The privacy paradox : a construal level theory perspective

(1)

The Privacy Paradox:

a Construal Level Theory perspective

Master thesis

Author: Jasmijn Koster (10963707)

University of Amsterdam, Faculty of Economics and Business MSc. in Business Administration – Marketing Track

Under supervision of:

Dr. J. Demmers, PhD candidate and Marketing Lecturer at the University of Amsterdam

(2)

Statement of Originality

This document is written by student Jasmijn Koster, 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.

(3)

ABSTRACT

In general, people claim to be anxious about the information that is available about them online, yet, they are still inclined to share personal data. This research searches for the underlying explanation of this privacy paradox by taking the potential influence of Construal Level Theory into account. Two studies examined the prediction that CLT influences consumer reliance on benefits and costs of online information disclosure. Respondents were allocated to three online information disclosure scenarios and exposed to either a proximal benefits/distant costs, proximal costs/distant benefits, or control condition (Study 1). Results showed that consumers rely on the benefits or costs of disclosing personal information that have the highest proximity of occurring. These results support the principle CLT, according to which the more proximal an activity is, the more concrete it is evaluated. The inclination to disclose personal information thus derives from the psychological distance. Subsequently, Study 2 was based on the hypothesis that temporal distance affects consumer reliance on benefits and costs of disclosing personal information, such that participants rely on consequences that are congruent with their mind-set. Respondents were allocated to two scenarios in which they had to disclose personal information. Simultaneously, they were exposed to a manipulation of the mind-set and a manipulation of temporal distance similar to the one executed in Study 1. However, results revealed no significant interaction effect of consumer mind-set and temporal distance on the inclination to disclose personal information. In conclusion, the results indicate that temporal distance is underlying for the inclination to disclose personal information, yet, mind-set does not moderate this outcome. Future research is required to expose whether an altered manipulation of CLT explains the privacy paradox.

Keywords: Construal Level Theory, privacy paradox, temporal distance, mind-set,

(6)

1. INTRODUCTION

The digital age has facilitated the accessibility of big data, providing companies and online networks with increased access to online consumer behaviour (Awad & Krishnan, 2006a). The availability of this information initiated from the integration of social media in everyday life, which has become an essential tool for consumers to stay in contact with close friends and acquaintances (Debatin, Lovejoy, Horn, & Hughes, 2009). Marketers can use this data-rich environment for implementing cookies, data mining, and phishing, providing themsel ves with the opportunity to specify their marketing tools and improve the targeting of messages (Arons, van den Driest, & Weed, 2014). The way marketers engage with customers has changed remarkably and Arons et al. (2014) accentuate that high performers’ marketing approaches outdo other marketers in their ability to leverage customer insights, to integrate data on why consumers are behaving in a certain way, and to deliver a rich customer experience in the long run. Even though it seems that the collection and usage of big data is always in the best interest of the consumer, the presence of personal information online might also be exploited negatively. Not only companies with marketing-related intentions have access to this data, the disclosed information can be misused for other purposes as well. Harassment, hacking, and identity theft pose severe risks to online users’ privacy (Debatin et al., 2009). Online consumer privacy is currently a debatable issue and consumers care more about their privacy now than they did several years ago. Prior research has shown that in 2009 33% of the Internet users was anxious about the amount of personal data that is available about them online, a percentage which has grown to 50% in 2013 (Rainie et al., 2013). These numbers would indicate that consumers behave accordingly, by being cautious in providing personal information on the Internet. However, even though consumers report high values of privacy concerns, they still display behaviour that contradicts their concerns (Rivenbark, 2010). Different explanations for this incongruity are provided in previous literature. Firstly, Dinev & Hart (2006) posit that a

(7)

consumer’s final decision on whether or not to disclose personal information online is determined by a privacy calculus. In addition, Fournier & Avery (2011) theorise that the social exchange theory, wherein consumers ‘objectively’ weigh benefits and costs when engaging in social interaction, provides insights in the privacy paradox. Secondly, Dinev & Hart (2006) and Metzger (2004) theorise that trust and previous information disclosing behaviour are the most important antecedents to privacy concerns. Lastly, Deuker (2010) suggests that the privacy paradox can be explained by consumers’ unawareness of the consequences of disclosing data and, accordingly, Gross & Acquisti (2005) suppose that consumers tend to underestimate the privacy dangers of disclosing personal data.

This study will start with a literature review to search for the fundamental theories underlying the issue of the privacy paradox. As a starting point, Acquisti & Grossklags (2004) propose theories on the differences between privacy attitudes and privacy behaviour. Subsequently, researching the concept of Construal Level Theory (CLT) will be of high importance as this paper aims to bridge the gap between privacy concerns and the inclination to disclose personal data. In their research, Trope, Liberman, & Wakslak (2007) studied construal levels and they focused on the influence of psychological distance on consumer behaviour, which might be a foundation of the privacy paradox. After bridging these two theories, the methodology for the survey will be set up and the approach for the research will be explained. In the first study, I will measure the effects of temporal distance on online information disclosure behaviour, and more importantly, the effect on behaviour when the temporal distance is manipulated. Thereafter, the second study will explore whether this effect is reliant upon the congruence between mind-set and psychological distance of benefits and costs. The collected data will be displayed, after which an analysis of the results will take place. This research will conclude with an in-depth discussion on the proposed hypotheses, followed by the implications, limitations, and recommendations for future research.

(8)

2. THEORETICAL FRAMEWORK

2.1. Privacy

According to Warren & Brandeis (1890), privacy can be defined as “the right to be let alone” (p. 205), and privacy concerns arise whenever personal information or other sensitive information is collected and stored. Consumers desire to control or manage their own information, but due to advances in digital technologies and the increasing information gathering by third parties, concerns about information privacy have risen (Bélanger & Crossler, 2011; Phelps, Nowak, & Ferrell, 2000). However, even though consumers report high values of privacy concerns, they still display behaviour that contradicts their concerns (Rivenbark, 2010). One possible explanation for this privacy paradox is that it is interpreted incorrectly. Literature illustrates that privacy attitudes are described on a broad, general level while privacy intentions and behaviour are defined narrowly. Thus, the comparison of these different entities might lead to deceiving results (Acquisti & Grossklags, 2004). This justification for the privacy paradox, however, is not entirely agreed upon and possible explanations for the incongruity are scrutinised extensively in prior literature.

