Construal level theory as a possible additional explanation for the privacy

(1)

Construal Level Theory as a possible

additional explanation for the privacy

paradox

Master Thesis

Student: Charlotte van den Belt

Student number: 11093447

Amsterdam, June 24

th

2016

Supervisor: J. Demmers

Program: MSc. Business Administration (Marketing track)

(2)

Construal Level Theory as a possible explanation for the privacy paradox C. van den Belt, University of Amsterdam

2

Statement of Originality

This document is written by Charlotte van den Belt 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)

3

Abstract

In today’s ‘Internet Age’, consumers’ concerns about their privacy online are growing (Brandimarte, Acquisti & Loewenstein, 2013). However, their behavior is not in line with their concerns, as people often disclose their personal information online (Norberg, Horne & Horne, 2007). This ‘privacy paradox’ will be explained in this paper based on the Construal Level Theory (CLT); a theory that explains how psychological distance can influence a persons’ behavior and thoughts (Liberman & Trope, 1998; Trope, Liberman & Wakslak, 2007). Two studies have been conducted to investigate if there is initial support for the CLT as an additional explanation for the privacy paradox. Study 1 investigates the main effect and study 2 the underlying mechanisms. The first study investigated if mindset influences three typical online decision situations (accepting or rejecting cookies, registering or not registering and accepting or rejecting a phonetrack option). The results demonstrated that people in an abstract mindset reject more cookies and more phonetrack options than people in a control mindset. There is no significant influence of mindset on the option to register or not. Furthermore, the results showed that in general, people in an abstract mindset are more willing to share personal information online than people in a concrete mindset. The second study took the consequence salience into account. This study found evidence for the hypothesis that people in an abstract mindset generate more cons and less pros than people in a concrete mindset.

(4)

4

Table of Content

1. Introduction 6

2. Theoretical Framework 9

2.1 Privacy 9

2.2 Consumers’ privacy behavior and the privacy paradox 10

2.3 Construal Level Theory 13

2.4 CLT and the privacy paradox 15

2.5 Conceptual Framework 18

3. Study 1 19

3.1 Method 19

3.1.1 Participants 19

3.1.2 Design 20

3.1.3 Stimuli and measurements 20

3.1.4 Procedure 21 3.2 Results 22 3.2.1. Hypotheses testing 22 3.3 Discussion 26 4. Study 2 28 4.1 Method 28 4.1.1 Participants 28 4.1.2 Design 28

4.1.3 Stimuli and measurements 29

4.1.4 Procedure 30

4.2 Results 31

4.2.1. Manipulation test 32

4.2.2. Hypotheses testing 32

(5)

5

5. General discussion 38

5.1 Findings 38

5.2 Limitations and directions for future research 41

6. References 44

7. Appendices 51

7.1 Survey study 1 51

(6)

6

1. Introduction

In the past twenty years, the internet has grown exponentially and has become more important (Cheung, Lee & Rabjohn, 2008). As a result of this growth, consumers’ concerns about privacy are growing in today’s ‘Internet Age’ (Graeff & Harmon, 2002; Brandimarte, Acquisti & Loewenstein, 2013). Online activities which used to be private or only shared with a few, are now easily revealed to commercial entities and the government and used by them. This often happens without consumers being aware of the fact that they are revealing their personal information (Acquisti, Brandimarte & Loewenstein, 2015). However, consumers do not try to control their personal information more carefully (Culnan & Armstrong, 1999; Acquisti & Grossklags, 2005).

Consumers are concerned about their privacy and ignorant about what will happen with the collected data (Graeff & Harmon, 2002; Brandimarte et al., 2013; Peltier, Milne & Phelps, 2009). Research even showed that online privacy concerns are one of the biggest fears in this ‘age of information’ (Acquisti & Grossklags, 2003). Nevertheless, consumers often disclose their personal information online. This contradiction is presented as the privacy paradox (Norberg, Horne & Horne, 2007; Taddicken, 2014). Until now, several factors on explaining the existence of the privacy paradox have been studied. First, consumers may not even realize that they share their personal information and what consequences are related to this information disclosure, as this process often happens invisible. Moreover, consumers outweigh the benefits and costs and base their decision on this trade-off (Acquisti, Brandimarte & Loewenstein, 2015; Culnan & Armstrong, 1999). On top of that, the present-bias, which causes people to overvalue these benefits due to hyperbolic temporal discounting, can be seen as a possible explanation for the existence of this paradox (Acquisti & Grossklags, 2003). Research by Acquisti and

(7)

7 Grossklags (2005) assumes that the decision making process regarding privacy is both a rational and an irrational process.

Additional research towards the drivers of this paradox is necessary to increase the understanding of the privacy paradox. Therefore, the Construal Level Theory is used. This theory describes how psychological distance dimensions (temporal, hypothetical, spatial and social and distance) have an influence on the representation of events (Trope, Liberman & Wakslak, 2007). It is important to research how this theory can be linked with the privacy paradox.

The CLT predicts a connection between psychological distance and mindset. This has not been researched before and, therefore, this research will try to close this gap in literature by investigating the following question; ‘How does mindset affect the level of sharing personal information online?’. Answering this research question will clarify if, and possibly how, the CLT can offer an additional explanation for the privacy paradox.

This research will contribute to existing literature in various ways. At first, studying the CLT as a possible additional explanation for the existence of the privacy paradox would be an added value to the current literature in better understanding the contradictions between consumers’ privacy concerns and their behavior. Furthermore, as personal customer data are important for marketers, being able to receive more personal consumer information could personalize their relationships with their consumers. Moreover, understanding how consumers think about their privacy concerns and act on this in this contemporary competitive marketplace is important for organizations (Jenkinson, 2006). Conversely, policy makers want to make sure that consumers understand the risks of disclosing their personal information and are willing to protect this.

(8)

8 This research will start with the theoretical framework, including the hypotheses and the conceptual framework. Then, the two conducted studies are discussed. Each study will explain and discuss the method, results and discussion. This is followed by the general discussion including the findings and limitations that provide interesting opportunities for future research.

