Construing people’s privacy information disclosure online

(1)

Construing people’s privacy information disclosure

online

Master Thesis

MSc in Business Administration - Marketing Track

Student

: Maartje Jans

Student number

: 10981934

Submission date : 26

th

of April, 2016

Version

: Final version

Program

: MSc. Business Administration - Marketing

Institution

: University of Amsterdam

(2)

Construing people’s privacy information disclosure online – M. Jans (University of Amsterdam) 2

Abstract

The current uses of consumer data create a fundamental tension between corporates’ economics interests and consumers’ privacy concerns. In the context of online transactions and consumers’ privacy, the privacy paradox is a well-established concept in the literature that refers to the inverse relationship between online users’ concern and actual behaviour. Since the exact contours and causes of the concept appear to be debatable, no previous study has adopted a theoretical explanation that explicitly underlies the privacy paradox. To address this gap, the present thesis will shed more light on CLT as an overarching theory that can account for the privacy paradox. In particular, the present work examines the direct relationship between one’s mindset and self-disclosure behaviour that is expected to be moderated by the temporal distance of costs and benefits of disclosing personal information in typical online situations.

This paper presents two experimental studies that were set up to provide users with realistic self-disclosure choices in online situations. The first study was conducted to measure participants’ actual self-disclosure behaviour, after their mindset was manipulated, on a website that was specifically designed for the purpose of this study. A similar mindset manipulation together with a cost/benefit framing technique was used to examine one’s tendency to disclose personal information online in the second study.

Although, the findings of both studies did not provide support for the proposed hypotheses, the current work adds to the existing models that are developed to understand the social and psychological mechanisms that underlie consumers’ irrational, economic behaviour. Both theoretical and managerial implications, as well as limitations, and directions for future research are discussed.

(3)

Statement of originality

This document is written by Student Maartje Jans 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.

(6)

1. Introduction

Since many people are spending a significant amount of their time on online activities, the Internet has become an integral part of our everyday life. Today, we are able to transact and communicate much faster than ever before through the exponential use of mobile devices. These devices offer unique combinations of utility, which are specially designed to offer social interaction, entertainment, convenience and personalization (Keith, Thompson, Hale, Lowry, Greer, 2013). Yet, these services cannot survive without users’ willingness to disclose personal information online (Smith, Dinev, Xu, 2011) since the success of online business increasingly depends on the analysis of consumer insights (Culnan & Armstrong, 1999). Increasingly sophisticated technologies together with the exponential growth of the online data-sharing environment have leveraged the ability of online businesses to target the right audience by tracing consumer behaviour online and exploiting massive databases of transaction data (Culnan & Bias, 2003; Norberg, Horne, Horne, 2007). Accordingly, when users click options to navigate to site-to-site, post comments, interact with each other or disclose personal information to access a Web site, companies are able to efficiently collect, store, or exchange large amounts of customer information. Often, this happens without the users’ consciousness or consent and a seemingly asymmetric relationship between company and consumer has been recognized in prior studies (Paine, Reips, Stieger, Joinson, Buchanan, 2007).

These uses of consumer data create a fundamental tension between corporates’ economics interests and consumers’ privacy concerns. On the one hand, companies are thirsty for more consumer information in order to protect their competitive advantage (Culnan & Bies, 2003). On the other hand, users’ concerns about privacy violations have grown in importance as people do not control the secondary use of their personal information online (Norberg et al., 2007).

(7)

Nevertheless, a so-called ‘privacy paradox’ has been widely documented by a large amount of research studies (Norberg et al., 2007; Keith et al., 2013; Spiekermann, Grossklag, Brandt, 2001; Taddicken, 2013). This phenomenon refers to the incongruence between people’s stated risks beliefs and their actual online self-disclosure behaviour and is well established in the literature (Norberg et al., 2007; Spiekermann et al., 2001; Taddicken, 2013; Trepte & Reinecke, 2011). In particular, in the realm of privacy online it seems that people do not always act rationally (Acquisti, 2004; Acquisti & Grossklag, 2005) and one’s privacy intentions are no precise predictors of actual behaviour. (Keith et al., 2013; Norberg et al., 2007). Several researchers have attempted to clarify the privacy paradox and demonstrated that privacy is a sensitive, multidimensional concept that is difficult to study. Though, none of these researchers adopted a theoretical framework that underlies the incongruence between people’s privacy concerns and actual privacy protecting behaviour. Therefore, future research in the privacy area is still essential and should be directed towards framing richer models that explain why this paradox exists (Acquisti, 2004; Keith et al., 2013; Trepte & Reinecke, 2011).

Drawing from work in the field of psychology and behavioural economics that examined the relationship between mental processing and changes in judgments and decisions over time, the present thesis concentrates on the Construal Level Theory (CLT) to address this gap. Moreover, CLT has been perceived as an underlying framework that have enhanced our understanding about the variety in preference inconsistencies over time (Fiedler, 2007; Trope et al., 2003). Previous research demonstrated that CLT has interesting effects on judgments and choices and clarified how different dimensions of psychological distance alters the way people mentally represent situations by shifts in weight given to high-level or low-level features (Liberman & Trope, 1998; Lynch & Zauberman, 2007).

Linking CLT to existing privacy literature will shed new light on the current dichotomy between people’s privacy concerns and actual self-disclosure behaviour in typical

(8)

online contexts. In particular, the emphasis of the present work is on the systematic change of people’s cognitive operations in the decision making process. Therefore, this thesis examines the following research questions:

How do mental representations (abstract vs. concrete) affect people’s decision-making process of disclosing personal information in typical online situations?

The current work presents the theoretical principles of CLT and applies these to typical online self-disclosure situation. Moreover, it will be inspected how CLT accounts for the dichotomy between people’s privacy intentions and behaviour over time. In doing so, I hope to expose additional value to the current CLT and privacy literature in several ways and stimulate new research that will further support our assumption that CLT can be perceived as the underlying framework of the privacy paradox. Besides, if this thesis demonstrates the effect of activating mindset it will make a novel contribution to extant models that are helpful in describing the decision making process of consumers. From a business perspective, if CLT can be perceived as an underlying mechanism that explains the privacy paradox, this can offer increased value to companies. In particular, it is fair to assume that if companies recognize the importance of the congruence between the consumers’ mental representations and the interface through which they interact with them, this might help companies to improve their services and products both online and offline.