2.1.1. Privacy paradox. Firstly, according to Fournier & Avery (2011), younger

generations seem to be quite comfortable with the exchange of privacy for benefits as they are more likely to act according to the social exchange theory. Social exchange theory states that consumers ‘objectively’ weigh benefits and costs when engaging in social interaction. A consumer grants certain value to individual beneficial and precarious aspects, forming a so -called privacy calculus (Dinev & Hart, 2006; Li, Sarathy, & Xu, 2010). When one driver outweighs the other, it would determine a consumer’s final decision on whether or not to disclose personal information online. Nonetheless, when researchers approached privacy concerns by assuming that consumers engage in a cost-benefit trade-off, it appeared that when it came to privacy, consumers behaved irrationally (Keith, Thompson, Hale, Lowry, & Greer,

(9)

2013). One reason for this finding is that privacy decision-making is only partially a rational trade-off; it is also affected by misperceptions of benefits and costs, emotions, and social norms (Acquisti & Grossklags, 2004). Again, this discrepancy between consumers’ attitudes and their actual disclosure behaviour is the foundation of the privacy paradox (Acquisti & Grossklags, 2004; Debatin et al., 2009; Gross & Acquisti, 2005; Norberg, Horne, & Horne, 2007; Wilson & Valacich, 2012). Although objectively irrational, the behaviour of individuals is considered to be subjectively rational within the given level of privacy awareness. Along the lines of subjective behaviour, Dinev & Hart (2006) theorise that trust is one of the most important antecedents to privacy concerns. Accordingly, Metzger (2004) posits that consumers value trust when interpersonal exchange situations take place. Expected trust is the inclination to believe that others act in a consumer’s best interest, thus when trust is perceived to be high, the perceived cost would be low and vice versa. Secondly, Metzger (2004) approaches the privacy paradox by affirming that consumers believe that after they have disclosed their personal information once, it would not be a problem to repeat their behaviour. This would indicate that future disclosure of personal information is not necessarily associated with perceived risk, but rather with the antecedent of past behaviour of having done it before (Metzger, 2004). Thirdly, an obvious source of privacy uncertainty arises from information asymmetry, wherein consumers are not fully aware of what information companies have and for what purpose it is used (Acquisti, Brandimarte, & Loewenstein, 2015). Most of the time, consumers solely want to disclose information for efficiency reasons, to receive tailored advertisements, and to enjoy personalisation of products. In these situations, consumers freely choose to exchange their personal information to get a benefit in return (Deuker, 2010). Privacy paradoxical behaviour can be explained by individuals’ limited capabilities in accessing and processing decision relevant information or due to different perspectives on privacy (Awad & Krishnan, 2006b). Accordingly, Gross & Acquisti (2005) suppose that consumers tend to underestimate the

(10)

privacy dangers of disclosing personal data, given that data protection guidelines are often accepted without being read. It can be proposed that this contradicting behaviour initiates from a mental representation of the benefits and costs, which is a consequence of the inclination to disclose personal information.

It has been shown above that several studies developed theories to explain why consumers’ inclination to disclose personal data is not consistent with their actual knowledge on the risks of sharing personal information. However, none of those studies bridged an element’s psychological distance and consumer mind-set, both elements of CLT, to the privacy paradox. Therefore, it will be of high importance to research whether consumer reliance on benefits and costs, and its subsequent information disclosure behaviour is dependent upon the proximity of these elements rather than on the benefits and costs itself.

2.2. Construal Level Theory

Construal Level Theory describes the relation between psychological distance and the extent to which a person’s mental state is abstract or concrete. The theory proposes that the more distant an object is from an individual, the more abstract the object will be perceived, while nearby objects will be processed in a low-level concrete mind-set (Bar-Anan, Liberman, Trope, & Algom, 2007; Liberman & Trope, 1998; Liberman, Sagristano, & Trope, 2002; Liberman, Trope, & Stephan, 2007; Liberman, Trope, McCrea, & Sherman, 2007; Trope & Liberman, 2003; Trope & Liberman, 2010; Trope, Liberman, & Wakslak, 2007). The differences in these mental states are acknowledged by CLT, and they occur because the amount of concrete or abstract information varies.

2.2.1. Low and high level construals. Events and objects can be described according

to their level of abstraction. Low-level construals are relatively contextualised and represent subordinate means (the “how” of the activity), whereas high-level construals are more abstract

(11)

and emphasise superordinate purposes (the “why” of the activity) (Trope et al., 2007). Prior research shows that accentuating the costs of a certain situation is effective when paired with a low-level, concrete mind-set, while emphasising the benefits is effective when paired with a high-level, abstract mind-set. These results show that when people are in an abstract mind-set, they are more likely to rely on the benefits of disclosing data than people in a concrete mind-set (White, MacDonnell, & Dahl, 2011). This demonstrates that psychological distance is a subjective representation of something close or far away from the self. For example, when benefits of disclosing data are presented in the distant future, it will be described with the use of higher level construals than when the benefits are expected to be proximal.

2.2.2. Psychological distance. In CLT, psychological distance refers to the distance

of an object from the direct experience, which comprises several interrelated dimensions; temporal, hypothetical, social, and spatial distance (Trope et al., 2007). The different dimensions of psychological distance affect mental construal and correspondingly guide predictions (Nussbaum, Trope, & Liberman, 2003), evaluations (Bar-Anan et al., 2007; Henderson, Trope, & Carnevale, 2006; Kim, Zhang, & Li, 2008), and behaviour (Henderson et al., 2006; Trope et al., 2007). One of the psychological dimensions, temporal distance, refers to distance in time; how much time separates the perceiver’s present time and the event (Herzog, Hansen, & Wänke, 2007). For example, the temporal distance of a certain element can be presented as a proximal situation or as a situation in the distant future. An important distinction can be made between desirability concerns, which involve the value of the end-state, and feasibility concerns which involve the means to reach the end-state (Trope et al., 2007). This implies that as psychological distance increases, a person will anticipate on desirability rather than feasibility (Liberman & Trope, 1998; Trope et al., 2007), which can be interpreted as seeing the distant future as more positive but maybe less realistic than the near future (Eyal, Liberman, Trope, & Walther, 2004; Mitchell, Thompson, Peterson, & Cronk, 1997; Savitsky,

(12)

Medvec, Charlton, & Gilovich, 1998; Trope & Liberman, 2003). A second relevant dimension of CLT is hypothetical distance. Hypothetical distance refers to the likelihood of an event occurring. According to CLT, an event that has a high hypothetical distance will be processed at a high level construal and is unlikely to occur, while a low hypothetical distance will be processed at a low level construal and is relatively likely to occur (Bar-Anan et al., 2007; Trope & Liberman, 2003; Trope & Liberman, 2010; Trope et al., 2007). Depending on their personal perception of hypothetical distance, people might perceive certain situations to be more probable than others.

