Introduction

In the Internet era, consumers can purchase all the goods they need by accessing different online shopping platforms or shopping APPs, the whole shopping process is simple and convenient, and the merchants who receive the orders will immediately organize personnel to deliver so that users can complete the entire shopping process without leaving home (Wang et al. 2018). Since e-commerce enterprises sell varieties and quantities of products in the online marketplace and provide conveniences for many consumers, this delivery market is rapidly growing. According to CNNIC (2022), China’s online shopping users have reached 841 million as of June 2022, accounting for 80% of the total number of Internet users, and e-commerce sales reached $6.3 trillion, a year-on-year increase of 3.1%. Due to the expanding market for online shopping, the delivery market is also growing (Tan et al. 2020; Wang et al. 2019), especially in the form of direct-to-consumer package delivery. According to the prediction of theWorld Economic Forum, compared to the current last-mile delivery volume, it is expected to increase by about 78% by 2030, and the vehicles required for last-mile delivery will also increase by 36%. According to the National Bureau of Statistics in China, the volume of express delivery business has rapidly increased from 9.19 billion in 2013 to 132 billion in 2023 in the past few years. The huge delivery volume has brought tremendous pressure to logistics and delivery enterprises.

To meet customers’ requirements for faster and higher delivery quality than traditional delivery methods, i.e. trucks. Online retail companies are attempting to apply intelligent logistics technologies to improve delivery services. Among the various innovation technologies emerging in the market, some online retail firms, i.e. Alibaba, JD, and Meituan, are attempting to apply autonomous delivery vehicles (ADVs) for delivery services to enhance delivery efficiency and reduce operating costs (Wu and Lin 2018). Statistics also show that in the past five years, the average annual growth of logistics robots exceeds 20%, by 2025, the expected market value of logistics robots is expected to be $127.39 billion, the future of this track will undoubtedly be very imaginative space (Liu 2023). ADVs have displayed many advantages over traditional delivery modes, such as higher delivery efficiency, lower maintenance costs, and energy-saving and environmentally-friendly characteristics (Kaur and Rampersad 2018; Manfreda et al. 2019; Tennant et al. 2019). ADVs integrate mainstream deep learning computing power, powerful general purpose computing power, and interface customization capabilities to adapt to complex user needs, providing computing power and interface support for autonomous driving perception, decision-making, and control (Tennant et al. 2019). Users can make an appointment to select delivery time and location through their mobile phone autonomously. After ADVs arrive, consumers are notified to complete deliveries via phone, the delivery can be completed by entering the pick-up code (Chen et al. 2021; Rai et al. 2022). ADVs enable fast and on-time delivery services with virtually no human error during delivery, and customers can enjoy multiple deliveries within a short period. With the use of ADVs, logistics service providers will be able to save up to 30% of delivery costs (GoFurther, 2023). The value of ADVs in providing competent delivery services, improving delivery efficiency, and developing community e-commerce is becoming increasingly significant. Moreover, ADVs are expected to have a broader impact on sustainability and provide long-term environmental benefits, such as improved road efficiency and safety, lower energy consumption and CO2 emissions, and relief from last-mile delivery issues (Butler et al. 2021; Chen et al. 2020; Kyriakidis et al. 2015; Shariff et al. 2017; Wang and Zhao 2019).

In addition, the usage of ADVs can greatly relieve the stress of last-mile delivery during the COVID-19 pandemic. In the case of the severe shortage of delivery force during the epidemic in Shanghai, the emergence of ADVs greatly relieved the pressure on community delivery personnel (Tan et al. 2020). To ease the pressure of last-mile delivery, several enterprises, including JD and Meituan, deployed more than 300 ADVs from across the country to serve communities, hospitals, temporary treatment centers, and other places with an urgent demand for logistical capacity. Another example, in the case of Guangzhou, many self-driving vehicles helped distribute materials and other goods to community residents, greatly easing pressure brought by manual operation (China News Network 2022). The launch of ADVs effectively mitigated the spread of the virus and significantly improved the efficiency of last-mile delivery services(Cugurullo et al. 2020; Di et al. 2020). As such, the delivery services provided by ADVs played a positive role in alleviating delivery pressure and effectively reducing the adverse impact of the epidemic (Chen et al. 2020; Tan et al. 2020; Zhang et al. 2020). However, many consumers perceive potential risks and hold skeptical attitudes toward this new technology (Kapser and Abdelrahman, 2020). Consumers have concerns about thse autonomous vehicles, for example, privacy invasion and misfunction of ADVs on public roads may prevent potential users from touching this technology (Kapser et al. 2021). Therefore, COVID-19 risk is introduced to investigate consumers’ ADVs adoption behaviors.

Although many studies have examined the factors that affected customer acceptance and adoption behaviors of innovative technologies, such as autonomous vehicles (Gill, 2020; Manfreda et al. 2019; Raj et al. 2020; Wang and Zhao 2019), limited studies on autonomous delivery services provided by ADVs have been conducted. Like autonomous vehicles (AVs), ADVs have huge economic and environmental advantages, such as the use of ADVs can improve consumer performance expectancy and reduce CO2 emissions and energy consumption (Kapser and Abdelrahman, 2020; Srinivas et al. 2022). The adoption of ADVs can be considered as a pre-environmental behavior, therefore we refer to the study of He and Zhan (2018), from the pre-environmental perspective, and apply the normative activation model (NAM) and add awareness of consequence and ascription of resbonsibility to explore the determinants consumers’ attitude and intention to use ADVs.

Perceived enjoyment plays a very important role and affects the continuous use of new technology (Natarajan et al. 2017; Nordhoff et al. 2020; Ribeiro et al. 2022). Perceived enjoyment, as defined by Natarajan et al. (2017), means the degree of pleasure that consumers feel when using an emegring technology. Some studies equate perceived enjoyment with hedonic motivation and reveal that perceived enjoyment positively affects consumers’ attitudes and adoption behaviors toward emerging technologies, i.e. autonomous vehicles (Nordhoff et al. 2020; Ribeiro et al. 2022). ADVs offer a range of interesting experiences when interacting with consumers, the high degree of automation enables consumers to have a better delivery experience compared to traditional delivery methods, which in turn generates a sense of entertainment and desire to try again. Therefore in our study, we introduce perceived enjoyment to explore consumers’ intention to use ADVs.

Consumer adoption behavior is easily influenced by word-of-mouth recommendation (WOMR) (Duarte et al. 2018). According to a large number of studies, an individual’s adoption behavior is easily influenced by others within a social circle (Kapser et al. 2021). Once a consumer has a good experience with an emerging technology, he may share the whole process through his social circle, and others in the social circle may also try and adopt it. WOMR will affect consumers’ purchasing behavior and adoption behaviors (Berger, 2014; Duarte et al. 2018). Meanwhile, WOMR is also easily affected by other factors, such as consumers’ attitudes. If consumers have a negative evaluation of a certain technology, they will not have recommendation intention and adoption behaviors (Yuen et al. 2022). At the same time, consumers are susceptible to the celebrity effect or fear of being isolated, so they respond to mainstream views within social circles and generate recommendation behaviors (Nian et al. 2021). Then we suppose subjective norm affects consumers’ WOMR. Consumers are easily affected by external conditions. For example, when necessary infrastructures are provided to facilitate the operation of ADVs, consumers with good experience will produce recommendation behaviors, and we suppose perceived behavioral control affects consumers’ word-of-mouth marketing. There are also other factors, such as self-promotion (Nian et al. 2021), altruism (Nian et al. 2021), personal experience (Berger 2014; Duarte et al. 2018), and other considerations causing individuals to engage in recommendation behaviors. In the study, we mainly explore how the TPB constructs (i.e. attitude, subjective norm, and perceived behavioral control) affect WOMR and the impact of WOMR on consumer intention to use ADVs.

