Governance in the Internet of Vehicles (IoV) Context: Examination of Information Privacy, Transport Anxiety, Intention, and Usage

Taylor’s Business School, Taylor’s University, 47500 Subang Jaya, Malaysia University of Economics and Human Sciences in Warsaw, Okopowa, Warsaw, Poland School of Business and Economics, Universiti Putra Malaysia, 43400 Serdang, Malaysia Malaysia and EIS-UPMCS Centre for Future Labour Market Studies, SOCSO, Kuala Lumpur, Malaysia Accounting Research Institute and Faculty of Accountancy, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia Accounting Research Institute and Faculty of Computer Science and Mathematics, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia Department of Accounting, College of Business Administration, Prince Sattam Bin Abdulaziz University, 11942 Al-Kharj, Saudi Arabia Islamic Business School, Universiti Utara Malaysia, Sintok, 06010 Bukit Kayu Hitam, Malaysia


Introduction
e modern age of the Internet of things has motivated the advancement of traditional Ad-hoc based network of vehicles into entirely a modern concept of the Internet of Vehicle. e Internet of Vehicle concept has emerged as the greater research value with a major interest in commercial technologies and computation as the prompt development [1][2][3]. In recent years, programmes related to smart cities have grown rapidly and users are switching from central computers to community data centres. As a result, the transportation markets are urgently in need of strategies that can enhance driving safety, vehicle safety, and perhaps even the cost of reducing prices [4]. ere is a related spike in the number of deaths caused by accidents when too many people drive automobiles and vehicles. Progressively linked to the Internet of ings (IoT), such automobiles create the Internet of Vehicles (IoV). e Internet of vehicle concept is the integration of IoT and Internet access for vehicles. For the motor industry, this is an emerging sector and an essential part of new technologies [5]. Today, IoT also makes it possible to connect cars, houses, and other objects equipped with sensors, electronic controls, and access points. e Network of Vehicles is known through the use of the Internet of ings (IoT). Similarly, the Internet of Vehicles (IoV) is used to create connectivity between objects, automobiles, and locations for the transmission between channels of data or information. [6]. Each automobile is conceived as just an autonomous object, configured with sensor applications, computer equipment, control units, including storage areas, as well as linked through vehicle-to-everything connectivity to any individual. Intelligent vehicles will take on multiple tasks, i.e., also being a customer and a database, taking and distributing big data resources, leading to several new applications, through aided driving through platooning, a safe collaboration of knowledge to vehicle control and implementation [7]. e proliferation of social media rewards individuals as well as organizations with unparalleled confidential information which was once difficult to obtain. Doubtlessly, privacy issues are becoming critically relevant on social networking sites as vendors also can eventually obtain access to a broad disclosure of personal data from users [8]. Various stakeholders, particularly corporate leaders, privacy advocates, academics, government officials, and individual users, are increasingly concerned about data privacy. Surveys show that Internet users are highly worried and confirm that privacy is the highest concern for corporations [9]. Privacy problems did not forecast Smartphone usage, inconsistent with the position including its privacy issue. Users concerned with privacy, nonetheless, were less likely to utilize resources and exchange information and were much more inclined to use precautions to protect confidentiality. Except for exchanging information, the interactions were equivalent in terms of intentions and behaviour [10]. Sharing is correlated with distinct types to privacy concerns relative to more conventional web services, which could exacerbate the degree of privacy concerns, requiring clear theoretical reasons for the sharing actions of users amid privacy concerns [11,12].
A key concern of the digital world is privacy. Advancement in information and communication technology (ICTs), as well as their broad acceptance, has increased the quantity of private information gathered by business and government agencies exponentially. While ICTs including fitness apps, headphone amplifiers, as well as social networks give alternative ways for users to connect and learn regarding themselves; these also present a variety of hazards to privacy [13].
Privacy is not a concern if individuals were not concerned with automated vehicles. Since people might be automated vehicle customers and buyers, it is crucial for the sustainability of these emerging forms of personal transportation to consider when and where confidentiality and autonomous driving will communicate. Whenever an entity is associated with such an autonomous car, the interests of privacy appear relevant [14]. Internet-of-things feature enables Telco's, such as private insurers, to collect large quantities of data about car users that are vulnerable to privacy. If a small financial reward is given, however, individuals are willing to sacrifice their privacy towards car insurers. Users consider behavioural privacy more important than local and spatial privacy [15]. e vehicle has become a powerful sensor network within the Autonomous Vehicle System, absorbing data from the world, from several other vehicles (as well as from the car owner) including feeding that to other vehicles and facilities to support in smooth operation, environment protection, and road traffic.
e Automobile Grid is another Internet of ings (IoT), that we refer to as the Internet of Vehicles (IOV), which is capable of making its very own choices about driving people to their targets [16,17]. e core idea of an Internet of ings (IoT) is always to equip computation, processing, and network control with physical objects to allow socialisation amongst them. e Internet of Vehicles (IoV) is indeed an IoT affiliate that has made substantial strides using networking technologies. Vehicles linked through the Web are able to exchange information which can greatly increase the efficiency of traffic congestion [18]. A Social Internet of Vehicle becomes the automobile application which has transformed the current Intelligent Transportation System (ITS) and Vehicle Call Services into the next level of Intelligent by incorporating socialising and constant communication aspects. Furthermore, as information is collected as well as processed at various layers of all its infrastructure, the task of privacy management appears important [19]. e present research study tries to explore the concern of information privacy for the Internet of the vehicle (CFIO-VIP). e study intends to explore a causal link between concern for the Internet of vehicle information privacy and behaviour intentions for the transportation community.

