ArticlePDF Available

Non-user Acceptance of Autonomous Technology: A Survey of Bicyclist Receptivity to Fully Autonomous Vehicles

September 2024
Computers in Human Behavior Reports 16(2):100490

DOI:10.1016/j.chbr.2024.100490

License
CC BY 4.0

Authors:

Peter André Busch

University of Agder

Research model of AV receptivity * The dashed lines show hypothesized moderating effects.

…

Factors influencing AV receptivity.

…

Descriptive information about the respondents (n = 219).

…

Summary of the structural model test results.

…

Figures - uploaded by Peter André Busch

Content may be subject to copyright.

Content uploaded by Peter André Busch

Content may be subject to copyright.

Non-user acceptance of autonomous technology: A survey of bicyclist

receptivity to fully autonomous vehicles

Peter Andr´

e Busch

Department of Information Systems, University of Agder, Kristiansand, Norway P.O. Box 422, 4604, Kristiansand, Norway

ARTICLE INFO

Keywords:

Trust

Autonomy

Interact

Innovation

Theory of planned behavior

Transport

ABSTRACT

Whereas the information systems literature mainly has focused on the individual user acceptance of technology,

this study focuses on non-user acceptance, termed technology receptivity. Autonomous vehicles (AVs) represent

an emerging technology that challenges individual technology user acceptance. Whereas AVs promise several

advantages, their success is conditioned on public trust. AVs impact not only those who use them but also those

who share the environment with them, such as bicyclists. Empirical research has been characterized by several

competing constructs aimed at explaining AV receptivity. In this article, we (1) review extant research on the

receptivity of AVs, and (2) develop and validate a model of AV receptivity. A cross-sectional survey of 219 bi-

cyclists showed empirical support for the model. The model can be a useful tool for the AV industry and poli-

cymakers in need of assessing the likelihood of success for a wider AV diffusion and help them understand

intervention strategies to address populations that may be less inclined to accept the new technology on public

roads. We offer several recommendations for future research on rening the measurement instrument and

enhancing our understanding of AV and autonomous technology receptivity.

1. Introduction

Technology acceptance is one of the most inuential and mature

research areas in the information systems (IS) domain (Mahmud et al.,

2022; Venkatesh et al., 2003; Williams et al., 2015). Its theories are

being widely used by researchers in several disciplines. Although IS

research has gained interest in additional perspectives of technology

acceptance such as rejection, discontinuance, and replacement, the

domain continues to focus much on individual technology acceptance

(Taherdoost, 2018; Venkatesh et al., 2003). These studies have mainly

used intention to use technology and actual usage behavior as depen-

dent variables.

More recently, the introduction of autonomous technologies chal-

lenges the traditional user acceptance perspective in IS. Examples of

these technologies include autonomous vehicles (AVs) and generative

articial intelligence (AI). Autonomous technology becomes part of the

societal fabric and may inuence people who have made no active

choice of being exposed to it but can still experience the consequences of

its use. Tasks performed by this technology involve information acqui-

sition, analysis, decision selection, and/or action implementation. For

example, if AI-driven diagnostic tools are used in healthcare, patients

may become uncertain about the quality of healthcare services. The

nature of autonomous technologies makes their acceptance different

from traditional technology acceptance in several respects. First, non-

users are exposed to technology usage inuencing them without their

consent. Second, whereas traditional technology acceptance considers

technologies that require certain levels of human control, autonomous

technologies can operate without human interference. Third, public

acceptance of autonomous technologies is inuenced by safety and

privacy challenges.

Third-party acceptance has been studied in other domains (e.g., Li

et al., 2023), however, IS research has focused less on this perspective. In

this study, we focus on AVs as an application of autonomous technology.

AVs, also referred to as self-driving, driverless, or automated vehicles,

encompass six categories (SAE International, 2021), ranging from level

0 (no automation) to level 5 (full automation). AVs at level 3 and above

can minimize or eliminate the need for human interaction in both

monitoring the driving environment and controlling the vehicle.

Whereas level 3 and 4 vehicles have certain requirements for humans to

resume control of the driving if needed, level 5 vehicles are expected to

be completely controlled by software and hardware (Zhou et al., 2022).

In this study, we specically refer to fully autonomous vehicles (FAVs)

(i.e., level 5 vehicles). FAVs are understood as robotic vehicles that drive

completely autonomously from the beginning to the end of a journey

E-mail address: peter.a.busch@uia.no.

Contents lists available at ScienceDirect

Computers in Human Behavior Reports

journal homepage: www.sciencedirect.com/journal/computers-in-human-behavior-reports

https://doi.org/10.1016/j.chbr.2024.100490

Received 31 May 2024; Received in revised form 12 August 2024; Accepted 19 September 2024

Computers in Human Behavior Reports 16 (2024) 100490

Available online 23 September 2024

without any need for a human operator (Kaur & Rampersad, 2018;

Paden et al., 2016).

AVs have received much societal attention for their potential bene-

ts. Around 1.2 million individuals lose their lives in trafc accidents

across the globe every year. Additionally, between 20 and 50 million

people sustain non-fatal injuries from these incidents (WHO, 2023).

Human error accounts for about 90% of all accidents (Deb et al., 2017;

Fagnant & Kockelman, 2015). AVs are projected to reduce accidents

signicantly (Kaur & Rampersad, 2018; Miskolczi et al., 2021). In

addition, AVs can enhance mobility for those unable to drive, decrease

pollution through improved efciency, and alleviate trafc congestion

(e.g., Krueger et al., 2016; Paden et al., 2016).

However, for AVs to deliver on their potential benets, they must

reach a certain degree of market adoption. Several substantial in-

novations fall short of meeting expectations and are often abandoned

before they reach the market (Story et al., 2011). The level of acceptance

among non-users could therefore be a signicant determinant of their

ultimate success or failure. The diffusion of AVs in the streets is complex.

AVs must be able to detect the external environment, such as road signs,

other road users, and trafc density, and respond accordingly. Never-

theless, these decisions hinge on the correct operation of all the com-

ponents that make up the technology, including cameras, lasers, sensors,

and radar scanners. This would involve anticipations of correct trafc

behavior across different conditions such as weather conditions (e.g.,

snow), different types of road users (e.g., children and distracted pe-

destrians), different road conditions (e.g., unrepaired roads), different

trafc rules (e.g., expected yielding behavior), and trafc density (e.g.,

urban or rural areas). Despite signicant efforts have been directed to-

ward transitioning vehicle control from human drivers to AVs, recent

research shows that the adoption of AVs by the public is relatively low

(Stilgoe & Cohen, 2021). Vulnerable individuals exposed to this tech-

nology, such as bicyclists and pedestrians who share the same urban

fabric with these vehicles, often seem to lack the condence and will-

ingness to accept them (Kenesei et al., 2022).

This study investigates technology acceptance from the perspective

of bicyclists who are exposed to AVs—which we term AV receptivity

(Deb et al., 2017). The bicyclist perspective is chosen since bicyclists are

more vulnerable in trafc accidents than, for example, motorists, who

are protected by the structure of their vehicles. Moreover, our review of

the literaturehas shown that most studies focus on technology recep-

tivity from the perspective of pedestrians, which is another reason why

the bicyclist perspective is important. This study has two main

objectives.

(1) To develop and validate a model of AV receptivity: Based on our

review, we developed a research model explaining non-users’

receptivity to AVs. The model development is described in Sec-

tion 4. An empirical test of the model is conducted, providing

support for its hypothesized relationships. The empirical valida-

tion of the model is presented in Section 6.

(2) To contribute to the IS literature: Whereas user acceptance is much

researched in the IS domain, the acceptance of non-users has not

achieved much attention, unlike in other domains. This study (a)

reveals potential misalignments between users’ and non-users’

acceptance of technology, (b) provides a theoretical model of AV

receptivity, which can serve as a starting point toward theorizing

autonomous technology receptivity, and (c) suggests further

research opportunities.

2. Non-user receptivity to autonomous technology

Studies on the acceptance and use of technology have encompassed a

wide variety of perspectives in terms of different technologies, contexts,

stakeholders, units of analysis, theoretical aspects, control factors, and

research methods (Williams et al., 2015). Building on the technology

acceptance model (TAM) or the unied theory of acceptance and use

(UTAUT), researchers have developed models explaining the acceptance

of car technology (Osswald et al., 2012, pp. 51–58) and autonomous

driving (Garidis et al., 2020, pp. 1381–1390). Often-used explanations

in this context are general attitudes to technology, perceived safety,

perceived privacy, trust, and social inuence (Deb et al., 2017).

However, technology usage can impact others than its users, and

even more profoundly. For example, injuries caused if a bicyclist is hit by

an FAV. Therefore, it is important to explore the receptivity of the non-

users to such technologies. In this study, receptivity is understood as the

willingness to interact with autonomous technology, that may be un-

certain, unfamiliar, or paradoxical (Deb et al., 2017), as opposed to

acceptance, understood as the willingness to embrace usage and/or begin

to use autonomous technology (cf., Venkatesh et al., 2003).

The willingness to interact with autonomous technology is different

from using it in terms of the level of technology control, goal orientation,

and technological knowledge. Actively using technology implies a

higher level of engagement and behavioral control. When individuals

actively use technology, they are consciously and deliberately engaging

with digital tools to accomplish specic tasks or goals. This requires a

more profound understanding and control over the technology and its

functions. For instance, a pilot operating the autopilot on a plane and an

individual using a smart home system both exemplify usage that ne-

cessitates active technological choices. Users in this context are not just

passive recipients of technology but provide initial input and supervise it

to achieve its desired outcomes.

By contrast, an individual interacting with autonomous technology

may have little or no knowledge and control of it. For example, a

bicyclist crossing the street in front of an FAV. The technology functions

independently of the individuals being exposed to it and operates ac-

cording to goals set by others. In these scenarios, individuals may not

always know how to anticipate the actions of autonomous technologies,

leading to uncertainty and discomfort (Deb et al., 2017). Despite algo-

rithms often outperforming humans in a variety of analytical tasks,

many users still lean toward human options when given a choice

(Mahmud et al., 2022). In their seminal article, Dietvorst et al. (2015)

termed this phenomenon “algorithm aversion”. Whereas algorithm

aversion presents a signicant challenge in the adoption of autonomous

technologies, this aversion becomes even more salient when the user is

exposed to the algorithms without making a conscious choice to do so.

When exposed to this technology, potential users might feel a sense of

lost choice and control (Kaur & Rampersad, 2018) due to perceptions

that such technologies carry security risks. These risks include a

perceived vulnerability to car hacking, unauthorized remote access or

control of the vehicle, and susceptibility to computer viruses (Kaur &

Rampersad, 2018; Schoettle & Sivak, 2014). The lack of control and

understanding regarding these highly technical and opaque systems can

exclude people from the technological decisions that affect their daily

lives.

Yet, the non-users are subjected to the consequences of autonomous

technology use. Autonomous algorithms are used in numerous areas,

ranging from recommendations for the next movie to watch to critical

systems performing tasks on behalf of humans across varied sectors.

However, the consequences of interacting with these systems differ

widely, warranting more attention from researchers (Hannon et al.,

2022). For example, the consequences of incorrect medical advice,

wealth management, and fatal pedestrian accidents far exceed those of

mistaken movie recommendations. Unintended interactions with tech-

nology can lead to feelings of alienation, confusion, skepticism, or

outright resistance (Chang & Wang, 2023; Egala & Liang, 2023; Hannon

et al., 2022; Renier et al., 2021). Media coverage of accidents involving

AVs contributes to the opposition to their adoption (Hegner et al., 2019).

A signicant obstacle to the widespread acceptance of FAVs is a lack

of trust in the technology (Kaur & Rampersad, 2018). For example,

pedestrians, being particularly vulnerable on the road, have been a

concern in previous studies, with reports indicating that a substantial

number of them (approximately 60%) lack condence that vehicles (or

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

drivers) will react appropriately to their presence (Deb et al., 2017;

Karsch et al., 2012). Several of the arguments in the algorithm aversion

literature are reected in the non-user perspective: worry about and/or

witnessing algorithms fail, the severity of algorithmic errors, unclear

algorithmic functioning, the context of use, domain experience, and the

inability of algorithms to address uniqueness (cf., Mahmud et al., 2022).

Whereas the IS domain has given technology receptivity little

attention, other domains have researched it. However, empirical

research on the receptivity of AVs is characterized by using several

models including a multitude of different constructs (e.g., Li et al., 2023;

Penmetsa et al., 2019; M. T. Rahman et al., 2021). Therefore, there is a

need to review and synthesize constructs that are used to explain AV

receptivity to progress our understanding of the phenomenon. This is the

purpose of this article. In the following, recent empirical research on AV

receptivity is presented.

3. A review of recent empirical research on AV receptivity

Existing research on AV receptivity is published in different journals,

mainly in the transportation, computer science, and psychology elds.

This review focuses on empirical studies with AV receptivity as the

dependent variable, excluding the perspectives of drivers. Appendix D

presents empirical results from AV receptivity research during the last

ten years.

Our analysis reveals the literature’s primary focus on pedestrians,

being vulnerable road users and vital for broad public AV acceptance.

This study focuses on bicyclists as another group of vulnerable road

users. Whereas pedestrians normally are the least protected, bicyclists

wear protection equipment that may be of little or no help if hit by an AV

at higher speeds. A variety of factors was investigated as predictors of

AV receptivity, which can be categorized as relating to specic AV en-

counters, characteristics of individuals, and individuals’ experience with

AVs. Of the constructs investigated in the studies, trust is salient. Several

of the studies used virtual experiments as research method, focusing on

specic AV encounters. Whereas such experiments bring forth several

advantages, they cannot fully grasp emotions and considerations made

in AV encounters. We further observed that the experiences of road users

are less researched as an explanation for AV receptivity. Considering the

current situation with a relatively low diffusion of AVs, this is under-

standable. Empirical research on factors explaining situational and

learned trust in AVs, is therefore nascent. Several of the studies used pre-

recorded videos to familiarize the respondents with AVs. Moreover,

most studies are cross-sectional.

Our review further suggests that whereas technology acceptance

models have been applied to investigate the user perspective, fewer

theoretical frameworks have been used in AV receptivity studies. The

uniqueness of AV introduces a new set of challenges that necessitates

more comprehensive and context-specic frameworks. For example,

trust in automation, perceived safety, and societal readiness are all

pivotal factors that could inuence AV receptivity. These factors are not

traditionally captured in technology acceptance models. Additionally,

AV receptivity involves diverse stakeholders such as bicyclists, passen-

gers, urban planners, policymakers, and the general public whose lives

can be reshaped by the advent of these vehicles. Therefore, building

theory to capture the multifaceted dimensions of AV receptivity may be

benecial in comparing studies within and maturing this stream of

research.