Although scholars have used different technology acceptance models, such as the technology acceptance model (Abrams et al. 2021), unified theory of acceptance and use technology, namely UTAUT (AlKheder et al. 2023; Kapser and Abdelrahman, 2020; Kapser et al. 2021), theory of planned behavior (Yuen et al. 2022), task-technology fit (Koh and Yuen, 2023) and value-belief-norm model (Ju et al. 2023) to reveal consumers’ intention to use ADVs or robots. Our study integrates the theory of planned behavior (TPB) and normative activation model (NAM) into the research model to better explain consumer adoption of ADVs from a multi-theoretical and pre-environmental perspective. This paper also examines determinants of TPB constructs such as perceived enjoyment, perceived COVID-19 risk, awareness of consequences, and ascription of responsibility to better understand the internal influencing mechanism of consumer WOMR intention and adoption of ADVs. In addition, the study takes China, a developing country in Asia, as the background, and the findings of the study have certain reference and practical value to other developing countries.

The rest of the study follows. Section 2 reviews the revelant literature, Section 3 introduces the development of hypotheses. Section 4 presents the research methodology. Section 5 indicates the data analysis, including reliability and validity, common method variance (CMV), hypothesis testing, and direct, indirect, total effect, and mediation analysis. Section 6 details discussions and conclusions. Section 7 reveals the theoretical contributions and practical implications. Finally, section 8 outlines the limitations and directions for future research.

Literature review

This section mainly summarizes the previous research on autonomous delivery and ADVs and then introduces the planned behavior theory and normative activation model in detail.

Autonomous delivery

Autonomous vehicles are one of the most disruptive technologies in the transportation industry. Lots of scholars have conducted research and exploration in this field (Acheampong and Cugurullo 2019; Qian et al. 2023; Yuen et al. 2022). Madigan et al. (2016) surveyed users’ willingness to use self-driving road transport systems. In Chinese cities such as Shanghai, and Shenzhen, autonomous delivery cars have independently deliveried meals or parcels to those supply-disrupted communities (Tan et al. 2020). In response to the lack of drivers and the need for contactless delivery services, Nuro in the United States has developed two self-driving vehicles to deliver goods to local retail partners (Crowe 2020). Heimfarth et al. (2022) noted that autonomous delivery robots (ADRs) can travel at the same speed as a pedestrian to reach a destination specified by the consumer and complete the last-mile delivery. Kapser and Abdelrahman (2020) suggested that ADVs have gradually become an effective delivery model in the last-mile delivery field.

Autonomous delivery vehicles (ADVs)

ADVs present many advantages compared to traditional delivery tools, e.g., vans or trucks.ADVs can provide faster and more environmentally friendly delivery services reduce delivery pressure and improve delivery efficiency, further improving customer experience and satisfaction (Kapser and Abdelrahman 2020; Kapser et al. 2021). In addition, autonomous delivery cars can effectively avoid human contact during a pandemic, preventing the further spread of the epidemic, and greatly facilitating people’s lives, while reducing road congestion and environmental pollution (Alkheder et al. 2023; Lu et al. 2023). The use of ADVs can reduce the stress of deliverers, provide contactless delivery services, automatic optimization of routing, intelligent navigation to adapt to different scenarios, energy-saving, and significant environmental benefits, then in our study we briefly introduce the significant characteristics aforementioned.

Reduce the stress of deliverers

The growing business volume has brought great pressure to the last-mile delivery link, the most obvious change is the serious shortage of delivery capacity (Chen et al. 2018). Therefore, some logistics companies have turned their attention to autonomous delivery modes, ADVs can undertake delivery, pick-up, and other functions, but also can lighten the load for couriers during peak times, these couriers only need to load the goods into the ADVs, without going out in the sun and wind, effectively reducing the labor intensity of couriers and fills the reality of manpower shortage (Kapser and Abdelrahman 2020; Rai et al. 2022).

Provide contactless delivery services

Contactless delivery refers to placing goods in the location specified by consumers, such as the front desk of the company and the door of the house, reducing face-to-face contact and ensuring the contact risk and safety of users and couriers. In the context of epidemic prevention and control, the delivery services provided by ADVs can effectively prevent social distancing and the spread of the epidemic, and effectively slow down the inner panic of consumers (Kapser and Abdelrahman, 2020; Kim et al. 2021a, 2021b; Tan et al. 2020). Compared with the mode of courier brother to call the user and the user to collect, the delivery process of ADVs is more intelligent and unmanned.

Automatic optimization of routing

The routing problem of ADVs has been studied by a large number of scholars (Boysen et al. 2021; Shalamov et al. 2019), these studies pointed out that ADVs can automatically select optimal route and delivery time through algorithm learning and avoid congestion and peak times, improve delivery efficiency and accuracy, and realize a more intelligent logistics delivery (Contini and Farinelli 2021; Li et al. 2022a, 2022b). At the same time, considering the delivery of large items and customers who cannot accept delivery robots, a hybrid delivery method integrating trucks and delivery robots has also been proposed by lots of scholars (Boysen et al. 2021; Heimfarth et al. 2022), when studying vehicle routing ploblem (VRP), scholars used integer programming model (Alfandari et al. 2022), genetic algorithm (Li et al. 2022a, 2022b), heuristic algorithm (Chen et al. 2021a; Moradi et al. 2023; Ostermeier et al. 2023), artificial neural network, simulated annealing algorithm, etc., some scholars explored the optimization problem of delivery path under time window conditions (Mourad et al. 2021; Ostermeier et al. 2022; Xiao and Konak 2016). For example, Ostermeier et al. (2022) proposed a time-window-based routing optimization method for joint the last-mile delivery of trucks and robotic systems, which focuses on minimizing the number of deliveries while considering available robots. The results show that combined truck and robot delivery can reduce the cost of the last mile delivery by 68% compared to using only truck transportation. Chen et al. 2021a propose a new vehicle routing problem with a time window and delivery robot. With the help of the delivery robot, the delivery machine provides service while the driver serves nearby customers, which can shorten the service time and improve work efficiency.

Intelligent navigation to adapt to different scenarios

Intelligent navigation technology has a very important impact on the driving performance of ADVs. By using internal and external location recognition sensors, ADVs can realize such functions as optimal route search, obstacle avoidance, and accurate navigation for delivery robots. (Bae et al. 2020; Lee et al. 2021). Traditional delivery methods are easily affected by extreme weather, and autonomous vehicles applied to the last-mile delivery fields can achieve a safe, free, and efficient user experience no matter what extreme weather it is. For example, when driving on snowy roads, heavy snow may cause the camera to blur, and road signs and routes to be blacked. With strong environmental perception ability and self-planning and decision-making ability (Yoon et al. 2018), ADVs can carry out effective algorithm processing to achieve calm driving on snowy days.