Present Measurement Instruments for the Concern of Information Privacy.
e concern for information privacy was first measured by Smith, Milberg, and Burke [20] using four-dimensional factors: improper access, unauthorised secondary usage, errors, and collection. e measurement and validation of four dimensions of concern for information privacy were confirmed based on first-order constructs. Later, a research study conducted by Stewart and Segars [21] confirmed and validated the concern for information privacy model as developed by [20]. However, in response to the firstorder model developed by Smith et al. [20], the study conducted by Stewart and Segars [21] tested and validated their research based on a model with more complex dimensions with second-order constructs. Similarly, a research study as carried by Malhotra, Kim, and Agarwal [22] tested and confirmed a first-order model for the information privacy for Internet users with three dimensions, namely, awareness, control and collection. In contrast, a study that was examined by Rose [23] as the cross-country regulatory comparison for the concern for information privacy indicated some differences for US and non-US data. e comparison validated one-factor information privacy concern model for US data while second-order information privacy concern was confirmed for non-US data. Similarly, a research thesis as investigated by Yau [24] revealed a significant difference between different groups of respondents for the concern for information privacy in the case of New Zealand research. e study concluded that upper level of respondents is more concerned with information privacy rather than lower-level respondents in terms of social status. In contrast, a study was conducted by Bélanger and Crossler [25] on the four dimensional model of concern for information privacy at individual, organizational, and societal levels with different order constructs. At individual level, the first-order model was validated, while on organizational and societal levels, the second-order model was confirmed and validated for the concerns of information privacy using the student-based data in the US environment.
Different studies also considered the limitation and usage of measurement and scale development of concern for information privacy in a different context, like the study of [26,27]. e study concluded with the limited use of information privacy concern's scale development with differences in dimensions for non-US studies and concluded with an argument that a single model and order is not viable for different context and nations. However, scale and dimension can be validated differently as per the requirements of the studies. Likewise, the study conducted by Lee, Chung, Han, and Song [28] considered the scale development and measurement of concern for information privacy in a medical context. ey confirmed a 2 nd order model for the concerns for information privacy in medical context based on three dimensions namely; errors, collections, and secondary usage and access. e study conducted by Osatuyi [8] considered the context of social media for testing the basic model of information privacy concern. e study confirmed the second-order construct based on the dimensions of errors, collection, secondary usage, and unauthorised access. Similarly, another piece of research study concluded with four dimensional 2 nd order construct for the concern of information privacy with the dimensions like unauthorised access, secondary usage, collection, and personal information providing concerns [29,30]. However, in response to the high expectations of South African citizen for online information privacy concern, it was not practical to ensure due to variation in regulatory requirement of the country to apply completely the second-order model for non-US data [31]. e study tested the model based on eleven different principles of online information privacy as opposed to the basic privacy of information concern as developed by [20].