This paper starts with a review of the existing privacy literature and, subsequently, combines this with the logic of CLT to formulate hypotheses. Two distinctive studies are conducted in order to test the proposed hypotheses. Next, the methodology of both studies are presented, followed by the results and a discussion section. In the last section of this paper the discussion of the overall findings, limitations, theoretical and managerial implications, directions for future research and final conclusion will be presented.

(9)

2. Literature review

2.1 Self-disclosure

Revealing personal information on the Internet can be defined as self-disclosure, which is ‘the process of making the self known to other persons’ (Jourard & Lasakow 1958, p. 91). In many cases, the disclosure of personal information is a precondition for the development of meaningful relationships, people’s social functioning and psychological well-being (Taddicken, 2013; Trepte & Reinecke, 2011). Furthermore, self-disclosure is a vital element of communication that serves to enhance bonds of trust between people and increases the mutual understanding (Taddicken, 2013). According to Berger’s (1979) uncertainty reduction theory, people like others who disclose a greater amount of personal information more and they have the intention to disclose more to those who they like.

Yet, the increased popularity of contemporary technologies that promotes the widespread and constant sharing of private data, together with the current engagement into commercial and social activities on the Internet, has transformed the ways in which information is disclosed or used. Compared with other modes of communication, the potential availability of self-disclosed information is significantly greater on the Internet (Joinson, Paine, Buchanan, Reips, 2008; Taddicken, 2013). A study by Smith et al. (2011) identified social adjustment and financial rewards as components of benefits that foster information disclosure. Often, self-disclosure serves to access services, to make online purchases, to join online groups or is requested to enjoy the benefits of personalized services (Joinson, Reips, Buchanan, & Schofield, 2010).

Regardless of the benefits and convenience of currently improved online services, the increased use of these technological advances has been offset by innovations that heighten the costs of self-disclosure (Beak, 2014). When online users visit, share or like a website they are

(10)

continuously encouraged to disclose personal information and, subsequently, leave digital footprints. Technological features such as cookies and beacons have enabled companies and other unknown parties to easily trace consumers’ behaviour by collecting, sharing and transmitting personal information (Sipior, Ward, Mendoza, 2011). Consequently, the more people disclose information, the more the risks of privacy violations will increase. For example, a study undertaken by Beak (2014) found that the use of popular mobile fitness and health applications might pose serious damage to the privacy of mobile users because these apps undoubtedly collect and store specific personal details like health information, lifestyle factors, and location. It can be argued that in general companies know a lot more about the way personal information is collected and used compared to users. Besides, for most users is it already difficult enough to clearly understand company’s privacy policies, detecting when their personal information is violated makes it even more complicated (Sholtz, 2001). As a result, negative privacy consequences are produced due to the outsized information asymmetries that exist between online users and companies.

All in all, people’s privacy has emerged as an important issue due to the tension between consumer and corporate interests. When people interact with each other either online or offline, they are continuously challenged to strike the right balance between the desire to protect their privacy and the desire to disclose information about the self (Taddicken, 2013). In other words, self-disclosure is an influential mechanism of social interaction that might contrasts people’s need for privacy. However, to put this issue into further perspective, it is vital to take a step back and to shed light on the definitions and subsets of privacy and people’s privacy concerns.

2.2 Information privacy

Despite the notion that privacy can be perceived as a universal human right – ‘the right to be left alone’ (Vasalou, Joinson, Houghton, 2015; Warren & Brandeis, 1890), research has

(11)

illustrated that there is no agreed-on definition of privacy (Paine et al., 2007; Phelps, Nowak, Ferrel, 2000; Smith. Masicampo, Ambady, 1996; Knijnenburg, Kobsa, Jin, 2013). Accordingly, contemporary privacy conceptions have built on the work of Alan Westin and Irwin Altman to (Loosen, as cited in Trepte & Reinecke, 2011;). Both have stood the test of time and set a foundation for the further understanding of the core of the privacy field by focusing on how people control and regulate the access to personal information. Both authors have stated that the need for privacy is an ongoing dynamic, based on our desirability and internal vs. external conditions to which we continuously adapt our levels of privacy. Altman proposes that individuals are looking for an ideal balance between being open and closed to interaction. He defines privacy as ‘the selective control of access to the self’ (Taddicken & Jers, as cited in Trepte & Reinecke, 2011, p.11). This control-based definition has since embedded in most conceptual definition and conceptualization of privacy and has been further applied to numerous attributes and measurements to operationalize privacy (Smith et al., 2011).

The evolving technologies of the past years, however, have challenged the conventional understanding of the privacy concept. Today, privacy has become a currency that we can exchange for free access to web content, personalized services or discounts (Motiwalla, Li, Liu, 2014). Therefore, it can be questioned if the more traditional approaches to understand privacy are applicable to the unique privacy problems contemporary technologies has posed. For example, Paine et al. (2007) argued that the Internet and technologies have created unique privacy problems that contrast the issues stated by previous privacy research. In particular, Social Networking Sites (SNSs) continuously encourage content provision by their users, because the functionality of these sites often relies on the disclosure of personal information. Furthermore, according to Selvadurai (2013), recognition technologies and the development of social networking databases have changed the paradigms

(12)

of privacy. Likewise, Paparcharissi & Gibson (as cited in Trepte & Reinecke, 2011) argued that contemporary social media are further blurring the boundaries between private and public, because relevant personal information can be shared with others over time and space (Schmidt as cited in Trepte & Reinecke, 2011).