(9)

9

2. Theoretical Framework

In this chapter a comprehensive review of the relevant literature for this research will be discussed in order to understand what is already known about privacy, consumers’ privacy behavior, the privacy paradox and the Construal Level Theory. Then, an explanation why the CLT has potential to be a possible explanation for the privacy paradox will be given by proposing two hypotheses. Finally, the conceptual framework is shown.

2.1 Privacy

Many attempts have been made to define privacy (Paine, Reips, Stieg, Joinson & Buchanan,

2007; Goodwin, 1991; Nowak & Phelps, 1995). Privacy in a legal context can be defined as the right to be let alone (Warren & Brandeis, 1980). Moreover, Westin (2003) defines information privacy as “the claim of an individual to determine what information about himself or herself should be known to others” (Westin, 2003, p.431). Mothersbaugh, Foxx and Wang (2012) defined privacy concerns of consumers as “consumers’ concerns about the use of their revealed information for marketing purposes”. Consumers may see online privacy as a basic right (Goodwin, 1991). However, it is difficult for scientists to create a unified definition of privacy. This could possibly be explained due to the fact that technology created new and unique privacy issues (Paine et al., 2007).

Consumers are mostly concerned about the question how organizations are collecting their personal information and how they use this information. It is unclear for consumers which information companies possess about them and for which purposes it will be used (Paine et al., 2007; Phelps, Nowak & Ferrell, 2000). Due to computer and database technology advances, marketers have better opportunities to use consumers’ personal information. This personal information can be used to develop and improve a personalized buyer-seller relationship

(10)

10 (Peltier, Milne & Phelps, 2009; Miceli, Ricotta & Costabile, 2007). Personalization of consumers’ information can be perceived as both positive and negative by consumers. Consumers may be interested in the more personalized advertisement content that organizations can offer after collecting consumers’ personal information (Lambrecht & Tucker, 2013). Marketers, on the one hand, could think that consumers want advertisements that are as relevant for them as possible (Turow, King, Hoofnagle, Bleakley & Hennessy, 2009). Consumers, on the other hand, may see this personalization as violation of their privacy (Stone, 2010).

Research by Goldfarb and Tucker (2011) and Hinduja and Patchin (2008) showed that varying age groups respond differently to privacy concerns online. Younger people are more willing to reveal and share information online compared to elderly people. This gap between younger and older people is growing. There are several explanations for this gap in difference between age groups, such as the amount of experience with information technology, the awareness of potential concerns related to online privacy and preferences for privacy change between different age groups. Interestingly, research by Taddicken (2014) demonstrated that age hardly has an impact on the level of sharing personal information online.

2.2 Consumers’ privacy behavior and the privacy paradox

Privacy notices provide consumers information about the risks of disclosing their personal information online. These notices inform consumers and, therefore, help them in deciding whether to disclose their personal information to marketers or even if they want to engage with the website at all (Milne & Culnan, 2004). These privacy notices try to promote the choice for consumers and reduce the risk of sharing this personal information online (Jensen & Potts, 2004; Milne & Culnan, 2004). However, these privacy notices are not read by every consumer.

(11)

11 There are several explanations why consumers do not read this privacy related information. When consumers have experience (online or offline) with an organization, a larger chance exists that they do not read the privacy notice due to the fact that they are already familiar with the organization. In addition, for some consumers the fact that a privacy notice exists, is already enough reason to trust the organization regarding their privacy (Milne & Culnan, 2004). Moreover, consumers may not take enough time to read the privacy notices (Pan & Zinkhan, 2006). Furthermore, research demonstrated that consumers who are more concerned about their privacy, are more likely to read this information, as they have more negative thoughts about sharing their personal information. This also shows that less concerned consumers are less likely to take the time to read these notices (Milne & Culnan, 2004). Research by Youn (2005) demonstrated that concerned consumers discover other ways to avoid privacy concerns, such as going to another website or providing the website with false, wrong or incomplete information. Interestingly, consumers who are aware of the risks of sharing their personal information often still share this information if they receive rewards in return. Rewards may involve discounts or free products and services (Hann, Hui, Lee & Png, 2002; Chorppath & Alpcan, 2013).

Nowadays, consumers are increasingly aware of the fact that disclosing personal information not only brings benefits, but involves risk as well (Zimmer, Arsal, Al-Marzouq, Moore & Grover, 2010). Consumers behavior is changing by, for example, enrolling in no-call lists or by not accepting information requests. As a result, they are ensured that their vulnerability to violations of their (online) privacy will decline (Peltier, Milne & Phelps, 2009). However, research indicated that consumers behave irrationally when they engage in a cost-benefit trade-off and have to make decisions regarding their privacy (Keith, Thompson, Hale, Lowry & Greer, 2013). Acquisti (2004) and Acquisti and Grossklags (2005) argue in their

(12)

12 research that it is not possible for an individual to estimate all the costs and benefits at all times. This can be explained by incomplete available information to individuals, an individual’s bounded rationality and at last, by deviation from utility maximization. This final argument shows that, when individuals would have all the available information and would be able to process this information, they may still behave irrational and against better judgments.

Although consumers’ privacy concerns are increasing, there are several studies that discovered an inconsistency between people’s disclosure behavior and their privacy concerns (Norberg, Horne & Horne, 2007; Taddicken, 2014; Acquisti & Grossklags, 2005; Wilson & Valacich, 2012). This discrepancy between privacy concerns and online disclosure behavior describes the privacy paradox. Son and Kim (2008) argue that the main response of individuals to protect their privacy is refusing to share personal information. However, according to Larose and Rifon (2007), although online privacy is a concern, it is questionable if it really influences individuals’ privacy protection behavior. Utz and Krämer (2009) state that it is possible that the privacy paradox arises, as it is important for individuals to find a trade-off between protection of their privacy and impression management. According to Utz and Krämer (2009), the trade-off between benefits and costs is one of the most important aspects of the privacy calculus. Awad and Krishan (2006) argue that the privacy paradox behavior can be explained either by the limited capabilities individuals’ possess to access and process information regarding decision making, or due to different privacy perspectives individuals possess.

The behavior of consumers regarding privacy concerns includes objective and subjective behavior. According to Dinev and Hart (2006), trust is one of the main antecedents of privacy concerns. Consumers value trust in situations where interpersonal exchange takes place (Metzger, 2004). Furthermore, Metzger (2004) argues that consumers, once they have disclosed their information before, are willing to disclose it again. Moreover, Metzger (2004) assumes

(13)

13 that information asymmetry is a source of privacy concerns. Consumers do not know what information companies posses and for what purpose they will use this information (Acquisti, Brandimarte & Loewenstein, 2015).