Concern for the Internet of Vehicle Information Privacy (CFIOVIP).
e concept of the Internet of the vehicle and its related issue of privacy concern found its origin from the study of Horswill and Coster [32], where the researcher aimed to analyse the impact of age, mileage, and gender of vehicle driver with car usual and expected performance. ey argued that if the vehicles are connected through a servicer like the Internet of vehicle, there are more chances of avoiding the risk associated with car driving. However, they did not use any proxy for estimating the Internet of a vehicle for their research, rather they suggested for its inclusion in future research. Similarly, another piece of research as conducted by Puri [33] considered a satellite-based technology named as "unmanned Arial vehicle," as the introductory application of Internet of vehicle concept in the transport industry for monitoring all types of vehicle and transports on the roads. However, a research study as examined by Barón and Green [34] considered some detailed measure for the performance of driving based on manual as well as speech to analyse their effect on the task performance of the vehicle.
A more detailed and comprehensive measure of the Internet of the vehicle was originated from the study of [35]. ey considered the IOV for the transport context with four different dimensions, namely, general information of the vehicle, safety information of the vehicle, motion control of individuals, and motion control of a group [36]. Similarly, a study conducted by Händel, Ohlsson, Ohlsson, Skog, and Nygren [37] suggested to include the smart-phone technology to automate the vehicle for traffic monitoring as well as for insurance purpose usage to measure the Internet of the vehicle concept. In contrast, some of the research studies considered knowledge management as the proxy for measuring the IOV to enhance the intelligent parking system [38]. More dynamic research for measuring the IOV through modern multidimensional categories includes an ad-hoc network of vehicles, smartphone networks, intervehicle network, intravehicle network, and network with beyond vehicle [39]. In contrast, a study as investigated by Ni et al. [40] argued about the measurement of the Internet of the vehicle under the dimensions of driver-assisted to automatic vehicles. Finally, a block-chain based system of the Internet of the vehicle can be more efficient in terms of information privacy concern for the transport context [41]. However, the block-chain technology has some loopholes with respect to the application with the Internet of the vehicle for considering the information privacy concern. e below model in Figure 1 is based on the present research for testing the transport anxiety using the Internet of vehicle concept by considering the concern for information privacy for transport context and its ultimate impact on the behavioural intentions of the ultimate users or driver of vehicles in the transport sector.

Development of Scale.
e present exploratory research investigation is aimed to explore the transport anxiety using the Internet of vehicle by considering the concern for Journal of Advanced Transportation information privacy and its ultimate impact on the behaviour intentions for the ultimate user or drivers of connected vehicles in the transport sector [42]. For this purpose, a survey-based questionnaire research instrument was developed as per the measurement and scales adopted by [8,20,21,28]. ese include the constructs like concerns of information privacy with all their four dimensions, namely, errors, collection, unauthorised access, and secondary usage along-with the other construct like behaviour intentions of users (Appendix A). e constructs measuring the concerns for information privacy for the Internet of the vehicle are measured on multiple item bases on a scaled from strongly disagree-strongly agree with five-point Likert values linear scale. Similarly, the measure for transport anxiety for the Internet of the vehicle (Appendix B) was adopted as the measurement and scale adopted from the studies of [35,39,43]. It was also measured on multiple item bases on a scaled from strongly disagree-strongly agree five-point Likert values linear scale [44]. However, the behaviour intention was measured on a linear scale of three points from highly likely to not likely. e first versions of survey questionnaire research instrument of this study were verified and approved from the expert and authorised persons from the transport sector as well as from the field of information privacy to validate the constructs and their relevant items related to the Internet of vehicle, concerns of information privacy, behavioural intentions, etc. Additionally, the remaining responses for the validity and reliability of item statement related to each dimension of the constructs were received from the student's community at undergraduate and graduate level for the clarity of question statements which helped towards the formation of the final version of the questionnaire.

Design of the Survey.
e present study used an online type of questionnaire survey for the collection of data related to the dimensions and constructs including their item statements. For this purpose, the questionnaire was shared through the personal e-mail address, including the social media sites of students studying the course of an information system at undergraduate and graduate level in different universities and using an automated and connected vehicle for the application of concept for the Internet of vehicle. Any data related to the personal information of the respondents were not demanded to maintain the privacy of the respondents as per the regulations of IRB (the data of surveybased questionnaire used to support the findings of this study are available from the corresponding author upon request). A total number of 425 online questionnaires were shared to targeted respondents but only 357 responses were received dully filled with all the required information with a rate of response of 84%. e demographic characteristics include age, gender, and Internet of vehicle (IOV) users.

Analysis of Data and Discussion on Results
e present exploratory study meant to analyse the validity and identification of structure for concern for the Internet of vehicle information privacy (CFIOVIP) and the evaluation of its nomological network. e process was executed on two phases: validation and identification of the structure based on the factors of CFIOVIP and evaluation of the model based on CFIOVIP's nomological network.

