User Acceptance of Internet of Vehicles Services: Empirical Findings of Partial Least Square Structural Equation Modeling (PLS-SEM) and Fuzzy Sets Qualitative Comparative Analysis (fsQCA)

Recently, IoV-based services and vehicles have come to the forefront as part of the growing market for the automobile industry. Since IoV-based services and vehicles were introduced, they have been expected to grow rapidly. However, contrary to optimistic expectations for future market growth, the IoV-based services and vehicles market has appeared to hit a roadblock and remains at an early market stage. *erefore, research of the determinants leading to consumers’ intention to accept and purchase IoV-based services and vehicles is significant for either academics or practitioners. Drawing upon the extended unified theory of acceptance and use of technology acceptance model (UTAUT2), the perceived risk theory, and the initial trust model, we developed an integrated conceptual model and explored what and how various determinant antecedent conditions fit together on consumer intention to accept IoV-based services and vehicles.*e proposedmodel and hypotheses were assessed by both symmetric (partial least square structural equation modeling, PLS-SEM) and asymmetric (fsQCA) approaches using online survey datasets with 362 Chinese consumers. *e findings suggest that PLS-SEM and fsQCA are complementary analytical techniques providing comparable results. PLS-SEM results indicate that performance expectancy, price value, habit, and initial trust have significant effects on behavioral intention to accept IoV services. Despite other determinants, e.g., effort expectancy, social influence, facilitating conditions, hedonic motivation, and perceived risk, have no significant effect. FsQCA results reveal twelve different configurations of determinants resulting in a high level of behavioral intention to accept IoV services, and eight causal paths equifinally leading to the negation of behavioral intention to accept IoV services. *ese findings suggest that several conditions that were not significant in PLS-SEM are sufficient conditions when combined with other conditions. *is study enriches relevant research studies on IoV-based services acceptance and provides relevant insights and marketing suggestions for incentivizing consumers to accept the IoV-based services.


Introduction
In recent years, connected autonomous vehicle (CAVs) or called intelligent connected vehicles, as the core component of the intelligent transportation system (ITS) and a node of Internet of Vehicles (IoV) system [1], have become one of the most popular research fields in network and intelligent transportation system, as well as attracted huge investments from the automotive manufacturers. Various IoV-based services and location-based services of the connected autonomous vehicle, e.g., car navigation systems, vehicle information systems, advanced driver assistance, humancomputer interaction, and car infotainment systems, play important roles in making transportation safer, cleaner, and more comfortable [1,2]. While IoV-based services have a wide range of benefits in terms of safety, energy efficiency, environment improvement, increased mobility, and more entertainment in driving [3][4][5][6][7][8][9][10], such benefits may not be realized until IoV-based services are widely accepted and used by a critical mass of consumers [11][12][13][14][15]. Recent surveys have shown that the consumers' intention to accept or purchase IoV-based services is generally low Wu et al. [3]. erefore, it is significant to explore the determinants of consumers' intention to accept and purchase IoV-based services [16].
With ongoing technological advances in automation and connectivity, several studies have been discussed concerning IoV-based services in recent years that have concentrated on different issues [1,17], such as the concept of IoV [18,19], architecture (four layers including vehicle network environment sensing and control layer, network access and transport layer, coordinative computing control layer, and application layer) [20,21], and key technologies for IoV [22,23], as well as barriers and determinants of IoV-based services and vehicles adoption [24,25]. Numerous studies have investigated key factors for consumers' willingness-to-pay (WTP) and the adoption of IoV-based services and vehicles from the perspectives of various innovation adoption theories [6,26]. However, almost all of the empirical studies have been grounded in the use of the conventional symmetric-based techniques, including multiple regression model [13,[27][28][29], structural equation modeling (SEM) [11,14,15,30,31], and partial least square (PLS) [10,12,[31][32][33][34][35], to explore the relationships between independent variables (IV) and dependent variables (DV). ese conventional techniques are variableoriented that focus on the "net effect" of IV on DV, while excluding possible asymmetric relations between variables [36], leading to the correlation and significance might vary depending on the variables the model includes. In real-life scenarios, a viable outcome often depends on combinations of several antecedents that collectively form what is referred to as an algorithm in the asymmetric method [37]. erefore, qualitative comparative analysis (QCA), as a holistic approach, has been recommended for facilitating the analysis of complex causality and logical relations among combinations of conditions and an outcome, allowing researchers to examine multiple causal paths that lead to the same outcome [38][39][40].
To fill this gap, in the study, we answer the calls for the application of the holistic approach to understand the determinants' configurations on consumers' behavioral intention to accept IoV-based services. erefore, based on the review of studies related to the acceptance of IoV-based services and vehicles, we develop an integrated theoretical model including the extended unified theory of acceptance and use of technology (UTAUT2) [41] with the constructs of perceived risk from perceived risk theory [42] and initial trust from the initial trust model [43]. ese factors employed have been extensively used in innovation adoption studies and are appropriate for explaining IoV-based services' acceptance. is integrated model makes up for the inadequate explanation of the individual behavior adopted by a single model. Using online survey datasets with 362 Chinese consumers, we use both symmetric (PLS-SEM) and asymmetric (fsQCA) methods, to explore the role of the above determinants on the intention to accept IoV-based services. Our study involves some antecedents not analyzed in the literature, and it starts by analyzing the net effects of each antecedent with PLS-SEM. Moreover, to provide a more accurate understanding of the complex reality associated with the various determinants and behavioral intention to accept IoV-based services, the configurations of causal factors are analyzed by fsQCA for explaining the complex intention that the PLS-SEM does not capture [44]. FsQCA can not only explore how the antecedent factors combine to produce multiple alternative paths that can successfully lead to this intention but also assume causal asymmetry to identify the paths that explain the negation of the intention. e findings of PLS-SEM and fsQCA provide relevant insights and marketing suggestions for incentivizing consumers to accept and purchase IoV-based services.
e study is organized as follows. Section 2 provides literature reviews for the research model. Section 3 presents our conceptual model and hypotheses. Section 4 explains the measurement and data used in the empirical analysis. Section 5 presents the results of the empirical analysis by PLS-SEM and fsQCA. e key findings, theoretical implications, managerial implications, and limitations of the present study are discussed in Section 6. Finally, Section 7 is the conclusions of this study.

