Merfeld, Wilhelms, & Henkel (2019)) distinguishes direct (left) and indirect (right) associations.
Drivers - Positive Barriers - Negative
- Social responsibility (22.43) - Safety (22.42) - Social connections (22.39) - Accountability (17.35) - Quality of life (20.42)
Merfeld, Wilhelms, Henkel, et al., (2019) state that researchers have previously investigated consumer behaviour and found positive effects driving acceptance towards autonomous vehicles due to elevated safety levels, convenience of usage, time of other purposes, and workplace flexibility. Negative effects causing adoption barriers have been identified as losing control as computers make decisions, individual acquisition costs, and loss of driving pleasure. In their Delphi study among industry experts they also researched motivational barriers and drivers for shared autonomous vehicles leading to 7 barriers, 7 drivers and 15 future developments (Appendix: Overview of study by Merfeld, Wilhelms, Henkel, et al., (2019)).
When it comes to addressing barriers, the study by Wintersberger et al. (2019) researched safety, transition, legal, cost and cyber concerns among consumers. Addressing drivers has been done by Haboucha et al. (2017) and resulting in the following factors: pro-autonomous vehicle sentiments, environmental concerns, technology interest, public transit attitude and enjoy driving. Since both studies by Merfeld et al. only focussed on being driven autonomously (quantitative study) as well as sharing an autonomous vehicle (industry expert’s opinions), this study will try to answer the question if these motivational structures apply for usage intentions of autonomous vehicles for consumers. Therefore, the following hypotheses in positive motivational patterns are distinguished:
H1 Motivational patterns (drivers) positively influence the usage intentions of autonomous vehicles:
I. Social responsibility positively influences usage intentions of autonomous vehicles II. Social connections positively influence usage intentions of autonomous vehicles III. Quality of life positively influences usage intentions of autonomous vehicles
This study also distinguishes these two negative motivational patterns:
H2 Motivational patterns (barriers) negatively influence usage intentions of autonomous vehicles:
I. Safety negatively influences usage intentions of autonomous vehicles II. Accountability negatively influences usage intentions of autonomous vehicles
3.3 TYPE OF OWNERSHIP
Recent studies have started researching the differences between consumers’ preferences towards private autonomous vehicles (PAV) and shared autonomous vehicles (SAV). In this subchapter the differences between private and shared autonomous vehicles will be explained.
3.3.1 PRIVATELY OWNED
Ownership as well as use of cars has increased the last decades (Douglas et al., 2011). Studies found people would rather own an autonomous vehicle than use one such as a Car2Go or Uber taxi (Zmud et al., 2016). According to Haboucha et al. (2017) usage of a privately-owned autonomous vehicle is similar to a regular car which is owned by the household. Except it can self-drive. When people do prefer a sharing service over owning an autonomous vehicle, it is because they want to experience driving or being driving by an autonomous vehicle before owning one. In their research Lee et al. (2019) examined psychological ownership on the usage intention of the respondents and discovered it has a positive influence on usage intentions. In their studies Lee et al. (2019) refer to a different study pointing out that (private) vehicle ownership has both positive and negative relationships with the use of the vehicle: positive since it will enhance convenience and comfort, and negative since it will cause parking problems and pollution.
Car ownership is changing for young adults since a decline in driving license holding is observed as well as car ownership and car use. What is clear is that changes in the socio-economic situation, demographic situations, living situations, value orientations and cost of owning and using a car play a role, while the role of mobile technology is less clear (Melia et al., 2018). In addition, now that you can order a vehicle on the go, car ownership is being questioned by urban customers. It will result in a reduction in car sales and therefore dealer sales volume and the increasing demand for MaaS will spike sales to fleet management companies and reduce parking issues (KPMG, 2019). Another perspective to consider is the fact that dependency on a private car could lead to health issues such as physical inactivity and injuries from crashes. Infrastructure is also considered a key factor for driving car dependency, due to the way that neighbourhoods were designed (Douglas et al., 2011).
Private car ownership has been described as an extension of your living room, as something you need. We are used to car ownership. According to Grush et al. (2016) the market dissemination model called
ECAN (Exclusivity, Choice, Access and Need) confirms that 50% of the vehicles will be automated by 2040 (90% privately owned, 10% shared), followed by 100% automation in 2060, and 70% still privately owned and 30% shared. Car ownership is also determined by car drivers’ perception (hedonic, symbolic, instrumental) of the vehicle and influences their intention to adopt or purchase an electric vehicle, a study by Schuitema et al. (2013) pointed out. In contrary to previous research, instrumental attributes (e.g.
driving range) was subordinate to hedonic attributes (e.g. pleasure of driving) or symbolic attributes (self- identification derived from ownership and usage). Research by Zmud et al. (2016) points out that it is more likely that people who already have a vehicle keep their own rather than adopt a new autonomous vehicle.
3.3.2 SHARED
Research often refers to shared mobility as shared autonomous vehicles (SAV). Haboucha et al. (2017) describes shared autonomous vehicles as a subscription to a system, where you can always access autonomous vehicles without owning it. The structure of this fleet of vehicles is comparable to e.g. Zipcar.
The difference being that this vehicle will come to pick you up and drop you off at your destination without you parking the vehicle. Sharing a vehicle potentially saves costs, provides convenience and reduces vehicle usage and ownership (Narayanan et al., 2020). Their study consolidated existing knowledge on shared mobility. According to them numerous studies point out that shared mobility services will grow due to autonomous vehicles, electrification and will lead to a more sustainable future. Shared autonomous vehicles can be assessed based on three categories: sharing system (car-sharing, ridesharing, mixed systems), integration type (special cases, integrated system, independent system) and booking type (on- demand, reservation-based).