Research by Acquisti and Gross (2006) only found a small link between individuals’ privacy concerns and their behavior related to social media; even if these individuals hold privacy concerns, they often join social media networks and share their personal information. This is illustrated by Acquisti and Grossklags (2005), who discovered that only a minimum amount of social media users change their standard privacy settings in order to minimize the visibility of their social media profiles. Moreover, Taddicken (2014) found only a small association between people’s privacy concerns and the amount of personal information disclosure on social media sites such as Facebook.

2.3 Construal Level Theory

The CLT presents itself as a dominant explanation for preference reversals in the near versus future distance by explaining how psychological distance can influence a persons’ behavior and thoughts (Liberman & Trope, 1998; Trope, Liberman & Wakslak, 2007). The theory uses two important concepts; the construal level and psychological distance. Construals can be divided in high and low-level construals. When an event, object or person is far away from the present (so, the construal shows a distant event) the construal is abstract. In this case a high-level construal is defined. High-level construals think in the way of a bigger picture. When an event, object or person is less far from the present (so, the construal shows a near event) the construal is concrete. This is defined as a low-level construal, which is more focused on details and thought processes(Trope & Liberman 2010; Trope, Liberman & Wakslak, 2007).

For example, an event such as children playing with a ball can be seen from different perspectives.A low-level construal might include more details such as the size and color of the

(14)

14 ball or the age of the children and are defined as concrete construals. A high-level construal might see the activity of children playing with a ball as “having fun” and is therefore a more abstract construal (Trope, Liberman & Wakslak, 2007).

Another dimension of the CLT is desirability and feasibility concerns in making decisions. Taking into account the desirability of an object, this is more focused on the end results itself and is a high-level construal. Contrary, feasibility is more focused on how to reach the end results (Liberman & Trope, 1998). This is a low-level construal. In sum, the ‘why are you performing the activity’ question is related to the high-level construal and an abstract mindset. Additionally, the ‘how are you performing the activity’ question is related to the low-level construal and a concrete mindset (Wakslak, 2012; Trope, Liberman & Wakslak, 2007; Trope & Liberman, 2010).

The other important concept of the CLT is the psychological distance. The four different dimensions of psychological distance are temporal distance, hypothetical distance, spatial distance and social distance (Trope & Liberman, 2003; Trope, Liberman & Wakslak, 2007). The first dimension is temporal distance, referring to distance in time. Events in the distant future are mentally represented as abstract construals. These events in the future, for example events occurring next year, are presented as high-level construals. On the other hand, events in the near future, for example events happening tonight, are concrete and low-level construals (Trope, Liberman & Wakslak, 2007). Hypothetical distance describes the likelihood of an event occurring. When an event is in the near future, the occurring of an event is more likely and therefore, this event will be linked with a low-level construal. At the opposite, an event in the distant future is less likely to occur and therefore, this event will be linked with a high-level construal (Wakslak, Trope, Liberman & Alony, 2006). Spatial distance refers to the physical distance between events (Trope, Liberman & Wakslak, 2007). When an event is located close

(15)

15 to a person, this is seen as more concrete and on a low-level. Likewise, when an event is located far away, this is seen as more abstract and on a high-level (Henderson, Fujita, Trope & Liberman, 2006). Social distance is the last discussed dimension, which refers to the measure of space between groups or individuals. This dimension involves the level of distance people feel towards other groups or individuals. It shows how (dis)similar they feel towards others and if they are able to compare themselves with other groups or individuals. If people feel similar to others, this is perceived as a close distance and a low-level construal. At the opposite, when people feel dissimilar, this is described as a far distance and therefore, as a high-level construal (Matthews & Matlock, 2015; Trope, Liberman & Wakslak, 2007).

2.4 CLT and the privacy paradox

An important assumption of the CLT is that construals that an individual uses to mentally represent events are dependent of perceived psychological distance (Trope & Liberman, 2010). Research by Kivets & Tyler (2007) found that a person’s goal or mindset determines whether a concrete or abstract representation predicts the behavior towards the event or object, in case this person holds concrete and abstract representations at the same time.

Hagel and Rayport (1997) argue in their research that individuals are only willing to share their personal information if they can profit from information disclosure. Moreover, Culnan and Armstrong (1999) state that individuals will only share their personal information when the benefits outweigh costs. Thus, the benefits and costs of sharing personal information online are important for individuals. When people make privacy decisions in online day-to-day situations, the benefits are often certain and immediate, while the costs are uncertain and delayed. Thus, there are often short-term benefits versus long-term costs. For example, receiving an immediate discount when subscribing for a newsletter or directly entering a website when accepting to disclose personal information are short-term benefits. In these situations, it is unclear whether

(16)

16 the company will use the consumers’ personal data, and if this will be in the consumers’ best interest. Consumers will not immediately experience these costs (Norberg, Horne & Horne, 2007; Acquisti, Brandimarte & Loewenstein, 2015). The expectation is that the benefits of disclosing personal information will be perceived as psychologically closer, and thus be construed in a more concrete way and at a lower level. Meanwhile, the costs of sharing personal information online will be perceived as psychologically more distant, and hence be construed at a more abstract and higher level. Immediate gratification from disclosure can trump the consequences in the long-term (Acquisti, Brandimarte & Loewenstein, 2015).

It is important to take the congruence between psychological distance and mindset into account. The temporal and hypothetical distance dimensions are the most relevant psychological distance dimensions in this research, as these dimensions refer to the distance in time and the likelihood of an event occurring (Trope, Liberman & Wakslak, 2007; Matthews & Matlock, 2015). Trope and Liberman (2010) defined psychological distance as “a subjective experience that something is close or far away from the self, here and now”. When consumers are exposed to online disclosure situations, they often have concrete goals in their mind. For example, they want to find information or buy something online. This decision in behavior situations is direct, thus has a short-term focus. People in a concrete mindset focus more on the short-term present (Trope, Liberman & Wakslak, 2007). The suggestion is that an individual’s mindset is in congruence with the benefits of sharing information online. Along these lines, the hypothesis is that people in a concrete mindset are more willing to disclose personal information online than people in an abstract mindset.