e Definition and Adoption of IoV-Based Services and
Vehicles. IoV as an important part of the wisdom city is a complex integrated network system [18], which connects different people within vehicles, different vehicles, and different environment entries in cities. As an important branch of IoT in the transportation field, IoV covers a wide range of technologies and applications, including intelligent transportation, vehicular information service, modern information and communications technology, and automotive electronics. However, owing to different understandings of the connotation of IoV in various research fields, there is no uniform definition of IoV [18]. IoV is different from telematics, vehicle ad hoc networks, and intelligent transportation, in which vehicles like phones can run within the whole network and obtain various services by swarm intelligent computing with people, vehicles, and environments.
IoV-based services and vehicles introduce all sorts of different benefits such as safety, energy efficiency, environment improvement, increased mobility, and more entertainment in driving [3][4][5]14]. However, there still exist several concerns about IoV-based services and vehicles, such as the relatively higher prices and maintenance costs compared with existing vehicles, performance, and safety issues when operating in complex conditions, impeding the adoption of IoV-based services and vehicles [14]. erefore, to ensure the social acceptability of IoV-based services and vehicles, numerous studies use technology adoption theories to explore the determinants of consumers' intention to accept and purchase IoV-based services and vehicles [6,26]. Table 1 summarizes previous studies on IoV-based services and vehicle acceptance based on a theoretical perspective of technology acceptance.
As shown in Table 1, previous studies have sought to explain the acceptance of various concepts of IoV-based services and vehicles, such as autonomous vehicles, connected vehicles, in-vehicle infotainment systems, and autonomous shuttle services. From a theoretical point of view, and SEM Environmental concern, green PU, PEOU, and BI Environmental concern, green PU, and PEOU positively affect BI; EC has an indirect effect on BI through mediators of green PU and PEOU [15] Trust and TAM Autonomous vehicles Online survey (369 German participants) and SEM Trust, concern of giving up control, PU, PEOU, driving enjoyment, personal innovativeness, and the adoption intention of AVs Trust in the technology/ concern about handing over control to a machine, PU, and personal innovativeness positively affects adoption intention, while driving enjoyment is a barrier to AVs adoption [34] Self-driving vehicles Survey (1355 participants in Tianjin and Xi'an, China) and PLS-SEM Demographic (familiarity, age, gender, education, and income) and psychological factors (PB, PR, perceived dread, and trust in SDVs) Younger and highly educated participants with higherincome were willing to pay more; participants who had heard about SDVs reported higher WTP and higher trust and perceived higher benefits, lower risks, and lower dread; trust and PB were positive predictors of WTP, while PR and perceived dread were negative [16] IDT and agentbased simulation modeling Connected autonomous vehicles the majority of empirical studies attempt to explore the determinants of acceptance using a single theory, such as the technology acceptance model (TAM) [2,12], the theory of planned behavior (TPB) [14], UTAUT [7,28,29,45], and innovation resistance model [33]. To provide a comprehensive understanding within the adoption decision-making of IoV-based services and vehicles [31], more and more researchers have called for integrating various theories or Mobile Information Systems 5 additional factors depending on the certain context to improve the model's explanatory power [31,46], such as the integration of TAM and initial trust theory [11,15,32,46]. erefore, we integrate various theories or constructs widely used and verified in diffident contexts for improving the model's explanatory power.
From a methodology of view, the empirical studies have been grounded mostly in the use of the conventional symmetric-based techniques, including multiple regression model [13,[27][28][29], SEM [11,14,15,30,31], and PLS [10,12,[31][32][33][34][35], to explore the relationships between variables. However, these techniques focus on the net effect of IV on DV, while excluding possible asymmetric relations between variables [36], leading to the correlation and significance might vary depending on the variables the model includes. For example, most of the empirical previous studies of perceived ease of use, perceived risk, social influence, or facilitating conditions have produced mixed results, and its effect on technology adoption has also been inconsistent. In real-life scenarios, a viable outcome often depends on combinations of several antecedents [37]. To understand the effects of determinants on the intention to accept IoV-based services and vehicles, QCA has been recommended for facilitating the analysis of complex causality and logical relations among combinations of conditions and an outcome [38,39]. erefore, we also answer this call and use fsQCA to provide a more nuanced understanding of how these antecedent conditions fit together to affect consumers' intention to accept and purchase IoVbased services and vehicles.

Unified
eory of Acceptance and Use of Technology (UTAUT). Information technology (IT) acceptance is an enduring topic in IS research, and a variety of popular models with different sets of acceptance determinant have been developed to explain individual's acceptance and usage of innovation technology, e.g., the theory of reasoned action (TRA) [49], TAM [50], the theory of planned behavior (TPB) [51], task-technology fit (TTF), motivational model, UTAUT [41,52], initial trust model (ITM) [43], diffusion of innovation theory (IDT) [53], and social cognitive theory. Many studies have used these traditional frameworks or added new constructs to develop models to conduct their research studies in various contexts of technologies.
Considering the choice of theoretical models for explaining user acceptance of new technology, Venkatesh et al. suggested a "need for a review and synthesis to progress toward a unified view of user acceptance" [52]. UTAUT integrating elements across the eight models (i.e., TRA, TAM, motivational model, TPB, a model combining TAM and TPB, the model of PC utilization, IDT, and social cognitive theory) explains more of the variance in user intentions to use new technology than each model [52]. UTAUT2 extends UTAUT with three variables explicitly proposed to be important in the consumer context (i.e., hedonic motivation, price value, and habit). Several studies in the IS discipline have confirmed the explanatory power of UTAUT in the acceptance and adoption studies. erefore, UTAUT is considered a robust and powerful model for investigating the determinants of technology adoption at the individual level, and it has generally been applied to understand the individuals' acceptance of various new IT innovations, e.g., IoT in the smart home [54] and mobile learning system [55]. UTAUT is also a commonly employed behavioral model in transport studies from a technology acceptance perspective [7,35,46]. erefore, it is considered theoretically and practically useful to utilize UTAUT2 as an adequate theoretical basis for exploring the effects of factors influencing consumer's acceptance of IoV-based services and vehicles.

