Delivering privacy-friendly location-based advertising over smartwatches : effect of virtual vser interface

(1)

Delivering Privacy-Friendly Location-Based Advertising over

Smartwatches: Effect of Virtual User Interface

AMSTERDAM BUSINESS SCHOOL

MSc. Business Administration Digital Business Track

Master Thesis

Author: Emiel Emanuel 10293000 emiel.emanuel@student.uva.nl Supervisor: Dr. Somayeh Koohborfardhaghighi s.koohborfardhaghighi@uva.nl Second reader: Prof. dr. Peter van Baalen

(2)

Abstract

The expansion of smartwatches and positioning technologies have offered marketers the opportunity to provide consumers with contextually relevant advertising messages. However, perceived privacy risk and a lack of privacy control are withholding consumers to use location-based advertising (LBA). This study explores how variation in the design of the user interfaces of smartwatches (regular user interface vs. virtual user interface) affects the perceived ease of privacy control, perceived privacy risk, and ultimately the intention of consumers to use LBA. A simple mediation analysis using PROCESS was conducted on data collected from a between-subjects experiment (N = 335). Results indicate that a smartwatch augmented with a virtual user interface increases perceived ease of privacy control. Also, this has a direct positive effect on perceived privacy risk and the intention to use LBA. However, perceived privacy risk does not mediate the relationship between perceived ease of privacy control and the intention to use LBA. Our results extend the growing literature on the topic of LBA and privacy concerns related to it. The findings of this study have important implications for various commercial players.

Keywords: Location-based advertising (LBA), Perceived privacy risk, Perceived ease of privacy control, Virtual user interface, Smartwatch, Smartwatch touch interface

(3)

Statement of Originality

This document is written by Emiel Emanuel who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Emiel Emanuel, Amsterdam 10293000

(4)

List of Figures

Figure 1. Our proposed conceptual model. ... 27

Figure 2. Model with path coefficients of the experimental group... 47

Figure 3. Model with path coefficients of the experimental group... 47

Figure 4. Scree plot ... 65

Figure 5. Homoscedasticity ... 66

Figure 6. Normal distribution: Histogram ... 67

(6)

List of Tables

Table 1. Literature research ... 23

Table 2. Measurement model ... 31

Table 3. Sample characteristics... 37

Table 4. Cronbach's alpha ... 39

Table 5. Correlations ... 40

Table 6. Factor loadings ... 42

Table 7. Effects experimental group ... 44

Table 8. Total, direct and indirect effects of experimental group ... 44

Table 9. Effects control group ... 45

(7)

1. Introduction

The expansion of mobile devices and positioning technologies has offered novel possibilities and challenges to revolutionize commerce applications for the mobile market. The producers of mobile devices in cooperation with marketers are presenting consumers the flexibility to make use of services considering their mobility patterns and demands. With the positioning technologies, marketers can provide mobile users contextually relevant advertising messages via mobile instruments on a geographic reference point, when they are at the edge of making a purchase. Wearable devices and smartwatches in particular, are perfect devices to execute these new marketing opportunities.

Worldwide sales of wearable devices are expected to grow from 347.53 million units in 2018 to respectively 504.65 million in 2021. It is known from a report of Smartinsight (2017) that the market is exponentially expanding with an annual growth of 23 percent, from 100 billion dollar in 2023 to 150 billion dollar by 2026. According to Allied Market Research (2018), the global smartwatch market has the potential to generate sales of $32.9 billion by 2020.

One of the main reasons for consumers to use a smartwatch may be due to their ability to track location. This feature can be beneficial for both companies and consumers. The Location-based Marketing Association (LBMA) argues that 75% of marketers believe location-based marketing is essential for their business (Salim, 2017). For marketers, the personalization of advertisements offers cost efficiencies in comparison to traditional mass marketing (Eastin, Brinson, Doorey & Wilcox, 2016). For example, using real-time intent data, a smartwatch user who strolls in front of a gym might receive an ad with an attractive offer from the nearby store which might be in line with his or her needs. As location tracking is not a new concept, location-based advertising can be executed more effectively using a

(8)

concerns are withholding the intention of consumers to use location-based advertisement due to the following reasons. First, Campbell & Park (2008) argue that users of smartwatches are mostly unaware of the continuous data collection because of its wearability and direct attachment. Second, they state that most internet users are ignorant of marketing practices by third parties, unauthorized secondary use and lack of knowledge regarding policy. Third, Park (2015) reported that high privacy concern exists on tracking location because consumers do not have the technical skills to protect and control their privacy. Park, Young, Skoric & Marko (2017) implicated that the use of smartwatches for marketing purposes and growth of data transactions have only intensified privacy concerns as consumers feel uncomfortable with the amount of data acquired by companies. Lastly, central in this study, Hargittai (2008) argued that an additional concern for consumers might be the shift from smartphone-based devices to the wearable ones. This shift causes difficulties in accessing, using and controlling the data due to the small interface of the smartwatch.

Specifically, location privacy is an all-or-nothing matter. Consumers can permit an application always to track their location, or they can completely turn this feature off. According to Kopfstein (2017), Apple and Google try to resolve this problem by giving the consumer more control over mobile location privacy. Besides, as stated by Hern (2018), Facebook is launching a variety of new tools to provide consumers with more control over their privacy. The smartwatch market is also responding to provide users with more control by eliminating the limitation of screen size.

The purpose of this study is to investigate how the existence of a virtual touchscreen on smartwatches increases the perceived ease of privacy control, perceived privacy risk and the intention of consumers to share their location information. This is essential for marketers who can convert location information into profits by creating a personalized and customized advertisement. On the other hand, privacy concerns do not enhance this process for users.

(9)

Therefore, this study proposes that it is essential to reduce perceived privacy risk by shifting the privacy control from third parties to the consumer. Developers of smartwatches are responding to this by augmenting smartwatches with a virtual touchscreen which can be projected on objects providing a larger display and eliminating the limitation of a small screen, stimulating personal privacy control. The primary purpose of this thesis is to research whether privacy control may be increased by the use of a virtual user interface to reduce perceived privacy risk and ultimately stimulate the willingness of consumers to share their location information in order to receive location-based advertising in return. Therefore, based on the line of argument presented so far the main research question in my thesis is:

How does a privacy control facilitated through a virtual user interface in smartwatches affect a consumer’s perceived privacy risk and intention to use location-based advertising?