First
Step: Factor Structure Identification of CFIOVIP. e first step of analysis includes the identification and validation of factor structures, scales, and measurements for the concern for the Internet of vehicle information privacy (CFIOVIP) as per the requirements of the study and based on previous historical studies of [22,45,46]. For achieving this objective, the study executed at first the analysis of confirmatory factors and secondly exploratory analysis of factors was done for the various items statements related to the factors of CFIOVIP.
For the estimation of factor analysis using exploratory characteristics, SPSS software of IBM was utilized. Using the EFA feature of SPSS, the technique of principal component analysis with the help of VARIMAX rotation option as well as the normalization of Kaiser was executed. For this purpose, CFIOVIP was initially estimated with 4 adapted dimensions as used by [45,46]. Table 1 indicates that 3 components were explored using this function with clean item loading and having no cross-loading at all. e dimensions like secondary usage and unauthorised access indicated the same component's factors as used in the previous historical literature stated above. e factor's reliability was estimated using the process of exploratory factor analysis as suggested by [47]. e reliability as shown from the estimated figures of Cronbach Alpha for the factors finalized for exploratory factor analysis process is above the minimum acceptable limit of 0.70 as suggested by [48]. As, secondary usage and unauthorised access, their factors were also combined in the same column to reflect the factors for a single construct as shown in Table 2.
e confirmatory factor analysis process is used evaluate the effectiveness of the structure of the proposed model of CFIOVIP containing 16 items to measure the relevant constructs as per the similar estimation scales and measurements used by [45,46]. e present research has executed the procedure of confirmatory factor analysis process based on covariance matrices as originally adopted by [45]. As per the study examined by Osatuyi [45], the confirmatory factor analysis process enable the scholars to respecify, estimate, and originally specify univariate and multivariate model based on related dependencies and relationships and also enable the researchers to estimate the previously unobserved constructs. Table 3 depicts the model construction fit for factors structures based on 1 st order and 2 nd order constructs for CFIOVIP. e models 1, 2, and 3 indicate the factor construction based on 1 st order constructs for concern for the Internet of vehicle information privacy. Models 4 and 5 indicate the factor construction based on 2 nd order constructs for CFIOVIP.
As per the table estimates, the first model indicates the first-order construction of CFIOVIP based on 16 items based on 4 dimensions, namely, collection, error, secondary usage, and unauthorised access. e first model is based on par the construct study for a similar domain for the unidimensional factors by [46]. If the study accepts this model, it can be inferred that the CFIOVIP can be measured as one single construct based on 16 item statements from the proposed research. e table also indicates that the second model comprises of the 16 item statements for CFIOVIP is comprised of two main dimensions of the study. One dimension is the combination of collection and error constructs while the other is comprised of secondary usage and unauthorised access to information. It is also consistent with studies of [22,45]. Table 3 also indicates the third model based on three constructs: collection, error, and unauthorised access + secondary usage collectively measures the concern for the Internet of vehicle information privacy by considering all the 16 item statement as equally important for the measurement and estimations. is model was also similar to that from the study of [45]. e table also indicates the 2 nd order construct for the variable concern for the Internet of vehicle information privacy (CFIOVIP) based on three-dimensional factors, namely, collection, error, and unauthorised access + secondary usage. e RMSEA value for this model is as low as 0.001 which increases the validity and acceptability of 4 th model 2 nd order construct of CFIOVIP and is also consistent with the similar results from [45]. e table also indicates the 5 th model 2 nd order construction of CFIOVIP based on four dimensions like collection, error, secondary usage, and unauthorised access.
is model was established to compare it with the 4 th model. However, its RMSEA value is slightly higher than the 4 th model which decreases the validation of the fifth model as compared to the 4 th . ese results also verify the similar findings from the study of [45]. e fit indices of the model indicate that 1 st and 2 nd models are not acceptable based on the values from NFI, GFI, AGFI, CFI, and RMSEA. However, models 3, 4, and 5 are acceptable based on the criteria Journal of Advanced Transportation 5 defined. Model 4 indicates superior acceptability and validity based on the lowest RMSEA value of 0.001. Table 4 indicates the standardized factor loadings, t values, average variance extracted, and composite reliability which indicate the convergent validity of the constructs. For a construct to be valid and reliable based on factor loading, its standardized values along-with composite reliability should exceed the minimum criteria of 0.70 and its average variance extracted should be greater than 0.50, while the t values should be greater than 1.96 for significance as per the suggestions of [47][48][49]. e estimated results indicated in the table infer that all the constructs measuring the CFIOVIP are reliable and valid for the study which indicates the convergent validity based on strength properties of the constructs. e item statement and relevant constructs measuring the CFIOVIP indicates strong properties for 4 th and 5 th model. However, the 4 th model's properties indicate better estimates as compared to model 5.
e discriminant validity for construct measuring CFIOVIP is displayed in Table 5 which indicates the correlation's coefficient between the constructs like collection, error, and information access (secondary usage + unauthorised access). As per the suggestion of Fornell and Larcker [49], for the constructs to discriminately valid, the correlation coefficient's value between them should be less than the average variance extracted square root of each construct. e estimated figures of the table indicate that all the constructs, namely, collection, error, and information access, are valid and reliable according to the criteria of discernment validity.