All in all, it has become impossible to separate the Internet from our ‘real world’ since it is more integrated with our daily lives than ever before. The distinction of what is public and what is private is in constant transformation and privacy has become a complex concept of multifaceted interests (Acquisti, 2004). Therefore, it is essential to put emphasis on a sub-dimension of privacy that can be applied to a specific context associated with the rise of the Internet that changed the landscape of information exchange and the related disclosure of personal information. To be more specific, this paper will deal directly with information privacy, which represents a fundamental aspect of privacy and refers to the individual’s ability to control the improper access, unauthorized use and disclosure of identifiable information about themselves (Keith et al., 2013; Smith et al., 2011; Selvaduari, 2013; Culnan & Bies, 2003).

Previous research has shown that the information privacy concept is increasingly affected by the way personal information is collected, accessed, and processed without one’s knowledge and has become an overall issue of online users (Saeri, Macchia, Smith, Louis, 2014). In particular, a wide majority of people is concerned about their personal data and consider it to be a vital aspect in their online decision making (Acquisti, 2004; Jensen, Colin Jensen, 2005; Paine et al. 2007; Phelps et al., 2000). For example, a survey conducted by Price Waterhouse Coopers (as cited in Acquisti, 2004) demonstrated that more than 66 percent of the respondents “would shop more online if they knew retail sites would not do anything with their personal information” (p.21). In addition, the findings of Acquisti’s study (2004) demonstrated that the three most common privacy concerns are: the theft of personal

(13)

information, organizations share personal information without permission and the risks of unconfident transactions. Since service providers, marketers, retailers or non-profit organization are able to collect, store, utilize, and exchange personal data of online users with increasingly frequency and in different context, people experience threats to their informational privacy because their control is lost or unwillingly decreased (Wu, Huang, Yen, & Popova, 2012; Phelps et al., 2000). In particular during online transactions, when an individual provides personal data to external parties, he/she loses control for an unpredictable time period (Acquisti & Grossklags, 2005). Consequently, violations of privacy occur and people only control the extent that they are able to manage their self-disclosure (Pan & Zinkhan, 2006).

2.3 The privacy paradox

Since the concept of information privacy and self-disclosure are theoretically related (Smith et al., 2011; Taddicken, 2013; Trepte & Reinecke, 2011), it is logical to expect that concerned individuals will disclose few or no personal information at all on the web. Yet, this does not seem to translate well into people’s online protecting behaviour. There are consistent research findings (Acquisti, 2004; Culnan & Milne, 2001; Park et al., 2012; Spiekermann et al. 2001) that demonstrate that people are less than selective or even careless in the protection of their own privacy. And that the high concerns that people express about their personal information, often, do not predict actual disclosure behaviour. For instance, Beak’s study (2014) presented that many people voluntary disclose personal data while they stated that they are highly concerned that online organizations misuse their private and personal information. In alignment, Acquisti & Grossklag’s (2005) linked individuals’ privacy concerns to the use of loyalty cards and indicated that more than 85 percent of the highly concerned individuals signed up for the loyalty card thereby disclosing personal information. Comparably, Spiekermann et al. (2010) observed participants during their online shopping trip and

(14)

compared respondents’ self-reported privacy preferences with their actual privacy protecting behaviour. This study’s findings demonstrated that the majority of participants revealed identifying information about themselves, which made it possible to construct personal profilers on the basis of only one shopping session.

All in all, an apparent dichotomy between individuals’ privacy attitudes and actual behaviour has been highlighted by prior research, surveys and experiments. The existence of this so called ‘privacy paradox’ have been confirmed by researchers (Acquisti & Grossklag, 2005; Norberg et al., 2007; Smith et al., 2011; Van de Garde-Perik, Markopoulos, De Ruyter, Eggen, Ijsselsteijn 2008; Park et al., 2012; Taddicken, 2013) from different perspectives independently. A relevant question here to be considered is why do concerned online users act irrational and engage in violating privacy behaviour on the Internet? Yet, some scholars have tried to address the inverse relationship between users’ concerns and behaviour and showed that the privacy paradox is a well-established concept, but the exact contours and causes of the concepts are debatable.

For example, Norberg et al. (2007) have used two prominent variables, trust and risk, to explain why the privacy paradox exists. They have found support that risk negatively affects individuals’ intentions to disclose personal information. But these intentions are not a truthful predictor of actual disclosure behaviour. Also, the influence of trust on actual disclosing behaviour remains a question since the authors did not find valid evidence. In connection with this emphasis on trust as an antecedent associated with privacy concerns, people might base their trust in technology on the presence of privacy policies that serve as “trust-marks” (Smith et al., 2011). Conversely, the findings of Jensen et al. (2005) suggested that users rarely read privacy policies and that those will not help them to make informed choices because online privacy policies are not easy to find or to understand. Furthermore, these authors showed that many online users often overestimated their own knowledge of

(15)

technology, their online behaviour, and potential risks. In this domain of online privacy protection, Saeri et al. (2015) tried to increase our understanding of people’s online privacy protection behaviour on Facebook and demonstrated that the there was no unique effect of trust on intentions, nor on behaviour.

Knijnenburg et al. (2013) made a distinction in the way individuals can be classified based on their information disclosure behaviour and developed a multidimensional measure to classify online users’ self-disclosure behaviour. As a result, they suggested that attitudes would be more accurate predictors of people’s privacy behaviours if they were related to more specific information types. However, these authors do not clarify how disclosure behaviours come about or how they can be affected.

Based on social psychological literature, also Baek (2014) has tried to solve the privacy paradox. An interesting outcome of his study (Study 3) demonstrated that the privacy paradox will be substantially diminished or even solved if participants were exposed to reasonable privacy related arguments, which let them think more about privacy. Although the author has attempted to solve the privacy paradox and makes an argument for why people’s privacy concerns fail to predict privacy protecting behaviour, it does not explain the underlying mechanism that can account for the privacy paradox. Furthermore, this study measured behavioural intentions instead of actual behaviour, which might affect the validation of the findings.