The contribution of this thesis is the following. First, we investigate the relationship between perceived ease of privacy control (facilitated through a virtual user interface) and the intention to use location-based advertising. Second, we also investigate the mediation role of the perceived privacy risk with the intention to use location-based advertising. Previous literature focused solely on the smartphones, however, we test our model within the context of smartwatches.

This research has the following implication; it is useful for marketers to know whether augmentation of a virtual user interface over smartwatches increases the sense of privacy control and stimulates the intention to use location-based advertising. This way, perceived privacy risk is reduced, and marketers can generate revenue by using location-based advertising and anticipating with their marketing strategies.

(10)

2. Literature Review

This chapter provides an extensive overview of the literature on wearable technology, personal privacy control through different devices and media, perceived privacy risk, and the intention to use location-based advertising (LBA). First, wearable technology is defined, its wide range of professional purposes are explained, and the growing market is illustrated. Then, the focus will be on smartwatches, mainly, on the pragmatic qualities and the accommodating constraints. Second, location-based advertising is defined and its objectives are explained. Third, perceived privacy risk is illustrated, its relationship with the intention to use location-based advertising and how these concepts are affected by privacy control. Finally, these constructs are linked together to shape the potential solution of implementing a virtual user interface on smartwatches to motivate users in using LBA.

2.1 Wearable technology

Technology is advancing every day and the use of mobile devices like smartphones is still increasing. They give access to information anywhere and anytime. Kim & Shin (2015) implicate that there seems to be a shift from carrying these mobile devices to wearing them. Thierer (2015), defined wearable technology as “networked devices that can collect data,

track activities, and customize experiences to users’ needs and desires.” Wearable

technologies consist of microchips, sensors and wireless communication competencies to operate. Casson, Logesparan & Rodriguez-Villegas (2010) state that wearable technology differentiates itself from smartphones by being designed to be worn casually in everyday life and their presence may be disregarded. Also, wearable technologies are attached to the skin and therefore data is continuously available (DuFour, Lajeunesse, Pipada, Xu & Nomee, 2017).

(11)

wide range of professional purposes like: surgery, emergency care, firefighting, law enforcement, entertainment services, airlines, financial services, political campaigning, sports and retailing. Likewise, the potential applications of wearable technology in observing health and medical conditions can lead to interesting practical implications. Examples include continuous monitoring of patients and measurement of medical data during sleep time. On top of that, wearable technology can be utilized to improve personal comfort in domestic environments like automatically altering lighting, temperature or entertainment preferences as users change locations (Thierer, 2015).

Kim & Shin (2015) implied that the market for wearable devices is growing exponentially and sales are forecasted to be 91.6 million by 2018 and respectively 373 million by 2020. The increase in sales can be explained by two factors. First, wearable technologies were perceived as clumsy and ugly, which could be a reason for the constrained adoption of wearable technology. However, “sensor-rich fabric” and “conductive fiber” technologies, meaning the fabric itself is the device, are growing. The conductive fiber technology will eliminate the current optical disadvantage by creating trendy clothes. Second, Thierer (2015) states that the costs to create such technologies are decreasing which stimulates the wearable technology business.

There are two main groups regarding wearable technology, namely; smart textiles and wearable computers. First, smart textiles have electronics woven into the fabric or material of a product. Second, wearable computers have electronics stored in a stylish accessory like a bracelet or a smartwatch. Since its function is hidden in an accessory, the performed tasks may go unnoticed which result in improved productivity and enjoyment. Hertleer, Langenhove & Schwarz (2012) argued that smart textiles are better for long-term monitoring as they are not skin attached, however, wearable computers have a broader range of functions and capabilities due to the presence of an interactive screen.

(12)

2.1.1 Smartwatch

In 1972, the first digital watch was released on the market, namely, the Hamilton Pulsar P1, which consisted of a LED screen. Nevertheless, the technology was expensive and its functionalities were not significantly more valuable than the non-digital watches. Then, at the beginning of the 21st century, the rise of mobile phones eliminated the digital wristwatch. The mobile phone was capable of performing a range of functionalities, including the display of time. In May 2012, Pebble watch became the most successful Kickstarter project. Preceding smartwatches were not considered independent as they needed a Bluetooth connection to a smartphone. Due to technological development, cheaper mass production of electronic parts and the success of the Pebble watch, various brands like Samsung, Sony, Android and Apple released their versions of the smartwatch. Smartwatches are not expected to replace smartphones but rather serve as a complementary device, as smartwatches provide faster and more convenient access to information. It was expected that 214 million units would be sold by 2018 (Rawassizadeh, Price & Petre, 2015).

Well known examples of smartwatches are Samsung Galaxy Gear, LG Watch Sport and Apple Watch. They are multifunctional and can be used for a range of interests like health observation, fitness purposes and location tracking (Kim & Shin, 2015). For developers and sellers of smartwatches, it is crucial to know the different factors influencing and stimulating the intention to utilize these devices. Therefore, the Technology Acceptance Model (TAM) has been used to analyze the acceptance and use among consumers. The TAM consists of multiple predictive factors. First, perceived usefulness analyzes to what extent a consumer assumes a smartwatch can stimulate his or her performance. Second, perceived ease of use looks at to what extent a smartwatch is easy to understand and operate (Page, 2015). Kim & Shin (2015), did research on the perceived ease of use of smartwatches. The perceived ease of use expresses whether the smartwatch is easy to operate. Their findings

(13)

imply that the ease of use is positively affected by mobility and availability. Third, mobility, analyzes to what extent the consumer is willing to use this smartwatch when moving from one place to another. Fourth, availability is about whether desired information can be accessed anytime and offers a sense of real-time connection (Kim & Shin, 2015). Perceived ease of use together with perceived usefulness positively affect attitude towards the technology which ultimately leads to increased intention to keep using smartwatches.