Although construal level and psychological distance are related, they do not refer to the exact same thing. Psychological distance refers to the perception of when or whether an event occurs, while construal level refers to the perception of what will occur (Trope et al., 2007). Nonetheless, the literature has reached a consensus that the relation between psychological distance and construal level is bi-directional, which indicates that manipulations of level of construal affect psychological distance in relatively the same way as psychological distance influences level of construal (Trope & Liberman, 2010; Trope et al., 2007). It is for this reason that the second study focusses on the congruence between the mind-set and the psychological distance of the benefits and costs of disclosing personal data. With regards to online privacy, when consumers are in a concrete mind-set and they are offered certain benefits with a low temporal distance, they will be more likely to act upon these temporally close benefits. Accordingly, when consumers are in an abstract mind-set and they are offered certain benefits with a high temporal distance, they will be more likely to act upon these temporally far benefits. Indeed, this provides evidence that a bidirectional link between temporal distance and mind -set exists (Wakslak & Trope, 2009). Wakslak & Trope (2009) revealed that when participants were asked to focus on abstract aspects rather than concrete aspects, they would see the activity

(13)

as temporally distant. These assumptions are an important starting point for this study when manipulating the temporal distance of the benefits and costs and the mind-set of the participants.

2.3. Online information disclosure

On a daily basis consumers are active in the online environment. They log in on websites with their personal information and they accept cookies, leaving a trail of data for companies to accumulate and store. This information is critical for companies for targeting individuals with customised products and, conversely, consumers benefit when they are receiving relevant information (Chen & Popovich, 2003). However, the collected information might be misused, after which consumers are exposed to the consequences of big data.

2.3.1. Costs and risks. Whenever consumers disclose personal information online,

they are exposed to potential risks and intrusion of their privacy. Youn (2005) developed relevant components of perceived risk that consumers can encounter during their presence on the Internet. First of all, time risk refers to the perception that time would be wasted in receiving, checking, and removing unsolicited spam mail. Consumers not only feel that they are wasting time, research also showed that consumers feel their privacy is being violated when they receive unsolicited e-mail messages from companies (Miyazaki & Fernandez, 2001). Secondly, consumers will have a feeling of physical risk when they expect their personal information to be misused by companies. Correspondingly, websites might not use it solely for their own gain, they might share or sell information to third-party companies, leading to unauthorised access to personal information (Miyazaki & Fernandez, 2001). The usage of data by third-party companies is seen as one of the most significant risks that consumers face, presenting companies with an important role to reassure consumers and offer benefits in return.

2.3.2. Intrinsic and extrinsic benefits. An increasing amount of companies are

(14)

theory posits that not only companies but also consumers can benefit from the accessibility of personal information online. In exchange for consumers’ personal information businesses can provide them with benefits to offset the costs and when the exchange is perceived to be beneficial, consumers are likely to enter into a relationship with a company (Hui, Tan, & Goh, 2006; Li et al., 2010). It has been shown that consumers’ willingness to provide information online may be influenced by the benefits they would obtain from voluntary disclosure (Youn, 2005). According to Hui et al. (2006), benefits can be categorised in those providing intrinsic and those providing extrinsic motivation.

When people are intrinsically motivated, they may perform a task simply because the act of performing the task offers them benefits. Pleasure and novelty are two of the intrinsic benefits that Hui et al. (2006) describe. Pleasure can be defined as gaining an enjoyable experience and novelty refers to the desire of gaining knowledge on a topic of interest. When people are extrinsically motivated, they may perform an action to obtain benefits that serve as a means to achieve other goals. Time saving is an important extrinsic benefit; it might offer the consumer the convenience of faster login. Likewise, the resource exchange theory posits tha t consumers are likely to trade privacy for speed and convenience of transactions. At the same time, monetary benefits such as receiving loyalty points directly affect the willingness of consumers to disclose personal information online (Hui et al., 2006).

According to the privacy calculus, a trade-off of these benefits and costs should lead consumers to make informed decisions (Acquisti & Grossklags, 2004), yet, prior research on the privacy paradox shows that consumers do not always act rationally. Several explanations have been put forward, but none of those posited CLT as a possible explanation for this irrational behaviour. Future research is needed to fully understand consumers’ intentions and actual behaviour concerning privacy related issues. Therefore, the following section puts forward a possible explanation for this discrepancy.

(15)

2.4. Relationship between Construal Level Theory and the privacy paradox

A consumer’s decision to disclose information is influenced by perceived benefits, risks, and privacy concerns. A consumer grants certain value to individual beneficial and precarious aspects, forming a so-called privacy calculus (Dinev & Hart, 2006; Li et al., 2010). When one driver outweighs the other, it would determine a consumer’s final decision on whether or not to disclose personal information online. Accordingly, the standard economics model assumes that inter-temporal preferences are time-consistent, which would indicate that the moment in time someone discloses information should not make a difference (O'Donoghue & Rabin, 2001). However, evidence shows that people exhibit time-inconsistent behaviour, which implies that a consumer’s preference for benefits gets stronger as the moment of gratification gets closer, confirming present bias (Hardisty, Appelt, & Weber, 2013).