Furthermore, they sum up models that have been built to predict the number of conventional vehicles that one shared autonomous vehicle will replace, which ranges from 1.17 up to 11 cars. Carsharing is a flexible mobility option. It complements public transportation, is as flexible as a private car without the
Shared autonomous vehicles are likely to reduce vehicle ownership mostly in compact, multi-modal urban areas, and will have little effect in exurban and rural areas (Litman & Litman, 2015).
According to Lee et al. (2019), the feeling of ownership, psychological ownership, changes for
autonomous vehicles since sharing services will enable us to use the services without owning the vehicle.
Yet Zmud et al. (2016) conclude that vehicle ownership was not significant when it comes to usage behaviour, while owning a vehicle with highly automated features is.
Studies differ on the percentage of penetration rates by shared autonomous vehicles, and until now agree that 100% will not be accomplished for at least the next 30 years (Narayanan et al., 2020). According to their study some might even suggest that privately owned vehicles continue to exist. For example, loss aversion will be hindering the transit towards shared mobility. Results from those studies contradict one another, since “61% of the respondents prefer ride–sharing shared autonomous vehicles over private autonomous vehicles,”(Stoiber et al., 2019). Research by Narayanan et al. (2020) classifies the shared mobility solution into car-sharing, ride-sharing and mixed systems. It can reduce costs, usage, ownership, kilometres travelled and at the same time provide convenience.
According to Wintersberger et al. (2019) high acquisition costs of autonomous vehicles will favour sharing the vehicle and will provide better mobility at a lower cost. It would enable new business models, with subscription options on a monthly or shared pay-by-the-mile bases. The state sharing vehicles could reduce ownership up to forty-three percent. This is also highlighted by Merfeld, Wilhelms, Henkel, et al., (2019) who in their paper share the thoughts of experts who think private autonomous vehicles need to be successful first, before carsharing will lift off. Yet another expert argues that since autonomous vehicles will be expensive, it is more likely to see fewer private autonomous vehicles and more use in carsharing, enabling a new form of mobility.
H3. Type of ownership positively influences the impact of motivational drivers and barriers on the usage intentions of autonomous vehicles.
3.4 USAGE INTENTIONS
Recently, a great number of studies have researched the willingness-to-pay, user acceptance, attitudes and perceptions towards electric vehicles, hybrid vehicles, alternative fuelled vehicles and autonomous vehicles (Hohenberger et al., 2016; Kyriakidis et al., 2015; Larson et al., 2015; Tsouros & Polydoropoulou, 2020; Zmud et al., 2016). They tested the relationship between personality traits and the willingness to purchase a type of vehicle. For example, people who idolize cars or see them as an indicator of prestige, are more likely to purchase a conventional vehicle over an alternative fuelled vehicle (Tsouros &
Polydoropoulou, 2020).
Research by Zmud et al. (2016) argues that while it is possible to research a person’s intent to use an autonomous vehicle, its actual usage behaviour can only be measurable when the technology becomes available. Although desire for control was insignificant according to their test results, perceived safety benefits and privacy are when it comes to usage intention. The study by Lee et al. (2019) points out that perceived risk influences usage intentions of autonomous vehicle, due to external factors such as system errors. Since users of autonomous vehicles are only involved in a few aspects of driving, they are hardly responsible for any accidents that occur. Their results also indicate that usage intentions are influenced by the perceived usefulness of autonomous vehicles.
3.5 TECH-SAVVY TARGET AUDIENCE
In their research Tsouros & Polydoropoulou (2020) distinguished the variables exuberance, eco-
friendliness and tech-savviness to purchasing an autonomous vehicle, when tech-savvy people proved to be the most interested in an autonomous vehicle. The same goes for social influence, which can be explained since the car is often regarded as a status symbol (Zmud et al., 2016). The correlation between early innovation adopters is broadly discussed in research papers, and how they will drive public
acceptance (Bansal et al., 2016; Fagnant & Kockelman, 2015; Noppers et al., 2015; Zmud et al., 2016). It has been proven that first adopters are more likely to use it too.
Preferences such as wireless internet, real-time information application and autonomous driving are considered to impact people’s opinions on vehicle technology (Shin et al., 2015). People who prefer these advanced technology options, are more likely to purchase an autonomous vehicle (Hohenberger et al., 2016).
Yet there are consistent differences according to their study when it comes to the two dominant biological sexes, their age, and their willingness to use autonomous vehicles. According to them, the differences when it comes to willingness to use by women can be explained when taking age into account. When age increases, the willingness to use decreases. They argue that older women worry more than younger women. Previous studies already pointed out that men compared to women can easier (imagine to) use an autonomous vehicle. Tech-savvy males in urban areas with higher-incomes are more likely to spend more money on a fully automated vehicle (Bansal et al., 2016). Kyriakidis et al., (2015) point out that men are more willing to pay for automation than women. Men are also less worried about autonomous
vehicles. That theory is supported by Wintersberger et al. (2019) claim that men purchase autonomous vehicles earlier, are willing to pay more for a higher level of automation and think they are safer.
While research emphasized autonomous vehicles will be beneficial for the senior population (those over age 65), they will also be beneficial for non-drivers and people with medical conditions that prohibit driving (Bansal et al., 2016; Harper et al., 2016; Zmud et al., 2016). It will enhance their individual mobility and thus participation in society. However, young adults appear to value autonomous driving more than older people (Shin et al., 2015). Interestingly, with regard to gender and age, studies have found that men are more willing to give up ownership than women. When it comes to age, both people between 18-25 years old and older people (over 60) have according to Silberg (2013) been the most keen to pay for the use of autonomous vehicles. Contradictory to those studies, the study by Zmud et al. (2016) points out that age nor household income was a significant factor for intent to use. They state that individuals with a higher level of intent to use autonomous vehicles are people who think it will decrease accident risk, are
unconcerned with data privacy, usage could be easy and fun, have physical conditions that prevent them from driving and value other people who would like to use an autonomous vehicle.