3.1. Factors inuencing receptivity to AVs

Drawing on our review of the literature and the categorization

scheme described by Hoff and Bashir (2015), factors explaining AV

receptivity are categorized (see Table 1). Among the different variables

explaining receptivity, trust is considered salient (Deb et al., 2017).

3.1.1. Disposition to trust technology

Individuals’ inclination to trust AVs varies widely. Whereas certain

reasons for human behavior can be attributed to context-dependent

factors, inherent traits inuence human behavior in most situations.

They constitute an individual’s propensity to trust technology and can

have both biological and environmental explanations. For example, role

models can explain personal innovativeness. Similarly, inherent traits

can explain technology mistrust. For example, algorithm aversion is

particularly prevalent among people who are cautious and prevention-

oriented (Chang & Wang, 2023).

In this study, disposition to trust technology refers to long-term

tendencies, meaning that these tendencies are expected to be preva-

lent in most situations. Thus, dispositional trust is a relatively stable trait

(Hoff & Bashir, 2015). Our review revealed ve primary explanations

for dispositional trust: age, gender, trust propensity, risk-taking ten-

dency, and personal innovativeness. For example, Hudson et al. (2019)

found young people to be more in favor of AVs than older adults, and

Deb et al. (2018) found that participants scoring more highly in personal

innovativeness spent less time waiting before crossing the road in front

of an FAV.

3.1.2. Situational trust in AVs

In addition to more stable dispositions to trust technology, in-

dividuals can be greatly affected by the specic situation in which they

are exposed to AVs. For example, research suggests that algorithm

aversion is more prevailing in situ, particularly in pressing situations

(Chang & Wang, 2023; Egala & Liang, 2023). Situational trust in AVs

can be linked to a specic individual (e.g., having a child as a passenger)

and to a specic environment (e.g., whether an AV can communicate its

intentions).

In this study, situational trust depends on specic encounters with

AVs. The socio-technical ensembles arising from these encounters

consist of a varying combination of individual and environmental fac-

tors, and are thus, expected to be different from situation to situation.

Contrary to dispositional trust, the dening characteristic of situational

trust is that it is unstable (Hoff & Bashir, 2015). Our review revealed

three primary sources of situational trust: characteristics of the AV

(trafc behavior, external human-machine interfaces (eHMI) status),

characteristics of the environment (e.g., weather, type of crosswalk),

and characteristics of the individual (e.g., mood, risk-taking). For

example, Rad et al. (2020) found that the presence of a zebra crossing

Table 1

Factors inuencing AV receptivity.

Factor Denition Example references

Dispositional

trust in

technology

An individual’s overall

tendency to trust new

technology, independent of a

specic context and/or

technology (based on Hoff &

Bashir, 2015).

Deb et al. (2018; Hudson

et al. (2019); Hulse et al.

(2018); Jayaraman et al.

(2019); Li et al. (2023)

Situational trust

in AVs

An individual’s trust

evaluation of AVs, dependent

on a specic context and/or

technology.

Hulse et al. (2018); Faas

et al. (2021, pp. 1–17); Rad

et al. (2020)

Learned trust in

AVs

An individual’s trust

evaluation of AVs, based on

prior experience and/or

knowledge (based on Hoff &

Bashir, 2015).

Liu et al. (2019); Penmetsa

et al. (2019); Rad et al.

(2020); Reig et al. (2018)

Social inuence The degree to which an

individual perceives that

important others believe s/he

should accept AVs (based on

Venkatesh et al., 2003).

Deb et al. (2017); Rahman

et al. (2019)

Perceived benet The degree to which an

individual believes that the

use of AVs will lead to benets

that are important to him/her.

Liu et al. (2019); Penmetsa

et al. (2019)

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

affected intended crossing behavior, and Faas et al. (2021, pp. 1–17)

concluded that a status eHMI may lead to pedestrians over-trusting AVs.

3.1.3. Learned trust in AVs

Learned trust refers to an individual’s perception of the performance

of AVs based on experience and knowledge. People tend to base de-

cisions on prior experience. Experience can change based on each

interaction, and even during interaction. Thus, learned trust in AVs may

vary. Learned trust can, however, also be linked to more stable per-

ceptions—for example, the reputation of a vehicle or technology brand

(e.g., Reig et al., 2018, pp. 198–209).

In this study, learned trust is linked to non-users’ experience with

and knowledge of AVs. For most people, these factors are not based on

professional knowledge but rather reect an ‘outside’ perspective. Our

review revealed ve primary sources of learned trust: perceived risk,

previous interaction (including experiences of malfunction occur-

rences), perceived safety (including compatibility), brand reputation,

and familiarity. For example, non-users’ trust in AVs could be affected

by their interpretations of vehicle brands (Reig et al., 2018, pp.

198–209) and familiarity with AVs can affect intended crossing behavior

(Rad et al., 2020).

3.1.4. Social inuence

As social beings, our behaviors are often shaped by those around us.

Social inuence can explain technology usage, reecting that in-

dividuals are inuenced by how others will view them after using the

technology (Venkatesh et al., 2003). Faced with new technology, in-

dividuals may look to others such as inuencing people (e.g., inuencers

and celebrities) and important people (e.g., family and friends) for cues

on how to react (M. M. Rahman et al., 2019). This endorsement, whether

explicit or implicit, serves as a form of validation, signaling that the

technology is useful, trustworthy, or trendy.

In this study, social inuence is considered important when non-

users are exposed to AVs that may be unfamiliar but also an important

change in trafc, requiring consideration about advantages and risks.

Others’ testimonies can be valuable. Receptivity may stem from argu-

ments for fewer accidents and better collaboration between trafc ac-

tors. On the contrary, Rahman et al. (2019)discovered that older adults’

road-crossing behavior was not inuenced by social norms.

3.1.5. Perceived benet

Since AV adoption could be associated with risk, safety concerns, and

a loss of control, the potential benets of AVs are important since in-

dividuals may be more positive if they consider potential risks to be

outweighed by perceived benets (Liu et al., 2019; Penmetsa et al.,

2019). In this study, perceived benet is understood as the extent to

which a non-user of AVs believes that its usage will lead to benets that

are important to him or her. Contrary to social inuence, in which an

individual acknowledges the inuence of important others, perceived

benet is a result of individual consideration. Whereas these benets

may be individually motivated, such as believing that AVs will lead to

fewer accidents involving bicycles, they may also be motivated by other

reasons such as the increased mobility of a parent. Our review revealed

six primary sources of perceived benet: increased safety, reduced

trafc congestion, reduced vehicle emissions and pollution, improved

fuel economy, reduced transport costs, and increased mobility. Research

suggests that perceived benets are more inuential in determining AV

receptivity than perceived risks (Liu et al., 2019; M. M. Rahman et al.,

2019).

3.1.6. Moderating effects

Key relationships in technology acceptance models are moderated

(cf., Venkatesh et al., 2003). Our review mostly revealed moderating

effects related to situational factors. The effects of crossing time, speed

of the approaching FAV, and the use of an external FAV interface are

moderated by age, with older adults being more hesitant to cross the

road than their younger counterparts, as they use more time in crossing

the street, are more skeptical of fast-approaching FAVs, and prefer to be

signaled by an external interface on the FAV (e.g., Deb et al., 2018;

Domm`

es et al., 2021; Jayaraman et al., 2019; Rad et al., 2020).

Furthermore, female participants were found to be more receptive to-

ward FAVs in terms of feature ratings compared to male participants

(Deb et al., 2018). Additionaly, street-crossing may be initiated sooner if

the eHMI shows intention in addition to status (Faas et al., 2021, pp.

1–17). Jayaraman et al. (2019) found that signalized crosswalks

moderated intent to cross the street. In addition to the situational fac-

tors, Li et al. (2023) suggested that prior experience with advanced

driving assistance systems (ADAS) could affect bicyclists’ behaviors to-

ward them, encouraging other researchers to control for this moderating

effect.

4. Model development

Based on the literature, we have developed a research model of AV

receptivity, reecting the factors of dispositional trust and learned trust

(through perceived reliability). While situational trust and social inu-

ence are salient in the broader understanding of AV receptivity, their

exclusion from the research model is based on the current context where

direct interaction with FAVs is not yet a common experience. Therefore,

we expect the effect of social inuence to be small (if present). Situa-

tional factors such as driving behavior and weather conditions are hard

to imagine, even considering detailed scenarios. Moreover, whereas

dispositional trust and learned trust can be explained by several factors,

this study has selected two salient explanations from the literature for

each of these constructs. Fig. 1 presents the developed and tested

research model. The rationale for the hypotheses is outlined below.

4.1. Receptivity explained by inherent dispositions (H1-3)

Personal innovativeness can be understood as an individual’s will-

ingness to try something new when faced with the unknown (Agarwal &

Prasad, 1998; Deb et al., 2017). Highly innovative individuals have

strong intrinsic motivations to try out new technologies, reecting their

innovative mindset. An innovative mindset is fueled by curiosity, which

in turn engenders a form of trust not contingent upon established fa-

miliarity or proven functionality but instead on the anticipation of

future gains and the intrinsic value seen in innovation itself. This trait

has been observed to impact attitudes, behaviors, and perceptions of

societal norms (Lee et al., 2007). Innovative individuals typically

possess more optimistic views about technology (Agarwal & Prasad,

1998). In their study, Iranmanesh et al. (2017) found that those who

score highly on personal innovativeness are better at coping with un-

certainties and tend to be rst-movers. Consistent with prior research,

Deb et al. (2017) discovered that heightened personal innovativeness

bolsters pedestrians’ receptivity to FAVs. We hypothesized that.

H1. Personal innovativeness is positively associated with the pro-

pensity to trust technology.

Risk propensity is understood as the predominant or prevailing

tendency of an individual to engage in risky behaviors (based on Hulse

et al., 2018). The psychological factors that explain an individual’s

willingness to engage with the unknown or to endure potential adverse

consequences such as fatal bicyclist accidents are also inuential in

shaping attitudes toward technology (e.g., Cristea & Gheorghiu, 2016).

The literature suggests that the perceived risk of using technology can

lead to resistance to adopting new technologies (Hong et al., 2020).

Individuals with a high propensity for risk are typically characterized by

a forward-looking optimism and a greater acceptance of potential failure

as a component of progress. This outlook naturally extends to their

perceptions of technology, where the benets of engaging with new

systems, tools, or platforms are often weighed against the potential risks.

Those who are more prone to risky behaviors will seek risk and

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

appreciate new technologies with uncertain outcomes. Conversely, a

prevention-oriented approach makes it difcult to accept autonomous

technology in situations that are characterized by uncertainty (Egala &

Liang, 2023). The argument here is that the same psychological traits

that incline an individual toward risk-taking also foster a propensity to

trust in the capabilities and promises of technology. In line with this

argument, we hypothesized that.

H2. Risk propensity is positively associated with the propensity to

trust technology.

Researchers suggest that in the same way people are predisposed to

trust or mistrust others, they might also be inclined to trust or be

skeptical of machines (Merritt & Ilgen, 2008; Muir & Moray, 1996). The

underlying idea is that those who naturally lean toward trusting others

are more likely to exhibit higher levels of trust when they rst encounter

a new entity. In this study, a propensity to trust technology describes a

bicyclist’s consistent, long-term inclination to trust FAVs, regardless of

the peculiarities of a specic situation (Hoff & Bashir, 2015; Zhou et al.,

2022). Unlike trust that is situation-dependent or acquired through

experience, this trust remains constant, shaped by both biological and

external factors (Hoff & Bashir, 2015; Lazanyi & Maraczi, 2017, pp.

135–140). The level of this inherent trust can inuence how individuals

perceive information and determine their readiness to interact with

FAVs (Lazanyi & Maraczi, 2017, pp. 135–140). Since dispositional trust

is situationally independent, the characteristics of FAVs are expected to

have little or no inuence on it. Based on this, we hypothesize that.

H3a. Propensity to trust technology is associated with a positive atti-

tude toward interaction.

H3b. Propensity to trust technology is positively associated with a

behavioral intention to interact.

4.1.1. Receptivity as a result of learning (H4-6)

Brand reputation has been shown to impact trust in AVs (Hengstler

et al., 2016; Reig et al., 2018, pp. 198–209). In this study, brand repu-

tation importance refers to the extent to which the reputation of FAV

technology brands is important for bicyclists (Li et al., 2023). Most bi-

cyclists are unlikely to have had interaction experiences with AVs, so

their perceptions of these vehicles may be inuenced by brand reputa-

tion (Zhou et al., 2022). Research in different areas has shown that a

strong and positive reputation can play a signicant role in fostering

trust (Forster et al., 2018, pp. 118–128).

One study examined how trust is shaped by the brand in both med-

ical devices and FAVs (Hengstler et al., 2016). Results revealed that trust

in autonomous technology is dual-faceted, consisting of trust in the

technology itself and trust in the company innovating it (Hengstler et al.,

2016). Another study supported these results, nding varied perceptions

of Uber affecting the perceived reliability of its technology (Reig et al.,

2018, pp. 198–209). Whereas some of the study participants emphasized

the size and established reputation of the company as explanations for

trust, others pointed out fast vehicle development processes as an

explanation for distrust. This study found that only one participant

mentioned the vehicle brand, indicating that trust mainly centers

around the technology producer rather than the vehicle manufacturer

(Reig et al., 2018, pp. 198–209). We hypothesized that.

H4. Brand reputation importance is positively associated with

perceived reliability.

New technology is open to many possible and plausible in-

terpretations. In this study, we understand familiarity with FAVs as bi-

cyclists’ recognition, awareness, or understanding of FAVs through

previous experience or exposure (Gefen, 2000). Bicyclists might acquire

previous knowledge about FAVs from their past experiences with similar

technologies or other modes of transportation, such as aviation (Hoff &

Bashir, 2015; Papadimitriou et al., 2020).

Research investigating non-users’ understanding of AVs found that it

is limited and, for many, conned to what they see on the streets or hear

through word of mouth (Reig et al., 2018, pp. 198–209). However,

others have noticed the driving patterns of AVs on the streets. Yet others

use popular media as a source of information or had a direct experience

with a company’s AV or representative (Reig et al., 2018, pp. 198–209).