Energy-saving and significant environmental benefits

The application of ADVs in last-mile delivery can significantly reduce operating costs, and CO2 emissions and improve environmental benefits (Jennings and Figliozzi 2019; Kemp et al. 2022; Li et al. 2020; Schlenther et al. 2020). For example, Jennings and Figliozzi (2019) went on to explore the findings of SADR (sidewalk autonomous delivery robots) and showed that SADR can save significant costs and time in some cases when compared to traditional delivery methods, greatly reduce the distance traveled per package. Kemp et al. (2022) confirmed that SADR is used for delivery with minimal CO2 emissions compared to other distribution methods. It also points out that halving shopping frequency can reduce greenhouse gas emissions by 44%. Schlenther et al. (2020) believe that the reduction of driver costs can reduce operators’ costs by about 74.5%. Siragusa et al. (2022) took Milan, Italy as an example and compared the last-mile delivery of electric vehicles(EVs) and internal combustion engine vehicles (ICEVs) from the economic and environmental perspectives. The results showed that the adoption of EVs for last-mile delivery could reduce greenhouse gas emissions by 17% (20 km per day) to 54%(120 km per day). The results of Reed et al. (2022) show that if the time spent searching for a parking space is ignored, the introduction of autonomous delivery robots (ADRs) will reduce delivery completion time by 0–33% for all customers. Figliozzi (2022) analyzed the performance of three types of autonomous vehicles (drones, sidewalk autonomous delivery robots (SADRs), and road autonomous delivery robots (RADRs)) in terms of vehicle miles traveled, energy consumption, and CO2 emissions. The results illustrate the significant environmental benefits of ADVs.

Consumer adoption of ADVs

A large number of technology acceptance models have been applied to the adoption behavior of ADVs by consumers, such as TAM (Abrams et al. 2021; Ganjipour and Edrisi, 2023; Yuen et al. 2022), TPB (Yuen et al. 2022), UTAUT (Kapser and Abdelrahman, 2020; Kim, 2021), and UTAUT2 (AlKheder et al. 2023; Kapser et al. 2021), DOI (Ganjipour and Edrisi, 2023; Lu et al. 2023). Kapser and Abdelrahman (2020) combined perceived risk and price sensitivity into the original UTAUT model while discarding habit and price value and changing price value to price sensitivity. Kapser et al. (2021) added innovativeness and trust in technology variables to the UTAUT2 model to explore adoption intention and further tested the moderating effect of gender between each variable and adoption intention. AlKheder et al. (2023) cited the UTAUT2 model and incorporated consumer acceptance to explore the willingness of Kuwaiti residents to adopt ADVs. In addition, Pröbster and Marsden (2023) used the Stereotype Content Model (SCM) to evaluate how different groups are perceived when using ADVs based on two dimensions: warmth and competence. Pani et al. (2020) investigated willingness to pay (WTP) based on ADRs delivery using a sample of 483 Polish customers, revealing the adoption behavior of different groups of consumers. Kim (2021) invited 76 participants to explore the impact of the anthropomorphism of delivery robots on the evaluation of delivery services by adult and adolescent consumers. The results showed that adolescent consumers have a higher evaluation of anthropomorphic delivery robots compared to non anthropomorphic delivery robots, while the impact of anthropomorphism on adult consumers is not significant, providing a reference for courier service providers to segment the market.

Meanwhile, some scholars use integrated models or incorporate situational variables to explore consumer adoption behavior towards ADVs. The Health Belief Model (HBM) is a model that changes people’s behavior by intervening in psychological activities such as perception, attitude, and belief and is widely applied in fields such as health education and health promotion activities. With the widespread spread of the epidemic in various regions, the health and life safety of many people are threatened. Therefore, some scholars have integrated the Health Belief Model (HBM) and traditional information technology adoption models such as TAM and TPB to explore consumer adoption behavior toward emerging technologies in the context of the pandemic (Koh and Yuen 2023; Yuen et al. 2022). For example, considering the impact of the epidemic, Yuen et al. (2022) combined TAM, HBM, and perceived trust to establish an integrated model to explore consumer adoption behavior of ADRs. Koh and Yuen (2023) combined the Health Trust Model (HBM) and Task Technology Fit (TTF) to establish a hybird model to explore the adoption of ADRs. The results showed that the constructs of the two theories had a significant impact on outcome expectation and task technology fit. And the outcome expection and task technology fit are powerful factors in predicting consumer willingness to use ADRs. Pettigrew et al. (2024) explored consumers’ perceptions of the self-service delivery service for a particular product (alcohol) through qualitative interviews and quantitative questionnaires. In addition to studying consumers who are willing to use or have used ADVs, the attitudes and opinions of pedestrians or observers (inCoP, incidentally co-present individuals) on the road also deserve attention (Abrams et al. 2021; de Groot 2019; Io and Lee 2019; Lee and Toombs 2020).

Compared with traditional delivery modes, ADVs have such functions as energy-saving and environmentally friendly. Ganjipour and Edrisi (2023) introduced the Normative activation model (NAM) and established an integrated model combining DOI and TAM to explore consumer willingness to adopt delivery robots. However, the drawback was that they did not consider the impact of social environment on ADVs operation. Therefore, this study combines the rational behavior model TPB with the NAM model to jointly explore consumer adoption of ADVs, enriching the last-mile delivery literature and consumer information technology acceptance and adoption area.

The theory of planned behavior(TPB)

The theory of planned behavior (TPB) has been widely used in the field of information technology and information system adoption (Kim et al. 2021a, 2021b; Wang et al. 2018; Wang and Zhao, 2019; Zhang et al. 2019) and pro-environmental behaviors (Liere and Dunlap 2010; Zhang et al. 2017). Studies have shown that TPB theory has good predictive and explanatory ability (Ajzen, 1991; Asif et al. 2022; Gkargkavouzi et al. 2019; Sarah et al. 2023). According to TPB theory, behavioral intention refers to the motivational factors that affect an individual’s behavior, indicating the extent to which an individual is willing to try a certain behavior and strive to achieve it (Morren and Grinstein 2021; Wan et al. 2022; Zhang et al. 2017). Generally, the stronger the intention to act, the more likely it is to take action (Ajzen 1991; Asif et al. 2022). The behavioral intention to use delivery services provided by ADVs refers to the effort consumers are willing to choose and pay after understanding the advantages and disadvantages of traditional delivery methods (i.e.trucks) and ADVs. Behavioral intention is influenced by behavioral attitude, subjective norms, and perceived behavioral control (Fishbein and Ajzen, 1975; Zhang et al. 2017). Attitude relates to the extent to which a person evaluates a certain behavior positively or negatively, and the subjective norm is based on the theory that human behavior is largely determined by the beliefs, attitudes, and perceptions of others about his or her behavior (Ajzen, 1991). It is a type of stress that influences an individual to perform or not perform behavior (Ajzen 1991). Perceived behavioral control refers to the extent to which people believe they have sufficient resources (e.g., time, money), skills, and opportunities to engage in behavior (Tucker et al. 2020).

The relationship between attitude, subjective norm, perceived behavioral control, and behavioral intention has been extensively explored in many contexts, including ADVs (Kapser et al. 2021; Rai et al. 2022) or autonomous delivery robots (ADRs) (Koh and Yuen 2023; Yuen et al. 2022). Similarly, Yuen et al.(2022) indicated that among the factors affecting consumers’ intention to use ADRs, attitude has the greatest impact, followed by subjective norm, and finally perceived behavioral control. ADVs offer many advantages, such as their accessibility and ability to provide environmentally friendly and contactless delivery services (Kapser and Abdelrahman 2020; Kapser et al. 2021). Therefore, it is meaningful and valuable to the last-mile delivery services. Some users may prompt other people to use and recommend ADVs to people who haven’t experienced it. In addition, modern consumers have basic skills in operating electronic devices (e.g., cell phones), and many complex technical and operational questions can be answered via the web. Consumers place orders through their cell phones, and then the ADVs are delivered to the location specified by the consumer, who can self-pick up by brushing his face or entering a verification code, etc. Consumers’ perceived ability to use ADVs increases their usage intention.