Also, younger people who have enjoyed higher education, and work either as an employee in the private sector or as a freelancer represent tech-savviness characteristics. Those positively affects the purchase of autonomous vehicles with higher levels of automation (Level >4). Automatic vehicle adoption by older people is more likely when their friends do, while younger people living in urban cities will adopt despite their friends’ behavior (Bansal et al., 2016).
People with high incomes would be willing to pay more for autonomous vehicles than people with low incomes (Shin et al., 2015). However, people with low incomes do seem to embrace smart vehicle applications, more so than people with high incomes. That could be explained since people with high
incomes already use their smartphones to access real-time traffic information. The global study by Kyriakidis et al. (2015) in 109 countries shows conflicting results with the previously mentioned studies, since their results show that neither age nor gender were significant factors, while they concluded that respondents with higher vehicle miles travelled and who used cruise control were more willing to pay for a higher level of vehicle automation. This theory is supported by Merfeld, Wilhelms, & Henkel (2019) where they refer to studies that have already found aspects of improved traffic flow and the ability to take longer commutes, which contribute to the adoption process.
3.6 CONTROL VARIABLES
For this study, socio-demographic control variables are added. To control for the target group, questions about respondent’s personality traits are integrated to examine a person’s exuberance, eco-friendliness, and tech-savviness. Especially tech-savviness is considered important for this study and will be used to define the sample. Other control variables are gender, age, education, occupation, net income per month per respondent, country and zip code, type of driver’s license, number of cars owned by the household, if they own/use an electric vehicle and the number of years of driving experience.
3.7 CONCEPTUAL MODEL
Limited research has been done on understanding what role the ownership model plays in the usage intentions of autonomous vehicle acceptance and purchase. Learning from the strengths and limitations of preceding research, I propose the conceptual framework of this thesis to be as follows:
Figure 1 Conceptual model for Direct and Moderating effects
HYPOTHESES
H1 Motivational patterns positively influence the usage intentions of autonomous vehicles I. Social responsibility positively influences usage intentions of autonomous vehicles II. Social connections positively influence usage intentions of autonomous vehicles III. Quality of life positively influences usage intentions of autonomous vehicles
H2 Motivational patterns negatively influence usage intentions of autonomous vehicles I. Safety negatively influences usage intentions of autonomous vehicles
II. Accountability negatively influences usage intentions of autonomous vehicles
H3 Type of ownership positively influences the impact of motivational drivers and barriers on consumers’
usage intentions of autonomous vehicles
4 RESEARCH DESIGN
This chapter will elaborate on the proposed measures of variables. Second, the process of collecting data is discussed and finally the chosen research methods are described.
4.1 MEASUREMENTS OF VARIABLES
A great number of items used in this research have been used in previous scientific research and therefore have a high degree of reliability. However, some constructs had to be created, since e.g. the construct of motivational patterns originated from a qualitative paper (Merfeld, Wilhelms, & Henkel, 2019).
Back translation was necessary as the survey was conducted in both Dutch and English and the items were originally developed in English. To limit the risk of bias in translation the formulation of items and instructions were reviewed by three master students.
4.1.1 MOTIVATIONAL DRIVERS
Hypothesis one proposes that motivational patterns such as social connections, social responsibility, and quality of life (=self-fulfilment) positively influence the usage of autonomous vehicles. The construct of motivational patterns is measured by both works of Merfeld et al. (2019) and Merfeld, Wilhelms, Henkel, et al. (2019). The construct contains fourteen items, of which five were selected (three positives, two
negatives) since they all contained a value of +17 for direct (gold) and indirect (black) associations. These associations are not the original questions asked, since only a list of barriers and motivators was
compiled. Therefore, I have transformed them into questions, where respondents can answer to with a Likert-scale 1-5. (Appendix: Overview of study by Merfeld, Wilhelms, & Henkel (2019))
Merfeld, Wilhelms, & Henkel (2019) split the questions into attributes, functional consequences, and psychosocial consequences for both drivers as well as barriers. This research focusses on the consequences of those attributes, to test how strongly people feel about them. Since there is overlap between the promotors and consequences in the attribution they bring, they have been summarized in the following items:
Drivers – social responsibility (D_SR)
Questionnaire items Adapted from
It creates cheaper transportation costs compared to other modes of transport (e.g. taxi, public transport)
Merfeld, Wilhelms, Henkel, et al., (2019)
It causes fewer accidents (due to e.g. advanced safety systems, intelligent driving)
Merfeld, Wilhelms, & Henkel (2019); Merfeld, Wilhelms, Henkel, et al. (2019)
It improves traffic situations (e.g. parking issues) Merfeld, Wilhelms, & Henkel (2019); Merfeld, Wilhelms, Henkel, et al. (2019)
It makes me feel safer than a conventional car Wintersberger et al. (2019)
Drivers – quality of life (D_QL)
Questionnaire items Adapted from
It enables mobility for those who cannot drive by themselves (elderly, handicapped, children)
Merfeld, Wilhelms, Henkel, et al., (2019)
It enhances convenience (e.g. time savings, comfort, ease of use)
Merfeld, Wilhelms, Henkel, et al., (2019)
It improves the ability to take longer commutes Merfeld, Wilhelms, Henkel, et al., (2019) Using it would increase my productivity Lee et al. (2019)
I would find it is useful Lee et al. (2019)
Drivers – social connections (D_SC)
Questionnaire items Adapted from
It nurtures social bonds with people outside the vehicle (e.g. social media time)
Merfeld, Wilhelms, Henkel, et al., (2019)
It nurtures social bonds with people inside the vehicle (e.g. fellow passengers)
Merfeld, Wilhelms, Henkel, et al., (2019)
It creates more time for my personal life Merfeld, Wilhelms, Henkel, et al., (2019) It creates more time for my professional life Merfeld, Wilhelms, Henkel, et al., (2019) It creates time for other activities while being in the
car (e.g. working, reading, sleeping)
Cunningham et al. (2019)
4.1.2 MOTIVATIONAL BARRIERS
Hypothesis two proposes that motivational patterns such as safety, reliability and accountability negatively influence the usage of autonomous vehicles.