In examining the underlying dynamics of the dichotomy between privacy attitudes and behaviours, Acquisti (2004) argued that it is important to recognize that online users are often willing to trade-off their privacy for convenience and relatively small rewards. The author focused on the individual decision making process with regard to privacy attitudes and behaviour based on theories of behavioural economics and demonstrated that a model of rational privacy behaviour appeared to be unrealistic when people had to make privacy

(16)

sensitive decisions. In particular, he suggests that individuals may not be able to act as rational agents concerning their personal privacy and that models of psychological distortions acknowledged in the behavioural literature offer a more realistic depiction of the dichotomy between privacy attitude and behaviour. Additionally, he argues that incomplete information, bounded rationality and discounting usually affect both naïve and sophisticated individuals when they make decisions in their best interest when it comes to privacy. In line with this, Leiser, Azar, Hadar (2008) acknowledged that a large body of behavioural research and experiments have identified that assumptions of full rationality do not always fit actual economic behaviour due to various biases. As a result, alternative models in the field of neoclassical economics and psychology are developed to understand the social and psychological mechanisms that underlie this irrational, economic behaviour. Yet, until now no unifying framework has been developed that accounts for the inconsistent privacy behaviour online.

Building on Leiser et al.’s thoughts, the present paper enhances the existing privacy paradox research by focusing on the psychological process that might explain why people do not follow a logical or rational course when disclosing personal information online. Recently, cognitive process theories have increased in popularity and I believe that it is interesting to examine whether the privacy paradox could be related to construal level, which is a processing style that has been applied to distinctive concepts in the economic behavioural and psychological literature (Dhar & Kim, 2006). In particular, the present work draws on the Construal Level Theory (CLT), which has risen to the forefront over the past years. Nira Liberman and Yaacov Trope initially introduce CLT and they showed that it is a leading comprehensive process theory that accounts for understanding a wide variety of preference reversals, which cannot be fully explained by other theories (Fujita, Trope, Liberman, Levin-Sagi, 2006; Liberman & Trope, 1989; Nussbaum, Trope, Liberman, 2003). This theory

(17)

focuses on the distinction between individual’s evaluations based on the consequences of psychological distance and related shift in weights of cognitive representation (Dhar & Kim, 2008). Furthermore, Trope, Liberman and Wakslak (2007) reviewed CLT as a foremost contemporary theory of mental construal with rich implication for explaining and predicting consumer behaviour. Therefore, a great deal of interest of this thesis is focused on CLT that accounts for a wide variety of temporal effects on choices by changing people’s mental representations. CLT will be linked to the privacy paradox to deepen our understanding about the psychological mechanisms and to explain the incongruence between people’s privacy intentions and behaviour over time.

2.4 The construal level theory: associations between construal level and

psychological distance

The main premise of CLT is that psychological distance influences the way individuals mentally represent events, which in turn influence people’s decision making process (Liberman, Sagristano, Trope, 2002; Lynch & Zaubermann, 2007). Trope & Liberman (2010) state that psychological distance is about “the perception of when an event occurs, where it occurs, to whom it occurs, and whether it occurs” (p. 442). In this context, it is assumed that we solely experience what is present and if an event is not part of the perceiver’s direct experience it is in some manner psychologically distant. If people imagine or think about a decision in the distant future, implement a third-person perspective, or define an evaluation as hypothetical, people psychologically distance themselves from a task and construe more abstract mental representations (Thomas & Tsai, 2012; Trope & Liberman, 2010). In other words, CLT posits that different dimensions of psychological distance (time, space, social, and hypothetically) exists that have similar effects on individuals’ level of construal, which in turn, influence the way people think and guide their evaluation, choices, and behaviour

(18)

(Bar-Construing people’s privacy information disclosure online – M. Jans (University of Amsterdam) 18

Anan, Liberman, Trope, Algom, 2007; Trope et al, 2003;). Conversely, CLT proposes that general attitudes, core values, and ideologies guide peoples’ decisions for psychologically distant conditions (Liberman, Trope, McCrea, Sherman, 2007). This indication of reverse inference demonstrates that the relationship between level of construal and psychological distance is automatic and bidirectional in that level of construal affects time perspective, and conversely, that time perspective affects people’s construal levels.

Explicating the relationship between construal level and distance in more detail, a wide range of prior studies have proved that similar objects of situations can be mentally represented at distinctive construal levels and people favour the event or information that is consistent with their construal level (Nussbaum et al., 2003). In particular, CLT proposes that construal level is affected by individual and environmental factors and that psychological distance is a primary determinant of construal level (Kim & John, 2008). People can represent situations or products either in concrete lower level terms or in abstract higher level terms dependent on the perceived psychological distance of the situation or products. Subsequently, psychological distance can bring changes in preferences and choices over time due to the alterations in the mental representations of people (Trope & Liberman, 2003). For instance, when one forms attitudes, predictions and behavioural intentions for distant events, he/she usually relies on high-level construals and internalizes increasingly more abstract mental representations (Kim & John, 2008; Leiser et al., 2008). These, high-level construals, tend to be relatively simple, decontextualized representations and include superordinate and central features (Nussbaum et al. 2003; Trope & Liberman, 2003; Trope & Liberman, 2010). Alternatively, for approximate events, people form more concrete mental models, or low-level construals, which consist of subordinate, contextual, peripheral aspects of events (Nussbaum et al., 2003). Consequently, these people focus more on the specific and detailed attributes of the situation when making a decision (Kim & John, 2008). To exemplify: a person thinking

(19)

about a holiday that takes place tomorrow can think about it in terms of superordinate goals, like “packing one’s suitcase”, whereas a person thinking about a holiday a year from now might be construing it in terms of more abstract and subordinate goals, such as “discovering new places” (Förster, Friedman, Liberman, 2004). Thus, concrete low-level construals are more specified and richer but less structured than abstract high-level construals (Liberman et al., 2007; Nussbaum et al., 2003).

In sum, the level of abstractness in the representations of outcomes is subjected to the different dimensions of psychological distance, since these are mentally associated and have similar psychological consequences (Bar-Anan et al., 2007; Förster et al., 2004). In the present work, temporal distance is of particular relevance for the explanation of the incongruence between people’s privacy intentions and actual privacy protecting behaviour and will be further indicated below.