Currently, smartwatches are capable of performing almost all tasks of a smartphone. They have two main advantages over smartphones. First, real-time access to location information and second, the proximity with the skin. Due to its light weight and location on the wrist, consumers have rapid access to messages and other digital information. Nevertheless, smartwatches are constrained by the small interactive user interface. Lim, Shin, Kim & Park (2015) implicated that this limitation makes it hard for the consumer to have real-time and full control of the device

2.1.2 Hedonic and pragmatic-qualities of Smartwatch

As mentioned, the smartwatch is held back by a major constraint, namely; small or no display (Rawassizadeh, Price & Petre, 2015). Kim (2017) studied whether screen shape and size have an effect on hedonic and pragmatic qualities of smartwatches and how they affect information processing. First, “hedonic qualities are associated with the emotional and

nonfunctional characteristics of a technology, such as the feelings of enjoyment, happiness, and sensuality, serving as personal aspirations that explain “why” interaction occurs (Hassenzahl & Monk, 2010).” Second, “pragmatic qualities are shaped by the ease of use, efficiency, and functionality of the technology, focusing on “how “ interaction with the technology occurs (Kim, 2017).” It is essential to distinguish between the two qualities

(14)

Especially, on user experience. However, this may not be generalizable to the context of smartwatches because the variation of their screen size is limited in range to a few centimeters.

Kim & Sundar (2014) explored the hedonic and utilitarian effects of a smartphone’s screen size on perceived control, affective quality and the components of the TAM mentioned in section 2.1.1. They collected data from 130 undergraduate students in Korea by means of an experiment and found that large screens have a positive effect on perceived control and perceived ease of use of smartphones. Smartphones possess touch-based interfaces which permit instinctive and dynamic interactions. Therefore, Kim & Sundar (2014) found that an increase in screen size contributes to a larger surface for interactions and a remarkable sense of control. Consequently, consumers who experience a strong sense of control are more likely to perceive that the technology is used effortlessly for realizing and satisfying their purpose and expectations. Also, larger screen size makes the exchange and access of information more convenient.

Next to the hedonic and pragmatic qualities of smartwatches, the size of user interfaces can predict the two modes of processing information, namely: systematic and heuristic. Systematic information processing is described as careful and analytic, whereas heuristic information processing is referred to as careless and rapid. Kim (2017) found that small-screen smartwatches are more likely to be processed systematically and consequently provide more content relevant information than smartphones with an extensive user interface.

When it comes to pragmatic qualities of smartwatches, Kim (2017) proved that smartwatches with a large screen, provide more perceived control for consumers than small screens. This study was conducted using an experiment with different touchscreen interface sizes. Kim (2017) argued that consumers who experience a greater sense of control perceive their smartwatches more useful and easy to use. Also, perceived control is related to

(15)

efficiency, which is about reducing the time and mental effort to achieve consumer’s purposes. On top of that, perceived control allows the consumer to determine the speed of their interactions. Liu, Bonazzi, Fritscher & Pigneur (2011) proved that a good relationship with consumers could be attained when they can control the distribution of personal information. Also, they argue that the perceived privacy risk increases when consumers have no control over their personal information. The focus of this thesis is particularly on the disclosure of personal location-based information, which is covered in the next section.

2.2 Location-based Advertising (LBA)

In 2001, location tracking functions were presented in Japan. Telecommunication operators and merchants use global positioning systems (GPS) for commercial purposes.

“GPS is a constellation of 24 well-spaced U.S. satellites that orbit the Earth and make it possible for people with ground receivers to pinpoint their geographic location (GeoCanada, 2001).” In this study, we argue that location tracking functions and utilization of its potentials

are essential in the marketing field as marketers can turn this into business profits. Marketers are able to target (potential) customers with customized services and offerings, both inside and even before entering a store. Firms are able to create a strong connection with their customers and deliver a personalized experience (Thierer, 2015). In other words, location-based information is essential for companies due to marketing reasons and providing location-based services (LBS) to their customers. Unni and Harmon (2007), defined LBS as services which rely on information about a mobile device’s location. A more specific description was given by Ratti and Frenchman (2006), namely: “comprising a set of

applications that use geographical position of a mobile contact device in order to provide services tailored to that information.” The purpose of LBS is to deliver contextual and

(16)

(Giwa, Broderick & Omar, 2015). Likewise, Bhaduri (2013) stated that the main purpose of LBS is to create value-adding services by using the information where a consumer is located and where they intend to go. LBS can be used in a range of fields like emergency triggers, entertainment, navigation, monitoring and mobile commerce. Furthermore, LBS can be operated through different devices like personal computers, smartphones, personal digital assistants and other mobile devices, including wearable technology, as long as it consists of a GPS receiver. Due to the development of information and communication technology (ICT), the LBS market is expanding rapidly. Heo & Kim (2017), implied that in the US, over 70% of the smartphone consumers utilized LBS in 2013. Also, the LBS industry is forecasted to have a global revenue of 43.3 billion US dollar by 2019.

Similarly, when service providers use GPS to create contextual advertising messages, it is called based advertising (LBA). Unni & Harmon (2017) described location-based advertising as a specific advertising initiative presented to a mobile device user (targeted through the location of the customer) from a determined party. LBA is a subpart which falls under the broad definition of location-based marketing (LBM). Xu & Teo (2005) proposed that LBA causes a five to ten times higher click-through rate when comparing to common internet advertising messages. According to a report by eWeek (2018), the LBA market is forecasted to be at almost 15 billion US dollar. Looking at this billion dollar industry and the rising annual expenses of companies on LBS and LBA, it is therefore essential that users be stimulated to use LBA. In conclusion, the willingness to provide location information is valuable to location-based service providers. However, the collection of location data is held back by the privacy right of the consumers. The privacy right is described as the consumer being free from excessive publicity. As smartwatches are commonly regarded as highly personal devices, LBA might be experienced as intrusive and

(17)

increase the perceived privacy risk of consumers. Consequently, this may slow down the effectiveness and expansion of LBA (Limpf & Voorveld, 2015).