In online concern situations, the psychological distance to the real decision is distant. This is related to the abstract mindset. People in this mindset focus more on the long-term future (Trope, Liberman & Wakslak, 2007). They will focus on the long-term costs, instead of the

(17)

17 short-term benefits in the concrete mindset. Therefore, people in an abstract mindset will be less willing to share this personal information online, as their mindset is in congruence with psychological distance of the costs of sharing this information and not in congruence with the benefits of disclosing information.

Acquisti (2004) discusses immediate gratification in his research. Immediate gratification assumes that people consider their current preferences as more important than their future preferences. This can be linked to this research. For example, people with an abstract mindset consider an online advertisement different than people in a concrete mindset. In an abstract mindset, they focus more on the costs of this advertisement. They can

experience the advertisement as annoying and think of the negative effects these

advertisements can have for them. So, they experience more cons than pros in this situation. In a concrete mindset, people see an advertisement and think more of the benefits. They see interesting information in the advertisement and want to benefit from this advertisement due to their current preferences. They consider the costs of this advertisement less than the benefits. These people experience more pros than cons.

As a result, the following hypotheses are composed:

Hypothesis 1: People with an abstract mindset are less willing to share their personal information online than people with a concrete mindset.

Hypothesis 2A: In an abstract mindset, people generate more cons than people in a concrete mindset.

Hypothesis 2B: In an abstract mindset, people generate less pros than people in a concrete mindset.

(18)

18

2.5 Conceptual Framework

Based on the developed hypotheses, the following conceptual framework has been designed:

H1

H2A H2B Concrete/Abstract

mindset information online Sharing personal

Consequence salience

(19)

19

3. Study 1

3.1 Method

In this chapter, the first study of this thesis will be discussed. This thesis conducted two studies, to indicate the main effect and the underlying mechanism. To investigate how mindset affects the level of sharing personal information online, data for two different studies have been collected. Study 1 investigates the main effect of the influence of mindset on the willingness to disclose information online in a real life setting and study 2 investigates the underlying mechanism. Study 1 will be discussed in this chapter and study 2 in the following chapter. This chapter will first discuss the participants of study 1. This is followed by the design and thereafter, the stimuli and measures. Lastly, the procedure is discussed.

3.1.1 Participants

In total, 149 participants filled in the experimental survey completely and were used for analysis. The amount of reached potential participants is not clear, due to the distribution of the experimental survey by social media (Facebook and LinkedIn), e-mail and snowball sampling (Saunders & Lewis, 2012). The average age of all participants in this study is 26.13 years old (SD=8.00). 37.4% (N=55) of the participants are male, and 62.6% (N=94) are female. Most participants are highly educated due to the fact that 93.3% completed their Bachelor or Master Degree (N=139). To be certain that participants’ demographic characteristics would not have an influence on the results of this research, the test controlled for the variables age and education.

(20)

20

3.1.2 Design

To test the first hypothesis of this study, an experimental survey was used. The primary aim of this study was to investigate if mindset affects the level of disclosing information online. This was tested for accepting or rejecting cookies, registering online and accepting or rejecting the phonetrack option. The independent variable of this study is mindset and the dependent variable is sharing personal information online. This research used a 3 (abstract, concrete, control mindset) by 3 (cookies, registering, phonetrack) between-subjects design to test hypothesis 1.

3.1.3 Stimuli and measurements

Mindset. The mindset manipulation, used before by Freitas, Gollwitzer and Trope (2004) and Fujita, Trope, Liberman and Levin-Sagi (2006), was included in this study. Participants were randomly assigned, either to the abstract or concrete mindset condition to manipulate their mindset, or to the control group, in which they were not manipulated at all. First, participants in the concrete and abstract mindset condition were asked to read some information about ‘how’ and ‘why’ people do certain things in their lives. Previous literature found that it is possible to relate ‘how’ questions to low-level construals and ‘why’ questions to high-level construals (Vallacher & Wegner, 1989). Therefore, people in the abstract mindset condition were asked ended ‘why’ questions and people in the concrete mindset condition were asked open-ended ‘how’ questions regarding money saving. Participants in the abstract mindset condition were asked ‘why’ they would save money. Their answer to this question was merged in the following question, where they were asked again ‘why’ they would do this. Then, their answer was again merged in the third and fourth ‘why’ question. For participants in the concrete mindset condition, the ‘why’ questions were replaced by ‘how’ questions. In the control mindset, participants did not receive additional questions and were directly redirected to the next part of the survey.

(21)

21 Disclosing information online. All participants were assigned to three typical online situations regarding disclosing information online. First, they had the option to reject or accept cookies which were shown on a cookie notification screen. By accepting cookies, people could directly experience the benefits of entering the website and they would not directly experience the costs of possible information sharing with others. Accepting the cookies would be the low-level disclosure option and rejecting the cookies would be the high-low-level option. Then, participants were asked if they wanted to register themselves. They could choose to register with their social media account or to develop a user name with a password. The main benefit of accepting to register was the ability to directly enter the website. The other possibility was to reject the option to register. Accepting the option to register would be the low-level disclosure option and rejecting the option to register the high-level disclosure option due to the immediate benefits versus the delayed costs. Finally, participants could choose to accept or reject the phonetrack option that popped up in their screen. They were asked to fill in their phone number, with as a results the possibility to win a tablet. Accepting this phonetrack option would be the low-level disclosure option and rejecting this option the high-level disclosure option, as accepting this option includes possible short-term benefits versus long-term costs.

3.1.4. Procedure

All participants were asked to take part in the online survey. They received a link which redirected them to the online survey, created in the program Qualtrics, to be able to start with the survey (Appendix I). First, a small introduction about the survey was given to explain to all participants the average amount of time this survey would take and explained that there are no right or wrong answers possible. Furthermore, it was emphasized that all responses would be handled confidentially. At last, all participants were thanked for participating in this research.