Importantly, a lack of familiarity with AVs led to mistrust and fear of

them as well as the creation of personal and creative explanations, such

as concerns about the unreliability of sensors and misunderstandings

about the vehicle’s functionality (Reig et al., 2018, pp. 198–209). In-

formation about fatalities has been shown to negatively inuence pe-

destrians and bicyclists with less familiarity with AVs, whereas

pedestrians and bicyclists with a high familiarity with AVs increase their

safety perceptions regarding AVs and demand less strict regulations (M.

T. Rahman et al., 2021). Similarly, Li et al. (2023) found that a group of

bicyclists who had heard about FAVs maintained their position in the

Fig. 1. Research model of AV receptivity

* The dashed lines show hypothesized moderating effects.

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

lane when exposed to an FAV, whereas the group of bicyclists who were

unfamiliar with FAVs, were more likely to move over. Khastgir et al.

(2018) emphasized that incorrect information from external sources,

such as media and marketing, could potentially result in both overtrust

and mistrust of AVs. Based on this literature, we hypothesized that.

H5. Familiarity is positively associated with perceived reliability.

Interacting with FAVs involves situations in which people may lack

complete volitional control over the behavior of interest. Although

experience and familiarity with technology have been shown to improve

trust in automated systems, trust beliefs are also dependent on how a

specic technology is performing (Muir, 1994; Muir & Moray, 1996).

Perceived reliability of FAVs refers to an individual’s perception of the

probability that FAVs will function correctly, effectively, and securely. If

failures occur, they still can continue to perform their function, although

with limited capacity. It indicates the reliability perception of bicyclists

prior to interacting with FAVs (Hoff & Bashir, 2015). Reliability is

paramount in an industry such as transportation, where technology is

central to critical functions. Failure or inconsistent performance can

have serious consequences, including injuries and endangering lives (cf.,

Renier et al., 2021). If non-users do not perceive FAVs as reliable, they

are less likely to be receptive to them. There is much empirical evidence

showing that trust declines when there is a defect in automation (Chang

& Wang, 2023; Dietvorst et al., 2015; Egala & Liang, 2023), leading to

further negative expectations of automation in subsequent encounters.

Understanding people’s readiness to interact with emerging technolo-

gies is crucial, especially in contexts where prevailing risk perceptions

need to be addressed (Hengstler et al., 2016; McKnight et al., 2002). To

summarize, perceived reliability affects the attitude of bicyclists as well

as their intention to interact with FAVs. Thus, we hypothesized that.

H6a. Perceived reliability is associated with a positive attitude toward

interaction.

H6b. Perceived reliability is positively associated with the behavioral

intention to interact.

4.2. Behavioral intention to interact explained by the attitude toward

interaction (H7)

Attitude toward behavior is dened as “an individual’s positive or

negative feelings (evaluative affect) about performing the target

behavior” (Fishbein & Ajzen, 1975, p. 216) and can include perceptions

of benets, such as increased safety (M. M. Rahman et al., 2019). Atti-

tude is a multifaceted construct that typically includes cognitive (beliefs

about the attributes of FAVs), affective (feelings related to the use of

FAVs), and behavioral (past experiences with or observations of FAVs)

components. In this study, we focus on behavioral intention to interact

with FAVs rather than actual interaction. The theory of planned

behavior posits that having a favorable or unfavorable evaluation of

behavior would inuence the intention to engage in this behavior

(Ajzen, 2002). Humans tend to make decisions to minimize costs and

maximize benet as much as possible (Liu et al., 2019). If individuals

hold favorable views of FAVs, believing that they are safe and efcient,

such attitudes are likely to translate into a willingness to engage with

this technology. Positive attitudes are thus indicative of an overall

appraisal that the implementation of FAVs into the urban fabric will lead

to positive outcomes, and thus, the interaction with them will be

desirable. We therefore hypothesized that.

H7. Attitude toward interaction is positively associated with the

behavioral intention to interact.

4.3. Moderating effects (H8)

The model also controlled for moderating effects. Following extant

research and suggestions for future research by Li et al. (2023), we

consider the moderating effect of the bicyclists’ prior experience with

ADAS on technology receptivity. There are smart vehicles with ADAS in

the market that aim to correct driving errors and support the driver (e.g.,

by measuring driver fatigue through face detection software) (Krügel

et al., 2023). Although the participants may have knowledge of FAVs

from before, ADAS are different in that they are not fully autonomous.

Having a positive experience with these systems may inuence an in-

dividual’s perception of how likely FAVs are to malfunction, leading to a

perception of fewer accidents, contrary to making this judgment based

on descriptive information only (e.g., based on media coverage or word

of mouth), which may increase the perceived probability that such

events will happen (Charness et al., 2018; Reig et al., 2018, pp.

198–209). On the contrary, experiencing shortcomings of ADAS may

lead participants to weaken their intention to interact with an FAV. We

therefore hypothesized that.

H8a. Prior positive experience with ADAS moderates the effect of the

propensity to trust technology on behavioral intention to interact.

H8b. Prior positive experience with ADAS moderates the effect of

perceived reliability on behavioral intention to interact.

5. Method

5.1. Measurement

The nal questionnaire consisted of three sections. The rst section

introduced the respondents to the topic and provided a video describing

FAVs. The second section measured age, gender, educational level,

residing area (urban/rural), occupation/work status, daily average

cycling distance, previous interaction with FAVs, and ADAS experience.

The respondents were asked if they watched the video. The nal survey

instrument is found in Appendix B. The questionnaire was designed after

a review of validated measurement instruments. The items were adapted

to the context of FAVs when necessary. A ve-point Likert scale with the

anchors (1) totally disagree and (5) totally agree was used to measure all

items.

To measure personal innovativeness (PI), we used four items from

Agarwal and Prasad (1998), adapted to this study context. The term

“information technology” was changed by removing “information” since

bicyclists may have other associations with this term, for example,

expecting the use of a personal computer. An example item is “Among

my peers, I am usually the rst to try out new technologies”.

We used ve items from the risk propensity scale (Meertens & Lion,

2008) to measure risk propensity (RP). An example item is “I take risks

regularly”. The scale originally consisted of seven items, however, we

removed two of them to avoid any domain-specic items in our ques-

tionnaire (e.g., “I do not take risks with my health”), instead measuring

RP as a general tendency. One of the items was adapted to be measured

using the same anchors as the rest of the items. Instead of having “risk

avoider” and “risk seeker” as anchors to “I view myself as a …”, the item

was changed to “I view myself as a risk seeker.”

Propensity to trust technology (PTT) was measured by using four

adapted items of the propensity to trust scale (Hagenzieker et al., 2020;

Merritt et al., 2013). The term “trust in machines” was changed to “trust

in technology”, as this better suited our study, and the terms have

similar meanings. Our adapted questionnaire included four out of the

original six questions (e.g., “I usually trust technology until there is a

reason not to”), with two items omitted due to their lack of relevance to

our study (e.g., “In general, I would rely on a technology to assist me”).

Familiarity with FAVs (FF) was measured by adapting three items

from K¨

orber (2019, pp. 13–30) and Gefen (2000) to the context of this

study. An example item is “I already know how fully self-driving vehicles

work.” Even though the respondents most likely had not interacted with

AVs before, they may have become acquainted with the concept through

sources such as media coverage.

Based on the literature, we developed three items to measure the

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

brand reputation importance of the technology (BRT). We used the

wording of similar studies (e.g., McKnight et al., 2002 on initial con-

sumer trust) to avoid unclear, complex, and ambiguous language. The

items were then included in a larger questionnaire in a preliminary test

for validation, using a sample of 57 bicyclists. An example item is “It is

important to me that well-respected technology brands are involved in

the production of fully self-driving vehicles.” The initial test revealed

that the items performed well; BRT showed high convergent and

discriminant validity as well as acceptable composite reliability (

0.898). An evaluation followed the preliminary test based on the

participant feedback on the questionnaire. After conrming the items’

reliability and validity, they were nalized for inclusion in the nal

questionnaire.

Perceived reliability (PR) was measured by adapting four items from

K¨

orber (2019, pp. 13–30) to measure reliability in the context of auto-

mated vehicles. Items that did not t the context of this study were

omitted (e.g., “The system is capable of taking over complicated tasks”).

An example item is “I believe fully self-driving vehicles can detect the

environment correctly.”

We measured receptivity to FAVs through two constructs: attitude

toward interaction (ATI) and behavioral intention to interact (BII). ATI

was measured using four items adapted from the theory of reasoned

action (Fishbein & Ajzen, 1975). An example item is “Crossing the road

in front of an approaching fully self-driving vehicle is a good idea”.

BII was measured through the willingness to cross the road in front of

an approaching FAV, following the description of a scenario. The sce-

nario the respondents faced was:

Background: You are a cyclist (alone) approaching a crosswalk in a city.

The streets are busy but well organized. There are functioning trafc

lights, and the crosswalks are clearly marked.

Situation: While approaching the crosswalk, you see that the trafc light

for crossing the street is green. At the same time, you notice a fully self-

driving vehicle approaching the crosswalk from the left side in the lane

closest to you. The fully self-driving vehicle seems to adjust its speed, but it

is not immediately clear whether it will stop completely at the crosswalk or

not.

As a cyclist, what will your action be at the crosswalk when the fully

self-driving vehicle approaches?

We used three items from UTAUT measuring behavioral intention to

use technology (Venkatesh et al., 2003), adapting them to the context of

this study. In addition, one reversed item was added to avoid response

bias. An example item is “I predict I will cross the street in front of the

approaching fully self-driving vehicle”.

5.2. Preliminary test

A preliminary test of the model was conducted to identify potential

issues with the questionnaire. The test used respondents (n =57)

recruited from social media groups focusing on amateur bicycling (posts

approved by moderators). They were asked to ll out the self-

administered questionnaire and could then comment on the wording

and instructions. No restrictions were made regarding age, gender, work

experience, etc., focusing on the representativeness of the respondents.

A gift certicate was offered to increase the quality of the responses and

the likelihood of participation. The gift certicate was handed out to one

of the respondents after a draw. The preliminary test showed that the

participants spent 7–9 min on average to complete the questionnaire.

The reliability of the model met the conventional standard of internal

consistency, with Cronbach’s alphas falling between 0.90 and 0.95 for

two of the constructs (FF and PTT), higher than the recommended level

of 0.70–0.90 (Diamantopoulos et al., 2012). The model met the re-

quirements for discriminant validity. The questionnaire was rened

based on the preliminary test (see Appendix A).

5.3. Participants

Participants were recruited from Prolic, a crowdsourcing platform

for online subject recruitment, servicing researchers. The participants

were informed about the research purposes of their participation, and

the service included several ltering options, such as occupation, resi-

dency, hobbies, and use of transportation services.

Prolic was chosen because of its proven quality, transparency,

ltering options, and population diversity (Palan & Schitter, 2018; Peer

et al., 2017). The service is a good option to receive relevant responses

within a reasonable time. Participants were ltered using the following

criteria: (1) approval rating (≥95%), (2) commuting to work by bike,

and (3) having English as a uent language. The cognitive bias checklist

by Draws et al. (2021) was used to inform our survey design. We used

online survey software to collect data and predicted £1.68 (£10.08/hr)

for a 10-min survey on Prolic. The average hourly compensation was

£9.85 and the median completion time was 10:14.

In total, 229 bicyclists completed the survey. 10 responses were

removed because of failed attention checks, of which nine belonged to

participants under the age of 30 years. Respondents’ ages varied from 18

to 74 years, with an average of 35.6 years and a median of 33 years. Most

of the respondents were male (n =152, 69.4%), whereas 29.7% (n =65)

were female, and 0.9% identied themselves as other. Table 2 provides

more details about the respondents.

6. Validation of the model

The hypotheses were examined using partial least squares structural

equation modeling (PLS-SEM), adhering to guidelines suggested by

Becker et al. (2023) and Hair et al. (2017, 2019). PLS-SEM is particularly

useful when applied in exploratory and/or prediction-oriented research

where the aim is to explain the variance in dependent variables (J. Hair

Table 2

Descriptive information about the respondents (n =219).

Measure Item Frequency Percentage

Gender Female 65 29.7%

Male 152 69.4%

Other 2 0.9%

Age Under 30 years 71 32.5%

30–39 years 80 36.5%

40–49 years 43 19.6%

50 years and older 25 11.4%

Education (highest completed) Primary school 0 0.0%

High school/upper

secondary school

21 9.5%

Some college/

university subjects

38 17.4%

Associates/

Bachelor’s degree

84 38.4%

Graduate/Master’s

degree or higher

76 34.7%

Work status Student/

unemployed

23 10.5%

Employed 194 88.6%

Retired 2 0.9%

Area Urban 18 8.2%

Rural 201 91.8%

Daily average cycling distance

(during the last two months of

the cycling season)

0 km - 3,99 km 49 22.4%

4 km - 6,99 km 67 30.6%

7 km - 9,99 km 52 23.8%

10 km - 13,99 km 29 13.2%

14 km or more 22 10.0%

Watched video? Yes 177 80.8%

No 42 19.2%

Interacted with a fully self-

driving vehicle?

Yes 20 9.1%

No 199 90.9%

Previous experience with

advanced driver assistance

technology?

Yes 38 17.4%

No 181 82.6%

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

et al., 2022). Due to the exploratory nature of this study, the limited

existing research on the subject, and the predictive objectives of this

research, PLS-SEM was considered appropriate method for data analysis

(J. Hair et al., 2022). The analysis was conducted in two steps: (1)

assessment of the PLS model’s reliability using all survey items (in-

strument validation) and (2) assessment of the hypothesized relation-

ships in the research model (structural model validation).

6.1. Instrument validation (outer model)

All the items in the survey instrument were reective. The instru-

ment validation procedure was conducted in two steps, checking (1)

construct validity and (2) construct reliability. Construct validity refers

to the degree to which a measurement tool measures the construct it is

intended to measure (O’Leary-Kelly & Vokurka, 1998) and is done by

calculating convergent and discriminant validity.

Convergent validity is understood as the extent to which different

items that are supposed to measure the same underlying construct, are

indeed related (J. Hair et al., 2022). The outer loadings (OLs) indicate

the associations between the constructs and respective items and are

calculated for each construct. A common rule of thumb is that the OLs

should be 0.707 or above (Fornell & Larcker, 1981), and showed satis-

factory values for all items except RP3 and RP4. The model was modied

by removing these items and after the modication, all OLs were satis-

factory. We further analyzed the average variance extracted (AVE) of the

constructs. AVE values must be above the recommended threshold of

0.50 (J. Hair et al., 2022), and all values satised this criterion.