2.3 Perceived Privacy Risk in sharing personal data

Xu et al. (2011), defined perceived privacy risk as: “a user’s concern about possible

loss of privacy, including collection, improper access and unauthorized secondary use.” The

concept of perceived privacy risk handles the rights of consumers whose data are shared (Okazaki & Hirose, 2009). There are several concerns related to information privacy. First, these concerns may be on accessing the device itself. For example, when a smartwatch gets stolen. Second, accessing the information which is shared with other devices, for example, over WiFi. Third, accessing personal data through the cloud or an alternative remote storage system (El Sacco, 2014). Fourth, a consumer ’s privacy concern is about the amount and sensitivity of data captured by the wearable tech device and even sharing it without consumers know about it. For instance, personal data might be revealed to legal authorities. Fifth, privacy policies can be altered or information might be sold due to a merger or acquisitions. Consequently, the information may be utilized by a third party for other purposes than what initially agreed on. Sixth, having difficulty deleting the data. Lastly, tractability of anonymized data makes it possible when data from multiple sources are combined (Thierer, 2015). All these factors increase the perceived privacy risk of consumers of smartwatches.

Liu et al. (2011) found that perceived privacy risk is a major predictor of the willingness to disclose personal information. Besides they found that personal data disclosed has a negative effect on the user’s payoff. Also, the authors proved that the intention of consumers to share location information is explained by obtaining personalized services in return.

(18)

2.4 Perceived Privacy Risk and the Intention to Use LBA

The relationship between smartphone users and LBA providers can be seen as an implicit social contract. The social contract theory implies that the consumers of LBA evaluate the possible benefits relative to the privacy costs. On the one hand, consumers benefit from personalization and customization which is provided by LBA. On the other hand, consumers share valuable personal data which may result in perceived privacy risk. Xu & Teo (2005) described the social contract as the consumer’s tradeoff between making personal information available and receiving economic or social benefits in exchange. In this thesis, in particular, the willingness of consumers to provide location information and receive value from LBA providers in return.

Perceived privacy risk includes the concern to determine when, how and to what extent personal information of consumers is shared with others. Okazaki & Hirose (2009), did research on privacy issues in the context of smartphones. Their findings implicate that users of mobile advertising are most concerned about the significant amount of data collected which is then stored indefinitely for potential use in the future.

Another perceived privacy risk discussed by Xu & Teo (2004) is about consumers having the feeling that they are always being watched by an unknown company when they are not aware of how their personal location information is collected. Privacy concerns can arise because of the chance that the provided LBA is not solely due to location information but also because of other personal data being collected, like social security number, address and credit card number. Inappropriate use of specified detailed information can result in the recognition and harmonization of location-based information and personal data to identify consumers. Consequently, heightening the visibility of consumer’s activities and behavior. This can result in the facilitation of personal potential awkward situations. Ultimately, this reduces the intention to use LBA.

(19)

To conclude, the collection and usage of location-based information may evoke privacy concerns. As a result, this may inhibit consumers from using LBA. Consumers are concerned about whether their data is collected and appropriately utilized. Zhou (2017) conducted research which showed that ultimately 35% of LBA users completely switched off the location tracking capability, which puts LBA providers at a disadvantage. Therefore, LBA providers need to understand the perceived privacy risk and take effective measures to reduce the negative effect of perceived privacy risk on the intention to use LBA (Zhou, 2017). As mentioned, a good relationship with consumers can be acquired when they can control the distribution of personal information. Also, Liu et al. (2011), found that the amount of control over personal information has a positive effect on the user’s payoff. Therefore, privacy control has to be stimulated.

2.5 Privacy Control

Xu, Dinev, Smith & Hart (2011) defined privacy control as: “privacy control reflects

an individual user’s perceived ability to manage his or her information disclosure.” The

findings of their research implied that: when privacy control is high, users feel in control of how their personal information is collected and used. As a result, this decreases perceived privacy risk and stimulates the adoption rate of a certain technology. The conceptualization of perceived control differs from the commonly used term control in that “perceived control is a

cognitive construct and, as such, may be subjective (Langer, 1975).” Particularly, “perceived control has been defined as a psychological construct reflecting an individual’s beliefs, at a given point in time, in one’s ability to effect a change, in the desired direction, on the environment (Greenberger & Strasser, 1986).” Skinner, Chapman & Baltes (1988) defined

perceived control as: ”the extent to which an agent can produce desired outcomes.”

(20)

consumer’s intention to use LBA from a privacy perspective. Especially, whether industry privacy self-regulation affects perceived privacy concern. Industry privacy self-regulation is described as using fair information practices (FIPs) to exert privacy control. FIPs is defined as the equilibrium of consumer’s privacy concerns and a firm’s need to use personal information. Control over personal information is obtained by assuring that firms will agree to a set of rules that most consumers find legitimate. Consequently, this will lower the perceived privacy risk. However, the privacy control is indirect because the privacy assurance is institution-based. Xu & Teo (2005) found that perceived privacy risk is even lower when consumers can exert personal control over location information.

2.5.1 Personal Privacy Control

Yamaguchi (2001) argued that there are different types of control. First, personal control, where an individual itself acts as the control agent. Second, proxy control, where an influential third party acts as the control agent. Third, collective control, where the collective acts as the control agent. This study focuses on personal control which is the most commonly used type. Individuals who value independence, find personal control most important. Tucker (2014) did research on the consequences of handling privacy issues by giving privacy control to Facebook users. As a result, they found that Facebook users were almost twice as likely to show positive reactions regarding personalized advertisements. Thus, publicly shifting control from companies to consumers, can also benefit the LBA providers. On the contrary, Goldfarb and Tucker (2011), state that giving privacy control to consumers, has a negative outcome related to personalized advertising.

There are several practical ways to provide privacy control to the consumers. Das (2017) researched privacy control on social platforms and introduced the use of privacy preferences, which are defined as: “access control rules that describe how a user wants to

(21)

information can be generalized, which means that the GPS location is not shared with its details. Instead, a more generalized position, or pseudo location, is acquired. Second, with whom the location is shared can be adjusted. This way, the user can treat social platforms differently. Third, the GPS location can be accessed between specified hours which are indicated by the user’s preferences (Das, 2017).

According to Xu & Teo (2004), perceived privacy risk reduces when consumers can use technology to exert direct control over personal information, and receive some benefits in return. Due to the fast development of technology and mobile communication devices, consumers are now able to utilize new features of a smartphone to raise the preservation of privacy. Smartphone users can exert direct personal control over their privacy preferences when utilizing LBS applications. Explicitly, a smartphone user can control when and where LBS providers can collect location information quickly and promptly. Mobile users are able to turn on and off the location-based services just by push buttons on their touchscreen. Furthermore, users can control the accuracy and to what extent location information is given free to LBS providers.