(22)

22 In the first part, several demographic questions were asked regarding age, gender and the level of education. After that, participants were randomly assigned to three different conditions; concrete mindset condition, abstract mindset condition and control mindset condition. Next, participants were redirected to a website that was created to appear similar to a real life setting. On this website, all participants were exposed to the same online disclosure situations. First, a cookie notification popped up asking participants if they wanted to accept or reject the cookies. Second, they were asked to register themselves. They had two possible options; to register themselves or to reject this option. They could either register by connecting with their social media account or by developing a user name and password. After that, a screen popped up giving participants the option to win a tablet if they filled in their phone number. At the end of the survey, participants were again thanked for their participation.

3.2 Results

3.2.1 Hypotheses testing

In this part, the hypothesis of study 1 is tested. SPSS was used to execute the statistical tests and to look at possible associations in the developed hypotheses. A significance level of p<.05 was used, which means that hypotheses with a p-value equal to or higher than .05 were rejected. Table 1, 2 and 3 show the acceptance rates of respectively the cookies, registering and the phonetrack option.

To test hypothesis 1, a logistic regression is performed. This test can be used to test the categorical independent variable mindset with the binary dependent variables cookies, registering and phonetrack. Table 4 shows the reference category. Table 5 provides the results of the different mindsets in relation towards cookies, table 6 towards registering behavior and table 7 towards the phonetrack option. First, a significant difference is found between people

(23)

23 in the abstract mindset in comparison with people in the control mindset towards accepting and rejecting cookies; B=-1.62, p=.009. However, there is no significant difference between people in the concrete mindset in comparison to people in the control mindset; B=.03, p=.972. Furthermore, the control variable age (p=.570) is not significantly associated to accepting or rejecting cookies. The control variable education shows a marginally significant association between the level of education and the decision to accept or reject cookies (p=.058). Second, the difference between people in the abstract mindset compared to people in the control mindset for the option to register or not to register did not reach significance; B=.583, p=.183. Likewise, there are no significant differences found for people in the concrete mindset in comparison to the control mindset; B=.644, p=.159. The control variable age (p=.119) and education (p=.669) are neither significantly associated to people their registering behavior. Finally, a logistic regression was used to look at mindset in comparison to the phonetrack option. The results show a significant difference between people in the abstract mindset in comparison to people in the control mindset; B=-2.82, p=.009. There is no significant difference between people in the concrete mindset compared to people in the control mindset; B=-0.36, p=.486. The results show for the control variable age (p=.182) and education (p=.191) no significant difference.

Table 1: Acceptance rates cookies Cookies Total Reject Accept Mindset Abstract 14 36 50 Concrete 3 46 49 Control 4 46 50 Total 21 128 149

(24)

24 Table 2: Acceptance rates registering

Registering

Total Not registered Registered

Mindset Abstract 13 37 50

Concrete 11 38 49

Control 17 33 50

Total 41 108 149

Table 3: Acceptance rates phonetrack Phonetrack Total Reject Accept Mindset Abstract 48 1 49 Concrete 40 9 49 Control 39 12 51 Total 127 22 149

Table 4: Categorical variables codings Frequency Parameter coding (1) (2) Mindset Abstract 50 1,000 ,000 Concrete 49 ,000 1,000 Control 50 ,000 ,000

Table 5: Logistic regression; cookies

B S.E. Wald df Sig. Exp(B)

Step 1a Mindset 10,088 2 ,006 Mindset(1) -1,618 ,621 6,783 1 ,009 ,198 Mindset(2) ,028 ,806 ,001 1 ,972 1,029 Age -,016 ,028 ,322 1 ,570 ,984 Education -,858 ,452 3,607 1 ,058 ,424 Constant 5,928 1,997 8,816 1 ,003 375,548

(25)

25 Table 6: Logistic regression; registering

Step 1a Mindset 2,648 2 ,266 Mindset(1) ,583 ,438 1,777 1 ,183 1,792 Mindset(2) ,644 ,457 1,987 1 ,159 1,903 Age -,031 ,020 2,425 1 ,119 ,969 Education -,108 ,254 ,183 1 ,669 ,897 Constant 1,767 1,081 2,672 1 ,102 5,852

Table 7: Logistic regression; phonetrack

Step 1a Mindset 6,928 2 ,031 Mindset(1) -2,822 1,075 6,898 1 ,009 ,059 Mindset(2) -,362 ,520 ,485 1 ,486 ,696 Age ,034 ,026 1,777 1 ,182 1,035 Education -,385 ,295 1,708 1 ,191 ,680 Constant -,852 1,267 ,452 1 ,502 ,427

To gain further insight into hypothesis 1, a one-way ANOVA has been performed to examine the effect of the independent variable mindset on the dependent variable sharing personal information online. The dependent variable sharing personal information online is a sum score of the cookies, registering and phonetrack online disclosure situations. In case of accepting the cookies, choosing to register and accepting the phonetrack, 1 point was given for every action. If a participant did not accept the cookies, choose not to register and did not accept the phonetrack, 0 point were assigned. So, this resulted in a sum score for every participant with a minimum of 0 points and a maximum of 3 points. First, the assumptions of the ANOVA were tested. The results of the Levene’s test showed that the assumption of homogeneity of variance

(26)

26 (p=.393) had not been violated regarding the mindset manipulation on the sum score of sharing information online. The results of the one-way ANOVA show a statistically effect of mindset on sharing; F(2,152)=3.41, p=.036. Between the concrete and abstract mindset, Tukey’s post hoc exposed a significant difference; p=.033. However, Tukey’s post hoc shows no significant difference between the control and concrete mindset (p=.758) and between the control and abstract mindset (p=.165).

3.3 Discussion

The primary aim of this study was to analyze the influence of mindset on sharing personal information online. The hypothesis stated that people in a concrete mindset would be more willing to disclose personal information online than people in an abstract mindset, due to the congruence between an individuals’ mindset and their psychological distance. Due to this congruency, people will rely more on the benefits and therefore, will be more inclined to share personal information online, even though they are concerned about their privacy. If it is possible to prove the congruency between an individuals’ mindset and the psychological distance, the CLT would show initial support for a possible additional explanation of the privacy paradox. However, further research would be necessary to research this.

Contrary to the expectations, the results did not completely support the first hypothesis. The outcomes of the logistic regression revealed partially significant differences between an abstract mindset and cookies as well as between an abstract mindset and the phonetrack option. This was in line with the expectations. Contrary to the hypothesis, the different mindsets had no influence on the decision whether to register or not. The control variables age and education did not show a significant influence on the level of sharing information online. There was only a marginally significant relation between education and the option to accept or reject cookies.