Normative activation model(NAM)

The normative activation model (NAM) is a popular model employed to predict the altruistic and environmental-friendly behavior of individuals (De Groot and Steg 2009, 2010; Schwartz 1977). Schwartz (1977) argued that the motivation of pre-environmental behavior was internal values and norms, and the activation of internal values and norms formed people’s sense of moral obligation, namely personal norm. The NAM contains three main variables, including personal norm (PN), awareness of consequences (AC), and ascription of responsibility (AR). Personal norm is mainly affected by the awareness of consequence (AC) and the ascription of responsibility (AR). The awareness of consequence refers to the consciousness that an individual has caused adverse consequences to others or other affairs when not performing altruistic behaviors (Hiratsuka et al. 2018; Schwartz 1977). Generally, the stronger the perception of the outcome of a particular situation, the stronger the sense of moral obligation, and the more likely the individual is to activate the personal norm to implement the corresponding altruistic behaviors. Ascription of responsibility means an individual’s sense of responsibility for adverse consequences (Schwartz, 1977). The stronger the individual’s sense of responsibility for the outcome, the more conducive to the implementation of the expected behaviors consistent with the personal norm (Hiratsuka et al. 2018; Le et al. 2021). Personal norms are internalized social norms and self-moral obligations, referring to the individual’s self-expectation to carry out specific behaviors under certain circumstances (Le et al. 2021). Adherence to personal norms can lead to increased people’s pride and dignity, while violations can lead to feelings of guilt, self-denial, or loss of self-esteem (D’Arco et al. 2023).

ADVs are more compatible with the environment than traditional delivery methods, and the purpose of our study is to investigate the environmentally friendly behavior of consumers, namely the adoption of ADVs. Therefore NAM is introduced in our study. In addition, other research has shown that the awareness of consequences positively affected the ascription of responsibility (De Groot and Steg 2010). In the same way, the awareness of consequence of not adopting ADVs can affect the ascription of consumers’ moral responsibility. Ascription of responsibility can also affect consumers’ attitudes toward ADVs.

Many studies have shown that environmental factors, i.e. environmental concern or environmental awareness, affect consumer attitudes and pre-environmental behaviors (Wang et al. 2020). Some scholars have combined NAM and TPB models to explore consumer pro-environmental intentions and behaviors, and the conclusion shows that the combination of the two models is more explanatory than using a single model (Zhang et al. 2017, 2018). Therefore, in our study, we also integrate TPB with NAM to explore consumers’ intention to use ADVs.

Hypotheses development

This section describes the hypothesized relationships between the variables, specifically the relationship between the effects of perceived enjoyment, ascription of responsibility, and COVID-19 risk on the relationship between the original TPB variables, as well as the relationship between each of the TPB variables on the willingness to adopt and WOMR of ADVs, and the relationship between WOMR on the intention to adopt ADVs.

Effects of attitude on intention to use and recommend ADVs

According to TPB, attitude is related to the degree to which a person positively or negatively evaluates a behavior (Ajzen 1991; Asif et al. 2022; Sarah et al. 2023). In this study, attitude denotes consumers’ overall evaluation of the delivery services provided by ADVs in the online shopping environment. In general, when a person has a positive attitude toward a certain behavior, she/he is more likely to act on it (Ismagilova et al. 2020; Mishra et al. 2021; Qian et al. 2023). As a novel delivery method, ADVs can help online shoppers relieve the psychological stress and tension of picking up packages by maintaining social distancing (Raj et al. 2020). In addition, the use of ADVs can provide environmental benefits by reducing energy consumption and greenhouse gas emissions, and other benefits include intelligent navigation, alleviating traffic congestion, and reducing collisions (Qian et al. 2023).

Extensive studies have emphasized the importance of attitude to behavioral intention. For example, in terms of green freight transportation (Schniederjans and Starkey 2014), mobile payment systems (Belanche et al. 2022), drones (Osakwe et al. 2022), private-sphere environmental behavior (Gao et al. 2017; Gkargkavouzi et al. 2019), attitude plays an extremely important role in the decision-making process of consumers. Likewise, several authors have repeatedly stressed the importance of consumer attitude when recommending autonomous vehicles to others (Yuen et al. 2022). As mentioned earlier, ADVs have many advantages over traditional modes of transportation (i.e. trucks), such as higher delivery and energy efficiency, as well as environmental friendliness, which drives people to use and recommend ADVs to others. Based on these arguments, we propose the following hypothesis:

H1a. Consumers’ attitude positively affects consumers’ intention to use ADVs.

H1b. Consumers’ attitude positively affects consumers’ word-of-mouth reconmmendation of ADVs.

Effects of Subjective Norms on Intention to Use and Recommend ADVs

Subjective norm is based on the theory that human behavior is largely determined by the beliefs, attitudes, and opinions of others about that behavior (Ajzen 1991; Asif et al. 2022). In our study, subjective norm refers to the social pressure perceived by the reference group (i.e. colleagues, friends or relatives) when deciding whether to accept or reject ADVs. When an individual’s behavior is recognized by the reference group, it will enhance the willingness of individuals to implement such behaviors (Berger, 2014; Mishra et al. 2021). Lots of studies have shown that the meaning of social influence is a structure equivalent to subjective norms (Benleulmi and Ramdani, 2022). A study by Acheampong and Cugurullo (2019) found that subjective norm positively affects consumers’ willingness to use ADVs. In addition, Kapser and Abdelrahman (2020) found a positive correlation between social influence and consumer adoption of ADVs. As a new and innovative delivery method, the delivery services provided by ADVs are easy to attract the attention of consumers with innovative and fresh ideas. Once these early adopters realize the benefits of the technology, they spread the word and promote it through personal social networks and platforms and recommend others to adopt it. Meanwhile, consumers who are slower to react to the market are easily influenced by early adopters to perform adoption behaviors (Hardman et al. 2019; Kapser and Abdelrahman 2020). Therefore, we propose the following hypothesis:

H2a. Subjective norm (SN) positively affects consumers’ intention to use ADVs.

H2b. Subjective norm (SN) positively affects consumers’ word-of-mouth recommendation of ADVs.

Effects of perceived behavioral control on intention to use and recommend ADVs

Perceived behavioral control (PBC) refers to the extent to which people believe they have sufficient resources (e.g., time, money), opportunities and skills to accomplish a certain behavior (Ajzen 1991; Tucker et al. 2020). In our study, PBC predominantly refers to the degree of difficulties individuals perceive when using contactless services provided by ADVs, which reflects their experience and anticipated barriers (Benleulmi and Ramdani 2022; Buckley et al. 2018). It has its origins in the work of Bandura (1977) and self-efficacy theory and is considered to be a predictor of intention to adopt emerging information technologies, i.e. ADVs (Yuen et al. 2022). Individuals with a high sense of perceived behavioral control are more likely to engage in certain behaviors (Zhang et al. 2020). The research of Buckley et al. (2018) suggested that perceived behavioral control positively affects consumer attitudes toward the adoption of autonomous cars. In addition, Benleulmi and Ramdani (2022) revealed that the desirability of control affects consumers’ willingness to use autonomous vehicles. Currently, ADVs are being put into practice in many universities and communities across the country (Lynn 2023). Convenient infrastructure for charging and parking of ADVs is available, and those early adopters can spread the service to others and recommend others to participate in this pre-environmental behaviors. Therefore, we make the following assumption:

H3a. Perceived behavioral control (PBC) positively affects consumers’ intention to use ADVs.

H3b. Perceived behavioral control (PBC) positively affects consumers’ word-of-mouth reconmmendation of ADVs.