Second
Step: Validation of CFIOVIP. e present study was aimed to test the validity of CFIOVIP construct based on nomological measures as suggested by [45]. e nomological approach requires a 2 nd order construct to be inserted between an input/explanatory variable and an outcome variable as per [45]. For achieving this objective, the 2 nd order   Journal of Advanced Transportation construct based on 4 th model of CFIOVIP was placed between the construct of transport anxiety and the behavioural intentions of transport users within the framework of the Internet of vehicle context. Table 6 reports the 2 nd order structural model based on models 6 and 7 based on 3 dimensions (collection, error, and information access) and 4 dimensions (collection, error, secondary usage, and unauthorised access), respectively. Table 6 reports the structural model goodness of fit indices in the form of NFI, GFI, AGFI, CFI, and RMSEA. e indices indicate that model 6 based on the 3-dimension structure for measuring CFIOVIP is stronger as compared to model 7.
e structure model based on models 4 and 6 is depicted in Figure 2 indicates the mediating relationship of CFIOVIP between transport anxiety and behaviour intentions of the users of the Internet of vehicle. Figure 2 above indicates a good fit and relevant loading values as per standardized threshold level for factor loading indicating a strong and positive relationship between the constructs like transport anxiety and behaviour intensions of IOV users which is strongly and positively mediated by CFIOVIP.

Conclusions
e present exploratory research study was mean to test the validity and measurement of concern for information privacy in transport context with the application of Internet of vehicle by applying and replicating the previous research studies basis like [22,45,46]. e previous study mainly considered the social media context for information privacy concern. e present research study tries to test the personal information concern in case of the Internet of vehicle context which may be a bigger issue in today's world of the connected vehicle. e information privacy domain requires a large set of different context to test their validity based on the existing model for different industries that requires the concern of information privacy for their users.
For this purpose, a survey-based online questionnaire was developed by carefully examining the item statements of previous research studies and adjusting those item statements based on the requirement of the present study by considering Internet of vehicle context for the constructs like transport anxiety, concern for information privacy, and behaviour intentions of the users. For this purpose, the exploratory and confirmatory analysis processes were performed for testing the validity of concern for the Internet of vehicle information privacy. A total number of 357 responses were received online which were complete in all respects from college and university students studying information system course at undergraduate and graduate levels.  Journal of Advanced Transportation e estimated results obtained from exploratory and confirmatory factor analysis revealed that 3-dimension construct containing collection, error, and information accuracy measuring concern for the Internet of vehicle information privacy (CFIOVIP) was validated through second-order construction with strong goodness of fit estimates. e relevant factors were reliable and valid as per their internal and external validity estimates of Cronbach Alpha, average variance extracted, composite reliability, discriminant validity, and factor loading values including their fitness indices for 16 items for 3-dimension construct of CFIOVIP. e factor loading values for all the 16-items for 3-dimensions like collecting, error, and information access were exceeding the minimum standard limit of 0.70. e Cronbach Alpha and composite reliability values for the same set of dimensions revealed the values exceeding the minimum required level of 0.70. Additionally, the average variance extracted was also exceeding the standardized minimum limit of 0.50. Similarly, the discriminant validity of the construct was also confirmed by having the values of coefficients of correlation for collection, error, and information accuracy less than the average variance extracted square root value. Finally, the model fit indices as indicated by NFI, GFI, AGFI, NFI, and RMSEA were also good and indicating stronger properties, especially for 2nd order construct of CFIOVIP, construct based on 3-dimension 4 th model as well as in case of 6 th model based on the same 3-dimension, and 2nd order construct for CFIOVIP using the structural equation modelling method.
e study further revealed a strong and optimistic link between transport anxiety and behavioural intentions for the use of the Internet of the vehicle as mediated strongly and positively by CFIOVIP. e t values were significant for the relationship between transport anxiety, concern for the Internet of vehicle information privacy, and behavioural intentions of the users of connected vehicles. e findings of the study indicate that Internet of vehicle users view secondary usage of information and unauthorised access of information as the same dimension with the title of information access for measuring CFIOVIP along with two other constructs like collection and error in information for the Internet of vehicle context. e study recommends a 2 nd order CFIOVIP construct based on three dimensions, namely, collection, error, and information access for testing the present domain of the study. e future research may, however, be applied to other domain using the same set of a statement with an amendment for items as per the desired context of the study.