Barriers – Safety (B_S)
Questionnaire items Adapted from
I fear the technology could be imperfect and could have possible malfunctions
Merfeld, Wilhelms, & Henkel (2019); Merfeld, Wilhelms, Henkel, et al. (2019)
I (would) get an uncomfortable feeling using it Merfeld, Wilhelms, & Henkel (2019); Merfeld, Wilhelms, Henkel, et al. (2019)
I think it may not perform well and cause accidents Wintersberger et al. (2019) I think that the transition from old to new technology
will lead to problems on the road Wintersberger et al. (2019)
Barriers – Accountability (B_A)
Questionnaire items Adapted from
I can be held personally accountable for something out of my control
Merfeld, Wilhelms, & Henkel (2019)
I fear data abuse (e.g. prone to hackers) Merfeld, Wilhelms, & Henkel (2019) My privacy could be exposed Merfeld, Wilhelms, & Henkel (2019) Being driven will make me only an observer Merfeld, Wilhelms, & Henkel (2019) I must hand over control to the car itself Merfeld, Wilhelms, & Henkel (2019) I fear others can take over control Merfeld, Wilhelms, & Henkel (2019)
4.1.3 TYPE OF OWNERSHIP
Hypothesis three is divided in two sections and proposes that the type of ownership acts as a moderator and positively influences the impact of motivational barriers on the usage of autonomous vehicles. To test this hypothesis questions and scales are used by Zmud et al. (2016), Lee et al. (2019), Wintersberger et al. (2019) and Haboucha et al. (2017). The questions were slightly altered. For example, Lee et al. (2019) ask if “I would sense an autonomous vehicle is my place”. This question has been taken out of the survey.
Questionnaire items (O) Adapted from Explanation:
• Privately-owned car
• Shared-autonomous car service o No ridesharing
o Ridesharing
Zmud et al. (2016)
Privately-owned autonomous car Created myself
Shared-autonomous car, without ridesharing Shared-autonomous car, with possible ridesharing
I would feel very a high degree of personal ownership for the autonomous vehicle
Lee et al. (2019)
I prefer a private autonomous car because I like it to be my own Haboucha et al. (2017) I prefer an autonomous car over a conventional car Haboucha et al. (2017) I prefer a private autonomous car over a shared autonomous car Haboucha et al. (2017) I want to experience driving or being driven by an autonomous car before
owning one
Hohenberger et al.
(2016)
4.1.4 USAGE INTENTIONS
Usage intentions of automotive vehicles, by either purchasing or sharing them, has been based on the work of Lee et al. (2019), using a Likert-scale.
Questionnaire items (U_I) Adapted from
Assuming I have access to an autonomous car, I would intend to use it Lee et al. (2019) Given I have access to an autonomous car, I predict I would use it Lee et al. (2019) In the future, I would not hesitate to use it Lee et al. (2019) I find using an autonomous car would be useful Created for this study When buying or using a new car, I want it to have automated features Created for this study
Usage Transport (U_T)
How often do you use the following means of transport in everyday life, for example, to commute and for free time activities?
- As a vehicle driver - As a vehicle passenger - By public transit - By telecommute - By biking - By walking
Zmud et al. (2016)
Their results also indicate that usage intentions are influenced by the perceived usefulness of autonomous vehicles. Since perceived ease-of-use does not affect the intention-to-use according to their results, they have been taken out of this study. Psychological ownership, as discussed in the previous paragraph, does affect the intention-to-use, and is therefore considered.
4.1.5 PERSONALITY TRAITS | TECH-SAVVINESS
For the personality traits items from these two studies were used:
Questionnaire items (PT_TS) Adapted from
I try new products before my friends and neighbors Haboucha et al. (2017) I often purchase new gadgets, even though they are expensive Haboucha et al. (2017) I am excited by the possibilities offered by new technologies Haboucha et al. (2017)
The internet is a big part of my everyday life Tsouros & Polydoropoulou (2020) I rely on technology to get things done Tsouros & Polydoropoulou (2020)
4.1.6 SOCIO-DEMOGRAPHIC VARIABLES
Questionnaire items Adapted from
In which country do you currently live? Haboucha et al.
(2017) Tsouros &
Polydoropoulou (2020)
Zmud et al. (2016) What is your zip code? (e.g. 1234 XX)
What is your age?
What is the highest degree, diploma, or certificate you have received?
To which gender identity do you most identify?
What is your occupation?
What is your approximate income per month (after tax)? Wintersberger et al. (2019) Does your household own or use a car and in what way? (multiple answers are
allowed)
Does your household own or use an electric car?
What kind of driver's license do you have?
How many years of driving experience do you have?
What is the highest level of automation of the car you own or use? Created for this paper
4.1.7 INTRODUCTION QUESTIONS
To give the respondents a broad idea of autonomous vehicles, seven introductory questions were asked first to provide them with more insights into the research topic. It was also for them to test their knowledge on the subject.
Questionnaire items Adapted from
Park itself
Cunningham et al.