Effects of environmental factors on original TPB contructs

Although TPB is an important model for predicting individual pre-environmental behaviors, it does not consider the impact of altruistic motivation on individual pro-environment behaviors, resulting in low explanatory power (Schniederjans and Starkey 2014). Therefore, some scholars have introduced the normative activation model (NAM) to improve the explanatory power of individual pro-environment behaviors(De Groot and Steg 2009, 2010; Hiratsuka et al. 2018; Le et al. 2021). Studies have shown that NAM mainly considers two pro-environment variables, namely, awareness of consequences and ascription of responsibility (Schwartz 1977; Liere and Dunlap 2010). The awareness of consequences means that when an individual perceives that not carrying out a certain behavior may bring adverse consequences to others, ascription of responsibility refers to an individual’s sense of responsibility for the negative consequences of not performing certain behaviors (Schwartz, 1977). Awareness of consequences affects the ascription of responsibility (Hiratsuka et al. 2018; Schniederjans and Starkey, 2014).When individuals realize that their behaviors hurt others, they tend to blame themselves and attribute responsibility for the consequences to themselves, which will further affect the individual’s attitude toward implementing certain behaviors (Li et al. 2022a, 2022b; Zhang et al. 2017). Ascription of responsibility affects normative factors such as personal norms and subjective norms (Li et al. 2018). In our study, people around are affected by those who adopted ADVs and then also did the same imitative behaviors. Once customers insist on recognizing the shortcomings of fuel delivery vehicles, they spread this influence to surrounding friends and colleagues, then we can assume that the ascription of responsibility affects consumer subjective norms. Furthermore, ascription of responsibility affects consumers’ confidence and ability to continue adopting pre-environmental technologies or services (Hiratsuka et al. 2018; Li et al. 2018), then we can believe that ascription of responsibility will affect consumers’ perceived behavioral control. In our study, ADVs are safer, faster, and more environmentally friendly than traditional delivery modes, especially during the pandemic, which has made last-mile delivery logistics safer and more convenient. Consumers realize that if they do not adopt environment-friendly ADVs, they may feel guilty about the environment and the whole ecology, which may prompt them to adopt a positive attitude toward ADVs and thus generate environment-friendly behaviors. To this end, we propose the following assumption:

H4. Awareness of consequences (AC) positively affects the ascription of responsibility (AR).

H5a. Ascription of responsibility (AR) positively affects consumers’ attitudes toward ADVs.

H5b. Ascription of responsibility (AR) positively affects consumers’ subjective norm to use ADVs.

H5c. Ascription of responsibility (AR) positively affects consumers’ perceived behavioral control to use ADVs.

Effects of COVID-19 risk on original TPB contructs

The premise of using TPB to analyze consumers’ intention to use ADVs is that consumers are rational in their approach to decision-making. But in reality, consumers are mostly emotional, therefore, contextual factors must be added to supplement and improve the original model (Kapser and Abdelrahman 2020). The epidemic threat will affect consumers’ inner mood, causing them to panic or fear (Ahorsu et al. 2020; Gawrych et al. 2020; Pandita et al. 2021). To avoid infection during the epidemic, consumers will try to avoid face-to-face contact with other people (including delivery personnel), and the contactless service model provided by ADVs just meets the demand of consumers. Therefore, we believe that the risk brought by the epidemic will affect consumers’ attitudes towards ADVs. Here, COVID-19 risk refers to the severity of COVID-19 on an individual’s physical, and mental health and the economy, as well as consumers’ susceptibility to infection (Envelope et al. 2021). The epidemic affects consumers’ decision-making process for offline shopping and receiving deliveries. People under quarantine policies are influenced by their surroundings and friends, and consumers are more receptive to the contactless delivery services offered by ADVs. Thus, we can argue that COVID-19 risk affects consumers’ subjective norms (Bavel et al. 2020; Kim et al. 2020; Pandita et al. 2021). Consumers with a strong perception of COVID-19 risk are more willing to receive packages safely without having to contact others, that is, they want more secure delivery modes and pick-up methods. As a result, the risk posed by the epidemic can have a significant impact on consumers’ perceived behavioral control. Therefore, we propose the following hypothesis:

H6a. COVID-19 risk positively affects consumers’ attitudes toward ADVs.

H6b. COVID-19 risk positively affects consumers’ subjective norm to use ADVs.

H6c. COVID-19 risk positively affects consumers’ perceived behavioral control to use ADVs.

Effects of perceived enjoyment on original TPB contructs

Like hedonic motivation, perceived enjoyment refers to the degree of pleasure or fun consumers perceive when using a particular technology (Herrenkind et al. 2019; Natarajan et al. 2017). Perceived enjoyment is considered to be an important factor affecting consumers’ attitudes and intentions toward information technology adoption (Nastjuk et al. 2020). Recent studies have found that perceived enjoyment positively influences autonomous vehicle adoption behaviors. Furthermore, Nordhoff et al. (2020) point out that hedonic motivation is the most critical factor in exploring consumer adoption of autonomous cars. Throughout the delivery process, ADVs provide consumers with smart ways to operate and interesting contractless services to consumers that increase perceived fun and positive attitudes. Consumers can schedule delivery times and locations via their smartphones. When the delivery is due to arrive, they will receive an SMS verification code that they can enter to select their own goods. In the whole process, the interaction between consumers and ADVs generates a range of perceptions (Kapser et al. 2021; Reed et al. 2022), for example, customers will find the new delivery method interesting and playful, which will encourage consumers to communicate through social channels and increase consumers’ subjective norm. In addition, consumers will also have positive control ability over this emerging delivery tool. However, lack of trust in technology, safety, and cost are the main reasons cited for being unlikely to use self-driving vehicles. In addition, some people are skeptical of the emerging technologies and remian distrustful of ADVs (Kapser and Abdelrahman 2020; Kapser et al. 2021). Overall, in our study, consumers have positive feelings and attitudes about using ADVs, which can enforce acceptance and the behaviors to recommend. To this end, we propose the following:

H7a. Perceived enjoyment positively affects consumers’ attitudes toward ADVs.

H7b. Perceived enjoyment positively affects consumers’ subjective norms.

H7c. Perceived enjoyment positively affects consumers’ perceived behavioral control.

Effects of Word-of-mouth (WOM) recommendations on intention to use ADVs

Word-of-mouth (WOM) refers to the spread of products or services to more potential customers through customer word-of-mouth in marketing, to improve brand recognition and brand acceptance (Brown et al. 2005; Cmkc and Mkol 2012). Word-of-mouth can be divided into positive word-of-mouth and negative word-of-mouth. Negative word-of-mouth has a greater impact than positive word-of-mouth (Zhang et al. 2020). Word-of-mouth recommendation intention refers to the possibility that users who have used a product or service take the initiative to recommend it to people around them. Word-of-mouth recommendation intention can affect consumers’ attitudes, adoption intentions, and actual adoption behavior. (AlKheder et al. 2023; Keaveney, 1995).

A large number of scholars have studied the motives and consequences of consumers to make WOMR (Berger 2014; Duarte et al. 2018). Among these studies, self-worth enhancement (Sundaram et al. 1998), personal experience (Berger 2014; Duarte et al. 2018), altruism (Sundaram et al. 1998), social interaction (Nian et al. 2021) are believed to be highly related with consumers’ intention to enforce WOMR behaviors. Social identity theory also states that individuals expect to be part of a common group, and WOMR can help consumers shape their own identity and enhance their self-image so as to be recognized by others (Udall et al. 2020). The research of Mensah and Mwakapesa (2022) indiciated that electronic word-of-mouth as an indenpent variable affects consumers’ adoption intention. Zhou et al. (2022) unveiled that online word-of-mouth acted as a mediating effect in the relationship between the perception of vedio images on tourists’ travel intentions. During the epidemic period, people respond to the quarantine policy and try to avoid contacting with others as much as possible, while ADVs provide contactless delivery services to meet customers’ contactless needs and greatly facilitate last-mile delivery. This contactless delivery method has been widely spread and proliferated by consumers through social media and other network platforms, attracting the attention of many enterprises and consumers, and increasing consumers’ intention to experience and use the delivery services provided by ADVs. Therefore, we can make the following assumption that:

H8. Consumers’ word-of-mouth recommendation (WOMR) positively affects consumers’ intention to use ADVs.