Discriminant validity evaluates whether a construct is truly distinct

from other constructs, meaning that each construct captures a phe-

nomenon that is not expected nor represented by any of the other con-

structs. We used the heterotrait-monotrait ratio of correlations criterion

(HTMT) to assess this (Henseler et al., 2015). Our analysis showed that

the HTMT values were satisfactory. Furthermore, we found that the

square root of each construct’s AVE was higher than the correlations

between the construct and other constructs, satisfying the For-

nell-Larcker’s test (Fornell & Larcker, 1981). Appendix C provides an

overview of the results.

In the second step, we tested construct reliability, understood as the

extent to which multiple items, intended to measure the same construct,

are reliable, that is, they produce consistent results. Construct reliability

was evaluated by studying Cronbach’s alpha (CA) and Composite

reliability (CR) using rho a (

A) (Dijkstra & Henseler, 2015; J. Hair et al.,

2022). The analysis showed that all CA and CR values were above the

recommended value of 0.70 (J. Hair et al., 2022). Some of the values

were above 0.90, which can harm content validity (J. Hair et al., 2022).

Appendix C shows the reliability and validity metrics, demonstrating

measurement quality.

6.2. Structural model validation (inner model)

To assess the structural model, we examined the statistical signi-

cance of the relationships between the constructs, along with the size of

the corresponding path coefcients. Fig. 2 shows the research model

with path coefcients (β), hypotheses, and the explained variance of the

endogenous variables (R

) without moderating effects. Table 3 presents

a summary of the results.

As shown in Table 3, eight out of nine hypotheses were empirically

supported. PI and RP had signicant positive impacts on PTT, explaining

31.1% of its variance. BRT and FF signicantly inuenced PR, explain-

ing 17.2% of its variance. For BII, PR, PTT, and ATI explained 38.6% of

the variance.

The model was further evaluated by examining effect size (f2). This

procedure assesses how the exogenous constructs contribute to the

endogenous constructs by simulating the inclusion and exclusion of the

exogenous constructs (Hair et al., 2022). The analysis showed different

effects on the endogenous constructs: large effects for PI→PTT, PR→ATI,

and ATI→BII, medium effects for FF→PR, BRT→PR, PR→BII, and

Fig. 2. Results of structural model tests.

Table 3

Summary of the structural model test results.

Hypothesized

path

βT

statistics

value

Hypothesis

supported?

Effect size

(f2)

H1: PI → PTT 0.475 7.958 0.000 Yes 0.301

H2: RP → PTT 0.204 3.605 0.000 Yes 0.056

H3a: PTT → ATI 0.109 1.748 0.040 Yes 0.012

H3b: PTT → BII 0.073 1.175 0.120 n.s. 0.006

H4b: BRT → PR 0.241 3.502 0.000 Yes 0.070

H5: FF → PR 0.333 5.609 0.000 Yes 0.134

H6a: PR → ATI 0.451 6.819 0.000 Yes 0.204

H6b: PR → BII 0.258 3.310 0.000 Yes 0.066

H7: ATI → BII 0.409 5.463 0.000 Yes 0.200

n.s. =non-signicant.

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

RP→PTT, and effects below the acceptable level for PTT→ATI and

PTT→BII.

We thereafter analyzed the results using the PLS

predict

procedure

(Shmueli et al., 2019). All the items of the endogenous constructs had a

smaller prediction error than the mean value prediction benchmark (i.e.,

predict

>0). Furthermore, PLS-SEM had a lower prediction error in

terms of the root mean square error (RMSE

PLS-SEM

) than the linear model

(LM) benchmark (RMSE

) for most of the items (see Appendix C).

Therefore, we concluded that the model had medium predictive power

(Shmueli et al., 2019). Moreover, the cross-validated predictive ability

test (CVPAT) showed that PLS-SEM demonstrated a signicantly lower

average loss than the LM benchmark (Sharma et al., 2022).

The study assessed the moderating effect of experience with ADAS

(EXP) on the relationship between PTT/PR and BII. The moderating

effects were tested sequentially and showed signicant effects. The re-

sults are presented in Appendix C. Our analysis showed negative and

signicant moderating effects of EXP on the relationship between PTT

and BII and PR and BII, supporting H8a and H8b. This means that with

an increase in negative experiences with ADAS, the bicyclists showed

less intention to interact with the approaching FAV when they had

experience with ADAS. Without the inclusion of the moderating effects,

the R

value for BII was 0.386. With the inclusion of the interaction

terms (EXP*PTT and EXP*PR), the R

values increased to 0.402 and

0.404, respectively. This shows an increase of 1.6% in variance

explained in BII when moderating the relationship between PTT and BII

and 1.8% when moderating the relationship between PR and BII.

7. Discussion

This study explored non-users’ receptivity to autonomous technol-

ogy, specically focusing on FAVs. It identied ve key factors inu-

encing technology receptivity: dispositional, situational, and learned

trust in technology, social inuence, and perceived benet. A theoretical

model excluding real-life FAV interaction factors (i.e., situational trust

and social inuence) was developed and empirically tested with a survey

of 219 bicyclists. The model is depicted in Fig. 1. The ndings supported

the model, highlighting three direct determinants of BII and two of ATI,

and showed the model could predict AV receptivity based on individual

characteristics and perceived AV reliability. When data were analyzed

without including prior experience with ADAS as a moderator, the effect

of the propensity to trust technology on behavioral intention to interact

was nonsignicant. The structural model analysis is depicted in Fig. 2

and Table 3 summarizes the ndings.

Prior to discussing the implications of this work, it is necessary to

recognize its limitations. First, the analysis showed low or negligible

effects for some of the constructs, suggesting that the way these con-

structs are measured or conceptualized should be revisited. In this

respect, the scales used to measure the constructs in the model constitute

a limitation. The measures included here should be considered pre-

liminary, and future researchers should develop and validate more fully

and appropriate scales for several of the constructs. For example, the

construct BII has high-loading items. Future research should oper-

ationalize the constructs, focusing more on content validity and then

revalidate the model or extend it accordingly with new measures (cf.,

Venkatesh et al., 2003). Second, this study omitted situational trust and

social inuence as explanations of autonomous technology receptivity.

Only 9.1% of the participants had experience with AVs, which is a

common challenge with current social science studies of AVs (Li et al.,

2023). The exclusion of some of the identied constructs calls for future

research. Situational factors are known to be important in settings where

risk is involved (e.g., Cristea & Gheorghiu, 2016). Given that experi-

ments can enhance our understanding of technology receptivity, future

research should investigate bicyclists’ receptivity toward FAVs using

participants who have been exposed to FAVs (Li et al., 2023). Third, this

study is cross-sectional, capturing the attitudes of individuals at early

stages of AV diffusion. The lack of longitudinal data limits the study’s

ability to assess changes in receptivity and behavioral intention over

time. Future research should target longitudinal studies to capture how

receptivity develops as non-users are more exposed to these evolving

technologies. And fourth, although some moderating effects (like

experience with ADAS) were considered, other potential moderators,

such as perceived situational risk, were not fully explored. This could

limit the understanding of how different factors interact to inuence AV

receptivity, and future research could explore additional, potential

moderating effects to better understand individual perceptions about

technology (cf., Venkatesh et al., 2003).

7.1. Theoretical implications

From a theoretical perspective, we have developed a model of AV

receptivity, representing early theoretical advancements regarding non-

user acceptance of technology. We project that this will be more

important in the future as autonomous technologies such as AVs and

generative AI are implemented and used.

Our model showed a nonsignicant relationship between PTT and

BII. We know from technology acceptance work that latent variables can

be signicant only under certain conditions (e.g., in cases of mandatory

use) or when used by specic groups of users (e.g., older adults).

Whereas this literature has identied several moderating effects (cf.,

Venkatesh et al., 2003), our review suggests that positive experiences

with ADAS may inuence the receptivity toward FAVs positively. Our

analysis showed that EXP moderated the relationship between PTT and

BII and that this relationship became signicant considering this

moderating effect. It further showed that the relationship between PR

and BII was moderated. Our ndings indicated that the experience with

ADAS was negative, weakening the effect of PTT and PR on BII (cf.,

Charness et al., 2018; Reig et al., 2018, pp. 198–209). Considering the

argument by Venkatesh et al.’s (2003) that “it is only when one con-

siders the complex range of potential moderating inuences that a more

complete picture of the dynamic nature of individual perceptions about

technology begins to emerge” (p. 470), understanding moderating ef-

fects may be important to better understand the complexity of AV

receptivity.

The analysis of effect sizes showed that certain constructs are crucial

in determining other constructs. For example, PI is a critical factor in

determining PTT, and changes in PI will likely lead to signicant

changes in PTT. Furthermore, the analysis indicates a substantial in-

uence of PR on ATI. This means that PR is important for ATI. However,

it may be valuable to identify other predictors or antecedents of PTT and

ATI. A substantial inuence of ATI on BII was also identied, meaning

that improvements in ATI would likely result in noticeable changes in

BII. Constructs with large effect sizes are of high practical signicance

and should be the focus of interventions or changes where the aim is to

inuence the associated endogenous constructs. However, small and

negligible effects were also identied. The analysis suggests that

changes in PTT will have little to no effect on ATI, implying that other

factors are more important in determining ATI. Furthermore, our anal-

ysis showed an insignicant or negligible inuence of PTT on BII, which

suggests that PTT does not play a meaningful role in shaping BII in this

model. Given the negligible effects of PTT on ATI and BII, it may be

worth revisiting how PTT is measured or conceptualized. There is a

possibility that PTT is not adequately captured in this study.

Whereas attitude toward behavior has been left out in technology

acceptance models (e.g., UTAUT), it may play an important role in

technology receptivity research since there is a difference between

having a positive attitude toward behavior and carrying out the

behavior, especially when the latter is associated with risk. Risk-taking

is a complex phenomenon, inuenced by several factors, including in-

dividual characteristics and the situations (Cristea & Gheorghiu, 2016;

Figner & Weber, 2011). The positive evaluation of bicyclists faced with

potentially risky behaviors could be an important explanation for BII

(Cristea & Gheorghiu, 2016). Furthermore, the model was built by

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

selecting some of the explanations for PTT and PR. These constructs

were selected based on their salience in the AV receptivity literature. As

our review shows, other explanations have also been posed, and future

researchers can explore these and others as antecedents to PTT and PR.

The dependent variable of the model, AV receptivity, is measured by

the potential to interact with FAVs through attitude and behavioral

intention. The level of AV receptivity is interesting from a theoretical

perspective. Researchers may investigate consequences of the AV

receptivity level. For example, it can help explain marketing strategies.

High AV receptivity among non-users may suggest that marketers can

focus on advanced features, safety improvements, and environmental

benets of AVs. On the contrary, among non-users with low AV recep-

tivity, marketing strategies might need to address basic concerns such as

reliability, safety testing, and the regulatory compliance of AVs. Another

potential avenue is the link to usage. For example, private use of FAVs

will most likely increase if non-users accept their presence in the streets.

Social inuence has repeatedly been identied as the least effective

predictor of behavioral intention (Armitage & Conner, 2001). Whereas it

has predicted technology acceptance in voluntary settings, technology

receptivity research may show that descriptive norms (i.e., what others

do without implying what a correct behavior would be) are more salient.

Descriptive norms have been identied as good predictors of risky be-

haviors in trafc (Cestac et al., 2014). If crossing behavior is perceived

risky, what others do could be a salient predictor of crossing behavior.

However, social inuence may affect perceptions about FAVs. Cristea

and Gheorghiu (2016) suggested that signicant others’ approval of

behavior may lead to a more positive attitude toward that same

behavior. Moreover, they found that the social inuence on attitudes

varied according to the situation in question (Cristea & Gheorghiu,

2016).

Theoretically extending the model to study other autonomous tech-

nologies offers a promising avenue for research. Other prominent ap-

plications of autonomous technology areas include generative articial

intelligence, algorithmic trading, patient diagnoses, and manufacturing.

However, the consequences of interacting with these systems differ

widely, warranting more attention from researchers (Hannon et al.,

2022). For example, the consequences of incorrect medical advice,

erroneous wealth management, and fatal bicyclist accidents far exceed

those of mistaken movie recommendations (Chang & Wang, 2023; Egala

& Liang, 2023; Hannon et al., 2022; Renier et al., 2021). One critical

aspect in this respect is the role of perceived reliability. For autonomous

technologies to be embraced, users must trust not only their current

capabilities but also their potential for consistent performance and

improvement. Considering the sometimes-grandiose promises of

generative AI, non-users need to trust not only its current capabilities

but also its potential for consistent performance. In healthcare, for

example, the ability of generative AI to analyze vast datasets could

signicantly advance diagnosis and treatment options, provided that

patients are willing to accept the recommendations of the technology.

Similarly, in the educational sector, generative AI is used experimen-

tally, for example, to provide grading explanations to students. This

practice will require the acceptance of students. An extension of the

model to other autonomous technologies may suggest other antecedents

to the main constructs of the model. For example, algorithmic bias may

be a worry of students if generative AI is used in education.

7.2. Practical implications

From a practical perspective, this research raises several interesting

questions. One important issue is the link between PR, ATI, and BII.

Whereas much of extant empirical research has investigated situational

trust through the use of eHMIs, this study shows that other explanations

also contribute to heightened AV receptivity, for example, PR explained

by factors such as BRT, and FF, as investigated in this study. As non-users

gain actual interaction experience with AVs, FF will become more

important. However, in an era where the knowledge of AVs is still low,

AVs must be introduced to the market in such a way that it allows society

to gain gradually learned trust in them. Increased learned trust in AVs

can be achieved through positive publicity about AVs as well as through

increasing the transparency of AV development processes (Zhou et al.,

2022).

Another perspective pertains to the importance of PTT. The rela-

tionship of the construct with BII was only signicant when a moderator

was introduced to the model: EXP. This nding could suggest that even

though an individual may have a positive attitude toward FAVs, an

actual interaction will be dependent on other factors such as EXP. This

experience could, of course, be both positive and negative, which ulti-

mately affects the attitude of the non-users. For example, positive ex-

periences with ADAS may reduce anxiety or increase comfort with

leaving control of technology, thereby increasing the willingness to

interact with FAVs. Marketing efforts could be directed toward non-

users with positive ADAS experiences if these individuals could be

identied. The marketing could then highlight similarities between

ADAS and FAV functionalities, leveraging trust and familiarity to

overcome barriers to acceptance among the public.