Consumers tend to over-value immediate rewards at the expense of long-term intentions (O'Donoghue & Rabin, 2001). This implies that when the consequences are immediate, a consumer should be more likely to put a higher weight on it. Correspondingly, as consequences are temporally further away their relevance reduces (Trope et al., 2007). Research on construal level shows that when people are in a concrete mind-set, they are more likely to rely on the benefits and costs that are more proximal. Likewise, when people are in an abstract mind-set, they are more likely to rely on the benefits and costs that are more distant. Both propositions provide evidence for bridging present bias to construal level theory, granting an explanation for the privacy paradox (Hardisty et al., 2013). This proposition indicates that psychological distance and the privacy paradox might not be as detached as initially seems. According to the privacy paradox, consumers disclose privacy related information for immediate benefits (Acquisti, 2004). If this is true, consumer reliance on benefits or costs of personal data disclosure should be influenced by the psychological distance of these benefits or costs (Trope et al., 2007). Consistent with literature on present bias, people tend to be impatient and simply

(16)

want to have gains or benefits immediately to satisfy their desire for positive outcomes (Hardisty et al., 2013). Research on CLT allows the researcher to enhance the principle of present bias by proposing that merely under a concrete mind-set consumers have the tendency to rely on proximal benefits or costs, proving a congruence between mind-set and psychological distance of these benefits and costs. Therefore, the effect of the temporal distance on consumer reliance on benefits and costs is assumed to be moderated by a consumer’s abstract or concrete mind-set. Literature on CLT and the privacy paradox lead to the assumption that in typical disclosure situations the benefits of disclosing personal data online are perceived as psychologically closer than the costs of disclosing personal data online, providing proof that consumers habitually rely on benefits (Trope et al., 2007).

Therefore, in this study, CLT will be applied to the existing literature on the privacy paradox to show its effect on consumer behaviour and decision making. This research is based on the proposition that manipulating the psychological distance of benefits and costs as well as a consumer’s mind-set alters his inclination to disclose personal data. These aspects will be the central issue in this research, as the results will provide evidence for the research question of this study:

“How does the temporal distance of benefits and costs influence consumer reliance on benefits and costs and its deriving information disclosure behaviour in the online environment?”

In order to support the research question, two consecutive studies will be executed.

The first study will focus on manipulating the temporal distance of benefits and costs to measure the effect it has on the inclination to disclose personal data. According to CLT, benefits and costs that have a low temporal distance will be proximal to occur, whereas benefits and costs that have a high temporal distance will be perceived as distant. In addition, literature on the privacy paradox shows that consumers are more likely to disclose personal data when they have to behave instantaneously, rather than when they express their future concerns in advance.

(17)

This study will find proof that manipulating the temporal distance influences consumer reliance on benefits and costs of disclosing personal data. The manipulation of this research will be performed on the temporal distance of benefits and costs, therefore the following hypothesis will be tested:

H1: Consumer reliance on benefits and costs is dependent upon the temporal distance of these

elements, such that people will base their decision more on the benefits when benefits are

proximal and costs are distant but more on the costs when costs are proximal and benefits are

distant.

The second study will compare the results of Study 1 to the effect of the consumer’ s mind-set on the reliance on benefits and costs. In the present study, the effect of the temporal distance on consumer reliance on benefits and costs will be moderated by the consumer mind-set. This study will indicate that when participants are in a concrete mind-set, they are more likely to rely on the benefits and costs that are more proximal. Likewise, when participants are in an abstract mind-set, they are more likely to rely on the benefits and costs that are more distant. This study will be executed in such a way that when benefits are temporally proximal and costs are temporally distant, people in a concrete mind-set rely on benefits and people in an abstract mind-set rely on costs, and when costs are temporally proximal and benefits are temporally distant, people in a concrete mindset rely on costs and people in an abstract mind -set rely on benefits. For this study, the researchers are expecting an interaction effect of mind-set and temporal distance. The manipulation of this study was performed on the mind-mind-set of the participant and on the temporal distance of benefits and costs. Therefore, the following hypothesis will be tested:

H2: The effect of the temporal distance on consumer reliance on benefits and costs is moderated

by the consumer mind-set. Such that participants rely on consequences that are congruent with

(18)

The following conceptual model provides a visual representation of the proposed hypothesis presented above. Consumer mind-set Temporal distance of benefits and costs Inclination to disclose personal data

Figure 1. Conceptual model – The effect of the temporal distance of benefits and costs on th e inclination to disclose personal data is moderated by the consumer mind-set.

(19)

3. STUDY 1

Manipulation of the temporal distance of benefits and costs

3.1. Method

To test whether consumer reliance on benefits and costs and its subsequent information disclosure behaviour is dependent upon temporal distance of these elements and not on the benefits and costs itself, this research manipulated the temporal distance of benefits and costs. To support the hypothesis, Study 1 consisted of three groups; 1) the first group of participants was exposed to temporally proximal benefits and temporally distant costs, 2) the second group of participants was exposed to temporally proximal costs and temporally distant benefits, and 3) the third group of participants functioned as a control group whereby the proximity of temporal distance of benefits and costs were open for the participant’s own interpretation. The information disclosure behaviour of the control group was expected to display a similar pattern as the first group of participants, which supports the assumption that in typical disclosure situations the benefits of disclosing personal data are perceived as psychologically closer than the costs, providing proof that consumers habitually rely on benefits.

3.1.1. Design and procedure. I conducted an experimental design to examine the

effect of the temporal distance of benefits and costs occurring on the inclination to disclose personal data. A 3x3 study was conducted: 3 (between-subjects: proximal benefits/distant costs vs. proximal costs/distant benefits vs. control) x 3 (within-subjects: a website visit vs. Facebook login vs. subscribing for a loyalty card). Each participant took part in all three scenarios, and for each of these scenarios they were randomly assigned to only one of the temporal distance conditions. Participants read a short story that was manipulated by the researcher; it emphasised either proximal benefits and distant costs or proximal costs and distant benefits, or the short

(20)

story did not emphasise the temporal distance (control condition of the manipulated temporal distance).