Methodology

Scale design and measures

In this study, a questionnaire survey and structural equation modeling (SEM) are used to test the hypothesis of the model. The questionnaire consists of three parts. The first part briefly introduces the features and functions of ADVs. The second part includes the measurement items of structural variables in the model. A seven-point Likert scale is used, where 1 stands for “strongly disagree” and 7 represents “strongly agree.” The third part is the demographic characteristics of the respondents. Samples were collected from November 2022 to December 2022, and a total of 615 responses were received. After eliminating inappropriate responses, 561 valid samples were left, with an effective rate of 91.22%. Every respondent who participated in the survey received informed consent, and the data obtained did not involve personal information such as the user’s name, email address, telephone number, geographical location, etc.

All measurements are taken from existing literature to improve content validity. Attitude, subjective norm, and perceived behavioral control are measured according to Kim et al. (2021a), (2021b), Osakwe et al. (2022), and Baig et al. (2022). The variables of awareness of consequences and ascription of responsibility are based on the modification of studies by Zhang et al. (2018). The measurement of perceived enjoyment is based on Kapser and Abdelrahman (2020). The measured items of COVID-19 risk are adopted from the scale of Envelope et al. (2021). The scale of behavioral intention is adopted from Chen et al. (2018) and Kapser and Abdelrahman (2020). The measurement of word-of-mouth recommendations(WOMR) is derived from the scale of Mishra et al. (2021). Constant revision and feedback from expert reviews, in particular, 40 MBA members are invited to conduct preliminary tests to ensure the accuracy and understandability of the scale. Due to incorrect analysis results, AC4, AR1, AR4, BI2, WOMR2 factors are excluded. The formal questionnaire is formed after repeated refining Table 1.

Table 1 Source of variables.
Full size table

Based on the above research analysis, contextual factors (COVID-19 risk), environmental factors (awareness of consequences and ascription of responsibility), and individual factors (perceived enjoyment) are introduced to the present study to establish an integrated model of influencing user intention to ADVs based on the theory of planned behavior and normative activation model (Fig. 1).

Fig. 1: The conceptual model.
Fig. 1: The conceptual model.
Full size image

AC awareness of consequences, AR ascription of responsibility, ATT attitude, PE perceived enjoyment, SN subjective norm, CR COVID-19 risk, PBC perceived behavioral control, BI behavioral intention, WOMR word of mouth recommendation.

Table 2 shows that among the participants in this survey, males accounted for 50.1%, and females accounted for 49.9%. The majority of respondents were 19–35 years old, accounting for 76.56%, and 70.59% of the respondents are mainly bachelors. Further, 36.36% of the respondents have an income of 1000–3000 yuan. The number of respondents who have used ADVs accounts for 39.93% of all samples.

Table 2 Demographic characteristics.
Full size table

Data analysis

Reliability and validity

Before the formal issue of the scale, expert consultation and questionnaire correction were performed to ensure good content validity. Then, the principal component analysis (PCA) method was used to select the maximum variation method. Finally, the remaining nine factors were preserved. The results showed that the explanatory variables of the remaining nine factors accounted for a 76.593 contribution rate, and each observed variable’s factor loading coefficients were greater than 0.6. This indicates that the scale has good structural validity. As such, this study’s scales exhibit good reliability and validity.

Reliability refers to the consistency and reliability of measurement results. Scholars generally agree that an indicator’s reliability is established when each indicator presents a coefficient value of 0.7 or higher. The results in Table 3 indicate that Cronbach’s alpha and the Composite reliability (CR) values of each factor ranged from 0.665 to 0.907 and 0.665 to 0.908, respectively. This indicates the suitable reliability of each scale (Hair et al. 2011).

Table 3 Reliability and validity indices.
Full size table

Validity comprises convergent validity, discriminant validity, and content validity. The KMO and Bartlett sphericity tests show that the P values are statistically significant and the KMO value was 0.912, above 0.8, indicating that the scale was suitable for factor analysis. The average variance extracted (AVE) values ranged from 0.4 to 0.711, indicating the convergent validity of the scale is acceptable (Hair et al. 2011). The result of Table 4 indicated that the values of the square roots of the average variance extracted (AVE) values were higher than the inter-construct correlations, validating good validity (Fornell and Larcker, 1981). Finally, the study revised the unsatisfactory initial model to obtain the final suitable model. The overall fitness of the structural model was tested, x2/df = 2.672, CFI = 0.944, NFI = 0.914, GFI = 0.907, NNFI = 0.932, and RMSEA = 0.055, indicating excellent model fit.

Table 4 Discriminant validity.
Full size table

Common method variance (CMV)

The Common method variance (CMV) is a systematic error and may occur in the same data source or measurement environment (MacKenzie and Podsakoff 2012; Podsakoff et al. 2003). To test possible common method biases, the Harman single-factor test was used (Podsakoff et al. 2003; Andrew et al. 2017). We found that the variance of each factor was less than 20%, indicating that there was no serious common methodology bias in the present study.

Hypothesis testing

In this study, AMOS software and bootstrapping with the 5000 subsamples method were used to estimate the significance level of each path. The structural model path analysis results of the study can be seen from Table 5. The results show that all hypotheses have been validated except that H1a, H3a, and H5c are not supported. In particular, subjective norm (SN) positively affects consumers’ word-of-mouth recommendation behavior and behavioral intention (BI), H2a and H2b are supported, respectively. Awareness of consequence (AC) affects ascription of responsibility (AR), and ascription of responsibility (AR) affects attitude(ATT) and subjective norm (SN), supporting H4, H5a, and H5b, respectively.COVID-19 risk affects attitude (ATT), subjective norm (SN) and perceived behavioral control (PBC), then H6a, H6b, and H6c are supported, respectively. Perceived enjoyment (PE) affects attitude (ATT), subjective norm (SN), and perceived behavioral control (PBC) and then H7a, H7b, and H7c are supported, respectively. Word-of-mouth recommendation (WOMR) positively affects behavioral intention (BI), and the H8 is supported.