Smartwatches are dependent upon other devices due to their small displays, which reduces the mobility, availability and ease of use. Ultimately, the intention to use LBA is diminished. Robosoftin (2018) stated that the dependence on other devices increases the lack of privacy control. As opposed to the current situation where a connection with another device has to be made first, Ghosh, Joshi, Finin, & Jagtap (2012) implicate that increased privacy control makes it possible to adjust privacy preference settings at runtime.

2.6 Virtual User Interface & Perceived Ease of Privacy Control

Lim, Shin, Kim & Park (2015) argued that controllability of a smartwatch can be enhanced by using a technology which makes use of unified infrared line array sensors

(22)

fixated along the side of the smartwatch. This way the location, swipe and click movements of the finger can be observed on the back of the hand. This is an example of an around device interface (ADI) which is increasingly being researched. ADIs are stimulating the pragmatic and controlling qualities of devices with small user interfaces. However, this innovation solves only a part of the problem. Still, the user interface of the smartwatch remains relatively small in comparison to the smartphone, which limits perceived ease of privacy control.

As mentioned, perceived ease of privacy control can be stimulated by making the smartwatch less dependent on other devices. A new Samsung patent file shows a projector integrated into the smartwatch so that the user can use objects as a touchscreen. The smartwatch would be able to project interfaces, a keyboard and menu options on the consumer’s hand or arm (Low, 2016). This new feature deals with the limited screen size of the smartwatch, making it independent of other devices. The virtual user interface of the smartwatch allows it to resolve the limitation of screen size and stimulate perceived ease of privacy control.

2.7 The Research Gap

Location-based information is essential for LBA providers due to the ability to create a strong connection with their consumers by creating personalized advertisements. Xu & Teo (2005) found that LBA leads to a five to ten times higher click-through rate in comparison to regular internet advertisements. However, smartwatches are regarded as highly personal devices which contain an immense amount of personal data. Therefore, Limpf & Voorveld (2015), argued that perceived privacy risk be increased and consequently slows down the effectiveness and expansion of LBA. Zhou (2017), found that ultimately LBA consumers will switch off their location tracking function completely, which puts LBA providers at a disadvantage. So far the literature covers the relationship between perceived ease of privacy

(23)

control, perceived privacy risk and the intention to use LBA in the context of mobiles and smartphones. This thesis aims to extend the literature by also applying the findings to smartwatches augmented with a virtual user interface. Lim et al. (2015) addressed the technical aspects of the design of smartwatches. They presented the enlargement of the touch area of the display to the back of the user’s hand. They made use of infrared technology to sense the touched finger location. The prototype was equipped with an existing smartwatch and it was validated that the sensed location information of the finger can be utilized to control the smartwatch user interface. To the best of my knowledge, no systematic empirical research exists addressing how the design of the user interfaces of smartwatches affects the perceived ease of privacy control, perceived privacy risk, and ultimately the intention of consumers to use LBA. The summary of the presented literature in this area has been depicted in Table 1.

(24)

Table 1. Literature research

Author(s)

Context Privacy LBS User Interface

Main findings M obil e Smar tphone Smar tw atch P rivac y R is k P er ce ive d P rivac y R is k P rivac y C ontr ol P er ce ive d P rivac y C ontr ol P er sonal P rivac y C ontr ol P er ce ive d E as e of P er sonal P rivac y C ontr ol R egulat or y C ontr ol L oc ati on -bas ed Se rv ice s L oc ati on -bas ed A dv er ti sing P er sonali ze d A dv er ti sing Siz e A round De vice Inter face V ir tual Us er Inter face Bhaduri, A. (2003). X X X X

“To create an operational vision supporting user controlled protection of privacy that can help direct technological efforts along appropriate paths.”

Das, P. K.

(2017). X X X

Practical privacy preferences lowers the perceived privacy risk.

Kim, K. J.

(2017). X X X

“Large screens positively influence information quality by simultaneously increasing both the hedonic and pragmatic qualities of smartwatches.”

Kim, K. J., & Shin, D. H. (2015).

X

“The AQ and RA of smart watches were found to be associated with perceived usefulness, while the sense of MB and AV induced by smart watches led to a greater perceived ease of the technology’s use.”

Kim, K. J., & Sundar, S. S. (2014).

X X X

“A large screen, compared to a small screen, is likely to lead to higher smartphone adoption by simultaneously promoting both the utilitarian and hedonic qualities of smartphones.”

Lim, S. C., et.al. (2015).

X X X

“Sensed positional information of the finger when it was used to touch the back of the hand could be used to control the smartwatch graphical user interface.” Limpf, N.,

et.al.

(2015).

X X X “Information privacy concerns influence consumers’

attitude only in the case of push but not pull LBA.” Liu, Z.,

Bonazzi, et.al. (2011).

X X X X

“(1) the personal data disclosed by users has a negative effect on user payoff; (2) the amount of personalization available has a direct and positive effect, as well as a moderating effect, on user payoff; and (3) the amount of control over a user's personal data has a direct and positive effect, as well as a moderating effect, on user payoff.”

(25)

et.al.

(2009).

advertising. Both perceived ubiquity and sensitivity of the information request further the negative impact of privacy concerns on trust.”

Rawassizad eh, R., et.al.

(2015).

X X

“Advantages and disadvantages of smartwatches compared to smartphones, e.g. small screen, reduced battery life.”

Thierer, A.

D. (2014). X X X

“The better alternative to top-down regulation is using a combination of educational efforts, technological empowerment tools, social norms, public and watchdog pressure, industry best practices and self-regulation, transparency, and targeted enforcement of existing legal standards as needed.”

Tucker, C.

E. (2014). X X X X

“After this enhancement of perceived control over privacy, users were nearly twice as likely to click on personalized ads."

Xu, H., et.al. (2004).

X X X X

“The technological assurance mechanism (i.e., mobile device in this study) played the most important role in assuring consumers’ perceived control over personal information.”

Xu, H.,

et.al.

(2005).

X X X X

“Consumers did regard self-regulation and legislation

on location information protection as the important factors affecting privacy concern in LBA.”

Zhou, T.

(2017). X X X X

“Privacy concern receives a dual influence from both

central cues and peripheral cues. Privacy control moderates the effects of privacy policy and privacy seals on privacy concern.”