Conceptual Model and Hypotheses
As outlined before, UTAUT2 will be utilized as a foundation to investigate user acceptance of IoV-based services. However, to examine the specific case of IoV-based services acceptance, the model needs to be modified and extended for improving the model's explanatory power and specificity [46]. IoV services, as the emerging technology, include both two-sidedness [26], that is, the technological advantages, and the possible negative consequences of IoV-based services, which should be investigated in parallel. Hence, two perceived factors (initial trust and perceived risk) were incorporated into the original UTAUT2 to form our model because of the uncertainty of the emerging technology. e proposed model is presented in Figure 1. In the following, the constructs of the proposed model and the developed hypotheses will be presented.

Performance Expectancy.
Performance expectancy is defined as the degree to which using technology will provide benefits to consumers in performing certain activities [41]. It is similar to the perceived usefulness of TAM and the relative advantage of IDT [53]. e effect of performance expectancy on behavior intention has been widely verified in previous studies, e.g., mobile learning [56], smart home [57], mobile health [58], and autonomous delivery vehicles [45]. According to IDT [53], users tend to accept new products or innovations if those innovations provide a unique advantage compared to existing ones. Compared with conventional vehicles, IoV-based services have functional benefits such as fewer accidents and reduced fuel consumption and emissions, as well as decreased congestion and driving time [3,10,48]. For example, IoV-based services can be synched with traffic signals and other vehicles by capturing the surrounding traffic conditions and environment to decrease travel time and cost [16]. IoV-based services are expected to decrease traveling time through anticipative driving, optimized routing, self-parking systems, and efficient usage of lanes [15]. More specifically, IoV-based services' integrated systems such as real-time navigation aim to provide the driver with a more comfortable and safer driving task [25]. Besides, ubiquitous connectivity allows consumers to be "always connected" or "always on," providing them with more freedom and access to information and services, such as infotainment [33]. Extant studies have shown that technological advantages increase the satisfaction level of consumers and affect the intention of consumers [47]. In this research context, the issue is the extent to which potential consumers will find IoV-based services better than traditional vehicles. erefore, the more performance expectancy is, the more likely consumers accept IoV-based services.

Effort Expectancy.
Effort expectancy is the degree of ease associated with consumers' use of technology [41]. It quite closely with the perceived ease of use of TAM and the complexity of IDT (the degree to which an innovation is perceived as relatively difficult to understand and use) [53]. Even if potential users believe that a given system is useful, meanwhile they may believe that the system is too hard to use and that the benefits of usage are out-weighed by the effort of using the system. Essentially, if IoV-based services were viewed to be difficult to use and complexity would act as a functional barrier, it would harm acceptance [14]. Effort expectancy has also been proven to be relevant with intention in various contexts, e.g., mobile learning [56]. Unlike conventional vehicles, IoV-based services do not require the driver to monitor and take back the operational control of the vehicle even when the system becomes failure [10]. Moreover, with the effortless communication with the human-car interaction system [35], the drivers just need to issue the instructions and confirm to IoV-based services, especially when they have a special preference or instant requirement. Analogously, we pose the following hypothesis.
H2: perceived ease of use positively affects IoV-based services' acceptance.

Social Influence.
Social influence is defined as consumers perceive that important others (e.g., family and friends) believe they should use the new technology [41,52], reflecting the effect of social environmental factors such as the opinions of a user's surrounding friends, relatives, and superiors on user behavior, which is similar to the subjective norm of TRA [48,49]. Previous studies also reveal the impacts of social influence on behavioral intention in various contexts, e.g., mobile payment [59], mobile banking [60], mobile learning [56], IoT in smart home [54], wearable technology [61], and autonomous delivery vehicles [45]. As a person in society [14], an individual's decision of technology adoption is very susceptible to their social networks [31] and social norms [35,62]. For IoV-based services, such a disruptive innovation in a burgeoning market, Talebian and Mishra [16] find that peer-to-peer communication (wordof-mouth) with a satisfied adopter from the social circles can not only convince a potential adopter that the resistances which he/she perceives are perhaps not important but also it is worthwhile to purchase IoV-based services. Besides, the use of autonomous vehicles by friends and family leads to a greater willingness to adopt the technology [63]. When friends, family, or other important ones start to use IoVbased services, consumers would be more likely to pay for an IoV-based service or vehicle [7].
H3: social influence positively affects behavioral intention to accept IoV-based services.

Facilitating Conditions.
Facilitating conditions are defined as a userʼs perception of disposable resources and support when performing a behavior [41,52], which are similar to perceived behavioral control of TPB [51]. It is significant in various technologies acceptance studies, e.g., mobile banking [60,64] and automated public transport [29]. IoV-based service/vehicle as new technology requires users to have certain skills such as configuring and operating smart applications to connect to the wireless Internet. It is believed that users have different levels in possessing information and resources that facilitate their use of IoV-based services (e.g., personal knowledge and help-hotlines). In  Mobile Information Systems general, consumers with a lower level of facilitating conditions will have lower intentions to accept or purchase IoVbased services [45].
H4: facilitating conditions positively affects behavioral intention to accept IoV-based services.
3.5. Hedonic Motivation. Hedonic motivation, considered as an intrinsic value, refers to the fun or pleasure derived from using a technology [41]. In IS adoption research, such hedonic motivation conceptualized as perceived enjoyment and has been found to play an important role in the acceptance and usage of various technologies, e.g., smart home [57], mobile health [58], and autonomous delivery vehicles [45]. e proposed system or technology must arouse the userʼs emotions and make the experience enjoyable and positive to motivate users to adopt it [57]. In the IoV context, hedonic motivation has also been found to be an important determinant of acceptance and purchase of autonomous shuttle services [29,45,46]. e attractiveness of new IoVbased services can be increased by the enjoyment experience provided by technology [46].
H5: hedonic motivation positively affects behavioral intention to accept IoV-based services.