The level of AV receptivity may signal when the market is ready for a

wider diffusion of AV technology. For AVs to deliver on their potential

benets, it is imperative that non-users develop a certain level of

receptivity to avoid an ultimate failure. For a successful diffusion of

FAVs, practice should further take into consideration the factors that

come into play when actual interaction takes place.

8. Concluding remarks

We can expect autonomous technology to be more prevalent in

future society. AVs represent a frontier of innovation that can redene

industries, job roles, and social norms. Therefore, it becomes imperative

to anticipate how this technology will affect us to mitigate potential

adverse consequences and address concerns among the public. Although

the model presented here has a parsimonious structure, it provides op-

portunities for researchers to further explore the antecedents and con-

sequences of AV interaction. Future research should identify constructs

that can add to the prediction of attitudes and intentions toward AVs,

beyond what is currently known and understood. Moreover, as AVs

become more integrated into our urban fabric, research should focus on

actual behavior and consequences in addition to attitudes and intention.

By studying actual behavior in naturalistic settings, researchers can gain

a clearer understanding of the real-world challenges and opportunities

that come with the widespread adoption of autonomous technology.

Whereas IS research mainly has been concerned with individual tech-

nology user acceptance, the present work advances research by focusing

on the acceptance of individuals who do not use technology, but still are

affected by it in their daily lives. Moreover, this research has reviewed

and synthesized empirical literature to address the use of multiple

constructs and to develop a parsimonious model that can progress our

understanding of AV receptivity. The model can also provide a foun-

dation for researchers focusing on other types of autonomous

technologies.

CRediT authorship contribution statement

Peter Andr´

e Busch: Writing – review & editing, Writing – original

draft, Validation, Project administration, Methodology, Formal analysis,

Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing nancial

interests or personal relationships that could have appeared to inuence

the work reported in this paper.

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Data availability

Data will be made available on request.

Appendices.

Appendix A: Measurement instrument renement

The results and evaluations of the preliminary test model suggested that some of the items should be removed or replaced by new items. In

particular, two issues motivated the renement of the measurement instrument: (1) low-loading reective items, and (2) some of the constructs having

too high ICR, threatening content validity (J. Hair et al., 2022; Venkatesh et al., 2003).

(1) Three of the items (PR2, RP3, and RP4) loaded too low, according to the recommended threshold of 0.707 (Fornell & Larcker, 1981). Two of the

items (RP3 and RP4) were deleted, and one item (PR2) was rened to address this issue.

(2) Two of the constructs had a high ICR, which often is not desirable (Diamantopoulos et al., 2012). Whereas a selection of items based on their

loadings or corrected item-total correlations is frequently suggested in the psychometric literature (Venkatesh et al., 2003), this approach

would favor a homogeneous instrument and may compromise content validity due to the reduced breadth of domain coverage (Venkatesh et al.,

2003). The items were rened to address this issue.

Table A1 provides a list of items that are either rened or removed in the developed model’s nal instrument.

Table A1

Rened and removed items compared to the preliminary test model

Construct Item

Original item Action Reason for removal/adjusted item

RP RP3 I really dislike not knowing what is going to happen. [R] Removed Low loading

RP4 I usually view risks as a challenge. Removed Low loading

PTT PTT2 For the most part, I distrust technology. [R] Removed Content validity

ATI ATI1 Crossing the street in front of an approaching fully self-driving

vehicle is a good idea.

Removed Content validity

FF FF2 I am unfamiliar with how self-driving vehicles work. [R] Modied Compared to others, I think I am quite familiar with how fully self-

driving vehicles work.

FF3 I am aware of how self-driving vehicles work. Modied I am familiar with fully autonomous driving.

PR PR2 I believe self-driving vehicles are likely to malfunction. [R] Modied I believe fully self-driving vehicles are likely to produce serious errors.

[R]

Appendix B: Final measurement instrument

Table B1Final measurement instrument

Construct and origin Item Wording M SD

Personal innovativeness (PI) Agarwal and Prasad (1998) PI1 If I heard about a new technology, I would look for ways to experiment with it. 3.63 0.99

PI2 Among my peers, I am usually the rst to try out new technologies. 2.99 1.08

PI3 In general, I am hesitant to try out new technologies. [R] 3.70 1.04

PI4 I like to experiment with new technologies. 3.68 1.01

Risk propensity (RP) Meertens and Lion (2008) RP1 I prefer to avoid risks. [R] 2.58 1.02

RP2 I take risks regularly. 2.48 0.98

RP5 I view myself as a risk-seeker 2.27 1.03

Propensity to trust technology (PTT) Adapted from Hagenzieker et al.

(2020) and Merritt et al. (2013)

PTT1 I usually trust technology until it gives me a reason not to. 3.77 0.93

PTT3 I am likely to trust technology even when I have little knowledge about it. 3.23 0.99

PTT4 My tendency to trust technology is high. 3.63 0.96

Brand reputation importance – technology (BRT) BRT1 It is important to me that well-respected technology brands are involved in the

production of fully self-driving vehicles.

3.68 1.04

BRT2 The reputation of the technology brands involved in producing fully self-driving

vehicles is important to me.

3.74 1.02

BRT3 It is important for me that reputable technology brands contribute in the

production of fully self-driving vehicles.

3.68 1.02

Familiarity with FAVs (FF) Adapted from Gefen (2000) and K¨

orber

(2019)

FF1 I already know how fully self-driving vehicles work. 3.17 1.01

FF2 Compared to others, I think I am quite familiar with how fully self-driving

vehicles work.

3.00 1.04

FF3 I am familiar with fully autonomous driving. 2.77 1.11

Perceived reliability (PR) Adapted from Fishbein and Ajzen (1975) and

K¨

orber (2019)

PR1 I believe fully self-driving vehicles can detect the environment correctly. 3.11 0.99

PR2 I believe fully self-driving vehicles are likely to produce serious errors. [R] 2.75 0.99

PR3 I believe fully self-driving vehicles are capable of interpreting situations

correctly.

3.19 0.90

PR4 I believe fully self-driving vehicles are reliable. 3.10 0.91

M =Mean, SD =Standard Deviation.

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Table B1

continued. Final measurement instrument

Construct and origin Item Wording M SD

Attitude toward interaction (ATI) Adapted from Fishbein and Ajzen

(1975)

ATI2 I like the idea of crossing the street in front of an approaching self-driving vehicle. 2.20 1.03

ATI3 The idea of crossing the street in front of an approaching self-driving vehicle is

pleasant.

2.07 0.92

ATI4 Crossing the street in front of an approaching self-driving vehicle is a foolish idea.

[R]

2.72 1.16

Behavioral intention to interact (BII) Adapted from Venkatesh et al.

(2003)

BII1 I imagine I will cross the street in front of the approaching self-driving vehicle. 2.70 1.26

BII2 I have no intention of crossing the street in front of the approaching self-driving

vehicle. [R]

2.78 1.33

BII5 I predict I will cross the street in front of the approaching self-driving vehicle. 2.84 1.26

M =Mean, SD =Standard Deviation.

Appendix C: Instrument and model validation results

Table C1

Crossloadings

Item ATI BII BRT FF PI PR PTT RP

ATI2 0.902 0.507 0.041 0.112 0.238 0.457 0.313 0.216

ATI3 0.917 0.492 0.080 0.217 0.260 0.476 0.345 0.208

ATI4 0.881 0.526 0.034 0.148 0.173 0.436 0.260 0.097

BII1 0.501 0.957 0.077 0.193 0.214 0.477 0.331 0.177

BII2 0.555 0.933 0.078 0.230 0.192 0.474 0.325 0.225

BII3 0.547 0.952 0.117 0.192 0.225 0.478 0.323 0.122

BRT1 0.038 0.089 0.894 0.014 0.028 0.204 0.043 −0.029

BRT2 0.040 0.091 0.910 0.040 0.038 0.226 0.084 0.037

BRT3 0.076 0.083 0.930 0.001 0.090 0.243 0.128 0.067

F1 0.151 0.170 0.037 0.886 0.280 0.290 0.070 0.140

F2 0.210 0.280 −0.030 0.916 0.452 0.355 0.213 0.259

F3 0.079 0.082 0.068 0.841 0.317 0.225 0.064 0.184

PI1 0.206 0.148 0.009 0.339 0.861 0.262 0.475 0.308

PI2 0.211 0.208 0.037 0.414 0.828 0.298 0.390 0.269

PI3 0.186 0.187 0.101 0.247 0.775 0.290 0.472 0.199

PI4 0.228 0.202 0.049 0.370 0.876 0.276 0.464 0.325

PR1 0.428 0.480 0.250 0.270 0.334 0.855 0.504 0.335

PR2 0.399 0.351 0.120 0.225 0.203 0.712 0.377 0.108

PR3 0.371 0.403 0.252 0.328 0.273 0.822 0.357 0.168

PR4 0.469 0.409 0.178 0.284 0.284 0.888 0.439 0.223

PTT1 0.311 0.315 0.136 0.126 0.513 0.532 0.914 0.280

PTT3 0.290 0.304 0.062 0.086 0.441 0.389 0.872 0.310

PTT4 0.319 0.313 0.061 0.171 0.512 0.463 0.920 0.385

RP1 0.129 0.175 −0.009 0.152 0.299 0.150 0.318 0.886

RP2 0.181 0.158 −0.001 0.214 0.233 0.233 0.313 0.919

RP5 0.212 0.168 0.085 0.242 0.356 0.316 0.347 0.909

Table C2

The heterotrait-monotrait (HTMT) ratio of correlations

ATI BII BRT FF PI PR PTT

BII 0.618      

BRT 0.064 0.104     

FF 0.191 0.223 0.062    

PI 0.286 0.248 0.080 0.467   

PR 0.591 0.564 0.280 0.387 0.394  

PTT 0.384 0.377 0.104 0.150 0.618 0.592 

RP 0.217 0.201 0.061 0.250 0.375 0.291 0.405

Table C3

Construct validity and reliability

Construct CA CR (

A) AVE Item OL

Personal innovativeness (PI) 0.856 0.859 0.699 PI1 0.861

PI2 0.828

PI3 0.775

(continued on next page)

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Table C3 (continued )

Construct CA CR (

A) AVE Item OL

PI4 0.876

Risk propensity (RP) 0.889 0.891 0.819 RP1 0.886

RP2 0.919

RP5 0.909

Propensity to trust technology (PTT) 0.886 0.890 0.815 PTT1 0.914

PTT3 0.872

PTT4 0.920

Brand reputation importance – technology (BRT) 0.898 0.905 0.831 BRT1 0.894

BRT2 0.910

BRT3 0.930

Familiarity (FF) 0.859 0.899 0.777 FF1 0.886

FF2 0.916

FF3 0.841

Perceived reliability (PR) 0.838 0.847 0.676 PR1 0.855

PR2 0.712

PR3 0.822

PR4 0.888

Attitude toward interaction (ATI) 0.883 0.883 0.810 ATI2 0.902

ATI3 0.917

ATI4 0.881

Behavioral intention to interact (BII) 0.943 0.944 0.898 BII1 0.957

BII2 0.933

BII3 0.952

Table C4

Predictive power analysis

Item Q

predict

RMSE

PLS-SEM

RMSE

PLS-SEM

<RMSE

ATI2 0.023 1.018 1.042 Yes

ATI3 0.062 0.892 0.920 Yes

ATI4 0.021 1.152 1.201 Yes

BII1 0.051 1.231 1.260 Yes

BII2 0.060 1.297 1.324 Yes

BII3 0.058 1.226 1.260 Yes

PR1 0.117 0.936 0.906 No

PR2 0.049 0.975 1.011 Yes

PR3 0.150 0.835 0.841 Yes

PR4 0.092 0.872 0.876 Yes

PTT1 0.259 0.807 0.830 Yes

PTT3 0.213 0.884 0.895 Yes

PTT4 0.298 0.806 0.840 Yes

Table C5

Moderation analysis

Relationship Beta SD T value P value

The moderating effect of EXP on the relationship between PTT and BII

PTT* EXP→BII −0.278 0.153 1.819 0.034

EXP→BII 0.176 0.161 1.089 n.s.

PTT→BII 0.125 0.064 1.937 0.026

The moderating effect of EXP on the relationship between PR and BII

PR* EXP→BII −0.301 0.134 2.239 0.013

EXP→BII 0.218 0.160 1.363 n.s.

PR→BII 0.305 0.084 3.651 0.000

Appendix D: Review of recent empirical AV receptivity research

Table D1

Recent empirical ndings from research on AV receptivity

Author(s) SAE

level

Context Independent variables Dependent variable

(s)

Primary ndings/outcome Perspective

(continued on next page)

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Table D1 (continued )

Author(s) SAE

level

Context Independent variables Dependent variable

(s)

Primary ndings/outcome Perspective

Deb et al.

(2017)

5 Development and validation of a

questionnaire to assess pedestrian

receptivity toward FAVs.

Attitudes, social norm, trust,

compatibility, system effectiveness.

Pedestrians’

receptivity toward

FAVs.

A validated questionnaire assessing

pedestrian receptivity toward

FAVs.

Pedestrian.

Deb et al.

(2018)

5 Determines external features on

the AV and investigates 30

pedestrians’ suggestions for which

features can affect their behaviors

and attitudes using an

experimental study.

External features tested on the FAV,

trafc behavior, gender, age,

personal innovativeness.

Waiting time before

crossing, crossing

time, rating of

different interfaces.

Pedestrians’ receptivity toward

FAVs signicantly increased with

the inclusion of external features.

Females and people 30+of age

reacted the most positively to the

features. Pedestrians who often

commit errors or who show

aggressive behaviors toward other

road users rated the

implementation of FAVs poorly.

Pedestrians who intentionally

violate trafc rules and get

distracted were found to be more

cautious in the presence of FAVs

and appreciated the inclusion of

external features.

Pedestrian.

Dey et al.

(2019)

5 Explores how the road-crossing

behavior of pedestrians may be

affected by driving modes, external

appearances, and driving

behaviors, using a video-based

experiment with 60 participants.

Driving mode, vehicle appearance,

driving behaviors.

Willingness to cross

the road in front of

an AV.

Suggests that in situations when the

intent of the vehicle was unknown,

there were differences in

pedestrians’ road-crossing

willingness between two types of

vehicles (manual/automated

mode) as a result of the differences

in their appearances.

Pedestrian.

Dommes

et al.