All constructs were measured using a multi-item, five-point Likert-type scale, and to ensure construct validity scales from previous studies were adapted wherever possible. All three groups of participants were exposed to the same questions so the data could be analysed on the same constructs. This was done to find support that CLT is the underlying mechanism for the privacy paradox. To measure the likelihood of disclosing personal data, a Likert scale was used with “very unlikely” and “very likely” as anchors (Sparks & Browning, 2011), and respondents had to describe on what information they based their decision. With the use of the same Likert scale, the likelihood of consequences and/or benefits of disclosing personal data was measured. The likelihood of benefits and costs occurring was examined to exclude that the hypothetical distance influenced the results. To measure if respondents believe disclosing personal information can be used to optimise a company’s products, to display relevant content, to make the browsing experience more efficient, or to save time the next time logging in, a Likert scale was used with “strongly disagree” and “strongly agree” as anchors (Jamieson, 2004). With the use of the same Likert scale, respondents were asked to what extent they agreed with the fact that accepting cookies, logging in with Facebook and subscribing for a loyalty card are used to pass information on to third parties or to target consumers with unsolicited spam. Existing forms of measurement of these constructs were used and adapted to the understanding of the constructs in this research (Miyazaki & Fernandez, 2001; Youn, 2005). Subsequently, the expected timespan of benefits and costs of disclosing personal data actually occurring were probed as a manipulation check of the temporal distance. CLT research uses time periods such as next year or two-to-six months to represent distant future events, whereas time periods such as one month, next week, or tomorrow represent near future events (Nussbaum et al., 2003).

(21)

3.1.2. Participants. The population for this study consisted of male and female

students from different levels of education in The Netherlands. The size of this population was retrieved from Centraal Bureau van de Statistiek (CBS)1_{and consists of 1.185.257 students in} 2014-2015. With a margin of error of 5% and a confidence level of 95%, the recommended sample size was 385 respondents, indicating that a sample of 128 participants per scenario was acknowledged to be sufficient2_{. Since the population is quite large and the researcher did not} have access to all the students in the population, a convenience sample was carried out. The survey was distributed online across various social media to reach the target group. For this study, students were chosen because they are naturally part of the population of interest as they have experience using the Internet. During the period of data collection, a total of 143 students were assigned to all three scenarios.

3.2. Results

3.2.1. Testing for assumptions of ANOVA

Missing values and outliers

All variables under investigation were checked for missing data. The amount of missing data was < 10% for all dependent variables, which implied that some respondents did not finish the questionnaire. For the correlation analyses, pairwise deletion of these cases was used. The outliers were examined to ensure no data entry or instrument errors were made and, fortunately, no outliers were detected.

(22)

Homogeneity of variances

Levene’s test indicated that the assumption of homogeneity of variance had not been violated with regards to the manipulation of the temporal distance for the inclination to disclose personal data (p = .199), the likelihood of costs (p = .882), the likelihood of benefits (p = .172), the timespan of the costs (p = .575), and the timespan of the benefits (p = .057).

3.2.2. Demographic variables

First preliminary analyses compared demographic variables to the main variable of interest; the inclination to disclose personal data. Bivariate correlations showed no significant correla tion between age (M = 24.5, SD = 5.33) and the inclination to disclose personal data, r(125) = .06,

p = .520. Also, there was no significant correlation between gender (40% male, 60% female)

and the inclination to disclose personal data, r(124) = -.07, p = .425. The results suggested no significant correlations between the study variables. Hence, none of the demographic variables were used as covariates in subsequent analyses.

Hypothesis 1: Consumer reliance on benefits and costs is dependent upon the temporal

distance of these elements, such that people will base their decision more on the benefits when

benefits are proximal and costs are distant but more on the costs when costs are proximal and

benefits are distant.

The results of Study 1 should indicate that respondents act according to the element (benefit or cost) that has the closest proximity of occurring. Accordingly, results should indicate that participants are more likely to disclose personal data in the proximal benefits/distant costs condition than in the proximal costs/distant benefits condition. Several manipulation checks had to be executed to confirm this hypothesis.

(23)

3.2.3. Manipulation checks of temporal distance on the timespan of benefits and costs

The manipulation of temporal distance should have significantly influenced the timespan of benefits and costs occurring. Indicating that participants who are in the proximal benefits/distant costs condition are more likely to disclose personal data than people in the proximal costs/distant benefits condition.

Two manipulation checks were conducted to find evidence that the manipulation of temporal distance had succeeded. Participants were asked within what timespan they expected to encounter benefits or costs. The researcher predicted that when consumers were participating in the proximal benefits/distant costs condition, they would indicate the temporal distance of the benefits to be significantly lower than the temporal distance of the costs. Likewise, in the proximal costs/distant benefits condition, participants were expected to evaluate the temporal distance of the costs to be significantly lower than the temporal distance of the benefits.

A one-way ANOVA showed that there was a statistically significant difference between the proximal benefits/distant costs condition (M = 2.65, SD = 1.34) and the proximal costs/distant benefits condition (M = 2.03, SD = 1.31) for timespan of the costs, F(1, 126) = 6.86, p = .010, η² = 0.05. However, a one-way ANOVA showed that there was no significant difference between the proximal benefits/distant costs condition (M = 1.66, SD = 1.10) and the proximal costs/distant benefits condition (M = 2.02, SD = 1.27) for timespan of the benefits,

F(1, 126) = 2.82, p = .096, η² = 0.02.

To confirm that these results are attributable to the manipulated temporal distance and not to the perceived hypothetical distance, the likelihood of benefits and costs occurring was measured. As expected, analyses did not yield significant results between participants in the proximal benefits/distant costs condition (M = 3.78, SD = 1.02) and participants in the proximal costs/distant benefits condition (M = 4.02, SD = 1.06) in reporting the likelihood of costs, F(1, 127) = 1.67, p = .198, η² = 0.01. Likewise, no significant results were found between

(24)

participants in the proximal benefits/distant costs condition (M = 3.87, SD = 0.92) and participants in the proximal costs/distant benefits condition (M = 3.65, SD = 1.03) in reporting the likelihood of benefits, F(1, 127) = 1.65, p = .202, η² = 0.01.

3.2.4. Temporal distance of benefits and costs on disclosure of personal data

The manipulation of temporal distance of benefits and costs should have significantly influenced the inclination to disclose personal information. Indicating that participants who are in the proximal benefits/distant costs condition are more likely to disclose personal data than people in the proximal costs/distant benefits condition. A one-way ANOVA yielded a significant main effect that participants were more likely to disclose personal data in the proximal benefits/distant costs condition (M = 3.23, SD = 1.31) than in the proximal costs/distant benefits condition (M = 2.72, SD = 1.41), F(1, 140) = 4.92, p = .028, η² = 0.03. This significant result supports the hypothesis, showing that as temporal distance of the costs increases, the inclination to disclose personal data will increase.