(27)

27 The results of the second test were in line with the hypothesis, meaning that people in an abstract mindset are less willing to share personal information online than people in a concrete mindset.

It can be concluded that in general, people in an abstract mindset are less willing to share information online than people in a concrete mindset. However, looking more specific at the three online decision situations, this conclusion is not applicable for people in an abstract and concrete mindset regarding the decision to register or not and for people in a concrete mindset regarding the cookies and phonetrack decision. Therefore, hypothesis 1 is not completely supported. This study can partially prove that privacy decisions online can be dependent of a persons’ mindset. However, the outcomes of study 1 provide no complete evidence for the hypothesis that people in a concrete mindset, whose mindset is in congruence with construal level of benefits, are more willing to share personal information online than people in an abstract mindset, whose mindset is in congruence with construal level of costs.

(28)

28

4. Study 2

4.1 Method

In this chapter, the second study of this thesis will be discussed. First, the participants of study 2 will be discussed. This is followed by the design. Then, the stimuli and measurements are discussed. This chapter ends with the procedure.

4.1.1 Participants

In total 153 participants filled in the experimental survey completely and were used for analysis. For this study the amount of potential participants who were reached is unclear, due to the use of social media, e-mail and snowball sampling (Saunders & Lewis, 2012). The average age of the in total 153 participants is 34.51 years old (SD=14.95). 26.1% (N=40) of the participants were male, and 73.9% (N=113) female. Most participants were highly educated due to the fact that 90.8% completed their Bachelor or Master Degree (N=139). The test controlled for the variables age and education.

4.1.2. Design

To test the second hypothesis of this study, an experimental survey is used. The primary aim of this study is to investigate if mindset has an influence on the amount of pros and cons people generate. The independent variable of this research is mindset and the dependent variable is the inclination to share information online. The mediator is the consequence salience. To test hypothesis 2, a between-subjects design is used.

(29)

29

4.1.3. Stimuli and measurements

Mindset. This research uses the mindset manipulation, previously used by Freitas, Gollwitzer and Trope (2004) and Fujita et al. (2006). Participants were randomly assigned to three different conditions; the abstract mindset condition, the concrete mindset condition or the control group condition. In the control group condition, participants were not manipulated. These participants were asked to answer several questions. First, an open-ended question was asked about the first thing a participant thinks of when thinking of working on keeping a healthy lifestyle. Then, participants had to rate on a 5-point Likert scale, from “very unimportant” till “very important”, how important working on keeping a healthy lifestyle is for them. Finally, they were asked to mention one advantage and one disadvantage of working on keeping a healthy lifestyle by answering two open-ended questions. In the concrete and abstract mindset condition, participants were manipulated in these mindsets. First, participants were asked to read some information about ‘how’ and ‘why’ people do certain things in their lives. According to Vallacher and Wegner (1987), it is possible to relate ‘how’ questions to low-level construals. In addition, it is possible to relate ‘why’ questions to high-level construals. Therefore, participants in the concrete mindset condition were asked open-ended ‘how’ questions regarding working on keeping a healthy lifestyle. They were asked ‘how’ they would work on keeping a healthy lifestyle. The answer to this question was merged in the second question, were they asked ‘how’ they would do this. Again, their answer was merged in the third and fourth question. In the abstract mindset condition, participants received four open-ended ‘why’ questions about working on keeping a healthy lifestyle.

Inclination to share information online. Participants were asked to indicate the possibility that they would choose to connect online with one of their social media accounts on a 5-point Likert from “very likely” till “very unlikely” (Sparks & Browning, 2011).

(30)

30 Before this question, participants were asked about potential consequences of connecting with one of their social media accounts by an open-ended question with a minimum of one consequence and a maximum of ten consequences. Hereafter, they were asked to rate their consequences on a 7-point Likert scale with “very large con” and “very large pro” as anchors.

Manipulation. This research included a manipulation check by using the Behavior Identification Form (BIF) as a measure of construal level. Participants were presented with ten activities and asked to match these activities with low-level (how) and high-level (why) alternatives (Liberman & Trope, 1998). These ten activities were used before by Vallacher and Wegner (1987). The questionnaire in their research originally contained 25 activities, but this research reduced the amount of activities to ten activities. For example, one of the activities is ‘picking an apple’ (Vallacher & Wegner, 1989). For this activity, the expectation is that people in the abstract mindset would choose the high-level alternative ‘pulling an apple of a branch’ and people in the concrete mindset the low-level alternative ‘getting something to eat’. Thus, this manipulation check could show whether participants evaluated these activities as more concrete or abstract, which then showed whether participants were holding a more concrete or abstract mindset.

4.1.4 Procedure

This experimental survey was conducted online. All participants received a link which redirected them to this survey, created in the program Qualtrics (Appendix II). First, the participants were requested to read a short introduction which gave them information about the average amount of time necessary to fill in the entire survey. Furthermore, they explained to the participants that no right or wrong answers are possible. It was emphasized that all responses

(31)

31 would be handled confidentially. At the end of the introduction, all participants were thanked for participating in this research.

This survey started with demographic questions regarding age, gender and level of education. In the second part, participants were randomly assigned to three different conditions; concrete mindset condition, abstract mindset condition and the control group condition.

The third part focuses on the consequence salience, which demonstrates how salient the positive and negative consequences are. It focuses on the consequences people experience, and if these consequences are positive (pros) or negative (cons). The hypothesis was that people in the concrete mindset condition would experience more pros and people in the abstract mindset condition more cons. This is tested by asking participants to indicate potential consequences of connecting with one of their social media accounts by an open-ended question with a minimum of one consequence and a maximum of ten consequences. Then, they were requested to rate these consequences, as a results that it would be clear if participants experience the consequences more positive or negative. After that, participants were asked to indicate how likely it is that they would connect online with their social media account to be able to test their level of inclination to share information online.

In the final part of this research, participants were exposed to the manipulation check. At the end of the survey, participants were again thanked for their participation.