(2021)

n.s. Compares 30 young and 30 older

adult pedestrians’ behavior when

crossing the street in front of

conventional and self-driving cars

through an experiment.

Age, trafc condition, speed of

approaching vehicle, target gap.

Decision to cross

the street in front of

an AV.

The introduction of AVs into

current road trafc poses potential

risks. Pedestrians tend to cross the

street by focusing their attention on

the trafc ow closest to them,

neglecting the speed of the

oncoming cars.

Pedestrian.

Dommes

et al.

(2024)

5 Explores the receptivity of 474

French pedestrians towards FAVs,

its effects on behavioral intention

when interacting with FAVs, and

acceptance of FAVs.

Attitudes, social norm, trust,

compatibility, system effectiveness.

Pedestrians’

receptivity toward

FAVs.

A validated questionnaire assessing

pedestrian receptivity toward

FAVs.

Pedestrian.

Eisele and

Petzoldt

(2024)

n.s. Explores the effect of a frontal

brake light on participants’ self-

reported willingness to cross an

AV’s path using a survey with 50

participants.

Presence of frontal brake light, AV

behavior, distance between AV and

pedestrian.

Willingness to cross

the road in front of

an AV or cross an

AV’s path.

When the AV activated a frontal

brake light, willingness to cross the

road was signicantly higher than

if not activated.

Pedestrian.

Faas et al.

(2021)

4+Investigates how different eHMIs

affect trust calibration and crossing

behavior of pedestrians using a

video-based laboratory study with

67 participants.

Occurrence of high-risk FAV

malfunction, status eHMI.

Trust, crossing

behavior.

A status eHMI may lead to

pedestrians overtrusting AVs so

additional messages were needed to

support trust calibration in AVs.

Pedestrian.

Feng et al.

(2024)

n.s. Explores the impact of risk

perception and trust in

autonomous vehicles on pedestrian

crossing decisions. Questionnaire

of 589 pedestrians.

Technological advancements,

personal attributes, risk perception,

trust in AV

Crossing choice

preferences.

Increase in AV speed and decrease

in AV distance increase

pedestrians’ tendency to not cross

the road. Risk perception and trust

in AVs are strong predictors.

Middle-aged and high-risk

perception-level pedestrians are

more conservative in crossing

behavior.

Pedestrian.

Harkin et al.

(2024)

Explores implicit communication

in cyclist-vehicle interactions,

using an experimental, video-

based study with 42 participants.

Driving dynamics. Cycle through the

intersection.

Weak intention to cycle through the

intersection under the crash

condition. The passive yield and

active yield conditions received

strong intentions, while the

intention inclined for the no yield

condition.

Bicyclist.

Hudson et al.

(2019)

n.s. Investigates people’s attitudes

toward AVs in the European Union

(EU), using secondary survey data

with approx. 1000 participants.

Underlying attitudes to new

technology, gender, age, age when

nishing education, location,

current occupation, prosperity, and

country dummy variables.

Attitudes toward

AVs.

In general, people tend to be

lukewarm to AVs. Some are hostile

whereas others are totally in favor.

Young, males, those in cities, and

those more educated are more in

favor. There seems to be more

support for AVs in countries with

high accident rates.

Public

opinion.

(continued on next page)

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Table D1 (continued )

Author(s) SAE

level

Context Independent variables Dependent variable

(s)

Primary ndings/outcome Perspective

Hulse et al.

(2018)

n.s. Explores perceptions of AVs among

925 road users, including drivers,

bicyclists, and pedestrians, using a

survey.

Gender, age, risk-taking

propensity, perceived risk, driver

status.

Attitudes towards

AVs.

AVs were perceived as a ‘low-risk’

form of transport, with little

opposition to future public road

use. Compared to human-operated

cars, AVs were perceived as riskier

when a passenger and less risky

when a pedestrian. Males and

younger adults displayed greater

acceptance.

Public

opinion.

Jayaraman

et al.

(2019)

4+Investigates whether driving

behaviors of AVs and trafc signals

affect pedestrians’ trust in AVs and

trust-related behaviors. Virtual

reality experiment with 30 people.

Driving behavior, crosswalk type Trust in AVs,

trusting behaviors.

Pedestrians’ trust in AVs was

inuenced by AV driving behavior

and the presence of a signal light.

Aggressive AV driving behavior

could signicantly diminish

pedestrians’ trust in AVs. People

expressed more trust in AVs at

signalized crosswalks.

Pedestrian.

Li et al.

(2023)

5 Investigates factors inuencing

bicyclists’ receptivity towards

sharing roads with FAVs and their

behavioral intentions in

interactions with FAVs. The study

used a survey with 314

participants.

Demographics (e.g., age, gender,

crash experience), cycling

behaviors (e.g., violations, errors,

positive behaviors).

Receptivity

towards sharing

roads with FAVs (e.

g., attitude, trust)

Younger and female bicyclists had

higher receptivity towards sharing

the roads with FAVs. Bicyclists

involved in a recent bicycle crash

and those reporting committing

more errors on the roads were more

willing to share roads with FAVs.

Having higher propensity to risky

behaviors and positive behaviors

were linked with taking less self-

protective behaviors during FAV

interaction.

Bicyclist.

Liljamo et al.

(2018)

n.s. Investigates people’s readiness and

concerns regarding the adoption of

AVs, using a survey of 2036

citizens.

Gender, age, education, area

density, driving controls,

technology safety, technology

reliability, privacy.

Attitudes towards

AVs.

Men, highly educated individuals,

people living in densely populated

areas, and those living in

households without a car had more

positive attitudes to AVs than other

respondents. Trafc safety and

ethical perspectives seem to have a

key role in the acceptance of AVs.

Public

opinion.

Liu et al.

(2019)

5 Explores the factors that inuence

people’s acceptance of fully

autonomous driving technology,

using a survey of 441 city

residents.

Perceived benet, perceived risk,

social trust.

General

acceptance.

Whereas perceived benet is a

signicant predictor of general

acceptance, social trust and

perceived risk played minor roles in

determining acceptance.

Public

opinion.

Palmeiro

et al.

(2018)

n.s. Explores the interaction between

pedestrians and AVs through an

experiment where 24 participants

were exposed to 20 different

scenarios.

Environmental factors (task

demands, road environment,

weather conditions), individual

factors (goals, preconceptions,

experience, training), situational

factors (situational awareness)

Crossing decision,

crossing behavior.

About half of the participants

thought that the vehicle was

(sometimes) driven automatically.

Most participants did perceive

differences in vehicle appearance

and reported to have been

inuenced by these features.

Pedestrian.

Penmetsa

et al.

(2019)

n.s. Examines interaction experiences,

perceived benets, and perceived

challenges of AVs among 798

vulnerable road users (pedestrians

and bicyclists) in Pittsburgh, PA.

Previous interaction, perceived

safety.

Attitudes towards

AVs.

Members of an organization,

promoting safe mobility options for

road users, had more positive

attitudes and beliefs regarding AVs

than the general public. Those with

interaction experiences had higher

safety benets expectations than

those with no experiences.

Furthermore, as the public

increasingly interacts with AVs,

their attitudes toward the

technology are more likely to be

positive.

Pedestrian,

bicyclist.

Rad et al.

(2020)

n.s. Explores the relation between

personal characteristics of

pedestrians and their crossing

behavior in front of an AV. The

study used a virtual reality

experiment with 60 participants.

Gender, age, time to collision, type

of vehicle, type of crosswalk,

vehicle appearance, AV familiarity.

Road crossing

behavior.

Besides the distance from the

approaching vehicle and the

existence of a zebra crossing,

pedestrians’ crossing decisions are

signicantly affected by the

participants’ age, familiarity with

AVs, the communication between

the AV and the pedestrian, and

whether the approaching vehicle is

an AV.

Pedestrian.

Rahman

et al.

(2021)

n.s. Investigates how pedestrians and

bicyclists perceive AVs based on

their knowledge and real-world

Familiarity, road-sharing

experience.

Perception of AVs. Perceptions of the respondents

were signicantly inuenced by

their views on AV safety,

Pedestrian,

bicyclist.

(continued on next page)

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Table D1 (continued )

Author(s) SAE

level

Context Independent variables Dependent variable

(s)

Primary ndings/outcome Perspective

road-sharing experiences with

AVs. A survey of 795 bicyclists and

pedestrians was conducted.

familiarity, extent of AV

informedness, and automobile

ownership. Negative perceptions

mostly included a lack of perceived

safety and comfort around AVs and

trust in the AV technology.

Respondents were also concerned

about AV technology issues while

sharing the road with AVs. AVs

following trafc rules appropriately

and AVs driving safer than human

drivers were the most notable

positive perceptions towards AVs.

Rahman

et al.

(2019)

5 Investigates how older adults

(60+) perceive AVs as pedestrians,

using a survey with participants

from the Amazon MTurk

participant pool.

Attitude, perceived usefulness,

social norms, trust, compatibility,

gender, familiarity, location.

Perception of AVs. Older adults, who are familiar with

AVs, are more likely to have a

favorable perception of them. Older

adults in suburban areas showed a

higher perception of trust in AVs.

Perceived usefulness was

associated with a positive

perception of AVs. However,

concerns about the interaction

between older pedestrians and AVs

were also raised.

Pedestrian.

Reig et al.

(2018)

4+Undertakes an interview-based

eld study of AVs with 31

pedestrians who have interacted

with Uber AVs.

Age, gender, education,

perceptions of autonomous driving

technology, brand reputation.

Trust in AVs,

acceptance of AVs.

Pedestrians’ trust in AVs is affected

by their favorable interpretations of

a vehicle brand and facilitated by

their knowledge of AVs.

Pedestrian.

Yokoi (2024) Experiment with 401 participants,

investigating the dynamics of trust

in AVs compared to human drivers,

particularly focusing on how errors

affect this trust.

Driving accuracy. Attitudes towards

AVs.

AVs were less trusted if there was a

slight possibility of making an

error. If errors are minor and do not

result in serious outcomes, people

are less likely to perceive a

signicant difference in trust

between AVs and human drivers.

Public

opinion.

n.s. =not specied.

References

Agarwal, R., & Prasad, J. (1998). A conceptual and operational denition of personal

innovativeness in the domain of information technology. Information Systems

Research, 9(2), 204–215. https://doi.org/10.1287/isre.9.2.204

Ajzen, I. (2002). Perceived behavioral control, self-efcacy, locus of control, and the

theory of planned behavior 1. Journal of Applied Social Psychology, 32(4), 665–683.

https://doi.org/10.1111/j.1559-1816.2002.tb00236.x

Armitage, C. J., & Conner, M. (2001). Efcacy of the theory of planned behaviour: A

meta-analytic review. British Journal of Social Psychology, 40(4), 471–499. https://

doi.org/10.1348/014466601164939

Becker, J.-M., Cheah, J.-H., Gholamzade, R., Ringle, C. M., & Sarstedt, M. (2023). PLS-

SEM’s most wanted guidance. International Journal of Contemporary Hospitality

Management, 35(1), 321–346. https://doi.org/10.1108/IJCHM-04-2022-0474

Cestac, J., Paran, F., & Delhomme, P. (2014). Drive as I say, not as I drive: Inuence of

injunctive and descriptive norms on speeding intentions among young drivers.

Transportation Research Part F: Trafc Psychology and Behaviour, 23, 44–56. https://

doi.org/10.1016/j.trf.2013.12.006

Chang, Y., & Wang, R. (2023). Conservatives endorse Fintech? Individual regulatory

focus attenuates the algorithm aversion effects in automated wealth management.

Computers in Human Behavior, 148, Article 107872. https://doi.org/10.1016/j.

chb.2023.107872

Charness, N., Yoon, J. S., Souders, D., Stothart, C., & Yehnert, C. (2018). Predictors of

attitudes toward autonomous vehicles: The roles of age, gender, prior knowledge,

and personality. Frontiers in Psychology, 9, 2589. https://doi.org/10.3389/

fpsyg.2018.02589

Cristea, M., & Gheorghiu, A. (2016). Attitude, perceived behavioral control, and

intention to adopt risky behaviors. Transportation Research Part F: Trafc Psychology

and Behaviour, 43, 157–165. https://doi.org/10.1016/j.trf.2016.10.004

Deb, S., Strawderman, L. J., & Carruth, D. W. (2018). Investigating pedestrian

suggestions for external features on fully autonomous vehicles: A virtual reality

experiment. Transportation Research Part F: Trafc Psychology and Behaviour, 59,

135–149. https://doi.org/10.1016/j.trf.2018.08.016

Deb, S., Strawderman, L., Carruth, D. W., DuBien, J., Smith, B., & Garrison, T. M. (2017).

Development and validation of a questionnaire to assess pedestrian receptivity

toward fully autonomous vehicles. Transportation Research Part C: Emerging

Technologies, 84, 178–195. https://doi.org/10.1016/j.trc.2017.08.029

Dey, D., Martens, M., Eggen, B., & Terken, J. (2019). Pedestrian road-crossing willingness

as a function of vehicle automation, external appearance, and driving behaviour.

Transportation Research Part F: Trafc Psychology and Behaviour, 65, 191–205.

https://doi.org/10.1016/j.trf.2019.07.027

Diamantopoulos, A., Sarstedt, M., Fuchs, C., Wilczynski, P., & Kaiser, S. (2012).

Guidelines for choosing between multi-item and single-item scales for construct

measurement: A predictive validity perspective. Journal of the Academy of Marketing

Science, 40, 434–449. https://doi.org/10.1007/s11747-011-0300-3

Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015). Algorithm aversion: People

erroneously avoid algorithms after seeing them err. Journal of Experimental

Psychology: General, 144(1), 114. doi.org/psycnet.apa.org/doi/10.1037/

xge0000033.

Dijkstra, T. K., & Henseler, J. (2015). Consistent partial least squares path modeling. MIS

Quarterly, 39(2), 297–316.

Dommes, A., Douffet, B., Pala, P., Deb, S., & Grani´

e, M. A. (2024). Pedestrians’

receptivity to fully automated vehicles: Assessing the psychometric properties of the

PRQF and survey in France. Transportation Research Part F: Trafc Psychology and

Behaviour, 105, 163–181.