(2019) Follow vehicle ahead at safe distance by itself
Avoid collisions with other vehicles and road users (e.g. pedestrians) by itself Stay within the lane by itself
Automatically adapt its speed to changing speed limits
Navigate itself to desired destination (find location and follow route) Change lanes by itself
Questionnaire items Adapted from
Imagine that you were riding by yourself in an autonomous vehicle on a trip to
the grocery store. What feelings come to mind? Zmud et al. (2016)
4.2 DATA COLLECTION
The survey was developed using the online survey platform Qualtrics and was online from April 15th until June 10th, 2020. This thesis will be based on a quantitative analysis to explain the strength of the
relationship of the type of ownership on usage intentions for autonomous vehicles. Since this is a
The sample is a convenience sample focussing on a tech-savvy target group, with an education level of high school or higher. The survey was offered via LinkedIn to participants in multiple groups with a strong focus on technology, innovation, A.I. and/or autonomous vehicles, since tech-savviness. The aim was to collect 360 respondents for the study.
4.2.1 SURVEY CONTENT
The survey consists of fourteen closed questions and one open question, each containing multiple sub questions. Also, twelve questions were asked for demographic purposes or as control variables. While predicted duration of the study was 11,2 minutes, the survey took an average of 1174 seconds (19
minutes) to complete. Out of 422 respondents 365 (86%) completed the survey and no questions were left unanswered. The unfinished records were filtered out of the results. The survey was finished in Dutch by 82,7% of the respondents.
4.2.2 ANALYSES
The analysis of the data was processed via IBM SPSS Statistics. The data was recoded, tested for
normality, skewness, kurtosis, outliers, missing values, correlations, factor analysis, scale reliability checks and T-test. Finally, a multiple linear regression analyses was performed to analyse the data, since the model consists of multiple independent variables.
5 RESULTS
In this chapter the strategy of analysis will be performed. First, preparation of the data is elaborated upon and first results are discussed. When testing for normality of the data, tests into possible outliers has been carried out. Test results of the reliability and validity checks are presented. The correlations between the independent, moderating, dependent and control variables are discussed using the correlation matrix. Finally, the outcomes of the Hierarchical Multiple Linear Regression Analysis are considered, and the hypotheses are evaluated.
5.1 PREPARING THE DATA
Once the data was collected, it was checked and prepared for statistical tests. Before downloading from SPSS, all questions were given question export tags by which they can easily be recognized. Frequencies were studied, and no errors were found in the data. As stated before, only completed surveys were used for analysis, by excluding uncompleted cases list-wise. A few items have been recoded, but recoding counter-indicative items turned out not to be necessary.
5.2 NORMALITY OF THE DATA
The variables were tested for normality. The p-value for the both the Kolmogorov-Smirnov test and the Shapiro-Wilk test is smaller than the 0.05 probability level, suggesting that the data is not normally distributed. Yet, with a sample size of over 300, these tests have been proven to be sensitive (Field, 2013). The histograms and Q-Q plots show a relative normal distribution and so do the absolute values of skewness and kurtosis (Appendix: Tests for normality). Since normality of the variables is not an
assumption of the multiple linear regression analysis, the data is not changed.
5.2.1 OUTLIERS
A handful outliers have been noticed by performing various tests. When using the univariate outliers check detecting Z-codes of outside the acceptable level of -3.0, as well as a boxplot and the IQR method (Tukey, 1977), seven cases were deleted (Appendix: Outliers analysis based on three tests). This resulted in 358 cases to be analysed.
The analysed respondents are 193 males (54,8%) and 45,2% females, and over 80% of the respondents are between the ages of 20 and 49. When it comes to education 6,5% of the respondents had a high school degree, 9,1% had a senior secondary degree, 1,4% had an associate degree, 45,2% had a
bachelor’s degree, 36,6% had a master and 1,1% had a doctorate. Two respondents had reported to have less than a high school diploma and had been taken out of the results. Four respondents that score 0 on any of the tech-savvy items. To increase the reliability of the data, they had been taken out of the results.
Out of the 356 respondents that filled out the survey completely a total of six respondents had been taken out since they do not match the target group, leaving total number of cases to be analysed at 352.
Table 1 Socio-demographics
Variables Count Percentage
Gender Male 193 54,8%
Female 159 45,2%
Age <19 4,0 1,1%
20-29 114,0 32,4%
30-39 123,0 34,9%
40-49 52,0 14,8%
50-59 41,0 11,6%
60-69 17,0 4,8%
>70 1,0 0,3%
Education High school 23,0 6,5%
Senior secondary 32,0 9,1%
Associate degree 5,0 1,4%
Bachelor 159,0 45,2%
Master 129,0 36,6%
Doctorate 4,0 1,1%
Occupation Employed full time 202,0 57,4%
Employed part time 50,0 14,2%
Self-employed / Freelance 29,0 8,2%
Student 62,0 17,6%
Unemployed 4,0 1,1%
Homemaker 1,0 0,3%
Retired 4,0 1,1%
Income < € 1.999 85,0 24,1%
€ 2.000 - € 2.999 109,0 31,0%
€ 3.000 - € 3.999 71,0 20,2%
€ 4.000 - € 4.999 21,0 6,0%
€ 5.000 - € 5.999 12,0 3,4%
€ 6.000 - € 6.999 4,0 1,1%
> € 7.000 13,0 3,7%
No answer 37,0 10,5%
Type of cars owned by household Owned 252,0 61,3%
Private lease 18,0 4,4%
Lease via company 85,0 20,7%
Shared via services 5,0 1,2%
Other 7,0 1,7%
I / my household do not use or own any cars 44,0 10,7%
Electric car No 310,0 88,1%
Yes 42,0 11,9%
Level of automation No car 41,0 11,6%
No automation 1,0 0,3%
Level 0 82,0 23,3%
Level 1 155,0 44,0%
Level 2 44,0 12,5%
Level 3 21,0 6,0%
Level 4 7,0 2,0%
Level 5 1,0 0,3%
License_TOT 0 14,0 4,0%
1 202,0 57,4%
2 79,0 22,4%
3 40,0 11,4%
4 11,0 3,1%
5 4,0 1,1%
6 2,0 0,6%
Type of drivers license Car (B, BE, B+) 336,0 58,9%
Motor (A, A1, A2) 62,0 10,9%
Scooter (AM) 98,0 17,2%
Truck (C, CE, C1, C1E) 12,0 2,1%
Bus (D, DE, D1, D1E) 6,0 1,1%
Tractor (T) 42,0 7,4%
I do not have a driver's license 14,0 2,5%
Driving experience (years) <1 5,0 1,5%
1 - 10 110,0 32,5%
11 - 20 125,0 37,0%
>21 98,0 29,0%
N = 352
Respondents were asked what emotions they would experience when going on a trip to the grocery store in an autonomous vehicle:
Table 2 Results on trip in autonomous vehicle
Male Female Total
Count Row N % Count Row N % Count Column N %
Carefree 53 63% 31 37% 84 11%
Convenient 133 55% 111 45% 244 31%
Having fun 32 73% 12 27% 44 6%
Independent 44 54% 37 46% 81 10%
Nervous 56 44% 72 56% 128 16%
Relaxing 69 68% 33 32% 102 13%
Stress 23 47% 26 53% 49 6%
Other 24 55% 20 45% 44 6%
N = 352
Table 3 Type of transport used by respondents
Never to rarely Sometimes to often (1-3 x per week) Very often
Count Row N % Count Row N % Count Row N %
Vehicle driver (car) 70 19,9% 90 25,6% 192 54,5%
Vehicle passenger (car) 160 45,5% 172 48,9% 20 5,7%
Vehicle driver (other than car)
232 65,9% 82 23,3% 38 10,8%
By public transport 220 62,5% 81 23,0% 51 14,5%
By biking 65 18,5% 139 39,5% 148 42,0%
By walking 39 11,1% 151 42,9% 162 46,0%
N = 352.