Current

Study X X X X X

“A smartwatch augmented with a virtual user interface increases perceived ease of privacy control. Also, this has a direct positive effect on perceived privacy risk and the intention to use LBA. However, perceived privacy risk does not mediate the relationship between perceived ease of privacy control and the intention to use LBA.”

(26)

3. Conceptual Model

So far the literature covers the relationship between perceived ease of privacy control, perceived privacy risk and the intention to use LBA in the context of mobiles and smartphones. This research aims to extend the literature by also applying the findings to smartwatches. Xu & Teo (2004) argue that perceived privacy risk will be reduced when applying direct control over personal information with the assistance of technology. Therefore, our first hypothesis proposes:

H1: Perceived Ease of Privacy Control (enhanced through a virtual user interface) reduces Perceived Privacy Risk in smartwatches.

Tucker (2014) proved that giving privacy control to the consumer stimulates their positive experiences regarding personalized advertisements. In the context of smartwatches, perceived ease of privacy control can be increased by making it less dependent on other devices. So far the literature covers the effect of perceived ease of privacy control on perceived privacy risk in the context of mobiles and smartphones. In addition, several kinds of research have been done on the effect of personal privacy control on LBS (Xu et al., 2005). To extend the literature of smartwatches and LBA, the second hypothesis proposes:

H2: Perceived Ease of Privacy Control stimulates the Intention to Use LBA in smartwatches

Next to perceived ease of privacy control, perceived privacy risk also has an effect on the intention to use LBA. Zhou (2017) proved that perceived privacy concerns inhibit LBA. Also, Limpf (2015) stated that information privacy concerns have a negative effect on the intention to accept mobile LBA. To extend the literature in the context of smartwatches, the third hypothesis proposes:

(27)

H3: Perceived Privacy Risk reduces the Intention to use LBA in smartwatches.

Furthermore, research has been done on the impact of the screen size of smartwatches on perceived privacy control. Kim (2017) found that smartwatches which possess larger screens have a positive effect on personal privacy control. However, the variation in screen size is limited to a few centimeters. Therefore, developers of smartwatches try to eliminate this limitation by proposing an innovative feature which deals with the limited screen size of the smartwatch. It enables smartwatches to project a touchscreen on objects including the consumer's hand, making it independent of other devices. The virtual user interface of the smartwatch allows it to resolve the screen size limitation, increase perceived ease of privacy control, reducing perceived privacy risk and ultimately stimulate the intention to use LBA. Therefore, the fourth hypothesis proposes:

H4: Perceived Privacy Risk mediates the relationship between Perceived Ease of Privacy Control and the Intention to use LBA in smartwatches.

The proposed conceptual model is presented in Figure 1 and is based on five theories. First, the technology acceptance model (TAM), which studies the acceptance process of consumers regarding new technologies. In this thesis, the new technology is the virtual user interface. Second, the innovation diffusion theory (IDT) which explains the various innovation attributes which lead to subjective assumptions about the innovation. According to Kim & Shin (2015), one of the attributes is relative advantage (RA) which analyzes whether the benefits outweigh the costs of innovation. In this thesis, the analysis is on whether the benefits of LBA outweigh the cost of perceived privacy risk. Third, the theory of reasoned action (TRA), which explains that consumers who have a generally positive attitude toward LBA, are more willing to use LBA than consumers who have a generally negative attitude toward LBA (Limpf & Voorveld, 2015). Fourth, the psychological reactance theory, which

(28)

explains that consumers who have a high perception of privacy risk, will respond defensively by gaining control over their privacy (Limpf & Voorveld, 2015). Fifth, the exchange theory, which explains, based on the concept of second exchange, that consumers are willing to provide location information to receive personalized offerings in return (Xu & Teo, 2005). Based on the theories and hypotheses stated above, the following conceptual model has been developed:

H4

Figure 1. Our proposed conceptual model.

PEOPC = Perceived Ease of Privacy Control. PPR = Perceived Privacy Risk.

INT = Intention to Use LBA. IV = Independent Variable. M = Mediator. DV = Dependent Variable. LBA INT (DV) PEOPC (IV) PPR (M) H2 H1 H3

(29)

4. Methodology

The purpose of this chapter is to discuss the methodology. First, the experimental design and manipulations of the independent variable will be presented. Second, the procedure of the experiment will be illustrated. Third, the variables to be analyzed of the hypothesized relationships including the control variables will be discussed. Last, an overview of the sample and statistical approach will be explained.

4.1 Design and Manipulations

To answer the proposed research question and hypotheses, a one-way between-subjects experiment with two conditions representing two levels of user interfaces (regular smartwatch user interface vs. virtual user interface) will be exploited. This design is used to analyze the causal relationships between the variables.

4.2 Procedure

To explore the proposed research question and to test the conceptual model, a quantitative between-subjects experiment will be conducted. Data will be collected online through Qualtrics using a cross-sectional design. The experiment survey design will consist of Likert scale questions in English. By providing an introductory text, the survey will enhance the participants to become more involved in the subject and to stimulate them to complete the survey. See Appendix 1 for the complete survey.

After signing an informed consent form, participants will be randomly assigned to either one of the following groups, namely: the regular smartwatch user interface group and the virtual user interface group. The actual goal of the experiment will not be explicitly disclosed. Participants will be provided with a short movie about a smartwatch, where after, they will be told to inform about their overall experience. Consequently, participant their

(30)

responses will not be influenced by the purpose of this research. The experimental group will be provided with a movie about a smartwatch augmented with the virtual user interface. The control group will be shown a movie about a smartwatch with a regular user interface. It is expected that the virtual smartwatch user interface group will have a lower perceived privacy risk, a higher sense of perceived ease of privacy control and a higher intention to use LBA than the regular smartwatch user interface group. This can be explained due to the fact that the smartwatch equipped with a virtual user interface has a larger screen and is less dependent on other devices. Consequently, the smartwatch is equal to a smartphone in its capabilities, including perceived ease of privacy control.

Furthermore, participants need to answer demographic questions (like age, educational background and gender) and personal questions (like their general attitude toward LBA, the overall level of innovativeness and previous privacy experience) to control for these variables. Finally, the research will be conducted according to the code of ethics of the University of Amsterdam. No personal information will be collected or used in this thesis.