Price Value.
Price value is defined as the consumersʼ cognitive tradeoff between a technology's perceived benefits and its monetary costs of usage [41]. In marketing research, the perception of price/cost is usually conceptualized together with the quality to determine the perceived value of products or services, reflecting the influence of the individually perceived price-performance and personal financial aspects. A positive price value occurs if the potential consumers' perceived benefits of using technology are greater than the associated monetary costs, which has a positive impact on consumersʼ adoption and usage of new technology [48]. Price value is a significant determinant of behavioral intention in various contexts, e.g., smart home [54]. Price value is expected to be particularly influential on IoV-based services acceptance as IoV technologies are evaluated against their comparable non-IoV objects. Although some studies have indicated that potential consumers are generally willing to pay a higher price for autonomous vehicles than conventional ones [63], however, they need to tradeoff price value when purchasing such a disruptive product [48,65,66].
H6: price value positively affects behavioral intention to accept IoV-based services.

Habit.
Habit is defined as the extent to which people tend to perform behaviors automatically as they learn how to use a new technology [41]. It usually reflects the consequences and results of prior experiences with a target technology [57]. at is when behavior is repeated, a habit is developed. Accordingly, the behavior is likely to be determined by habit strength [67]. e habit has been shown to have a direct effect on technology use, e.g., smart home [57]. erefore, the greater the habit of consumers to use the IoVbased services, the greater will be their intentions to adopt.
H7: habit positively affects behavioral intention to accept IoV-based services.

Perceived Risk.
Perceived risk relates to the uncertainty regarding the occurrence of adverse consequences [16,68,69]. It is defined as the degree to which consumers feel uncertainty and problems about the possibility of negative consequences from accepting a technology and drawing from perceived risk theory [42]. Consumers often weigh benefits and risks beliefs about technology through cognitive processes that inform their decisions to accept or reject it [34]. Numerous studies use perceived risk in analyzing consumer behavior related to innovative technology and find it is negatively related to intention to use such innovative technology as IoT [70], smart home services [68], e-government [71,72], social commerce [73], Internet banking [74][75][76], mobile payment [77], particularly automated vehicles [11,47], and autonomous delivery vehicles [45]. Dialectically, everything has two sides. Many surveys have reported that while consumers acknowledge the potential benefits of IoV-based services, they have also expressed great concerns about safety and privacy risks associated with the acceptance of IoV-based services [3,6]. Safety risk includes the malfunctioning risk due to operating system/equipment failure/crash, virus attack, or disconnection from the Internet [16], as well as vehicle performance degradation when operating in such complex conditions as poor weather, night, limited visibility, and in areas with low Internet coverage [3]. Privacy risk originates from the possibility that travel data, behavioral data, or personal information could be transmitted to other organizations for misuse without notice or be hacked by others [11,48]. Previous studies show that users seem to have privacy concerns when being confronted with the concept of connected vehicles because users of connected vehicular services may be concerned with the service providers' handling of data in terms of information collection, storage, and usage [35]. Such risks are considered as the most important barriers for acceptance of IoV-based services [10,16]. erefore, the more perceived risk consumers assess, the less consumers are willing to accept IoV-based services.
H8: perceived risk negatively affects behavioral intention to accept IoV-based services.
3.9. Initial Trust. Initial trust refers to the trustorʼs subjective knowledge and confidence of trustee without experience and knowledge [35,43]. In technology adoption research, trust is the expectation of adopter toward the adoptee to satisfy a certain service or fulfill a certain promise. Under risky or uncertain situations, trust is treated as a vital component of a relationship [8,78]. Numerous studies have extensively explored trust as an important determinant to affect user acceptance of various technology services, e.g., mobile banking [64,79], e-government [71], IoT in agriculture [70], particularly autonomous vehicles [11,32], and autonomous shuttle [46,48]. An individual tends to accept an innovation only if they believe that the specific attributes of this innovation will help achieve one's goals [80]. If trust is absent, even individual who tends to make adoption decision may start to defer adoption or hold a wait-and-see attitude until the market of radical innovation (IoV-based services) matures. At the early stage of the marketization of IoV-based services and vehicles, potential consumers need to shape sufficient trust that can help to increase the expectation of successful application of relevant IoV-based services, overcome perceptions of risks and uncertainties, and form a positive attitude towards it [11,43].
H9: initial trust positively affects behavioral intention to accept IoV-based services.

Measurements.
A survey questionnaire was conducted to collect data for this study. Based on an extensive review of the literature, each construct and its items were adapted from validated instruments of previous studies, and wording was modified to fit the context of IoV-based services and vehicles [72], to improve content validity [50,81]. e method of translation and back translation was used for ensuring equivalence between the source and translated versions [82]. Appropriate modifications were made based on expert reviews by two IS experts. Several postgraduate students pretested the instrument, identifying and revising ambiguous or poorly worded items to improve the clarity and understandability. e instrument was then pilot tested with 30 undergraduate students who were not included in the main survey.
e results of pilot test confirmed the reliability, validity, and translational equivalence of the scales. e final items and their sources are listed in Table 2. All items were measured using a five-point Likert scale ranging from "strongly disagree" to "strongly agree."