2.5 Temporal distance

Temporal distance is perhaps the single most important determinant of psychological distance, since CLT has initially been developed to explain intertemporal choices (Leiser et al., 2008). Therefore, the primary focus of CLT has been on the psychological consequences of temporal distance from distant events for many years. However, like many other theories that were originally studied in one field, the notion of temporal distance has been generalized to other antecedent of psychological distance and outlined the wide range of implications of CLT (Dhar & Kim, 2007; Lynch & Zaubermann, 2007; Trope & Liberman, 2003). Förster et al. (2004) defined temporal distance as ‘the perceived proximity of an event in time that changes people’s responses to future events by altering their mental representations of those events’ (p.177). Past implications of CLT have demonstrated that temporal distance is directly associated with level of construal (Nussbaum et al., 2003). People use more abstract representations, or higher-level construals to predict behaviour in the more distant future than

(20)

the near future (Trope & Liberman, 2003). Furthermore, Nussbaum et al. found that temporal distance reduces the sensitivity and spontaneous search for contextualized information that is externally provided and they showed that distant future predictions are often based on more global information. In addition, Trope & Liberman (2003) argued that decontextualized global traits, core values and ideologies, appear to be more prominent when predicting behaviour in the distant future because these comprise high-level construals of behaviour. Contrariwise, these authors stated that people give more weight to concrete aspects when they think about their behaviour in near future situations because they rely more heavily on their mental states, such as expectations and feelings that constitute low-level construals. Besides, people are expected to anticipate to the unforeseen contingencies of the context in more proximal situations, whereas in distant future circumstances it is suggested that people express their personality and character. To exemplify this, a person who is perceived to be a humble or shy individual is expected to always behave demurely in distant situations. Nevertheless, in near future situations this person might behave contrarily as a consequence of unpredictable situational factors (Nussbaum et al., 2003).

When combining the effects of temporal distance and construal level to enhance our understanding about the inconsistency between people’s self-disclosure attitudes and behaviour online, it can be reasoned that a time lag occurs between the formation of intentions and the actual self-disclosure behaviour in typical online situations. According to CLT, intertemporal choices might lead to a systematic difference in the structure of one’s cognitive operations (mindset) (Freitas, Gollwitzer, Trope, 2004) and subsequently influence people’s initial evaluations or choices (Trope et al., 2003). To be more specific, if people think about online situations, and thus considering whether or not to engage in self-disclosure behaviour, it is likely that the event is distant from direct experiences and limited reliable information is accessible. Besides, Acquisti & Grossklag (2004) argued that even if people had access to

(21)

complete information, it is unlikely that people are able to estimate the probability distributions of the event and alternative scenarios. Accordingly, people’s intentions or attitudes seem to be affected by bounded rationality that refers to the people’s inability to process all relevant information. Therefore, people tend to rely on more abstract, schematic mental representations and base their self-disclosure intentions on broader prototypical experiences and the likelihood of fulfilling overarching goals. Likewise, Nussbaum et al. (2003) proposed that this reliance on oversimplified representations of distant future situations give rise to a prediction bias because people fail to integrate the diluting effect of contextual factors and misinterpret their own behaviour. In addition, Trope & Liberman (2000) argued that people underrate task completion times and base their predictions of future behaviour on the basis of information and inferences at higher construal levels.

Yet, these general features of self-disclosure online only should hold when the activity is still perceived as distant. As the online self-disclosure situation nears, it becomes more important to actually plan the self-disclosure behaviour and a processing shift towards more low-level construals takes place, according to the principles of CLT. Subsequently, this means that people’s mental representations become more concrete and it is likely that one will focus on the contextualized and incidental aspects of the online disclosure situation, because the individual is directly experiencing the situation. Moreover, this increased concreteness will enhance the value of the immediate outcomes of online self-disclosure, because people tend to prefer the information and inferences that is congruent with their mindset (Kim & John, 2008), such as the direct access to the website or the convenience to communicate directly with others.

Thus, based on the generalized implications of CLT, it is fair to assume that people’s privacy intentions are based on abstract, high-level construals, whereas people’s actual

(22)

self-Construing people’s privacy information disclosure online – M. Jans (University of Amsterdam) 22

disclosure is based on concrete, low-level construals. According to the logic of CLT, this shift in construal level alters attribute weights and might lead to inconsistent choices over time.

2.6 Mental representations and the decision making process

The various alternatives in typical and complex multidimensional decisions can differ in terms of features and attributes on which we rely our decision on. Some are more central to the end value of an action, whereas other features load on the mean of an action (Fujita et al., 2006). Likewise, Liberman & Trope (1998) have made a distinction between feasibility and desirability considerations in the decision making process and applied this to CLT. They suggest that this distinction is related to the variance between the high-level and low-level construals of goal-related activities. Accordingly, feasibility raises the attention to the effort someone has to make to reach the ultimate goal, whereas desirability refers to the value of the ultimate goal. From a CLT perspective, this shift can create temporally inconsistent preferences since people seem to prefer outcomes that are unfeasible, but highly desirable in the distant future. While highly feasible, but less desirable option outcomes appear to be more salient in the near future (Trope & Liberman, 2003; Zhao, Hoeffler, Zaubermann, 2007).

Along similar lines, the action identification theory of Vallacher & Wegner (1989) demonstrated that any action could be illustrated in a cognitive hierarchy, ranging from low-level to high-low-level identities. When an action is internalized at a rather higher low-level, it is proposed that people construe abstract mindsets that refer to the general meaning or the “why” aspects behind the action. Whereas a low-level identification of an action is related to concrete mental constructs, that in turn focus on the logistic procedures or the “how” aspects of an action (Torelli & Kaikati, 2009). In addition, these opposing identifications of an action appear to be related to the desirability versus feasibility of an outcome. Feasibility is related to the subordinate, contextual ‘how’ facets of the action and compromise low-level identities, while desirability is related to the superordinate, and more abstract ‘why’ facets of the action

(23)

that compromise rather high-level identities (Liberman & Trope, 1998). Consistent with the earlier literature reviewed in this paper, the results of the research on Action Identification Theory illustrated that when an individual have the intention to engage in an action, it is likely that he or she naturally adopt higher-level identities by focusing on the larger, overarching (why) purpose of the act. However, when the action’s process become more prominent it is possible that the performance of the activity can be disturbed due to the unexpected increased complexities of the situation (Vallacher & Wegner, 1989). In these situations, one will move to simpler, low-level (how) identities and engage in activities that are of secondary concern that might have no connection to overarching values (Freitas et al., 2004).