Dommes, A., Merlhiot, G., Lobjois, R., Dang, N.-T., Vienne, F., Boulo, J., Oliver, A.-H.,

Cretual, A., & Cavallo, V. (2021). Young and older adult pedestrians’ behavior when

crossing a street in front of conventional and self-driving cars. Accident Analysis &

Prevention, 159, Article 106256. https://doi.org/10.1016/j.aap.2021.106256

Draws, T., Rieger, A., Inel, O., Gadiraju, U., & Tintarev, N. (2021). A checklist to combat

cognitive biases in crowdsourcing. Proceedings of the AAAI Conference on Human

Computation and Crowdsourcing, 9, 48–59. https://doi.org/10.1609/hcomp.

v9i1.18939

Egala, S. B., & Liang, D. (2023). Algorithm aversion to mobile clinical decision support

among clinicians: A choice-based conjoint analysis. European Journal of Information

Systems, 1–17. https://doi.org/10.1080/0960085X.2023.2251927

Eisele, D., & Petzoldt, T. (2024). Effects of a frontal brake light on pedestrians’

willingness to cross the street. Transportation Research Interdisciplinary Perspectives,

23, Article 100990.

Faas, S., Kraus, J., Schoenhals, A., & Baumann, M. (2021). Calibrating pedestrians’ trust

in automated vehicles: Does an intent display in an external HMI support trust

calibration and safe crossing behavior?. Proceedings of the 2021 CHI conference on

human factors in computing systems. https://doi.org/10.1145/3411764.3445738

Fagnant, D. J., & Kockelman, K. (2015). Preparing a nation for autonomous vehicles:

Opportunities, barriers and policy recommendations. Transportation Research Part A:

Policy and Practice, 77, 167–181. https://doi.org/10.1016/j.tra.2015.04.003

Feng, Z., Gao, Y., Zhu, D., Chan, H.-Y., Zhao, M., & Xue, R. (2024). Impact of risk

perception and trust in autonomous vehicles on pedestrian crossing decision:

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Navigating the social-technological intersection with the ICLV model. Transport

Policy, 152, 71–86.

Figner, B., & Weber, E. U. (2011). Who takes risks when and why? Determinants of risk

taking. Current Directions in Psychological Science, 20(4), 211–216. https://doi.org/

10.1177/0963721411415790

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to

theory and research. Addison-Wesley.

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with

unobservable variables and measurement error. Journal of Marketing Research, 18(1),

39–50. https://doi.org/10.1177/002224378101800104

Forster, Y., Kraus, J., Feinauer, S., & Baumann, M. (2018). Calibration of trust

expectancies in conditionally automated driving by brand, reliability information

and introductionary videos: An online study. Proceedings of the 10th international

conference on automotive user interfaces and interactive vehicular applications. https://

doi.org/10.1145/3239060.3239070

Garidis, K., Ulbricht, L., Rossmann, A., & Schm¨

ah, M. (2020). Toward a user acceptance

model of autonomous driving. Proceedings of the 53rd Hawaii international conference

on system sciences (HICSS).

Gefen, D. (2000). E-Commerce: The role of familiarity and trust. Omega, 28(6), 725–737.

https://doi.org/10.1016/S0305-0483(00)00021-9

Hagenzieker, M. P., Van Der Kint, S., Vissers, L., Van Schagen, I. N. L. G., De Bruin, J.,

Van Gent, P., & Commandeur, J. J. F. (2020). Interactions between cyclists and

automated vehicles: Results of a photo experiment. Journal of Transportation Safety &

Security, 12(1), 94–115. https://doi.org/10.1080/19439962.2019.1591556

Hair, J., Hollingsworth, C. L., Randolph, A. B., & Chong, A. Y. L. (2017). An updated and

expanded assessment of PLS-SEM in information systems research. Industrial

Management & Data Systems, 117(3), 442–458. https://doi.org/10.1108/IMDS-04-

2016-0130

Hair, J., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2022). A primer on partial least

squares structural equation modeling (PLS-SEM) (3rd ed.). Sage Publications.

Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to

report the results of PLS-SEM. European Business Review, 31(1), 2–24. https://doi.

org/10.1108/EBR-11-2018-0203

Hannon, O., Gal, U., & Dar-Nimrod, I. (2022). Aversion vs. Abstinence: Conceptual

distinctions for the receptivity toward algorithmic decision-making systems within

value-laden contexts. Proceedings of australasian conference on information systems

(ACIS).

Harkin, A. M., Mangold, A., Harkin, K. A., & Petzoldt, T. (2024). Implicit communication

in cyclist-vehicle interaction: Examining the inuence of driving dynamics in

interactions with turning (automated) vehicles on cyclists’ perceived safety,

behavioral intention, and risk anticipation. Journal of Cycling and Micromobility

Research, 2, Article 100028.

Hegner, S. M., Beldad, A. D., & Brunswick, G. J. (2019). In automatic we trust:

Investigating the impact of trust, control, personality characteristics, and extrinsic

and intrinsic motivations on the acceptance of autonomous vehicles. International

Journal of Human-Computer Interaction, 35(19), 1769–1780. https://doi.org/

10.1080/10447318.2019.1572353

Hengstler, M., Enkel, E., & Duelli, S. (2016). Applied articial intelligence and trust—the

case of autonomous vehicles and medical assistance devices. Technological

Forecasting and Social Change, 105, 105–120. https://doi.org/10.1016/j.

techfore.2015.12.014

Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing

discriminant validity in variance-based structural equation modeling. Journal of the

Academy of Marketing Science, 43, 115–135. https://doi.org/10.1007/s11747-014-

0403-8

Hoff, K. A., & Bashir, M. (2015). Trust in automation: Integrating empirical evidence on

factors that inuence trust. Human Factors, 57(3), 407–434.

Hong, A., Nam, C., & Kim, S. (2020). What will be the possible barriers to consumers’

adoption of smart home services? Telecommunications Policy, 44(2), Article 101867.

https://doi.org/10.1016/j.telpol.2019.101867

Hudson, J., Orviska, M., & Hunady, J. (2019). People’s attitudes to autonomous vehicles.

Transportation Research Part A: Policy and Practice, 121, 164–176. https://doi.org/

10.1016/j.tra.2018.08.018

Hulse, L. M., Xie, H., & Galea, E. R. (2018). Perceptions of autonomous vehicles:

Relationships with road users, risk, gender and age. Safety Science, 102, 1–13.

https://doi.org/10.1016/j.ssci.2017.10.001

Iranmanesh, M., Zailani, S., Moeinzadeh, S., & Nikbin, D. (2017). Effect of green

innovation on job satisfaction of electronic and electrical manufacturers’ employees

through job intensity: Personal innovativeness as moderator. Review of Managerial

Science, 11, 299–313. https://doi.org/10.1007/s11846-015-0184-6

Jayaraman, S. K., Creech, C., Tilbury, D. M., Yang, X. J., Pradhan, A. K., Tsui, K. M., &

Robert Jr, L. P. (2019). Pedestrian trust in automated vehicles: Role of trafc signal

and AV driving behavior. Frontiers in Robotics and AI, 6, 117. https://doi.org/

10.3389/frobt.2019.00117

Karsch, H. M., Hedlund, J. H., Tison, J., & Leaf, W. A. (2012). Review of studies on

pedestrian and bicyclist safety, 1991–2007. United States: National Highway Trafc

Safety Administration. https://doi.org/10.21949/1525710

Kaur, K., & Rampersad, G. (2018). Trust in driverless cars: Investigating key factors

inuencing the adoption of driverless cars. Journal of Engineering and Technology

Management, 48, 87–96. https://doi.org/10.1016/j.jengtecman.2018.04.006

Kenesei, Z., ´

Asv´

anyi, K., K¨

ok´

eny, L., J´

aszber´

enyi, M., Miskolczi, M., Gyulav´

ari, T., &

Syahrivar, J. (2022). Trust and perceived risk: How different manifestations affect

the adoption of autonomous vehicles. Transportation Research Part A: Policy and

Practice, 164, 379–393. https://doi.org/10.1016/j.tra.2022.08.022

Khastgir, S., Birrell, S., Dhadyalla, G., & Jennings, P. (2018). Calibrating trust through

knowledge: Introducing the concept of informed safety for automation in vehicles.

Transportation Research Part C: Emerging Technologies, 96, 290–303. https://doi.org/

10.1016/j.trc.2018.07.001

K¨

orber, M. (2019). Theoretical considerations and development of a questionnaire to

measure trust in automation. Proceedings of the 20th congress of the international

ergonomics association (IEA). https://doi.org/10.1007/978-3-319-96074-6_2

Krueger, R., Rashidi, T. H., & Rose, J. M. (2016). Preferences for shared autonomous

vehicles. Transportation Research Part C: Emerging Technologies, 69, 343–355. https://

doi.org/10.1016/j.trc.2016.06.015

Krügel, S., Ostermaier, A., & Uhl, M. (2023). Algorithms as partners in crime: A lesson in

ethics by design. Computers in Human Behavior, 138, Article 107483.

Lazanyi, K., & Maraczi, G. (2017). Dispositional trust—do we trust autonomous cars?.

IEEE 15th international symposium on intelligent systems and informatics (SISY).

https://doi.org/10.1109/SISY.2017.8080540, 2017.

Lee, H. Y., Qu, H., & Kim, Y. S. (2007). A study of the impact of personal innovativeness

on online travel shopping behavior—a case study of Korean travelers. Tourism

Management, 28(3), 886–897. https://doi.org/10.1016/j.tourman.2006.04.013

Li, X., Afghari, A. P., Oviedo-Trespalacios, O., Kaye, S.-A., & Haworth, N. (2023). Cyclists

perception and self-reported behaviour towards interacting with fully automated

vehicles. Transportation Research Part A: Policy and Practice, 173, Article 103713.

https://doi.org/10.1016/j.tra.2023.103713

Liljamo, T., Liimatainen, H., & P¨

oll¨

anen, M. (2018). Attitudes and concerns on automated

vehicles. Transportation Research Part F: Trafc Psychology and Behaviour, 59, 24–44.

https://doi.org/10.1016/j.trf.2018.08.010

Liu, P., Yang, R., & Xu, Z. (2019). Public acceptance of fully automated driving: Effects of

social trust and risk/benet perceptions. Risk Analysis, 39(2), 326–341. https://doi.

org/10.1111/risa.13143

Mahmud, H., Islam, A. K. M. N., Ahmed, S. I., & Smolander, K. (2022). What inuences

algorithmic decision-making? A systematic literature review on algorithm aversion.

Technological Forecasting and Social Change, 175, Article 121390. https://doi.org/

10.1016/j.techfore.2021.121390

McKnight, D. H., Choudhury, V., & Kacmar, C. (2002). The impact of initial consumer

trust on intentions to transact with a web site: A trust building model. The Journal of

Strategic Information Systems, 11(3–4), 297–323. https://doi.org/10.1016/S0963-

8687(02)00020-3

Meertens, R. M., & Lion, R. (2008). Measuring an individual’s tendency to take risks: The

risk propensity scale 1. Journal of Applied Social Psychology, 38(6), 1506–1520.

https://doi.org/10.1111/j.1559-1816.2008.00357.x

Merritt, S. M., Heimbaugh, H., LaChapell, J., & Lee, D. (2013). I trust it, but I don’t know

why: Effects of implicit attitudes toward automation on trust in an automated

system. Human Factors, 55(3), 520–534. https://doi.org/10.1177/

0018720812465081

Merritt, S. M., & Ilgen, D. R. (2008). Not all trust is created equal: Dispositional and

history-based trust in human-automation interactions. Human Factors, 50(2),

194–210. https://doi.org/10.1518/001872008X288574

Miskolczi, M., F¨

oldes, D., Munk´

acsy, A., & J´

aszber´

enyi, M. (2021). Urban mobility

scenarios until the 2030s. Sustainable Cities and Society, 72, Article 103029. https://

doi.org/10.1016/j.scs.2021.103029

Muir, B. M. (1994). Trust in automation: Part I. Theoretical issues in the study of trust

and human intervention in automated systems. Ergonomics, 37(11), 1905–1922.

https://doi.org/10.1080/00140139408964957

Muir, B. M., & Moray, N. (1996). Trust in automation. Part II. Experimental studies of

trust and human intervention in a process control simulation. Ergonomics, 39(3),

429–460. https://doi.org/10.1080/00140139608964474

O’Leary-Kelly, S. W., & Vokurka, R. J. (1998). The empirical assessment of construct

validity. Journal of Operations Management, 16(4), 387–405. https://doi.org/

10.1016/S0272-6963(98)00020-5

Osswald, S., Wurhofer, D., Tr¨

osterer, S., Beck, E., & Tscheligi, M. (2012). Predicting

information technology usage in the car: Towards a car technology acceptance

model. Proceedings of the 4th international conference on automotive user interfaces and

interactive vehicular applications. https://doi.org/10.1145/2390256.2390264

Paden, B., ˇ

C´

ap, M., Yong, S. Z., Yershov, D., & Frazzoli, E. (2016). A survey of motion

planning and control techniques for self-driving urban vehicles. IEEE Transactions on

Intelligent Vehicles, 1(1), 33–55. https://doi.org/10.1109/TIV.2016.2578706

Palan, S., & Schitter, C. (2018). Prolic. ac—a subject pool for online experiments.

Journal of Behavioral and Experimental Finance, 17, 22–27. https://doi.org/10.1016/j.

jbef.2017.12.004

Palmeiro, A. R., van der Kint, S., Vissers, L., Farah, H., de Winter, J. C. F., &

Hagenzieker, M. (2018). Interaction between pedestrians and automated vehicles: A

wizard of oz experiment. Transportation Research Part F: Trafc Psychology and

Behaviour, 58, 1005–1020. https://doi.org/10.1016/j.trf.2018.07.020

Papadimitriou, E., Schneider, C., Tello, J. A., Damen, W., Vrouenraets, M. L., & Ten

Broeke, A. (2020). Transport safety and human factors in the era of automation:

What can transport modes learn from each other? Accident Analysis & Prevention,

144, Article 105656. https://doi.org/10.1016/j.aap.2020.105656

Peer, E., Brandimarte, L., Samat, S., & Acquisti, A. (2017). Beyond the Turk: Alternative

platforms for crowdsourcing behavioral research. Journal of Experimental Social

Psychology, 70, 153–163. https://doi.org/10.1016/j.jesp.2017.01.006

Penmetsa, P., Adanu, E. K., Wood, D., Wang, T., & Jones, S. L. (2019). Perceptions and

expectations of autonomous vehicles–A snapshot of vulnerable road user opinion.