Data Collection.
e data were collected through an online survey by Wenjuanxing (http://www.wenjuan.com), a professional online questionnaire survey platform in China, from 22 to 26 June 2020. e survey targeted consumers who are using or intend to use an IoV-based service and vehicle. e survey includes a short explanation of the definition and functions of IoV-based services to help respondents understand this concept. e questionnaire was distributed randomly to actual respondents by Wenjuanxing. A total of 420 questionnaires were collected. Excluding invalid questionnaires with unusually short completion time, incomplete data, or the same options, there were 362 valid questionnaires for this study. e demographic characteristics and driving-related information of the respondents are shown in Table 3.
As seen in Table 3, the demographic data indicate that the sample was balanced regarding gender (51.38% of the respondents are male) and age (51.66% of the respondents are less than 30 years old), consisting primarily of bachelor degree and above (93.37%).
Nonresponse bias generally occurs when some of the target individuals are unwilling or unable to participate in the survey and, consequently, causes an unreliable representation of the selected sample and limit a study's external validity [84]. Following the recommendations of Urbach and Frederik [85], we took measures to address the issue of nonresponse and make sure we had a representative response rate, both before, during, and after data collection [86]. To minimize nonresponse before and during the data collection, all respondents were provided with a monetary incentive to participate in the questionnaire. Besides, following the initial invitation to participate in the questionnaire, all of the respondents were recontacted to remind them to complete the survey. After the data collection, nonresponse bias was assessed by verifying that the responses of early and late respondents were not significantly different [84,87]. In both samples, the early and late responses had insignificant differences in the means of any demographics based on T-tests using SPSS. erefore, nonresponse bias is not a major concern in this study.
Taking into consideration that all data were collected from a single source at one point in time and that all data were perceptions of key respondents, we used two approaches, i.e., Harman's one-factor test [88] and full collinearity assessment approach [89], to test the existence of common method bias (CMB), based on the guidelines of previous studies [44,54,90]. First, Harman's single-factor test shows that the highest variance explained by the single factor is 44.95%, which is less than the threshold value of 50%, indicating the absence of CMB in the dataset. Second, CMB is examined through a full collinearity assessment approach in PLS-SEM [89]. e variance inflation factor (VIF) values of all the latent variables are less than 4.58 (see Table 4), which is below the acceptable threshold of 5. Taken together, these results indicated that CMB was minimal in this study.

Data Analysis and Results
Data were analyzed using both the PLS-SEM method with Smart PLS 3 for validating the measurement model and structural model [81], and the fsQCA 3.0 software for exploring the set relations of the casual and outcome conditions [38]. ese two methods have different focuses and rely on different principles [91]. e PLS-SEM is a variableoriented technique that focuses on the net effect of the independent variable (IV) on the dependent variable (DV). It treats independent variables as competing to explain the variation in the dependent variables, and it relies on the principles of additive effects, linearity, and unifinality [40]. On the contrary, the fsQCA is a case-oriented technique that focuses on combinatorial effects. It assumes complex causality and focuses on asymmetric relationships between conditions (IV) and outcome (DV) [91]. FsQCA also allows for multifinality in which identical conditions can lead to different outcomes [39]. us, fsQCA is considered as an appropriate complementary analysis to PLS-SEM when detecting effects caused by unobserved heterogeneity [40,[92][93][94].

PLS-SEM Analysis.
e variance-based PLS-SEM method, a symmetric approach for modeling, is widely used in various studies [90,95]. PLS-SEM enables to estimate the complex models with many constructs, indicator variables, and structural paths without imposing distributional assumptions on the data, which is often recommended when the focus of research is prediction rather than hypothesis testing when the sample size is not large, or in the presence of noisy data [81,95]. Following the two-step approach recommended by Zhou [79], the measurement model was used to assess reliability and validity, and the structural model was examined to test hypotheses and model fitness.

Measurement Model.
A measurement model was used to assess the reliability and validity [81]. e internal consistency reliability of the scales was measured by Cronbach's alpha (CA) and composite reliability (CR) [96]. Table 4 shows that CA and CR of all constructs are more than 0.7, thus confirming the excellent reliability of scale [97]. e indicator reliability was evaluated based on the criterion that the loadings should be greater than 0.70. As shown in Table 4, the loadings (in bold) are greater than 0.708 [81], meaning that the instrument presents good indicator reliability.
Construct validity includes two fundamental aspects, i.e., convergent validity and discriminant validity [97]. Convergent validity, the extent to which the construct converges to explain the variance of its items, is determined by average variance extracted (AVE) [96]. Table 4 shows that AVE is greater than 0.5 for all of the constructs, meaning that the instrument presents good convergent validity [81].
Discriminant validity is the extent to which a construct is empirically distinct from other constructs in the structural model. e discriminant validity of the constructs is examined by using two criteria, i.e., the Fornelle-Larcker criterion and cross-loadings' criterion. Fornell-Larcker criterion requires that the square root of AVE be higher than Given the chance, I predict that I should use IoV services in the future BI3 I likely have the intention to use IoV services to conduct driving the correlations between the constructs [96], and the crossloadings' criterion requires that the factor loading must be higher than all cross loadings [98]. As seen in Tables 5 and 6, both criteria satisfy the discriminant validity of the constructs.

Qualitative Comparative Analysis.
FsQCA provides suitable methods to accommodate complex complementarities and nonlinear relationships among constructs [37]. FsQCA has attracted the attention of researchers in several fields, e.g., strategy and organization research [99,100] and information management [94,101]. erefore, this study employs fsQCA for a detailed analysis of how causal conditions contribute to an outcome [39,100]. e key stages of a fsQCA study contain model building, sampling, calibration, data analysis, reporting, and interpretation of findings [99]. Before analysis, data should be calibrated first from the original five-point Likert scale into a dataset suitable for data analysis by fsQCA.

Calibration.
A five-point Likert scale provides information to calibrate these variables; however, the actual sample is not normally distributed [92]. erefore, the mean of each variable was chosen as the crossover point [100], rather than the median of a five-point Likert scale [58]. Using the fsQCA 3.0 Software, calibration is automated completed and shown as Table 7.

Analysis of Necessary Conditions.
is study begins with the analysis of necessary conditions [91,102]. An analysis of necessary conditions determines whether any causal condition can be regarded as a necessary condition for the outcome to occur [103]. Drawing upon the recommendations [44,58,99,103], a condition is necessary when its consistency score is greater than the threshold of 0.9. Table 8 presents the results of the analysis of necessary conditions considering both the presence and the absence (∼) of the conditions for two outcome variables, i.e., behavioral intention to accept IoV services ("BI") and the negation of behavioral intention to accept IoV services ("∼BI"). e results show that none of the conditions alone, except PE (consistency is 0.932) and EE (consistency is 0.931), is necessary conditions for the outcome "BI." Besides, none of the conditions alone is a necessary condition for the outcome "∼BI," except "∼HA."