Table 5 Summary of results of structural relationships.
Full size table

Direct, indirect, total effect, and mediation analysis

This section tests the effect of each antecedent variable on the outcome variables. From the analysis rsults in Table 6, it can be concluded that ascription of responsibility (AR) partially mediates the relationship between awareness of consequences (AC) and attitude(ATT), attitude (ATT) completely mediates the paths of perceived enjoyment (PE)-attitude (ATT)-behavioral intention(BI) and ascription of responsibility (AR)- attitude (ATT)- behavioral intention (BI) and COVID-19 risk (CR)- attitude (ATT)- behavioral intention (BI). Subjective norm (SN) plays a complete mediating role in the path of perceived enjoyment (PE)-subjective norm(SN)- behavioral intention (BI) and ascription of responsibility (AR)-subjective norm (SN)- behavioral intention (BI). Perceived behavioral control (PBC) completely mediates the relationship between perceived enjoyment (PE)- Perceived behavioral control (PBC)- behavioral intention (BI) and COVID-19 risk (CR)- Perceived behavioral control (PBC)- behavioral intention (BI). Attitude (ATT) plays a partial mediating role in the path of COVID-19 risk (CR)/ perceived enjoyment (PE)/ ascription of responsibility (AR) - attitude(ATT) - word-of-mouth recommendation (WOMR). Subjective norm plays a partial mediating role in the path of ascription of responsibility (AR)/ perceived enjoyment (PE)- subjective norm (SN)- word-of-mouth recommendation (WOMR). Perceived behavioral control plays a partial mediating role in the path of perceived enjoyment (PE)- Perceived behavioral control (PBC) - word-of-mouth recommendation (WOMR). As for the direct effects of the model in our study, first of all, the main predictors of attitude are perceived enjoyment (PE)(a32 = 0.826), COVID-19 risk (CR) (a42 = −0.601), and ascription of responsibility(AR)(a22 = 0.274). Secondy, the predictors of the subjective norm (SN) are perceived enjoyment (PE) (a33 = 0.518), COVID-19 risk (CR) (a43 = −0.292), and ascription of responsibility (AR) (a23 = 0.13). Thirdly, the direct predictors of perceived behavioral control are perceived enjoyment (PE) (a34 = 0.67), COVID-19 risk (CR) (a44 = 0.134), and ascription of responsibility (AR) (a24 = −0.018). The main direct predictors of word-of-mouth recommendation (WOMR) are attitude (ATT) (a55 = 0.751), subjective norm (SN) (a65 = 0.281), and perceived behavioral control (PBC) (a75 = 0.102). Finally, the main direct predictors of behavioral intention (BI) are, in descending order, word-of-mouth recommendation (WOMR) (a86 = 0.734), subjective norm (a66 = 0.158), attitude (a56 = 0.149), and perceived behavioral control (PBC) (a76 = 0.011).

Table 6 Direct, indirect and total effects.
Full size table

Regarding the indirect effects of the model in our study, perceived enjoyment (PE), ascription of responsibility (AR), and COVID-19 risk (CR) indirectly influence consumers’ willingness to use ADVs and word-of-mouth recommendation (WOMR). In this study, the bootstrap analysis method is applied to test the mediation effect. The results show that perceived enjoyment (PE) has the largest indirect effect on consumers’ intention to use ADVs(b36 = 0.067), followed by COVID-19 risk (CR) (b46 = −0.04) ascription of responsibility (AR) (b26 = 0.013). As shown in Table 6, the ascription of responsibility (AR) has an indirect effect on word-of-mouth recommendation (WOMR) through attitude (ATT) and subjective norm(SN), while perceived enjoyment (PE) has an indirect impact on consumer word-of-mouth recommendation (WOMR) through all TPB constructs.

Finally, in terms of total effects, word-of-mouth recommendation (WOMR) has the greatest total effect on consumers’ behavioral intention (c86 = 0.734). This is followed by subjective norm (SN) (c66 = 0.158), attitude (ATT)(c46 = 0.149), perceived enjoyment (PE) (c36 = 0.067), then COVID-19 risk (CR) (c56 = −0.04), ascription of responsibility (AR) (c26 = 0.013), and lastly, perceived behavioral control (PBC) (c76 = 0.011).

Discussion and conclusions

COVID-19 has had a huge impact on the logistics and express industry while autonomous delivery technologies have shown great potential in the field of last-mile delivery during the pandemic. Based on the planned behavior theory (TPB) and normative activation model (NAM), the study explores significant antecedent factors that affect consumers’ intention to use ADVs. In addition, the study investigates the determinants that affect original TPB constructs (i.e. attitude, subjective norm, and perceived behavioral control). The results indicate that the integrated model has more explanatory power than the separate planned behavior model or normative activation model. The results also show that perceived enjoyment, COVID-19 risk and ascription of responsibility all affect TPB constructs,i.e. attitude, subjective norm and perceived behavioral control, except for ascription of responsibility (AR) does not affect perceived behavioral control. Based on the total effects analysis, WOMR affects consumers’ intention to use ADVs the most, followed by subjective norm, attitude, perceived enjoyment, COVID-19 risk, ascription of responsibility.The perceived behavioral control has the weakest effect on consumers’ intention towards ADVs.

The results of the study show that subjective norms affects WOMR. This demonstrates that consumers can be affected by groups within social networks, surrounding friends and relatives can motivate consumers to recommend and spread ADVs to others.

While the relationship between attitude or perceived behavioral control on intention are not significant. However, the results show that only subjective norm affect consumers’ intention, which is coincident with the studies of Schniederjans and Starkey (2014) and Sarah et al. (2023) that subjective norms positively affect consumer adoption of emerging technologies. The findings also indicate that the relationship between perceived behavioral control and intention is not statistically significant, which is violated by the study of Zhang et al. (2017) and Belanche et al. (2022) that perceived behavioral control has a positive effect on consumers’ intention to use emerging technologies. This may be because although the external necessary conditions are already in place, it does not prevent consumers from making internal decisions, indicating that there are other reasons not found in this study to prevent consumers from adopting ADVs.

The findings of the study show that pro-environmental factors affect consumers’ attitudes and subjective norms, and thus affect consumers’ final adoption intention. This is consistent with the research of Zhang et al. (2018) that awareness of consequences and ascription of responsibility play critical roles in pro-environmental attitudes and behaviors. It has also been verified that pro-environmental motives, namely, awareness of consequences affect ascription of responsibility, and ascription of responsibility positively affects electric vehicle adoption behaviors, which confirms the impact of consumers’ environmental perspectives on attitudes and intention behaviors (Adnan et al. 2018; Li et al. 2022a, 2022b). Autonomous delivery cars have the potential to revolutionize last-mile delivery in a more efficient, environmentally friendly, and customer-focused way. The environmental features of ADVs could spur consumer adoption of this emerging technology. However, a surprising finding shown in our study is that the relationship between ascription of responsibility and perceived behavioral control is not statistically significant. This may be becsuae the ascription of responsibility mainly focuses on the internal moral restriction of consumers, and does not directly affect the external objective environment and resources.

COVID-19 risk affects orginal TPB constructs (i.e. attitude, subjective norm, and perceived behavioral control), and thus affects consumers’ WOMR and adoption of ADVs. The COVID-19 risk variable is added to the study to demonstrate the influence of contextual factors on consumers’ willingness to use ADVs and word-of-mouth marketing. The results show that COVID-19 risk negatively affects consumers’ attitudes, and subjective norms and has a positive impact on perceived behavioral control. This is consistent with the studies of Kapser et al. (2021) and AlKheder et al. (2023). The study of AlKheder et al. (2023) noted that consumers’ perceived COVID-19 risk negatively affects consumers’ attitudes and intentions to use ADVs. The research of Kapser et al. (2021) also indicated that perceived risk negatively affects electronic vehicle purchase intention. Maintaining appropriate social distancing and responding to quarantine policies prevent people from socializing and traveling elsewhere, and further mitigate the spread risk of the epidemic(Envelope et al. 2021; Kapser et al. 2021; Kim et al. 2020). Similarly, these behaviors have also extended to consumer’s choice of the last mile delivery services (Kapser et al. 2021). The risk posed by the pandemic not only affect consumers’ perceptions and emotions but also their confidence and ability to adopt emerging information systems and information technologies. This negative impact spreads to other consumers through word-of-mouth, thus discouraging them from further adoption of ADVs.