5.3 RELIABILITY & VALIDITY
The interrelationship between items is tested with a factor analysis to explain the variance. With the Kaiser-Meyer-Olkin Measure (KMO) the assumptions to perform a Principal Component Matrix (PCA) are tested. The measure of sampling adequacy is 0.867. With Bartlett’s test of sphericity, the variables are tested if they are unrelated and unsuitable for structure detection. The test results show the Chi square is 3679,203 with 276 degrees of freedom, leading to 0.000 probability level. Therefore, that hypothesis is rejected, indicating that the data is related and suitable for structure detection.
The correlation between the variables is examined and resulted in the use of the Confirmatory Factor Analysis (CFA), with a Principal Component analysis (PCA), extracting seven components. The components were rotated using the OBLIMIN method to increase the interpretability of the results.
Together these components explain 69.16% of the variance.
The consistency of the measures was re-examined since some items were adapted or created for this research. This was done to determine if the items are truly measuring the constructs. The reliability of the constructs is tested by using Cronbach’s Alpha (Cronbach, 1951).
To further improve the (loading of the) scale some items were removed (Appendix: 8.5.3.1), resulting in a Cronbach’s Alpha of 0.708 and higher (table 4) which is sufficiently high. However, Drives – Quality of Life scale is lower than the intended Cronbach’s Alpha of 0.700, since two items were loading onto different components and had to be removed. But the results show that the corrected item-total correlation was well above 0.300, which indicates a good correlation with the total score of the scale.
Table 4 Summary of Reliability Results
Construct Number of
items
Cronbach’s Alpha
Scale
improvement Skewness Kurtosis
Drivers – Social Responsibility 4 0.708 No -0,262 0,377
Drivers – Quality of Life 3 0.667 Yes -0,211 -0,135
Drivers – Social Connections 4 0.827 Yes -0,037 -0,154
Barriers – Accountability 3 0.803 Yes -0,250 -0,030
Barriers – Safety 3 0.823 Yes -0,361 -0,280
Ownership 3 0.773 Yes -0,551 0,812
Usage Intentions 4 0.860 Yes -0,401 0,201
5.3.1 CORRELATION ANALYSIS
The Pearson Correlation analysis is performed to see how strongly and in what direction the components correlate.
Table 5 Mean, Standard deviation, Correlations
M SD 1 2 3 4 5 6 7 8 9 10 11 12 13
1 UI 3,650 0,758 1,00
2 D_SR 3,447 0,723 **,52 1,00
3 D_QL 3,715 0,690 **,52 **,46 1,00
4 D_SC 3,114 0,795 **,42 **,39 **,56 1,00
5 B_S 3,317 0,850 **-,49 **-,52 **-,34 **-,23 1,00 6 B_A 3,371 0,631 **-,23 **-,24 -0,10 0,01 **,35 1,00 7 O 3,585 0,581 **,28 **,18 **,22 **,17 **-,20 0,03 1,00 8 Gender 0,450 0,498 **-,23 -0,08 *-,13 **-,16 0,09 -0,06 **-,15 1,00 9 Age 3,190 1,196 0,09 -0,02 -0,09 -0,07 -0,03 0,06 -0,07 **-,26 1,00 10 Education 5,000 1,175 0,05 0,01 0,08 0,00 -0,05 0,00 0,01 -0,01 0,07 1,00 11 Income 3,040 2,192 0,08 0,01 0,01 0,03 0,01 0,07 -0,03 **-,32 **,34 **,14 1,00 12 Electric car 0,120 0,325 **,21 *,14 0,09 0,06 **-,18 **-,18 0,01 -0,07 0,06 0,05 0,08 1,00 13 Level
autom.
2,727 1,346 **,15 0,08 0,03 0,04 **-,16 **-,14 0,05 **-,24 **,24 -0,02 **,22 **,45 1,00
N= 352 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed)
The variables in column 1 represent the correlations with the dependent variable Usage intentions. The correlation matrix indicates that all values have a significant correlation at the 1% level. However, that does not apply for the variables Age, Education, and Income. There is no significant relation between those variables and Usage intentions on any level. The variables do not show strong correlations between themselves (>0,80), giving a first indication that multicollinearity will not be a problem.