Analysis of Sufficient Conditions for Behavioral
Intention to Accept IoV Services. To explore which conditions might consistently result in the intention to accept IoV-based services, the truth table algorithm described by  [38] was used. e truth table algorithm proceeds in a twostage analytic procedure, to empirically complete this identification of causal processes. e first step is creating a truth table with 2 k (k is the number of causal conditions) rows from the fuzzy data, reflecting all possible combinations of causal conditions that lead to a specific outcome ("AI") [58]. e second step involves specifying the causal conditions and outcomes to minimize. e number of rows is reduced in line with two principles: (1) the minimum number of cases (frequency) required for a solution to be considered and (2) the minimum consistency level of a solution [38]. In this study, we set frequency thresholds to be 5 and consistency thresholds to be 0.8 (the actual value was 0.867) to exclude less important configurations; then, the configurations selected captured 82% of the cases, which meet the recommendation that a threshold is chosen that retains at least 75-80% of the cases [99]. Based on the "Standard Analyses" procedure, the complex solution, intermediate solution, and parsimonious solution are automatically derived based on the different treatment of the remainder combinations [38], and these solutions can be distinguished based on "easy" and "difficult" counterfactuals [37]. e complex solution includes all possible configurations of conditions (i.e., includes neither easy nor difficult counterfactuals). e parsimonious solution offers vital conditions that can be either easy or difficult counterfactuals. e intermediate solution offers vital conditions based on easy counterfactuals [37]. e notion of causal conditions belonging to the core or peripheral configurations is based on these parsimonious and intermediate solutions: "core" conditions are decisive causal ingredients (those that are part of both parsimonious and intermediate solutions), and "peripheral" conditions are those that are eliminated in the parsimonious solution and thus only appear in the intermediate solution [100]. Table 9 shows the results of fsQCA providing the solution for the outcome (high-level intention to accept IoV-based services) that distinguishes between "core" and "peripheral" conditions. e results shown in Table 9 indicate none of the isolated antecedent is a sufficient condition for the presence of "AI" and reveal seven equifinal configurations (combinations of conditions linked to the outcome) grouped by their core conditions [100]. All configurations are sufficient because all of the consistency scores resembles the notion of significance in statistical models [92,102] and exceeds the threshold of 0.8 in the sufficiency analysis [38,99]. Besides, each configuration's coverage, such as an analogous measure of R 2 in regression analysis [92], is greater than 0.

Behavioral intention to accept IoV-based services
us, all configurations are empirically relevant [38,44]. According to the solution coverage value, the overall solution accounts for 86.9% of the cases associated with high intention to accept IoV-based services.

Analysis of Sufficient Conditions for the Negation of
Behavioral Intention to Accept IoV Services. Contrary to most traditional techniques as SEM [104,105] and regression model [94,101], fsQCA focuses on causal asymmetry [38,44,91,100]. FsQCA can be used to analyze separately the configurations for the presence and the absence of an outcome [99]. erefore, to check this asymmetry, fsQCA conducts another set of analyses in which the negation of behavioral intention to accept IoV services represents the outcome and is coded as the inverse (∼BI). To explore which conditions consistently lead to this outcome, the same frequency (5) and consistency (0.8, the actual value was 0.858) thresholds are set. e solution for the negation of behavioral intention to accept IoV services is shown in Table 10.

Mobile Information Systems
As shown in Table 10, seven identified solutions comply with the recommended consistency and unique coverage thresholds [38]. e overall solution coverage indicates that causal conditions account for 84.4% of cases in the solution.

Key Findings.
is study aims to show how the analysis of net and combinatory effects of specific antecedent variables can improve the understanding of behavioral intention to accept IoV services [91].
e PLS-SEM results show that performance expectancy (β � 0.150) has a significantly positive effect on consumers' behavioral intentions. e analysis of necessary conditions of fsQCA also confirms that performance expectancy (consistency � 0.932) is a necessary condition for high behavioral intention. Moreover, PE is indeed core conditions in nine of the twelve configurations for behavioral intention, reinforcing the findings of PLS-SEM. e relevance of performance expectancy is consistent with the previous studies that highlight the importance of the users' expectancy of performance to achieve the behavioral intention, e.g., autonomous car [7] and autonomous delivery vehicles [45]. Such smart technology usage should be simple and obvious, surprisingly, effort expectancy, as an important functional characteristic of technology, has no significant effect on the behavioral intention from PLS-SEM results. is result is in line with previous research on the acceptance of emerging technologies as smart home [57] and autonomous delivery vehicles [45], but contradicts results from some studies on the adoption of autonomous car [7]. Nevertheless, the results from fsQCA confirm that effort expectancy (consistency � 0.931) is a necessary condition of consumers' behavioral intention. Moreover, it is indeed core conditions in eight of the twelve configurations for behavioral intention, complementing the findings PLS-SEM failed to detect. In such circumstances, IoV-based services and vehicles are not well understood by the users, and they did not Table 9: Sufficient configurations for high intention to accept IoV-based services. Solutions  1a  1b  2  3a  3b  3c  3d  4  5  6  7a 7b Perceived expectancy   have sufficient information and they were not familiar with the technology behind the concept. e level of information provided could reduce the perceived complexity, as the less the technology is perceived as complex, the more likely it is that the innovation will be adopted [57]. Different from the significant effect of SI and FC in previous studies [29,45], PLS-SEM results show that neither SI nor FC has a significant "net effect" on behavioral intention to accept IoV, which is aligned with previous studies that find a nonsignificant direct effect of SI on behavioral intention [55,57], and nonsignificant direct effect of FC on behavioral intention [54,76]. Nevertheless, in the fsQCA result, SI is present as a peripheral condition in six of the twelve configurations for BI and is absent as a core condition in four of the eight configurations for the negation of behavioral intention to accept IoV-based services, suggesting the relevance of SI on AI.