Technological Forecasting and Social Change, 143, 9–13. https://doi.org/10.1016/j.

techfore.2019.02.010

Rad, S. R., de Almeida Correia, G. H., & Hagenzieker, M. (2020). Pedestrians’ road

crossing behaviour in front of automated vehicles: Results from a pedestrian

simulation experiment using agent-based modelling. Transportation Research Part F:

Trafc Psychology and Behaviour, 69, 101–119. https://doi.org/10.1016/j.

trf.2020.01.014

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Rahman, M. M., Deb, S., Strawderman, L., Burch, R., & Smith, B. (2019). How the older

population perceives self-driving vehicles. Transportation Research Part F: Trafc

Psychology and Behaviour, 65, 242–257. https://doi.org/10.1016/j.trf.2019.08.002

Rahman, M. T., Dey, K., Das, S., & Shernski, M. (2021). Sharing the road with

autonomous vehicles: A qualitative analysis of the perceptions of pedestrians and

bicyclists. Transportation Research Part F: Trafc Psychology and Behaviour, 78,

433–445. https://doi.org/10.1016/j.trf.2021.03.008

Reig, S., Norman, S., Morales, C. G., Das, S., Steinfeld, A., & Forlizzi, J. (2018). A eld

study of pedestrians and autonomous vehicles. Proceedings of the 10th international

conference on automotive user interfaces and interactive vehicular applications. https://

doi.org/10.1145/3239060.3239064

Renier, L. A., Mast, M. S., & Bekbergenova, A. (2021). To err is human, not

algorithmic–Robust reactions to erring algorithms. Computers in Human Behavior,

124, Article 106879. https://doi.org/10.1016/j.chb.2021.106879

SAE International. (2021). Taxonomy and denitions for terms related to driving automation

systems for on-road motor vehicles. J3016 Standard, April 2021 (pp. 1–41) htt

ps://www.sae.org/standards/content/j3016_202104/.

Schoettle, B., & Sivak, M. (2014). A survey of public opinion about autonomous and self-

driving vehicles in the US, the UK, and Australia. Ann Arbor, Transportation Research

Institute: University of Michigan.

Sharma, P. N., Liengaard, B. D., Hair, J. F., Sarstedt, M., & Ringle, C. M. (2022).

Predictive model assessment and selection in composite-based modeling using PLS-

SEM: Extensions and guidelines for using CVPAT. European Journal of Marketing, 57

(6), 1662–1677. https://doi.org/10.1108/EJM-08-2020-0636

Shmueli, G., Sarstedt, M., Hair, J. F., Cheah, J.-H., Ting, H., Vaithilingam, S., &

Ringle, C. M. (2019). Predictive model assessment in PLS-SEM: Guidelines for using

PLSpredict. European Journal of Marketing, 53(11), 2322–2347. https://doi.org/

10.1108/EJM-02-2019-0189

Stilgoe, J., & Cohen, T. (2021). Rejecting acceptance: Learning from public dialogue on

self-driving vehicles. Science and Public Policy, 48(6), 849–859. https://doi.org/

10.1093/scipol/scab060

Story, V., O’Malley, L., & Hart, S. (2011). Roles, role performance, and radical innovation

competences. Industrial Marketing Management, 40(6), 952–966. https://doi.org/

10.1016/j.indmarman.2011.06.025

Taherdoost, H. (2018). A review of technology acceptance and adoption models and

theories. Procedia Manufacturing, 22, 960–967. https://doi.org/10.1016/j.

promfg.2018.03.137

Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of

information technology: Toward a unied view. MIS Quarterly, 36(1), 425–478.

https://doi.org/10.2307/30036540

WHO. (2023). Road trafc injuries. https://www.who.int/news-room/fact-sheets

/detail/road-traffic-injuries.

Williams, M. D., Rana, N. P., & Dwivedi, Y. K. (2015). The unied theory of acceptance

and use of technology (UTAUT): A literature review. Journal of Enterprise Information

Management, 28(3), 443–488. https://doi.org/10.1108/JEIM-09-2014-0088

Yokoi, R. (2024). Trust in self-driving vehicles is lower than in human drivers when both

drive almost perfectly. Transportation Research Part F: Trafc Psychology and

Behaviour, 103, 1–17.

Zhou, S., Sun, X., Liu, B., & Burnett, G. (2022). Factors affecting pedestrians’ trust in

automated vehicles: Literature review and theoretical model. IEEE Transactions on

Human-Machine Systems, 52(3), 490–500. https://doi.org/10.1109/

THMS.2021.3112956

P.A. Busch

Computers in Human Behavior Reports 16 (2024) 100490

Podcast about the article

Data

October 2024

Peter André Busch

Download

Investigating Robotaxi Crash Severity Using Geographical Random Forest

Preprint

Full-text available

May 2025

This paper quantitatively investigates the crash severity of Autonomous Vehicles (AVs) with spatially localized machine learning and macroscopic measures of the urban built environment. We address spatial heterogeneity and spatial autocorrelation, while focusing on land use patterns and human behavior. Our Geographical Random Forest (GRF) model, accompanied with a crash severity risk map of San Francisco, presents three findings that are useful for commercial operations of AVs and robotaxis. First, spatially localized machine learning performed better than regular machine learning, when predicting AV crash severity. Bias-variance tradeoff was evident as we adjust the localization weight hyperparameter. Second, land use was the most important built environment measure, compared to intersections, building footprints, public transit stops, and Points Of Interests (POIs). Third, it was predicted that city center areas with greater diversity and commercial activities were more likely to result in low-severity AV crashes, than residential neighborhoods. Residential land use may be associated with higher severity due to human behavior and less restrictive environment. This paper recommends to explicitly consider geographic locations, and to design safety measures specific to residential neighborhoods, when robotaxi operators train their AV systems.

Research on the evaluation system of autonomous vehicle driving behavior based on riding experience

Conference Paper

Apr 2025

Steering decisions: Exploring the impact of perceived value and perceived risk on consumer intentions for adopting driverless cars in China

Article

Feb 2025
TRANSPORT RES F-TRAF

Trust in self-driving vehicles is lower than in human drivers when both drive almost perfectly

Article

Full-text available

Apr 2024
TRANSPORT RES F-TRAF

Ryosuke Yokoi

Studies have investigated the determinants of trust in self-driving vehicles (SDVs) and confirmed that the ability to execute the driving task flawlessly is essential to promote trust. However, little is known about the extent to which errors decrease trust in SDVs. This study conducted four experiments (N = 2221) and tested whether people’s trust in SDVs was lower than that in human drivers when they made errors without causing negative events. In Experiments 1 and 2, which manipulated the driving accuracy of the drivers, the participants checked nine different pieces of information that showed accuracy. The results demonstrated that the SDV was less trusted than humans only when there was a slight possibility of making an error. This study did not identify factors explaining lower trust in the SDV. Experiments 3 and 4 consisted of participants watching videos showing that the SDV and human driver made minor errors, such as taking a long time to park. This study showed that the minor errors largely reduced trust, regardless of whether the vehicle was self-driven or driven by humans. These findings imply that errors without describing severe accidents are less likely to cause a gap in trust between SDVs and human drivers.

Effects of a frontal brake light on pedestrians' willingness to cross the street

Article

Full-text available

Dec 2023

Effects of a frontal brake light (FBL, a potential external human–machine interface for automated vehicles) on participants’ self-reported willingness to cross a vehicle’s path were investigated. In a mixed design online study (vehicles in the experimental group were equipped with FBLs, there were no FBLs in the control group), participants observed videos of a vehicle approaching at different speeds from the perspective of a pedestrian standing at the curb. The vehicles exhibited either yielding behavior (braking onset 55 m or 32 m before standstill in front of the pedestrian ’s position) or non-yielding behavior (approach speed was maintained). Participants specified their willingness to cross the vehicle’s path at different distances. When the vehicle yielded (i.e., FBL was activated), willingness to cross was significantly higher in the experimental group than the control group. Notably, we further observed a significantly lower willingness to cross in the experimental group than the control group when the vehicle did not yield (i.e., FBL was deactivated). Novel external human–machine interfaces might therefore influence the interaction with vehicles not only when they are activated but also when they are deactivated.

Aversion vs. Abstinence: Conceptual Distinctions for the Receptivity Toward Algorithmic Decision-Making Systems Within Value-laden Contexts

Conference Paper

Full-text available

Dec 2022

Whilst algorithmic decision-making systems (ADMS) become increasingly pertinent across several contexts, many remain reluctant to adopt such systems, preferring human alternatives – often explored as “Algorithm Aversion”. However, the associated literature primarily frames this tendency in a utility-focused fashion, based on users’ perceptions of efficacy or accuracy. This framing offers a narrow scope of “aversion” that neglects emotional and experiential elements that may be at play, as well as especially prominent in “value-laden contexts” (e.g., medicine). This study uses an inductive approach to identifying various concepts and themes emerging from open-ended responses to the potential use of a future ADMS in such a context. Different reactions (both reluctant and receptive) of ADMS are then discussed and offered conceptual distinctions that may inform future examinations of the resulting biases. In doing so, we start to respond to the call for qualitative research examining the underlying motives related to Algorithm Aversion.

Algorithm aversion to mobile clinical decision support among clinicians: a choice-based conjoint analysis

Article

Full-text available

Aug 2023

Cyclists perception and self-reported behaviour towards interacting with fully automated vehicles

Article

Full-text available

May 2023
TRANSPORT RES A-POL

Fully automated vehicles (FAVs) have the potential to improve road safety and reduce traffic congestion and emissions. Most studies of acceptance of FAVs have focused on motor vehicle users, largely ignoring other road users, such as cyclists. This study investigates the factors that influence cyclists' receptivity towards sharing roads with FAVs and their behavioural intentions in interactions with FAVs. The online survey collected information on participant demographics (e.g. age, gender, crash experience), self-reported on-road cycling behaviours (e.g. violations, errors, positive behaviours) and their receptivity towards sharing roads with FAVs (e.g. attitude, social norms, trust). Three typical cyclist-vehicle interaction scenarios were presented to test the cyclists' intention to engage in self-protective behaviours (e.g. giving a hand signal, giving way or moving over) during the interaction with a FAV. Three hundred and fourteen Australian adults (106 females vs 208 males) who had ridden a bicycle at least once in the past year completed the survey. The results show that older cyclists and male cyclists had a lower receptivity towards sharing roads with FAVs than younger cyclists and female cyclists, respectively. Cyclists who reported being involved in a bicycle crash in the last two years and those who reported committing more errors on roads were more willing to share roads with FAVs. Cyclists who had a higher propensity to risky behaviours and positive behaviours were less likely to take intended self-protective behaviours during interaction with FAVs. Findings of the study provide some insights from the cyclist's perspective to facilitate the development and implementation of automated vehicles.

Pedestrians’ receptivity to fully automated vehicles: Assessing the psychometric properties of the PRQF and survey in France

Article

Aug 2024
TRANSPORT RES F-TRAF

Implicit Communication in Cyclist-Vehicle Interaction: Examining the Influence of Driving Dynamics in Interactions with Turning (Automated) Vehicles on Cyclists' Perceived Safety, Behavioral Intention, and Risk Anticipation

Article

May 2024

Successful and safe management of interactions between cyclists and motorized vehicles often includes implicit communication, such as vehicle motion signals. While the introduction of automated vehicles (AVs) is anticipated to increase safety for cyclists by eliminating human error, it is still unclear how they should communicate implicitly in complex traffic scenarios. The turning maneuver constitutes a complex and ambiguous scenario, as it involves vehicle deceleration even when there is no intention to halt for the cyclist, proceeding straight ahead. A video-based laboratory study incorporating a bicycle setup examined the influence of four driving dynamics and automation status on cyclists' risk anticipation, perceived safety, and intention to continue to cycle through an intersection. Forty-two participants viewed videos of an oncoming vehicle making a left turn at an intersection from the perspective of a cyclist traveling straight through the intersection. The results indicated that participants perceived early and continuous braking of the vehicle ("passive yield") as the most favorable approach (reduced risk anticipation, higher perceived safety, and greater intention to continue to cycle through the intersection), even when compared to later but stronger braking ("active yield"). The objectively riskiest maneuver that would result in a collision ("collision") was evaluated as the least desirable. A rule-violating yet objectively safe maneuver (accelerating to exit the conflict area before the cyclist; "no yield") led to increased risk anticipation but not to lower perceived safety or reduced intention to continue to cycle through the intersection compared to active yield. The displayed automation status had minimal influence overall, although participants in the passive yield condition expressed greater uncertainty regarding their intention to continue to cycle through the intersection in interactions with AVs compared to MVs. It appears that natural driving dynamics exhibited by "good" human drivers could represent a promising approach to ensuring the safety and comfort of cyclists in interactions with AVs. Here, early and continuous braking is preferable to later and stronger braking.

Impact of risk perception and trust in autonomous vehicles on pedestrian crossing decision: Navigating the social-technological intersection with the ICLV model

Article

May 2024
TRANSPORT POLICY

Conservatives endorse Fintech? Individual regulatory focus attenuates the algorithm aversion effects in automated wealth management

Article

Jul 2023
COMPUT HUM BEHAV

A Checklist to Combat Cognitive Biases in Crowdsourcing

Article

Oct 2021

Recent research has demonstrated that cognitive biases such as the confirmation bias or the anchoring effect can negatively affect the quality of crowdsourced data. In practice, however, such biases go unnoticed unless specifically assessed or controlled for. Task requesters need to ensure that task workflow and design choices do not trigger workers’ cognitive biases. Moreover, to facilitate the reuse of crowdsourced data collections, practitioners can benefit from understanding whether and which cognitive biases may be associated with the data. To this end, we propose a 12-item checklist adapted from business psychology to combat cognitive biases in crowdsourcing. We demonstrate the practical application of this checklist in a case study on viewpoint annotations for search results. Through a retrospective analysis of relevant crowdsourcing research that has been published at HCOMP in 2018, 2019, and 2020, we show that cognitive biases may often affect crowd workers but are typically not considered as potential sources of poor data quality. The checklist we propose is a practical tool that requesters can use to improve their task designs and appropriately describe potential limitations of collected data. It contributes to a body of efforts towards making human-labeled data more reliable and reusable.

Non-user Acceptance of Autonomous Technology: A Survey of Bicyclist Receptivity to Fully Autonomous Vehicles

Figures

Supplementary resource (1)

Recommended publications

Non-User Acceptance of Autonomous Technology: A Survey of Bicyclist Receptivity to Fully Autonomous...

What influences vulnerable road users’ perceptions of autonomous vehicles? A comparative analysis of...

Examining pedestrians' trust in automated vehicles based on attributes of trust: A qualitative study

Users’ acceptance of private automated vehicles: A systematic review and meta-analysis