(25)

4. STUDY 2

Manipulation of the mind-set and temporal distance of benefits and costs

4.1. Method

Study 2 examined to what extent consumer reliance on benefits and costs and its subsequent inclination to disclose personal information was dependent upon these elements, by manipulating mind-set and temporal distance of benefits and costs. Under a concrete mind-set, consumers should have the tendency to rely on proximal benefits or costs, proving a congruence between mind-set and psychological distance of these benefits and costs. This study explored if indeed the effect of the temporal distance on consumer reliance on benefits and costs was moderated by the consumer’s abstract or concrete mind-set. Results should indicate whether this effect is reliant upon the congruence between mind-set and psychological distance of benefits and costs.

Study 2 consisted of three groups of mind-set manipulations; 1) concrete mind-set, 2) abstract mind-set, and 3) a control group. Furthermore, the study consisted of three manipulations of temporal distance; 1) the first group of participants was exposed to temporally proximal benefits and temporally distant costs, 2) the second group of participants was exposed to temporally proximal costs and temporally distant benefits, and 3) the third group of participants functioned as a control group whereby the proximity of temporal distance of benefits and costs was not emphasised.

4.1.1. Design and procedure. We conducted an experimental design to examine the

effect of the manipulation of the mind-set and the manipulation of temporal distance of benefits and costs on the inclination to disclose personal data. A 3x3x3 study was conducted: 3 (between-subjects: concrete mind-set vs. abstract mind-set vs. control) x 3 (between-subjects:

(26)

subjects: a website visit vs. Facebook login vs. sharing data with app). Each participant took part in two of the three scenarios, and for each of these scenarios they were randomly assigned to only one of the manipulated mind-set conditions and one of the manipulated temporal distance conditions. The outcomes of this testing should support H2 by showing a relation between 1) concrete mind-set & proximal benefits/distant costs, 2) concrete mind-set & proximal costs/distant benefits, 3) abstract mind-set & proximal benefits/distant costs, and 4) abstract mind-set & proximal costs/distant benefits.

All constructs were measured using multi-item Likert-type scales, and to ensure construct validity scales from previous studies were adapted wherever possible. All participants were exposed to the same questions to analyse the data on the same constructs to find support that CLT is the underlying mechanism for the privacy paradox. In Study 2, participants read a short story manipulated by the researcher by either stimulating the participant’s concrete mind-set, abstract mind-mind-set, or by narrating a story on grasshoppers (control condition of the manipulated mind-set) (Fujita, Trope, Liberman, & Levin-Sagi, 2006). Thereafter, participants were asked several probing questions to make certain that they had been brought in the right mind-set. Subsequently, participants got the task to select the identification that best described their behaviour. Participants were confronted with a statement of an action followed by two options; an action in terms of either how it was performed, which is consistent with lower-levels construals or an option of why it was performed, which is consistent with higher-level construals (Fujita et al., 2006). These questions were based on the Behaviour Identification

Form, containing questions assessing the level at which individuals construe certain activities

(Vallacher & Wegner, 1989).

The second manipulation of this research was performed on the temporal distance of benefits and costs. Participants were presented a short story on three different scenarios; a website visit, using Facebook to log in, and sharing data with an app. Participants were

(27)

presented questions on their likelihood to disclose personal data and they had to describe to what extend they actually perceived benefits and costs of disclosing data to be disadvantageous or advantageous. To measure the likelihood of disclosing data, a seven-point Likert scale was used with “very unlikely” and “very likely” as anchors (Sparks & Browning, 2011). Open-ended questions were used to record advantages and disadvantages of disclosing personal information. In addition, a five-point scale was used to verify whether participants saw the potential positive and negative aspects of sharing data to be disadvantageous or advantageous. Last of all, the timespan of consequences and/or benefits of disclosing personal data occurring was measured. The expected timespan of benefits and costs actually occurring was probed as a manipulation check. CLT research uses time periods such as next month, half a year, and a year to represent distant future events, whereas time periods such as immediately, tomorrow, or next week to represent near future events (Nussbaum et al., 2003).

4.1.2. Participants. The population for this study consisted of male and female

participants from different levels of education from all countries. Since the population is quite large and the researcher did not have access to all the students in the population, a convenience sample was carried out. The survey was distributed online across various social media to reach the target group. During the period of data collection, a total of 178 participants were assigned to two of the three scenarios.

4.2. Results

4.2.1. Testing for assumptions of ANOVA

Missing values and outliers

All variables under investigation were checked for missing data. The amount of missing data was < 10% for all dependent variables, which implies that some respondents did not finish the

(28)

questionnaire. The outliers were examined to ensure no data entry or instrument errors were made and, fortunately, no outliers were detected.

Homogeneity of variance

Levene’s test indicated that the assumption of homogeneity of variance had not been violated with regards to the manipulation of the temporal distance for the inclination to disclose personal data (p = .545) and the timespan of the costs (p = .054). However, the assumption of homogeneity of variance was violated for the timespan of the benefits (p < .001). Levene’s test indicated that the assumption of homogeneity of variance had not been violated with regards to the manipulation of the mind-set for the inclination to disclose personal data (p = .746).

4.2.2. Demographic variables

First preliminary analyses compared demographic variables to the main variable of interest; the inclination to disclose personal data. Bivariate correlations showed a significant correlation between age (M = 25.7, SD = 8.08) and the inclination to disclose personal data, r(160) = -.21,

p = .007. This might be explained by the fact that younger generations seem more comfortable

with the exchange of privacy as they are more likely to act according to the social exchange theory. Also, consumers of the current generation are more likely to disclose personal data because they believe that both authenticity and reputation originate from their online presence (Sprague, 2007). There was no significant correlation between gender (36% male, 64% female) and the inclination to disclose personal data, r(162) = -.13, p = .110. No significant correlation was found between education and the inclination to disclose personal data, r(162) = -.02, p = .805, and the same was true for time spend online and the inclination to disclose personal data,

r(162) = .13, p = .103. Hence, the demographic variables above were not used as covariates in

(29)

Hypothesis 2: The effect of the temporal distance on consumer reliance on benefits and costs

is moderated by the consumer mind-set. Such that participants rely on consequences that are

congruent with their mind-set.