Causal condition
e result is consistent with numerous studies in IoT [106], Internet banking [75,76], e-government [71], autonomous vehicle [4,7,13], and automated public transport [28]. Besides, the presence of FC is a peripheral condition in four of the twelve configurations for behavioral intention to accept IoV-based services and the absence of FC is a core condition in four of the eight configurations for the negation of behavioral intention to accept IoV-based services, showing the existence of causal asymmetry in a complex context. e result confirms results from previous research carried out on technology acceptance such as mobile banking [60,64] and automated public transport [29]. Overall, the absence of SI and FC as core condition in four solutions for the negation of behavioral intention to accept IoV-based services, especially in solution 3 and solution 4 (the absence of SI or FC, even if all else conditions are present, will determine the outcome the absence of BI), indicates that the lower SI and FC about potential consumers, the lower the acceptance of IoV-based services. e results show that IoV-based service/vehicle is a new technology not yet widely adopted and the intention to purchase an innovation is stronger when the innovation is already used by others. Once the concept becomes part of daily life, the impact of SI could be higher. e PLS-SEM results show that hedonic motivation (HM) does not significantly impact behavioral intention (BI) to use IoV-based services. Nevertheless, fsQCA results indicate that HM is present as a core condition in seven of the twelve configurations for BI and is absent as a peripheral condition in three of the eight configurations for the negation of BI to accept IoV-based services, suggesting the lowlevel significance of HM on BI. e finding is contrary to previous studies on the acceptance of autonomous delivery vehicles [45], automated road transport systems [29,46], social networking service [107], wearable technology [61], and mobile health [58], while it is aligned with some studies such as smart home acceptance [57] and consumer e-loyalty [108]. In summary, IoV services are perceived by connected autonomous vehicle consumers as more of an intelligent utilitarian solution rather than a hedonic one.
Price value (PV) (β � 0.153) is found to be a significant determinant of behavioral intention (BI) in SEM-PLS results, which is in line with previous research on the acceptance of emerging technologies, e.g., autonomous delivery vehicles [45], autonomous electric bus [31,48], Internet of things [54], and electric vehicles [65]. is means that consumers consider the benefits of IoV-based services against the monetary value. Furthermore, fsQCA results indicate the absence of PV, as a core condition in four of the eight configurations for the negation of BI to accept IoVbased services, suggesting the low price-performance of consumers' perception on IoV-based services is the barrier of accepting IoV-based services. ere is a price premium on IoV-based services compare with the conventional vehicle. Consumers who believe that the price of IoV-based service is justified by the potential benefits have a stronger behavioral intention to adopt. Talebian and Mishra suggest that the automobile will be nearly homogenous in about 2050 only if prices decrease at significant rates (15% or 20% annually) [16].
e PLS-SEM results show habit has the strongest effect (β � 0.34) on behavioral intention, as confirmed in previous studies on the acceptance of emerging technologies, e.g., smart home [57]. Besides, the fsQCA results also confirm that the negation of habit (∼HA) is a necessary condition (consistency � 0.930) for the negation of behavioral intention (∼BI), and the absence of habit is indeed a core condition in four of the eight configurations for behavioral intention, indicating the lower the habit to use IoV-based services, the less likely it is to accept IoV-based services. erefore, for the emerging and not widely used technologies, the lack of habit may be the biggest obstacle to technology adoption.
Despite perceived risk has been frequently cited as one major concern in accepting IoV-based services in several studies [3], in this study, PLS-SEM has failed to identify its significant net effect on behavioral intention to accept IoVbased services.
is result is aligned with the previous studies, e.g., social commerce [73], e-government [72], autonomous vehicle [32,34], and connected vehicle [35]. Interestingly, fsQCA results show that the absence of perceived risk is indeed a core condition in six of the twelve configurations for behavioral intention, which corroborates previous studies that observe a negative effect of perceived risk on the adoption of innovative technologies, e.g., mobile payment [59,77,109], Internet banking [74][75][76], IoT in agriculture [70], social media purchase [110], electric vehicles [111,112], and autonomous vehicle [8,10,45,47]. Previous empirical findings are rather mixed, resulting in the argument that perceived risk cannot be seen as a steady predictor of IoV-based services acceptance [8,45]. FsQCA results show that perceived risk has been proven as a core condition (an important determinant) in six configurations, indicating the lower the risk perception by potential users, the higher the acceptance of IoV-based services.
It is found from PLS-SEM that the role of initial trust (β � 0.41) is much stronger in influencing behavioral intention to accept IoV-based services than other perceptual factors. Although "the presence of initial trust" is just a peripheral condition of configuration for behavioral intention, however, "the absence of initial trust" is indeed a core condition in two of the seven configurations for the negation of behavioral intention (i.e., the presence of BI), suggesting that the lack of initial trust in IoV-based services will also reduce the acceptance intention. e significance of initial trust in shaping behavioral intention corroborates previous studies, e.g., mobile banking [79,113], cloud technology [114], e-government [72], autonomous vehicles [8,15,32,34], and autonomous shuttle/bus [31,46,48]. is finding suggests that trust, as "a tool for the reduction of cognitive complexity," can help to simplify and facilitate the decision-making process, especially in situations with risks and uncertainty [11,115]. erefore, insufficient trust has been thought to be one of the major psychological barriers to the wide adoption of IoV-based services [15,34].
In summary, results from PLS-SEM reveal that higher levels of performance expectancy, price value, habit, and initial trust increase behavioral intention. Moreover, results from fsQCA which is used as a supplementary analysis technique [93] and reinforces the symmetric findings of PLS-SEM, as well as offer additional novel interesting, and more nuanced insights that indicate a combination of the conditions needs to be taken into account to explain the outcome of behavioral intention. Five configurations for "BI" and seven configurations for "∼BI" are identified. is synergetic effect could not be captured by PLS-SEM since it examines the condition in isolation from the other conditions [44,92].