Combining these principles of outcome (why) vs. process orientation (how) with the principles of level of mental representations and choice over time, it can be argued that the relative importance of particular types of information and attributes differ in each stage of the decision making process of disclosing personal information online as people prefer the information that is consistent with their mindset (Kim & John, 2008). According to the CLT perspective, it can be expected that people construe more abstract, high-level mental representations when they form self-disclosure intentions. Therefore, it is likely that they base their attitudes on the abstract, desirability aspects and the overarching long-term aims of the activity, such as ‘protecting their privacy’ since it appeared that many people are concerned about their privacy and one’s privacy can be perceived as an abstract concept. In addition, this lead to a lower use of even ignorance of concrete information (Köhler, Breugelmans, Delleart, 2015). However, when the online situation comes physically closer, it is likely that the action’s process becomes more relevant and low-level mental construals are activated. This should lead one to focus on the concrete, feasibility aspects of the self-disclosure situations in a natural way. Hence, this shift in mental processes could explain the incongruence between

(24)

people’s privacy attitudes and behaviour because the way in which an individual construes a situation can determine in part one’s behaviour.

2.7 Construal level and self-control

Multiple CLT studies have shown that the pattern of inconsistent choices, behaviour, or preference over time can be reduced by a numerous means through which construal level can be manipulated (Dhar & Kim, 2007; Liberman & Trope, 1998; Irmak, Wakslak, Trope, 2013). For example, Freitas et al. (2004) demonstrated that manipulating cognitive procedures or mindsets creates the tendency to construe situations in either low- or high-level terms. Priming an abstract mindset led people focus on relevant high-level goals (e.g. self-improvement), while activating a concrete mindset make one experience the same situation in lower level terms (e.g. immediate comfort) (Torelli & Kaikati, 2009). Additionally, Freitas et al. (2004) stated that people prefer the situation that matches with their mindset, because the ease of processing translates into the perceived ease of engaging in the behaviour. Along similar lines, Slepian, Masicampo, Ambady (2015) showed that thinking about the concrete step-by-step process will induce low-level construals and concrete feasibility-focused mental representations play a dominant role. This should lead one to prefer the most comfortable means of action, given that the immediate outcomes, which can be either positive or negative, associated with the action process are most salient. In contrast, an abstract processing style facilitates the pursuit of overarching, long-term aims over short-term gains (Slepian, Masicampo, Ambady, 2015). Hence, the way in which an individual construes a situation can determine in part one’s behaviour.

Furthermore, previous work that examined the conceptualization of self-control has demonstrated that self-control can be promoted as a function of construal level (Fujita et al., 2006). According to Fujita et al., self-control involves making decisions that are in line with one’s primary, long-term goals, which reflect rational preferences instead of

(25)

psychological-Construing people’s privacy information disclosure online – M. Jans (University of Amsterdam) 25

based responses to situations. Fujita et al.’s work examined self-control questions such as how and when someone fails to do what he/she initially wanted to do. Consistent with CLT, these authors have measured shifts in construal level by focusing on temporal difference in the structure of self-control. The authors demonstrated that the activation of high-level construals over low-level construals led one to act in accordance with long-term outcomes that reflected greater self-control. Besides, these findings suggest that increased psychological distance decreases the likelihood of self-control problems. Also, these findings support the assumption that self-control issues can be perceived as conflicts between low-level behavioural and high-level behavioural consequences of an identical situation (Bar-Anan et al., 2007). Therefore, the core essence of self-control is to find a balance between immediate, concrete experiences and long-term, abstract aims (Freitas et al., 2004)

Drawing on previous research that demonstrates that construal level operates at the level of mindset (Trope & Liberman, 2010; Thomas & Tsai, 2011), it makes sense to argue that encouraging an individual to induce an abstract mindset, he/she will construe a near online disclosure situation in terms of relevant high-level goals. It is likely that this led one to focus on the desirability aspects of the behaviour and he/she would seek information about more global characteristics by focusing on the overarching aim of the behaviour (e.g. protecting privacy). Consequently, it is expected that they behave more consistently between the different phases in the decision making process over time since no shift in weight of construal level takes place, which automatically decrease self-control issues. If this is true, then it can be proposed that in typical online situations, people’s mindset directly affects their actual self-disclosure behaviour. Therefore, it will be interesting to test the following hypothesis.

(26)

H1: In typical disclosure situations, people that hold an abstract (high-level) mindset

less likely disclose personal information online compared to people that hold a concrete (low-level) mindset.

2.8 Cost and benefit calculations

It appears that people often perform a cost-benefit analysis or privacy calculus by assessing the potential outcomes they obtain as a consequence of disclosing personal information to others (Culnan & Bies, 2008; Holland, 2008, Smith et al. 2011, Wu et al. 2010). One is prepared to reveal personal, and thus accept the loss of privacy, as long as the benefits outweigh the potential risks of disclosure. On the one hand, the benefits of self-disclosure online, such as financial rewards, personalized services, social adjustment (Motiwalla et al., 2014) are commonly direct and immediately noticeable. On the other hand, the costs of online self-disclosure, such as the secondary usage, loss of control, identity theft, and spam are spread over an unpredictable time period, not directly quantifiable and frequently perceived as vague (Motiwalla et al., 2014, Taddicken, 2013).

Acquisti (2004) further indicated that individuals obtain difficulties in estimating the pay-offs. Particularly the long-term consequences are hard to predict, as these are examined or ascertained solely through actual experience. This makes it debatable how likely it is that someone is able to correctly monitor all the values associated with the payoffs, since it is already difficult enough to acquire, memorize, and process all the available information that is relevant to the decision-making process. Given that people have limited foresight perspectives, it is unclear who and how many people are included in the transaction context and it seems difficult to estimate how long the information is available or searchable as a result of the perceived temporal and spatial distance from direct experience (Taddicken, 2013). Moreover, it is easier to deal with effects and actions that are in the near future and it appears that people

(27)

have the tendency to rely on simple heuristics instead of cognitive values when making decisions in the context of transactions in privacy-sensitive situations (Acquisti, 2004).