4.2 Results

The hypothesis of study 2 is tested in this chapter. To execute statistical tests and to look at possible associations in the developed hypotheses, SPSS was used. A significance level of p<.05 was used, meaning that hypotheses with a p-value equal to or higher than .05 were rejected.

(32)

32

4.2.1 Manipulation test

The Behavioral Identification Form (BIF) was included in this study to test the effectiveness of the manipulation. To test this manipulation check, a one-way ANOVA was used. First, the BIF data had to be recoded to assign to all the higher level alternatives a 1 and all the lower level alternatives a 0. Thereafter, a sum score of all the higher level options of the ten BIF questions was developed. This resulted in the following outcome; F(2,150)=5.57, p=.005. This shows a significant difference between the abstract and concrete mindset in relation to the BIF score, meaning that people in the abstract mindset have a significant higher BIF score than people in the concrete mindset. This shows that the manipulation check was effective. The means of the BIF score for the abstract, concrete and control mindset are respectively; 7.04; 5.41 and 6.28.

4.2.2 Hypotheses testing

To test hypothesis 2, a one-way ANOVA was performed to examine the effect of the independent variable mindset on the dependent variable inclination to share information online. Reverse coding of the variable inclination to share information online was necessary to enable that a higher value indicates a higher likelihood to share. First, the assumptions of the ANOVA were tested. The results of the Levene’s test indicated that the assumptions of homogeneity of variance (p=.155) had not been violated regarding the mindset manipulation on the inclination to share information online. The results of this test show a statistically effect of mindset on inclination to share information online; F(2,150)=5.43, p=.005. Tukey’s post hoc exposed a significant difference between the concrete and abstract mindset; p=.004. Furthermore, the results of this test show no significant difference between the control and concrete mindset (p=.121) and between the control and abstract mindset (p=.371).

To further test hypothesis 2, a mediation analysis is performed using the PROCESS Macro by Hayes (2012). The number of bootstrap samples used in this research is 5000 instead

(33)

33 of the default of 1000, due to recommendation by Hayes (Hayes, 2012). Moreover, 95% confidence intervals were used.

First, the effect of mindset on consequence salience is tested. The mediator consequence salience is the total score per participant of their rating of the potential consequences. Participants rated their consequences on a 7-point Likert scale from “very large con” till “very large pro”. This was recoded in a score from a -3 score (“very large con”) till a +3 score (“very large pro”) to create the new variable for consequence salience. A negative total score indicates that cons are more salient than pros, while a positive score indicates that pros are more salient than cons. The results show a negative significant difference between the concrete and abstract mindset in relation to the consequence salience, as the confidence interval is completely below zero; [95% C.I. : -4.73;-.06]. Between the concrete and the control mindset is a negative significant difference in relation to the consequence salience as well; [95% C.I. : -4.91;-.22]. When adding the control variable age, the results do not indicate significance due to the including zero in the confidence interval; [95% C.I. : -.07;.06]. For the control variable education, the results do not indicate significance either; [95% C.I. : -1.08;1.28].

Second, the effect of the consequence salience on inclination to share information online is tested. The results indicate a positive significant effect; [95% C.I. : .10;.17]. When adding the control variables age and education, the results show a negative significant effect for age; [95% C.I. : -.05;-.02]. However, the results do not indicate significance for education; [95% C.I. -.05;.41].

Finally, the effect of mindset on inclination to share information online is tested. The difference between the concrete and abstract mindset in relation to inclination to share information online shows a negative significant effect; [95% C.I. : -1.51;-.39]. The difference between the concrete and control mindset in relation to inclination to share information online

(34)

34 is not significant; [95% C.I. : -1.10;.02]. The results indicate a negative significant effect for the control variable age; [95% C.I. : -.05;-.02]. For the control variable education the results do not indicate a significant effect; [95% C.I. : -.09;.48].

Table 9 shows the statistics of the indirect effect. The comparison between the concrete and abstract mindset has a negative significant indirect effect on inclination to share information online; [95% C.I. : -.66;-.01]. The comparison between the concrete and control mindset has a negative significant indirect effect on inclination to share information online; [95% C.I. : -.70;-.04]. Thus, the indirect effect can provide support for mediation. Moreover, in table 10, the results demonstrate a negative significant direct effect on the comparison between the concrete and abstract mindset in relation to the inclination to share information online; [95% C.I. : -1.09;-.16]. There is no direct significant effect for the comparison between the concrete and control mindset on inclination to share information online; [95% C.I. : -.66;.28]. Table 11 shows the total effects of mindset on inclination to share information online. There is a negative total significant effect of the comparison between the concrete and abstract mindset on inclination to share information online; [95% C.I. : -1.50;-.39]. No total significant effect is found for the comparison between the concrete and control mindset on inclination to share information online; [95% C.I. : -1.10;.02].

(35)

35 Table 8: Overview outcomes Process

Consequent

M(salience) Y(inclination)

Antecedent Coeff. SE p Coeff. SE p

X(D1*) a -2.392 1.182 .045 c’ -.623 .236 .000 X(D2**) a -2.565 1.186 .032 c’ -.189 .237 .001 M --- --- --- b .136 .016 .000 Constant i1 -.412 2.330 .860 i2 4.032 .458 .000 R² =.039 R² =.457 F(4,148)=1.447 p=.222 F=(5,147)=23.921 p=.000

*D1=Comparison between concrete and abstract mindset **D2=Comparison between concrete and control mindset

Table 9: Statistics of indirect effect of mindset on inclination to share information online

BC 95 % CI

B SE Lower Upper

Indirect effect of comparison concrete and

abstract mindset on inclination to share -.33 .17 -.66 -.01

Indirect effect of comparison concrete and

control mindset on inclination to share -.35 .17 -.70 -.04

Table 10: Statistics of direct effect of mindset on inclination to share information online

BC 95 % CI

B SE p Lower Upper

Direct effect of comparison concrete and

abstract mindset on inclination to share -.62 .24 .009 -1.09 -.16

Direct effect of comparison concrete and

control mindset on inclination to share -.19 .24 .428 -.66 .28

Table 11: Statistics of total effect of mindset on inclination to share information online

BC 95 % CI

B SE p Lower Upper

Total effect of comparison concrete and abstract

mindset on inclination to share -.95 .28 .001 -1.50 -.39

Total effect of comparison concrete and control

(36)

36

4.3 Discussion

The main objective of this study was to investigate the underlying mechanisms by investigating the relationship between the different mindsets and inclination to share information online, with consequence salience as a mediator. The hypothesis was that people in an abstract mindset would generate more cons than people in a concrete mindset. Furthermore, people in an abstract mindset would generate less pros than people in a concrete mindset. This can be explained due to immediate gratification, indicating that people consider current preferences as more important than distant preferences (Acquisti, 2004).