eoretical Implications.
is study has two key theoretical contributions. Firstly, this study takes an integrated approach towards technology acceptance in the context of the IoV-based services, which complements and extends technology acceptance studies. As different theories emphasize the different insights regarding technology acceptance, most researchers have called for a holistic and comprehensive approach that integrates more than one single theoretical perspective for understanding the acceptance of innovative technologies [31,46,64,116], to make full use of the advantages of different models, and make up the deficiency of the one-sidedness from a single model. erefore, to provide broad coverage of factors for IoVbased services acceptance, this study answers this call by developing an integrated model incorporating UTAUT, perceived risk theory, and initial trust model. e net effect analysis of this integrated model shows an explained variance (R 2 ) of 78% for behavioral intention to accept IoV services, indicating a moderate explanatory power.
Secondly, this study contributes to the IoV-based services' acceptance literature at the methodological level. Existing studies on the topic generally rely on multiple regression models, SEM, and PLS methodology and view consumer's intention to accept IoV-based services primarily as the outcome of several isolated antecedents. Besides the PLS-SEM, the study uses the fsQCA to investigate which configurations of determinants lead to the intention to accept IoV-based services and which ones lead to the negation of intention to accept IoV-based services. To the best of our knowledge, this study is the first to use this type of analysis to explore the intention to accept IoV-based services.
is study demonstrates that the fsQCA offers much in terms of understanding how various determinants explain the acceptance of IoV-based services, more so than the PLS-SEM. Besides, this study by fsQCA also answers the call for the application of this technique to complex behavior research, since it can offer new insights into understanding the IoV-based services acceptance phenomena [44,94]. erefore, the value of this study lies in the effort to describe combinatorial complexities assuming asymmetrical relationships between variables, rather than symmetrical net effects that PLS-SEM estimates [105].
6.3. Managerial Implications. From a practical point of view, this study's findings provide several managerial implications for automobile manufacturers, marketers, and policymakers. Based on our findings of PLS-SEM and fsQCA, they should focus on different configurations of antecedents that give rise to an outcome, rather than the impact of these antecedents isolation from each other.
First, as suggested by our findings of PLS-SEM and fsQCA, both the utilitarian benefits (i.e., performance expectancy, and effort expectancy) and hedonic benefits are favored by potential users. Hence, automobile manufacturers and marketers should focus more on the development and the marketing communication activities of IoV-based services on the utilitarian benefits and hedonic benefits compared to conventional vehicles. Moreover, lower perceived risk is also found to contribute to IoV-based services acceptance to a large extent. Besides, if potential risks are not well understood thoughtfully, they could slow IoV-based services adoption rates to socially suboptimal levels. erefore, marketers should take perceived risk reduction into consideration when promoting IoV-based services in society. Most importantly, high quality and low price, the best cost effective. Price is the most stimulating and sensitive factor influencing consumers' purchasing behavior. Sell IoV-based services at a premium decreased consumers' willingness to pay, hence, manufacturers and sellers should give consumers discounts and cut prices to more reasonable levels.
Second, habit and initial trust is a major construct for explaining the adoption of IoV-based services, with the first and second-largest effect, respectively, and meanwhile, the lack of habit and initial trust is likely to lead to the lower intention to accept IoV-based services. As IoVbased services are in its initial stage, manufacturers, marketers, and policymakers should be more concerned about the measures to increase the individuals' habit and initial trust toward this innovative technology. erefore, consumers' habit formation and trust building become an urgent mission for these stakeholders [8]. For instance, exhibitions and user experience activities of IoV-based technologies and services should be held to help consumers observe and directly experience IoV-based services, such experiences and knowledge are conducive to the formation of habit and trust toward IoV-based services [80] and are critical ingredients for widespread and rapid technology diffusion [53].
Finally, the lower level of social influence and facilitating conditions also lead to a lower level of behavioral intention to accept IoV services. When the individuals' knowledge levels about a specific technology are low or the technology is still in its initial stages, and public communication significantly influences societal acceptance [8]. For the acceptance of IoV-based services, it is of particular relevance to what the social environment (i.e., family, friends, and acquaintances) thinks, what the media reports, and what opinion experts reflect [16]. Hence, marketers might use social influence when promoting IoV-based services during the market introduction stage.

Limitations and Future Directions.
Our study has several limitations that could be addressed in the future. First, we did not take into account other factors that might affect consumers' intention to accept or purchase IoV-based services. Our model based on UTAUT2, perceived risk theory, and initial trust theory only considers some important factors, but it is far from reaching a comprehensive explanation. In the future, the extent to which consumers' purchase intentions are affected can be explored based on other consumer behavior and technology adoption theories that match the characteristics of IoV-based services. Second, this study uses data collected through a purposive sample in a single country, China, which implies that our findings reflect only the situation in this nation, and may affect the generalization of our findings due to differences in the cultural environment and political system of countries. erefore, a future study may consider a comparative study across different countries, e.g., the U.S. of North America, Germany of Europe, Japan, South Korea, and India of the Asian-Pacific region, and Brazil of South America.

Conclusions
IoV-based services, as radical innovations for changing transportation fundamentally, introduce all sorts of different benefits such as safety, energy efficiency, environment improvement, increased mobility, and more entertainment in driving [3-5, 9, 14]. IoV-based services are not widely accepted by consumers without understanding IoV-based services diffusion. us, investigation of the determinants leading to consumers' intention to accept and purchase IoVbased services is significant for both academics and practitioners. is study identifies the determinants of behavioral intention to accept IoV services by using an integrated model that combines UTAUT, perceived risk theory, and initial trust theory. is study uses both symmetric (PLS-SEM) and asymmetric (fsQCA) methods to explore the role of determinants in consumers' intention to accept and purchase IoV-based services. Specifically, the net effects of each antecedent factor on intention are analyzed by conventional correlational techniques (PLS-SEM). e direct effects of performance expectancy, price value, habit, and initial trust on intention are found to be significant. Despite the determinants (e.g., effort expectancy, social influence, facilitating conditions, hedonic motivation, and perceived risk) are found to be nonsignificant effects on intention, however, it cannot be said they are not important to intention to accept IoV-based services, due to the existence of causal complexity. For the high levels of causal complexity, fsQCA provides a more nuanced understanding of how these antecedent conditions fit together to affect consumers' intention to accept and purchase IoV-based services. e results from fsQCA provide twelve different configurations to achieve high levels of behavioral intention to accept IoV services and eight causal paths equifinally to lead to the negation of behavioral intention to accept IoV services. e findings provide relevant insights and marketing suggestions for incentivizing consumers to accept IoV-based services.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.