4.3 Stimuli

Perceived ease of privacy control. To measure the perceived ease of privacy control,

participants will be randomly assigned to either one of the following two groups: a regular smartwatch user interface group and a virtual smartwatch user interface group. The regular smartwatch user interface will be the control group who are provided with a movie of a regular smartwatch. The virtual smartwatch user interface group will be the experimental group who are provided with a movie of a smartwatch with a virtual user interface. Furthermore, the participants in both groups will be assigned to the items related to perceived privacy risk, intention to use LBA and perceived ease of privacy control. The complete list of items are expressed in Table 2.

(31)

4.4 Measures

The various constructs will be measured using a seven-point Likert-type scale, ranging from 1 (strongly disagree) to 7 (strongly agree). To ensure construct validity, scales from previous studies were neglected and were analyzed all over again. According to Lewis & Saunders (2012), experimental survey design is used due to its relatively fast method. A survey which is less time consuming is expected to lead to higher response rates. The complete list of experimental survey items utilized in this research is expressed in Table 2. The minimal sample size needed for the one-way between-subject experiment design was 200.

(32)

Table 2. Measurement model

Constructs

With References Items

Measures of Constructs (measured on a seven-point

Likert scale) Cronbach’s Alpha (CA) Perceived Privacy Risk (PPR)

(Dinev & Hart, 2004)

PPR_1 PPR_2

PPR_3

Please indicate to what extent you agree with the following statements:

 I think service providers may keep private location information in a non-secure way  I think service providers may use my location

information for other purposes, like tracking my daily activities to obtain information about me  I think service providers may share my location

information with other firms without my permission .878 Intention to Use LBA (INT) (Gefen et al., 2003) INT_1 INT_2 INT_3

 I would provide the LBA service provider with my location information it needs to better serve my needs

 I intend to use LBA in the next 12 months  The benefits of LBA outweigh the cost of my

privacy concerns .840 Perceived Ease of Personal Privacy Control (PEOPC) PEC_1 PEC_2 PEC_3

 I can exactly see what data is disclosed to LBA providers and I am able to set a specified and limited amount of options regarding the disclosure  I can alter the privacy preferences where and

whenever I want

 I can exercise personal control of my privacy

.899 General Attitude toward LBA (ATT) (Wang, 1988) ATT_1 ATT_2 ATT_3

 In general, LBA are attractive  In general, LBA are useful  In general, LBA are valuable

.856 Consumer Innovativeness (INNV) (Joseph & Shailesh, 1984) INNV_1 INNV_2 INNV_3

 I like to try new things and do them differently  I am the one who try new things before my friends

and colleagues do so

 I like to experiment with new ways of doing things .799 Previous Privacy Experience (PPRV) (Smith et al., 1996) PPRV_1 PPRV_2 PPRV_3

 I have experienced incidents whereby my personal information was used by some service provider without my authorization

 I have been the victim of what I felt was an improper invasion of privacy

 I have heard and read the last year about misuse of consumers personal privacy information

(33)

 Perceived ease of privacy control

The independent variable ‘perceived ease of privacy control’ will be measured with three validated items using a seven-point Likert scale. A Cronbach’s alpha of .899 is going to be used to measure to what extent the consumer of a smartwatch experiences personal privacy control. The conceptualization of the construct perceived control differs from the commonly used term control in that “perceived control is a cognitive construct and, as such,

may be subjective (Langer, 1975).” Particularly, “perceived control has been defined as a psychological construct reflecting an individual’s beliefs, at a given point in time, in one’s ability to effect a change, in a desired direction, on the environment (Greenberger & Strasser, 1986).” Skinner et al. (1988) define perceived control as: “the extent to which an agent can produce desired outcomes.” Personal privacy control is operationalized, for

example, by the following item: “I can exactly see what data is disclosed to LBA providers and I am able to set a specified and limited amount of options regarding the disclosure.”

• Perceived privacy risk

The mediator ‘perceived privacy risk’ will be measured with three validated items using a seven-point Likert scale (1=strongly disagree, 7=strongly agree). A Cronbach’s

alpha of .878 is going to be used to measure the privacy risk a consumer of a smartwatch

perceives. The conceptualization of the construct perceived privacy risk is defined as: it can arise because of the chance that the provided LBS is not solely due to location information but also because of other personal data being collected like; social security number, address and credit card number. Xu & Teo (2004) argue that this is due to the possibility to identify individuals when combining personal information with matching location data. Perceived privacy risk is operationalized, for example, by the following item: “When I share location information with an LBA provider I believe I cannot be personally identified.”

(34)

 Intention to use LBA

The dependent variable ‘intention to use LBA’ will be measured with three validated items using a seven point Likert scale. A Cronbach’s alpha of .840 is going to be used to measure to what extent the consumer of a smartwatch intents to use LBA. Xu & Teo (2005) state that the construct intention to use LBA is conceptually defined as: the tradeoff between perceived benefit and perceived risk of location information disclosure which leads to the intention to use LBA. The intention to use LBA is operationalized, for example, by the following item: “I am very likely to provide the LBA service provider with my personal information it needs to better serve my needs.”

 Control variables

Previous studies on perceived privacy risk and personal privacy control proposed a number of supplementary factors which should be added due to their potential influence on perceived privacy risk, personal privacy control and intention to use LBA.

Consumer’s general attitude towards LBA. According to Phelps et al. (2001),

consumers who have a better general attitude towards LBA, are also less concerned about privacy issues. Consumer’s general attitude towards LBA will be measured with three validated items using a seven point Likert scale. A Cronbach’s alpha of .856 is going to be used to measure whether and to what extent consumer’s general attitude towards LBA affects the variables to be researched. Consumer’s general attitude towards LBA is operationalized, for example, by the following item: “In general, LBA is attractive.”

Previous privacy experience. According to Smith et al. (1996), consumers who have

had a bad experience regarding their personal information privacy, will have a stronger perceived privacy risk. Previous privacy experience is measured with three validated items using a seven point Likert scale. A Cronbach’s alpha of .676 will be reconsidered and item

(35)

PPRV_3 will be deleted. Thereafter, two items will be used to measure whether and to what extent previous privacy experience affects the variables to be researched. Previous privacy experience is operationalized, for example, by the following item: “I have experienced incidents whereby my personal information was used by some service provider without my authorization.”