Perceived enjoyment positively affects orginal TPB constructs (i.e. attitude, subjective norm, and perceived behavioral control), and thus affects consumers’ WOMR and adoption of ADVs. Based on consumer factors, the study adds perceived enjoyment to explore the relationship between perceived enjoyment and TPB constructs and then consumers’ behavioral intention, and WOMR. The results show that perceived enjoyment plays the most significant influence on consumers’ attitudes(c32 = 0.832). second perceived behavioral control (c34 = 0.669), and last subjective norms (c33 = 0.519), which indicates that consumer factors are primarily responsible for consumers’ attitudes and behavioral intentions and WOMR. This is consistent with the research of Kim et al. (2021a), (2021b) and Kapser and Abdelrahman (2020), which suggest that perceived enjoyment positively influences consumers’ adoption of autonomous vehicles. In addition, perceived enjoyment can affect consumers’ usage intention through attitude, subjective norm, and perceived behavioral control, namely, attitude, subjective norm, and perceived behavioral control play an intermediary role in the relationship between perceived enjoyment and intention, respectively.

Consumers’ word-of-mouth recommendation (WOMR) positively affects their intention to use contactless delivery services provided by ADVs. This is consistent with the research of Kim et al. (2009) that positive word-of-mouth communication enhances the revisit intention of university dining customers. The research of Spence et al. (2014) also suggests that store atmosphere can guide consumers towards visiting and positive word-of-mouth intention. Mensah and Mwakapesa (2022) indicated that electronic word-of-mouth as an independent variable affects consumers’ adoption intention. Zhou et al. (2022) reveal that online WOM acted as a mediating effect in the relationship between the perception of video images and tourists’ travel intentions. This phenomenon is understandable because once consumers’ positive confirmations are formed, they would spread and speak loudly to surrounding friends or family members to increase usage intention. The safety of contactless delivery services provided by ADVs alleviates the psychological stress of consumers and attracts more attention than before, some potential consumers may be influenced by early adopters to choose this contactless delivery method.

Implications

Theoretical implications

The study has great theoretical contributions. Firstly, the study explores the determinants of consumer adoption of ADVs by integrating the theory of planned behavior and normative activation model into the theoretical framework to better understand consumers’ intention to use ADVs. The study bridges a gap and provides insights for a better understanding of the interrelationship between multiple theories and their influence on consumers’ intention to use ADVs. The integration of the two theories has shown greater explanatory power than signal one. The findings of our study indicate that attitude, subjective norms, and perceived behavioral control all positively affect consumers’ word-of-mouth recommendations while only subjective norm affects consumers’ intention to use ADVs, the relationship between attitude or perceived behavioral control and intention is not significant. Ascription of responsibility, perceived enjoyment, and COVID-19 risk all affect orginal TPB constructs, except that ascription of responsibility does not affect perceived behavioral control.

Secondly, the study enriches previous studies by adding perceived enjoyment, ascription of responsibility, and COVID-19 risk to explain consumers’ intention to use ADVs. COVID-19 may affect consumers’ perception of the pandemic and their intention to use ADVs (Elliott 2021). To address this, COVID-19 risk is added in our study to describe the perception of the likelihood of contracting COVID-19 and its consequences. ADVs act as environmentally-friendly products and the adoption of ADVs can be pre-environmental behaviors, then our study includes pre-environmental factors, i.e.awareness of consequence and ascription of responsibility, to explore consumers’ intention to use ADVs. Perceived entertainment, as one of the very important individual attributes of consumers, plays an important role in consumers’ emerging technology adoption behaviors. Then our study also adds perceived enjoyment into the research framework to explore consumers’ intention to use ADVs.

Finally, the total effect analysis in Table 6 reveals that WOMR has the greatest total effect on intention (c86 = 0.734), followed by subjective norms (c66 = 0.158), attitude (c56 = 0.149), perceived enjoyment (c36 = 0.067), COVID-19 risk (c46 = −0.04), ascription of responsibility (c26 = 0.013), and perceived behavioral control (c76 = 0.011). The results show that before adopting innovative technologies, consumers should collect and analyze online reviews promptly and pay attention to the impact of online word-of-mouth marketing. The nature of entertainment and fun will also attract consumers to stop and try it to a large extent. In addition, consumer attitudes also influence their adoption and use of emerging technologies, so positive evaluation and feedback are extensively important for potential users.

Practical implications

The study has profound practical implications for different stakeholders, including logistics companies, ADVs technology service providers, and government departments.

From the perspective of logistics companies, logistics managers can take different marketing measures to promote positive images and attitudes of contactless delivery services provided by ADVs to the public. Moreover, operational functions and environmental attitudes about reducing CO2 emissions and improving energy efficiency should be repeatedly emphasized to increase consumers’ trust and positive attitude to this emerging technology. Logistics companies can improve consumers’ experience level and service quality by promoting the contact-free attribute and technical performance of these particular services, increasing consumer recognition and acceptance of these new technologies, whereby enhancing and improving consumer attitudes and adoption behaviors. Industry experts or society celebrities can be invited to conduct public speeches or declarations to present the environmental and innovative benefits of ADVs and play the leading role of excellent examples. Ordinary consumers may be inspired and glad to follow in the footsteps of celebrities to experience and adopt this new delivery technology.

From the perspective of ADVs technology providers, they should continue to improve technical performance and optimize service functions, increase the entertainment and curiosity experienced by consumers, and reduce consumers’ perception of epidemic risks. At the same time, the operational steps at the consumer level should be simplified to enable consumers to learn the process of using ADVs to pick up packages faster. The R&D enterprises of ADVs should add appealing or interesting attributes and functions in the design process to attract consumers’ attention to this emerging technology. such as intelligent voice interaction and smart perception of ADVs. Consumers should be involved to participate in the design and development of ADVs so that R&D companies can produce distinctive ADVs that are more suitable for practical use.

From the central government’s perspective, it is very necessary to establish a sound and comprehensive infrastructure for the smooth implementation of ADVs. For example, relevant government departments should stipulate whether autonomous vehicles can travel on motor vehicle lanes and the maximum speed limit. When ADVs are officially put into operation, corresponding road rights tests should be conducted. At the same time, relevant laws and regulations on ADVs should be formulated and improved as soon as possible, such as how to solve the problem of ADVs colliding with pedestrians or buildings. These measures will increase consumers’ perceived behavioral control, thereby increasing their willingness to recommend behaviors and adopt them. Finally, the government can use official website channels to promote the delivery advantages of ADVs, so that people in the entire social circle will recognize the advantages of this delivery method, and thus promote consumers’ adoption and recommendation behavior. Local government administrations can provide specific and appropriate subsidies to those representatives of ADVs technology services providers. In addition, infrastructure construction and standardization problems should be highly valued by related authorities and industry associations.

Limitations and directions for future research

Although the study has made significant contributions, it still has some shortcomings. First of all, the acquisition of the questionnaire is based on cross-sectional data, which will affect the accuracy of the research to some extent. Longitudinal studies can be used to compare consumer behavior before and after the epidemic. Secondly, the study mainly selected Chinese consumers for investigation, in the future, the sample scope and scale can be expanded to other countries and regions to expand the generalizability of the results. Thirdly, among the individual characteristics selected in this study, only the perceived enjoyment is selected. In the future, other variables such as consumer innovativeness or Big Five Personality can be set to explore the consumers’ intention to use ADVs. This study chooses the contextual variable of COVID-19 risk to explore the adoption intention, and other contextual variables can be selected for further study in the future. Fourthly, the study only looks at ADVs as an alternative form of traditional human delivery. Some of the latest available delivery methods (e.g., drone delivery) are not considered in this study. Therefore, future research could explore other innovative delivery methods.