The results of Study 2 should indicate that respondents act according to the element (benefit or cost) that is congruent with their mind-set. Accordingly, the results should indicate that when participants are in a concrete mind-set, they are more likely to rely on the benefits and costs that are more proximal. Several manipulation checks had to be executed to confirm this hypothesis.

4.2.3. Manipulation checks of temporal distance on the timespan of the costs and benefits

A one-way ANOVA on timespan of the costs showed that there was a statistically significant difference between the proximal benefits/distant costs condition (M = 3.67, SD = 2.25), the proximal costs/distant benefits condition (M = 2.42, SD = 1.88), and the control condition (M = 3.78, SD = 2.21), F(2, 161) = 6.96, p = .001, η² = 0.08. The significant outcome of this ANOVA indicates that there is a difference between the three conditions, however planned contrasts had to be carried out to find the origin of the significant results.

Planned contrasts revealed that any manipulation of the temporal distance of the benefits and costs significantly influenced the perceived time span of the costs compared to the control group, t(161) = 2.07, p = .039. Predominantly, contrasts revealed a significant difference between the proximal benefits/distant costs condition and the proximal costs/distant benefits condition, t(161) = -3.10, p = .002, d = 0.60. This indicates that in the proximal benefits/distant costs condition, costs are perceived to be temporally further away than in the proximal costs/distant benefits condition, meaning that the manipulation of temporal distance was successful. It is assumed that in typical disclosure situations (control condition) people rely more on the proximal benefits than on the proximal costs. As expected, planned contrasts

(30)

showed no significant difference between the proximal benefits/distant costs condition, and the control condition t(161) = 0.26, p = .796, d = 0.05.

A one-way ANOVA on the timespan of the benefits showed that there was a statistically significant difference between the proximal benefits/distant costs condition ( M = 1.60, SD = 1.36), the proximal costs/distant benefits condition (M = 2.42, SD = 2.07), and the control condition (M = 1.41, SD = 1.09), F(2, 161) = 6.45, p = .002, η² = 0.08. The significant outcome of this ANOVA indicates that there is a difference between the three conditions, however planned contrasts had to be carried out to find the origin of the significant results.

Planned contrasts revealed that any manipulation of the temporal distance of the benefits and costs significantly influenced the perceived time span of the benefits compared to the control group, t(161) = -2.32, p = .022. Predominantly, contrasts revealed a significant difference between the proximal benefits/distant costs condition and the proximal costs/distant benefits condition, t(161) = 2.74, p = .007, d = 0.47. This indicates that in the proximal benefits/distant costs condition, benefits are perceived to be temporally closer than in the proximal costs/distant benefits condition, meaning that the manipulation of temporal distance was successful. It is assumed that in typical disclosure situations (control condition) people rely more on the proximal benefits than on the proximal costs. As expected, planned contrasts showed no significant difference between the proximal benefits/distant costs condition, and the control condition t(161) = -0.64, p = .521, d = 0.15.

4.2.4. Main effects and interaction effect of the mind-set, temporal distance, and scenario on the inclination to disclose personal data

For the reason that all manipulation checks were successful, a factorial ANOVA was conducted to measure the effect of the independent variables on the dependent variable and to find

(31)

evidence to support the hypothesis. The first three analyses focused on the main effect of all three independent variables on the inclination to disclose personal data (DV). Subsequently, analyses were performed to find an interaction effect of mind-set and temporal distance on the inclination to disclose personal data.

The main effect of mind-set on the inclination to disclose information was not significant, F(2, 140) = 0.02, p = .980. Accordingly, the concrete mind-set, abstract mind-set, and the control group did not differ on the reported amounts of the inclination to disclose personal data. These results confirm that there is no main effect of mind-set on the inclination to disclose personal data. The main effect of temporal distance on the inclination to disclose information was not significant, F(2, 140) = 0.92, p = 0.402. In view of that, the conditions of proximal benefits/distant costs, proximal costs/distant benefits, and the control group did not differ on the reported amounts of the inclination to disclose personal data. The main effect of scenario on inclination to disclose personal data was not significant either, F(2, 140) = 1.38, p = .256. Likewise, website visit, Facebook login, and sharing data with an app did not differ on the reported amounts of inclination to disclose personal data. All the analyses above yielded non-significant results on the main effects, confirming the expectations of the researchers (table

1).

The manipulations of mind-set and temporal distance were expected to significantly influence the inclination to disclose personal information. This would indicate that when benefits are temporally proximal and costs are temporally distant, people in a concrete mind -set rely on benefits and people in an abstract mind--set rely on costs, and when costs are temporally proximal and benefits are temporally distant, people in a concrete mind-set rely on costs and people in an abstract mind-set rely on benefits. This assumption was expected to be true to indicate an interaction effect of mind-set and temporal distance on the inclination to disclose personal data.

(32)

A twoway betweengroups analysis of variance was conducted to explore the impact of mind -set and temporal distance on the inclination to disclose personal data. Participants were divided into three groups according to their manipulated mind-set (Group 1: abstract mind-set, Group 2: concrete mind-set, Group 3: control) and they were allocated to a manipulated temporal distance condition (Group 1: proximal benefits/distant costs, Group 2: proximal costs/distant benefits, Group 3: control). The interaction effect between mind-set and temporal distance was not statistically significant, F(4, 140) = 1.01, p = .405 (table 1).

Table 1. Main effects and interaction effect of the mind-set, temporal distance, and scenario on the inclination to disclose personal data.

Inclination to disclose personal data

Variable F p η² Mind-set 0.020 .980 .000 Scenario 1.375 .256 .019 Temporal Distance 0.918 .402 .013 Mind-set*Scenario 0.829 .509 .023 Mind-set*Temporal Distance 1.009 .405 .028 Scenario*Temporal Distance 1.629 .170 .044 Mind-set*Scenario*Temporal Distance 0.847 .563 .046 R2_{= .137 (Adjusted R}2_{= -.023)}

These results imply that hypothesis 2 was not supported as the mind-set did not significantly moderate the effect of the temporal distance of benefits and costs on the inclination to disclose personal data (figure 2).