When it comes to significant relationships, Motivational Drivers such as Social Responsibility (0.52), Quality of Life (0.52) and Social Connection (0.42) have a moderately strong positive correlation with Usage intentions. This is in line with the first three hypotheses. The Barrier Safety (-0,49) has a moderately strong negative correlation with Usage intentions, as was expected in hypothesis 2.1. Although Accountability (-0,23) also has a negative correlation, the relationship is weaker. Ownership (0,28) had a significant but very weak positive correlation with Usage intentions, which is in line with hypothesis 3.
As for the control variables, Gender (-0,23) has a significant negative weak correlation with Usage intentions. Owning or using an electric car in the household does have a weak positive correlation on Usage intentions, as does the Level of automation (0,15).
5.4 MULTIPLE REGRESSION
The hypotheses are tested through a four-stage hierarchical linear regression with Usage intentions as the dependent variable. At the first stage the control variables Gender, Electric car, Level of automation were entered to control for socially desirable responding. Both Gender and Level of automation are not
converted into dummies since the steps between the levels go up in the same frequency. Independent variables Drivers (3x) and Barriers (2x) were entered at stage two of the regression. The moderating variable Ownership was entered in the third stage. Interaction variables were entered in the fourth stage of the model. Table 6 shows both the unstandardized and standardized beta coefficients, R2 and R2 Change.
For the analysis, the centered means of the independent variables were used. The output can be found in the Appendix: Output Linear regression analysis.
Table 6 Summary of Regression Analysis with Usage intentions as Outcome
Model 1 Model 2 Model 3 Model 4
Variable B SE B β B SE B β B SE B β B SE B β
Gender ** -0,326 0,080 -0,214 ** -0,213 0,063 -0,140 ** -0,194 0,063 -0,128 ** -0,210 0,064 -0,138 Electric car ** 0,447 0,134 0,191 * 0,215 0,105 0,092 * 0,224 0,104 0,096 * 0,227 0,105 0,097 Level of automation 0,007 0,033 0,012 0,001 0,026 0,002 -0,001 0,026 -0,001 0,002 0,026 0,003 D_SR_centered ** 0,212 0,053 0,202 ** 0,205 0,053 0,195 ** 0,199 0,053 0,189 D_QL_centered ** 0,273 0,056 0,249 ** 0,258 0,056 0,235 ** 0,271 0,056 0,246 D_SC_centered * 0,116 0,047 0,122 * 0,112 0,047 0,118 * 0,120 0,047 0,125 B_S_centered ** -0,195 0,044 -0,218 ** -0,179 0,044 -0,201 ** -0,176 0,044 -0,197 B_A_centered -0,090 0,052 -0,074 * -0,104 0,052 -0,086 * -0,109 0,052 -0,091
O_centered ** 0,152 0,054 0,117 ** 0,143 0,055 0,110
D_SRxO ** 0,189 0,099 0,108
D_QLxO -0,058 0,096 -0,035
D_SCxO -0,095 0,083 -0,064
B_SxO 0,050 0,070 0,034
B_AxO 0,090 0,088 0,046
R2 0,092 0,461 0,473 0,481
R2 change 0,092 0,369 0,012 0,008
F ** 11,695 ** 36,623 ** 34,117 22,341
F change 11,695 46,947 8,046 1,077
N = 352, * p < .05; ** p < .01. Gender: Male was coded “0”, Female was coded “1”.
The first model is statistically significant (F(3, 348) = 11.695, ρ <0.000), with an R2 of .092. Meaning that the demographic variables Gender (β = -0.214, ρ .000), Electric car (β = 0.191, ρ .001) and Level of automation (β = 0.012, ρ .842) in this model explain only 0,09% of the variance. Gender and Electric car show to have a significant relationship with the dependent variable, while Level of automation does not. However, these relationships were not in the aim of this research and therefore these results will not be further discussed.
The second model includes the independent variables Drivers (Social Responsibility, Quality of Life and Social Connection) and Barriers (Accountability and Safety) was also found to be statistically significant F(8,242) = 36.623, ρ <0.001, with an R2 of .461. It means that the independent variables reliably predict the dependent variable Usage intentions. This model explains 46,1% of the variance in the Usage intentions, which is a change of 36,9%.
The first hypothesis states that motivational drivers positively influence the Usage intentions of autonomous vehicles and has been split into three sub hypotheses: Social Responsibility, Social Connections, and Quality of Life. The regression analysis shows that Social Responsibility (β = 0.202, ρ .000) is significantly related to the Usage intentions, and therefore evidence is found to support hypothesis 1.1. This means that when Social Responsibility increases with one unit, Usage intentions will increase with 0.202 unit. When it comes to Social Connections (β = 0.122, ρ .014) evidence is found to support hypothesis 1.2. The result is that if Social Connections increase with one unit, Usage intentions will increase with 0.122. With Quality of Life (β = 0.249, ρ .000) evidence is found to support hypothesis 1.3. It means that if Quality of Life increases with one unit, Usage intentions will increase with 0.249.
The second hypothesis states that motivational barriers negatively influence Usage intentions of autonomous vehicles and has been split into two sub hypotheses: Safety and Accountability. The regression analysis shows that Safety (β = -0.218, ρ .000) is significantly related to the Usage intentions, and therefore evidence is found to support hypothesis 2.1. It means that if Safety goes up by one unit, Usage intentions will go down by 0.218 unit. When it comes to Accountability (β = -0.074, ρ .088) there is no significant relationship and therefore no evidence is found to support hypothesis 2.2, which is therefore rejected.
The third model with the moderating variable Ownership is found to be statistically significant (F(9,342) = 34.117, ρ <0.000), with an R2 of .473. This model explains 47,3% of the variance, with an R2 change of 0,01%. This model will be used to determine the moderating effects of Ownership on Usage intentions.