These findings can be explained by psychological distortions, such as hyperbolic discounting, self-control issues, and under insurance, which make it more complicated to perform a rational cost-benefit analysis (Acquisti, 2004). These distortions negatively affect one’s ability to make concrete decisions that chase overarching, future desires to protect their privacy (Holland, 2008). Additionally, it can be argued that psychological distortions can account for the tendency people have to trade-off their identifiable information in exchange for immediate rewards or non-monetary values, despite the privacy concerns they have (Acquisti, 2004; Acquisti & Grossklag, 2005; Park, Campbell, Kwan, 2012).

According to the theoretical principles of CLT, this form of immediate gratification can be explained by the temporal changes in the value of the situation. The perceived value of an event is originated from its construal and, therefore, it is suggested that the value of a low-level aspect of an event is different from the value of its high-low-level aspects (Liberman et al., 2002). In particular, temporal distance systematically change the level of representation of the situation, which result in a corresponding change in one’s preference towards the value associated with the high-level construal of a choice (high-level value) and the value associated with low-level construal of a choice (low-level value). Consequently, the phenomenon time discounting takes place if the subjective value of a large distant reward is delayed over time and, therefore, people have the tendency to prefer immediate rewards and care less about the future consequences (Frederick, Loewenstein, O’Donoghue, 2002; Klapproth, 2008). For example, Leiser et al. (2008) stated that if people are asked to choose between option A: receive $20 directly, or option B: receive $50 in one year, the majority of people have a preference for option A. In response to this example, Klapproth (2008) argued that such

(28)

myopic behaviour can be explained by people’s inclination to place more value on short-term than on long-term desires when they decide to act.

Converging evidence suggests that the level of abstractness of mental representation influence the perceived value and relevance of future distant outcomes and that people prefer the outcome that is congruent with their mindset. Prior CLT research have shown that desirable central, goal related features received more weight than feasible secondary, goal-irrelevant features in the more distant future (Liberman & Trope, 1998; Trope & Liberman, 2000) and that payoffs are more salient than probabilities in choosing bets (Leiser et al., 2008; Liberman et al., 2002). Linking these findings to the privacy paradox, it can be argued that, if the prior proposition about the direct relationship between people’s mindset and self-disclosure behaviour holds, people induced to construe an online self-disclosure situation in higher-level, abstract terms have in general a higher degree of future orientations and exhibit more thoughtful behaviour. Subsequently, it is more likely that they exert self-control, act in congruence with their initial intentions and will give more weight tot the long-term consequences because they prefer the higher-level, long-term value and outcomes that match their abstract mindset. In contradiction, it is likely that people who strongly devalue future outcomes and discount the consequences that could occur in the distant future hold concrete, low-level construals and have short time horizons. Those people appear to be short-sighted in their actual disclosure behaviour and seem to have a systematic bias to overvalue the present over the future. Therefore, it can be expected that people that hold a concrete mindset prefer the most comfortable means of action and base their self-disclosure behaviour decisions on the immediate, low-level values that matches with their mindset.

To deepen our understanding about the underlying mechanism that influence people’s online self-disclosure behaviour, it will be interesting to further explore the moderating role of people’s reliance on the costs and benefits of revealing personal information online by

(29)

extending these assumptions Therefore, the present research will explore whether changes in the perceived temporal distance of costs and benefits affects people’s self-disclosure behaviour and the following proposition will be tested:

H2: the perceived distance of the costs and benefits of online self-disclosure

moderates the effect of people’s mindset on their actual self-disclosure behaviour such that participants rely on the consequences that are congruent with their mindset.

(30)

3. Conceptual model

The aim of this thesis study is to enhance our understanding of the privacy paradox, which is a common element that is identified in the findings of various privacy studies. Although preliminary interpretations have illustrated privacy as a complex concept and focused on related concerns and predictions in privacy-protecting behaviour, no previous study has adopted a theoretical explanation that explicitly underlies the privacy paradox. To address this gap, the present thesis will shed more light on CLT as an overarching theory that can account for the privacy paradox. To be more specific, this thesis is part of a greater comprehensive study that aims to increase our understanding of the privacy paradox by explaining the underlying psychological mechanism. Within the framework of CLT (Liberman and Trope, 1998, Liberman et al., 2007), it has been suggested that construal level (abstract vs. concrete) influences people’s judgments and decisions. Therefore, the emphasis of the present thesis studies is on specific implications of CLT to offer additional evidence to the assumption that CLT is the underlying framework that clarifies the existence of the privacy paradox. In particular, this thesis explores the relation between people’s mental representations and their privacy behaviour in typical disclosure situations as well as the moderated role of temporal distance of perceived costs and benefits. The hypothesized relationships between the variables are illustrated in the conceptual model below (figure 1). To further clarify this model, people’s ‘mindset’ condition (abstract vs. concrete vs. control) is the independent variable and people’s ‘online self-disclosure behaviour’ is the dependent variable. Additionally, the ‘temporal distance of the perceived costs and benefits of disclosing personal information’ moderates the relationship between these two variables.

This thesis incorporates two singular studies that examine how people construe their privacy concerns and act up on respectively based on well-established approaches used by studies on temporal construal manipulation (see, e.g. Liberman and Trope, 1998, Liberman et

(31)

Construing people’s privacy information disclosure online – M. Jans (University of Amsterdam) 31 H1

al., 2007, Zhao et al., 2007, Thomas & Tsai, 2011). In the following section the distinctive studies are described in more detail. For both studies, it is first outlined how the independent and dependent variables are operationalized. Second, the sampling description is highlighted, followed by a clear description of the studies and procedure of the research. Next, the analysis of the data and outcomes will be drawn in the next chapter.