(33)

Figure 2. Interaction effect between mind-set and temporal distance on the inclination to disclose personal data 3,000 3,200 3,400 3,600 3,800 4,000 4,200 4,400 4,600

Abstract Concrete Control

T e m p o r a l d is ta n c e o f b e n e fi ts a n d c o st s Mind-set

Inclination to disclose personal data

(34)

5. DISCUSSION AND CONCLUSION

Since online privacy is a controversial topic and no research has been conducted on whether CLT accounts for the privacy paradox, this study will be of high value to academics. Even though not all the aspects of the hypotheses were supported, the results have relevant implications for academic literature and marketers. The fact that consumer reliance on benefits and costs is dependent upon the temporal distance of these elements is a novel finding, offering opportunities for marketers to apply this knowledge in their business strategies.

5.1. Study 1

The first study manipulated the temporal distance of benefits and costs to understand the effect it has on the disclosure of personal data. This effect was expected based on Construal Level Theory, which reveals that people rely on the element that has the highest proximity of occurring (Trope et al., 2007). The results of the present study found proof for hypothesis 1, which predicted that consumer reliance on benefits and costs is dependent upon the temporal distance of these elements, such that people will base their decision more on the benefits when benefits are proximal and costs are distant but more on the costs when costs are proximal and benefits are distant. In sum, consumers will rely more on the benefits (costs) of disclosing personal data when these benefits (costs) are presented as temporally close.

Generally, based on the results described in Study 1 it can be concluded that participants were more likely to disclose personal data when the benefits of disclosing personal data were temporally close in the proximal benefits/distant costs condition than when the costs of disclosing personal data were temporally close in the proximal costs/distant benefits condition. The results confirmed the expectations of hypothesis 1, which stated that an increase of the temporal distance of the costs leads to an increase of the inclination to disclose personal data.

(35)

However, results did not find evidence that an increase of the temporal distance of the benefits leads to a decrease of the inclination to disclose personal data.

Overall, these results provide valuable insights for academics and marketers in practice. Previous research showed that the privacy paradox exists when consumers report high values of privacy concerns, but still display behaviour that contradicts their concerns. Several explanations for this paradox have been put forward by academics, but none researched the influence of psychological distance on the privacy paradox. Psychological distance is widely acknowledged to demonstrate a subjective representation of something close or far away from a consumer, presenting marketers opportunities to act accordingly. Firstly, it is advisable for marketers to emphasise a consumer’s expected benefits of disclosing personal data to amplify their inclination to disclose information. This supposition arises from the fact that consumers tend to disclose more personal information when the benefits are expected to be occurring soon. Secondly, results showed that temporal distance guided consumer predictions and behaviour, demonstrating that the temporal distance affected consumer reliance on benefits and costs. Implications for marketers consist of opportunities to reassure consumers that they are acting in the consumer’s best interest. So, based on CLT, marketers should emphasise immediate benefits sharing data if they want consumers to disclose personal information. The collection and processing of this information by companies in turn provides consumers with further benefits. These findings imply that privacy conscious consumers are more likely to disclose personal data when the benefits of disclosing personal data are temporally close, providing evidence that CLT can be an underlying mechanism for the privacy paradox.

The outcomes of the first study gave the researcher input for conducting a second study. The researcher proposed that the temporal distance of benefits and costs, when moderated by the mind-set of consumers, influences the inclination to disclose personal data.

(36)

5.2. Study 2

This study manipulated the mind-set of participants and the temporal distance of benefits and costs to understand the effect it has on the disclosure of personal data. The results of the second study indicated that in the proximal benefits/distant costs condition, costs were perceived to be temporally further away than in the proximal costs/distant benefits condition. Likewise, in the proximal benefits/distant costs condition, benefits were perceived to be temporally closer than in the proximal costs/distant benefits condition. As expected, results confirmed that there is no main effect of mind-set, temporal distance of benefits and costs, or scenario to which the participants were exposed on the inclination to disclose personal data. These results were predicted because participants were exposed to a combination of two manipulations. They were first exposed to a manipulation of the mind-set and subsequently to a manipulation of temporal distance. This information is valuable because it was expected that the combined manipulation of both mind-set and temporal distance yielded significant results, and not each independent variable by itself. Unfortunately, the study did not find concrete evidence to support hypothesis 2, which predicted that there would be an interaction effect of mind-set and temporal distance on the inclination to disclose personal information. Nevertheless, these results do provide marketers and academics with some valuable insights in the behaviour of privacy conscious consumers.

Prior research revealed that consumers are becoming progressively more anxious of the amount of personal information that is available online. Yet, they display contradicting behaviour. Consumers are continuously exposed to consequences of sharing data, which makes it difficult for them to make informed decisions on whether or not to share data. Therefore, the role of marketers is to make consumers feel at ease and eliminate the contradiction between attitude and behaviour. Firstly, it is advisable for marketers to clearly communicate the benefits of sharing personal data. Literature on CLT and the privacy paradox propose that the benefits

(37)

of disclosing personal data online are perceived as psychologically closer than the costs of disclosing personal data online. Hence, priming the immediate benefits would positively influence the consumer’s inclination to disclose personal data. Secondly, implications for researchers consist of guiding consumers into an adequate mind-set when they are making decisions on disclosing information. When consumers have to share information immediately, they act according to their concrete mind-set and habitually rely on the benefits of disclosing personal data. This assumption arises from the fact that consumers tend to disclose more personal information when the benefits are expected to be occurring soon. Whereas when consumers express their privacy concerns they behave according to their abstract mind-set and rely on the future costs of disclosing personal data.

The present study did not provide sufficient evidence that there is an actual linkage between mind-set, temporal distance of benefits and costs, and the inclination to disclose personal information. This meant that there is no proof that when benefits are temporally proximal and costs are temporally distant, people in a concrete mind-set rely on benefits and people in an abstract mind-set rely on costs, and when costs are temporally proximal and benefits are temporally distant, people in a concrete mind-set rely on costs and people in an abstract mind-set rely on benefits. Even though the results were not evaluated significantly, the outcomes did show tendency towards supporting the hypothesis.

In conclusion, the results of both studies indicate that temporal distance has a fundamental influence on the inclination to disclose personal information, however, consumer mind-set does not moderate this effect. So, further research is required to expose whether an altered manipulation of theories on construal level explains the privacy paradox. Other dimensions of psychological distance, combined with a manipulated mind-set, might accentuate the observed results, leading to opportunities for future studies.

The privacy paradox : a construal level theory perspective