The results of the first test showed that, in general, people in a concrete mindset have a higher inclination to share information online than people in an abstract mindset. This is in line with the hypothesis. The consequence salience was included in the second test, to further test this hypothesis. Results demonstrated that people in a concrete mindset experience more pros and less cons than people in an abstract mindset. This result is in line with the hypothesis. Moreover, consequence salience and inclination to share information online show a positive relation, meaning that people who experience more pros than cons have a higher inclination to share information online. Finally, people in an abstract mindset show a lower level of inclination to share information online than people in a concrete mindset. The results provide evidence for the indirect effect. As a result, there is evidence for hypothesis 2; it can be concluded that people in an abstract mindset generate more cons and less pros than people in a concrete mindset.

Another interesting finding is related to the difference between age groups. Age is significant in the relationship between the consequence salience and inclination to share information online and in the relation between mindset and inclination to share information online. This first result means that younger people who experience more pros and less cons,

(37)

37 have a higher inclination to share information online than older people. The second result shows that younger people in a concrete mindset have a higher inclination to share information than older people in this mindset. This is in line with research by Goldfarb and Tucker (2011) and by Hinduja and Patchin (2008). However, age is not significant in the relation between the different mindsets and consequence salience. These results can be related to research by Taddicken (2004), saying that age hardly impacts the level of sharing personal information online.

(38)

38

5. General discussion

5.1 Findings

The primary aim of this research was to explain the discrepancy between people’s online disclosure behavior and their online privacy concerns using the Construal Level Theory. Two studies were performed to investigate this paradox. The hypothesis of the first study was that people in an abstract mindset would be less willing to share personal information online than people in a concrete mindset due to the congruence between mindset and psychological distance. This study tested the main effect of the relation between mindset and sharing personal information online in a real life setting. The second study investigated the underlying mechanisms. It included consequence salience as a mediator, to test the hypothesis that people in an abstract mindset generate more cons and less pros than people in a concrete mindset.

The first study found that people in an abstract mindset were less willing to accept cookies and the phonetrack option than people in the control mindset. However, people in the concrete mindset were not more willing to accept cookies and the phonetrack option than people in the control mindset. Both mindsets showed no significant relationship with the online option to register or not to register. The total effect of mindset on sharing information online showed a significant association, meaning that in general, people in an abstract mindset are less willing to share personal information online than people in a concrete mindset. This study made a relevant contribution to the existing literature by investigating various online disclosure decision situations in a real life setting.

Taking into account all findings from the first study, there can be partially concluded that consumers act on their mindset when making decisions towards sharing personal information online. There is partially evidence for the hypothesis that people in a concrete mindset, whose mindset is congruent with construal level of benefits, are more willing to

(39)

39 disclose personal information online than people in an abstract mindset, whose mindset is congruent with construal level of costs. These findings align with the expectation that people perceive benefits of disclosing personal information online as psychologically closer than costs of disclosing this information.

The second study took the consequence salience into account. The findings from this study provided support for mediation. Therefore, this study confirmed the hypothesis that people in an abstract mindset would generate more cons and less pros than people in a concrete mindset due to immediate gratification (Acquisti, 2004).

After conducting these two studies, we can answer our research question ‘How does mindset affect the level of sharing personal information online?’. This research is the first in investigating the privacy paradox by using the CLT. Therefore, this research can contribute to literature from both the privacy paradox and the CLT. It is the first research which analyzes the differences between an abstract and a concrete mindset in relation to the inconsistency between people’s privacy concerns and their amount of personal information disclosure online. The results of this research provide partially evidence for the assumption that mindset affects personal information disclosure online. In general, people in a concrete mindset are more willing to share information online than people in an abstract mindset due to the congruence between mindset and psychological distance. Furthermore, this study shows that people in a concrete mindset generate more cons and less pros than people in an abstract mindset. To conclude, this research demonstrates initial support for the CLT as possible additional explanation of the privacy paradox.

The CLT presents itself as a dominant explanation for preference reversals by explaining the influence of psychological distance on a persons’ behavior and thoughts (Liberman & Trope, 1998; Trope, Liberman & Wakslak, 2007). Preferences reversals are

(40)

40 important, as the goal of advertisements and campaigns is to change consumers’ preferences. Changes in consumers’ preferences are related to psychological distance dimensions. Some options are more feasible, while others are more desirable (Fiedler, 2007). This can be linked to the choices consumers make online regarding their personal information disclosure. When making decisions online, consumers often take into account the temporal and hypothetical distance. The distance in time, the likelihood of an event occurring and their preferences are important when making the decision to disclose personal information online.

Prior research identified various possible explanations for the existence of the privacy paradox, including the unawareness of consumers about their information sharing, the trade-off consumers make between benefits and costs, and the existence of the present-bias (Acquisti, Brandimarte & Loewenstein 2015; Culnan & Armstrong, 1999; Acquisti & Grossklags, 2003). This research assumes that when consumers make a trade-off in an abstract mindset, they are more focused on the distant and uncertain costs, while in a concrete mindset they experience more close and immediate benefits. The present-bias refers to people’s tendency to overvalue current benefits compared to delayed costs when considering a trade-off. The findings of this research argue that the existence of the privacy paradox may be explained due to the fact that an individual’s mindset affects the decision to disclose personal information online. An individual’s mindset takes into account the benefits and costs of disclosing personal information online. Thus, the present-bias explanation can be partially linked to this research.

Research by Culnan and Armstrong (1999) showed that people make a trade-off between cost and benefits. Therefore, their decision to disclose personal information is a rational decision. Acquisti & Grossklags (2005) disagree, by assuming that the process of disclosing personal information is influenced by rational and irrational factors. This current research explains the privacy paradox mostly in line with the vision of Acquisti and Grossklags