Hypothesis 3 states that the type of Ownership positively influences the relationship of motivational drivers and barriers on Usage intentions of autonomous vehicles. The beta coefficient (β = 0,117, ρ <.005) shows that there is a significant moderating effect at the 1% level.
Therefore, evidence is found to support hypothesis 3 indicating that if Ownership goes up with one unit, Usage intentions increase with 0,117. In conclusion, it also means that if private ownership increases, so does the usage of autonomous vehicles. While this does not apply for shared autonomous vehicles.
The fourth model, with dependent variables, control variables and interaction variables, is found to be statistically significant (F(14,337) = 22.341, ρ <0.000). It means that the group of independent variables when used together reliably predict the dependent variable Usage intentions. With an R2 of .481 this interaction model explains 48,1% of the variance, which is 0,008% more than a model that contains only main effects. However, when examining the interaction terms individually, one can only conclude that Ownership has a significant impact on Social responsibility’s relationship with Usage intentions. The beta coefficient (β = 0.189, ρ < .001) shows that there is a significant moderating effect at the 1% level. It means that the relationship between Social responsibility and Usage intentions is increased by Ownership with 0.189.
5.4.1 ASSUMPTIONS OF MULTIPLE LINEAR REGRESSION
There are four assumptions when it comes to multiple linear regression. The first assumption is the linear relationship between the independent and dependent variables. The second assumption is a normal distribution of the residuals. The third assumption states that there should not be any multicollinearity present among the independent variables. The final assumption is that the data should be homoscedastic.
The results will be discussed in the subchapters below.
5.4.2 LINEAR RELATIONSHIP BETWEEN INDEPENDENT AND DEPENDENT VARIABLES
To determine the linear relationship between the dependent and independent variables a scatterplot is used (Appendix: Linear relationship between dependent and independent variables). All scatterplots include a best fitting line to provide a clear picture of the relationship. It can be concluded that for every independent variable there is a linear relationship with the dependent variable Usage intentions since they show a linear fit with the data points. This means that the assumption of linearity is met.
5.4.3 NORMALLY DISTRIBUTED RESIDUALS
To determine if the residuals are normally distributed, a scatterplot with the standard predicted value on the x-axis and standardized residuals on the y-axis is used to analyse if the errors are normally distributed.
If a normal distribution is the case, no patterns should be visible, and the dots should be scattered around the line. For the analysis a histogram and probability plot are presented. The histogram (figure 2.I) shows the data generally following the line shaped like a bell.
Figure 2 .I Histogram and II. Probability Plot of the Standardized Residuals
The probability plot (figure 2.II) shows the dots are broadly following the line as well. Therefore, based on figures 2 and 3 we can conclude the assumption of normality of the residuals to be met.
When checking for autocorrelations with the Durbin-Watson test, the value must be between 0 and 4. For the fourth model a value of 1,766 is detected which indicates a positive correlation.
Figure 3 Scatterplot of Standardized Residuals
5.4.4 NO MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES
Two methods are being used to test for multicollinearity. Multicollinearity should not be present in the data, meaning that none of the independent variables can be strongly correlating (≥0.80) between themselves.
To test if the assumption is met, a Pearson Correlation matrix was previously presented in table 5. There are no values greater than 0.7, meaning that there is no multicollinearity among predictor valuables.
Secondly, the Variance Inflation Factors (VIF’s) are analysed and should not exceed 10 and the average should not be bigger than 1. Also, the tolerance values of each variable in the regression analysis are checked and should be greater than 0.2. All four regression models show that the VIFs are below 10 and an average of 1.217 for Model 1, 1.420 for model 2, 1.392 for model 3 and 1.556 for the final model. The tolerance level is greater than 0.2 for all models. It is concluded that multicollinearity will not be a problem
Table 7 The Variance Inflation Factors (VIF) and Tolerance Values
Variables Regression 1 Regression 2 Regression 3 Regression 4
VIF Tol. VIF Tol. VIF Tol. VIF Tol.
Gender 1,061 0,942 1,103 0,907 1,114 0,897 1,137 0,879
Electric car 1,260 0,793 1,296 0,771 1,298 0,771 1,306 0,765
Level of automation 1,328 0,753 1,349 0,741 1,350 0,741 1,359 0,736
D_SR_centered 1,646 0,607 1,650 0,606 1,667 0,600
D_QL_centered 1,666 0,600 1,682 0,595 1,702 0,588
D_SC_centered 1,554 0,644 1,555 0,643 1,579 0,633
B_S_centered 1,539 0,650 1,565 0,639 1,585 0,631
B_A_centered 1,204 0,831 1,214 0,823 1,229 0,814
O_centered 1,100 0,909 1,144 0,874
D_SRxO 2,090 0,478
D_QLxO 2,153 0,464
D_SCxO 2,027 0,493
B_SxO 1,469 0,681
B_AxO 1,341 0,746
Average VIF 1,217 1,420 1,392 1,556
N = 352
5.4.5 HOMOSCEDASTICITY
The fourth assumption is that the data should be homoscedastic, meaning that error terms are the same across every value of the independent values. When evaluating Cook’s Distance, providing an overall measure for the influence of individual cases, the maximum value is 0.065 which is well below the point of concern (1.0). When performing the Levene’s Test to test if the population variances of the two groups (male and female) are equal, that null hypothesis is rejected (p >0.05). This means that the equal variances are not assumed and there is a significant difference in the mean between males and females (t342.639 = 4,447, p < .001). The difference in Usage intentions for males was 0,35 more than for females (Appendix: Levene’s test and Independent samples T-test).
With the four assumptions of the hierarchical multiple linear regression were all found to be met, the conclusion is drawn regarding the results of the hypotheses being all valid.