Consumer innovativeness. According to Joseph & Shailesh (1984), consumers who

are more innovative, have the intention to use LBA more often. Innovativeness is measured with three validated items using a seven point Likert scale. A Cronbach’s alpha of .799 is going to be used to measure whether and to what extent a consumer’s innovativeness affects the variables to be researched. Innovativeness is operationalized, for example, by the following item: “I am the one who tries new things before my friends and colleagues do so.” 4.5 Sample

The target group of this research are potential and actual users of smartwatches and LBA. The research will be conducted using a non-probability convenience sampling technique which includes: self-selection and snowball sampling. It will be based on anonymous respondents. Participants will be gathered via social media (like Facebook), WhatsApp, e-mail and face-to-face. This is the easiest and fastest way to reach the defined target group. Demographic information will be obtained to measure the similarity between the control and experimental group. To calculate the minimum number of participants needed for the results to be generalizable, the rule of thumb is used. This research measures a total number of two different conditions (regular smartwatch interface vs. virtual user interface) so a minimum of 200 participants are needed.

(36)

4.6 Statistical procedure

Data will be collected online using the Qualtrics software. For the statistical analyses, scores will be collected and transferred into SPSS. Dummy variables will be composed for the two conditions regarding the independent variable. Descriptive statistics, skewness, kurtosis and normality tests will be conducted for all variables. The outliers will be evaluated to guarantee that no data entry or instrument errors are generated.

A simple mediation using PROCESS will be executed to investigate the mediating role of perceived privacy risk between the relationship of perceived ease of privacy control and the intention to use LBA. A confidence interval of 95% will be established

(37)

5. Results

In this section, the results of the data analysis will be presented. First, an overview of the characteristics of the participants will be given. Second, the preparation of the data will be explained. Third, the hypotheses testing and prior assumptions will be provided.

5.1 Data Collection

The time period to fill in the survey ranged from the 4th of May 2018 until the 23rd of

May 2018. Out of the 446 participants who started the experimental survey, 335 completely finished the experimental survey. This equals a completion rate of 75.1%. Out of the total participants, 51.9% represented males, where 48.1% of the participants represented females. Approximately 51.3% (N=172) of the participants were between 22 and 25 years old with a minimum of 17 and a maximum of 67. Around 81.5% (N=273) of the participants had an education of at least HBO or university. Furthermore, 90.4% (N=303) of the respondents were European. Lastly, 90.4% of the participants did not own a smartwatch. The demographics mentioned above are summarized in Table 3.

(38)

Table 3. Sample characteristics

5.2 Data Preparation

In order to be confident in the extracted and adjusted constructs we assessed them again. The internal consistency of the control variable regarding previous privacy experience was found to be insufficient (Cronbach’s alpha = 0.69) and therefore item PPRV_3 was deleted. After the item was deleted, the Cronbach’s alpha of the control variable regarding previous privacy experience equaled .804. See Table 4 for the items and their related Cronbach’s alphas. The mean, standard deviation and Pearson correlation coefficients of the independent, dependent, mediating and control variables were computed to assess their relationships. The control variables include: general attitude toward LBA (ATT), innovativeness of the consumer (INNV) and previous privacy experience (PPRV). The

Characteristic Amount (N) Percentage (%)

Gender Male 174 51.9 Female 161 48.1 Age 17-21 38 11.3 22-26 192 57.4 27-31 37 11.0 32-36 27 8.1 37-67 41 12.3 Education High school 43 12.8 Vocational 13 3.9 Bachelor’s 54 16.1 Master’s 82 24.5 PhD 10 3.0 Other 6 1.8 Origin Europe 303 90.4 North-America 12 3.6 Asia 16 4.8 Australia 1 .3 South-America 3 .9 Smartwatch Owner 32 9.6 Non-owner 303 90.4 Condition

Regular User Interface 174 51.8

(39)

Pearson correlation coefficient is analyzed to test the strength of the linear correlation between two variables. The results are depicted in Table 5, which suggest that 11 out of 15 correlation were statistically significant and were greater or equal to r(333) = -.14, p<.01,

two-tailed. These outcomes indicate that a strong correlation exists between the variables,

except for perceived ease of privacy control and innovativeness of the consumer; perceived privacy risk and general attitude toward LBA; perceived privacy risk and innovativeness of consumer; and innovativeness of consumer and previous privacy experience. Participants who started the survey equaled 446. However, due to incomplete surveys, 111 respondents were listwise deleted. In this method, a full record is eliminated from analysis when a single value is absent. Due to the central limit theorem, normality is assumed (n>30). To measure the reliability using the Cronbach’s alpha, there was no need to recode counter indicative items. Most constructs which are used in this survey were extracted from past studies. The reliability of these constructs was all tested to possess a Cronbach’s alpha of .70 or higher.

(40)

Table 4. Cronbach's alpha Item Cronbach’s Alpha Cronbach’s Alpha if Item Were Deleted

Perceived Ease of Personal Privacy Control .90 I can exactly see what data is disclosed to LBA providers and I am

able to set a specified and limited amount of options

.91 I can alter the privacy preferences where and whenever I want .84

I can exercise personal control of my privacy .81

Intention to Use LBA .84

I would provide the LBA service provider with my location information it needs to better serve my needs

.80

I intend to use LBA in the next 12 months .76

The benefits of LBA outweigh the cost of my privacy concerns .77

Perceived Privacy Risk .88

I think service providers may keep private location information in a non-secure way

.83 I think service providers may use my location information for other

purposes, like tracking my daily activities to obtain information about me

.83 I think service providers may share my location information with

other firms without my permission

.83

General Attitude Toward LBA .86

In general, LBA are attractive .86

In general, LBA are useful .78

In general, LBA are valuable .75

Consumer Innovativeness .80

I like to try new things and do them differently .75

I am the one who try new things before my friends and colleagues do so

.71 I like to experiment with new ways of doing things .71

Previous Privacy Experience .69

I have experienced incidents whereby my personal information was used by some service provider without my authorization

.40 I have been the victim of what I felt was an improper invasion of

privacy

.37 I have heard and read the last year about misuse of consumers

personal privacy information

.80*

Delivering privacy-friendly location-based advertising over smartwatches : effect of virtual vser interface