Temporal distance of costs and benefits

Mindset Self- disclosure

behaviour

H2

(32)

4. STUDY 1

4.1 Methodology

4.1.1. Research strategy and design

The purpose of this study is to investigate what role level of construal may play on online self-disclosure behaviour. This thesis is based on positivism research psychology (Saunders & Lewis, 2012), which is characterized by the law of cause and effect to predict people’s self-disclosure behaviour based on a specific theory, namely CLT. Accordingly, an experimental 3 (abstract mindset vs. concrete mindset vs. control) x 3 (acceptation cookies vs. registration vs. participation) between-subjects design was chosen to be appropriate for testing the previously described causalities and effects between people’s mindset level and their actual online disclosure behaviour. Compared to descriptive or exploratory designs, Field (2005) argued that experimental designs allow for some level of control over independent variables and to test the effects of it by comparing the variety in outcomes. This explanation of causal relationships between variables and the use of a structured method to enable replications are key characteristics of deduction, which refers to a ‘top down’ approach (Saunders & Lewis, 2010). To enhance the experimental and quality control participants were directed through various systematic steps, which are described in more detail in the following sections.

4.1.2. Sample

In November 2015, potential respondents were reached randomly via Facebook and mail with a link to the experimental online questionnaire. In total 191 participants opened the link of which 125 completed it. Nevertheless, an error was found in the abstract mindset condition and the data of 15 participants was unusable and deleted. Consequently, in December 2015 an additional survey was distributed in which participant were solely assigned to the abstract

(33)

mindset condition to overcome the prior error and to generate a representative total sample. From the 24 participants that took part in the second survey, 19 participants completed the survey. Eventually, a total sample of 117 participants was left (41.9% male and 58.1% female) with an age ranged from 19 to 60 years (75.2% was aged between 19 and 25). The average age was 26 with a standard deviation of 8.119. Participants were randomly assigned to one of the three manipulation groups (Group 1: abstract; Group 2: concrete; Group 3: control).

4.1.3 Stimuli and measures

In this section, I shed light on the different variables that are measured in this study. When available, relevant measures from other studies were adapted and, when necessary, the wording and order were changed to better fit into the present thesis study.

Mindset inductions

People’s mindset was the hypothesized independent variable in this thesis study and refers to people’s own cognitive operations (Freitas et al., 2004). As stated before, the manipulation techniques used in this thesis are built on the study of Freitas et al. (2004) that provided strong support for the role of construal level in the decision making process. In particular, Freitas et al. demonstrated that it is possible to manipulate participants’ mindset to construe situations either at low or high levels. Comparable to experiment 1 and 2 conducted by Freitas et al., participants in the present study were randomly assigned to complete either the abstract/high-level (Group 1) concrete/low-abstract/high-level (Group 2) or control (Group 3) version of the mindset manipulation treatment. Participants who were assigned to Group 1, were presented a passage that directed participants’ attention to why they do things the way they do it. Subsequently, they were instructed to use the passage as a guideline to list four ways why they would engage in an activity (saving money) that is related to one’s ultimate goal. Participants allocated to Group 2 read a comparably short passage and the same activity was used. Contrarily, this description focused participants’ attention on the process of how we do things, designed to

(34)

provoke more concrete thought. After reading the passage, all the participants in this treatment condition were faced with the question “How do you save money?”. Identical with the abstract condition, they need to fill out four successive ways how they would go about this activity. Accordingly, the activity ‘saving money’ was chosen such that the abstract and the concrete description would differ in abstractness solely, without conveying different information about the activity in question. Participants assigned to the control group, were not induced to attain a certain mindset at all and were directly guided to the next experimental task.

Manipulation check

After the allocation to one of the three mindset treatments, participants were guided to the manipulation check based on the study Vallacher & Wegner (1989). In this section participants in all conditions completed an adapted version of Vallacher & Wegner’s (1989) Behavioural Identification Form (BIF), which is initially designed to measure how abstractly or concretely one is representing an action. Originally, the BIF included 25 activities (e.g. locking a door) each followed by two alternative descriptions of this activity that represent the action more abstractly (securing the house) or concretely (putting a key in a lock). In addition, in the present study participants were presented a forced-choice questionnaire that contained 10 activities adapted from the original BIF, followed by a high-level or a low-level identification. These identifications correspond either to the “why” (high-level) aspect of the activity reflecting the value of a goal, or to the “how” (low-level) aspect of the activity, reflecting the process of achieving the goal. The participants were asked to choose one of the two identifications that best described the activity. Accordingly, the manipulation check proved to be effective if the adapted BIF in the presented study demonstrated that the manipulation systematically primed the tendency to construe the unrelated activities at either

(35)

a high- or low-level construal. Furthermore, the order of the two options was counterbalanced, to ensure that it did not affect the choices of the participants.

Actual online self-disclosure behaviour

In contrast to privacy attitudes and concerns, comparatively limited studies have been conducted to test actual privacy disclosure behaviour (Joinson et al. 2010; Keith et al. 2013; Knijnenburg et al., 2013). Therefore, this thesis study was set up to provide participants with realistic privacy decisions in an online context situation. A mock-up website was specifically designed for this research to measure the dependent variable, participants’ actual self-disclosure behaviour in privacy sensitive online choice situations. During their online journey on this website, participants were confronted to disclose personal information in three different ways that were as realistic as possible. The three situations incorporated both the positive and negative consequences of self-disclosure and are described below in more detail.

1. Accepting cookies

At first, people were confronted with a pop up screen that contained a cookie notification. Cookies are sophisticated tracing-technologies that are initially designed to improve users’ interactions with the Web. However, often they need to be accepted otherwise the online user is not able to access a website or view specific information, such as images or videos. If they are activated, user specific details such as registration information, preferences, transactions, and downloading could be stored. In the present experimental task, the cookie contained the following information: ‘This site would like to use cookies. Although you will still be able to

use this site when you reject cookies, accepting cookies will make the site run faster and smoother and will enhance your general user experience. However, please note that accepting cookies allow us to track all your onsite behaviour and that your personal

Construing people’s privacy information disclosure online