Cultural differences in the privacy-paradox on social networking sites : a research on online privacy self-disclosure behaviour of German and Dutch Facebook users.

Cultural Differences in the Privacy-Paradox on Social Networking Sites

Submitted by: Vanessa Schanzmann, 10460381 Submission date: 26th of February 2014

Version: Final

Qualification: Msc. In Business Studies – International Management Institution: University of Amsterdam

First supervisor: Mr Joris Demmers Second reader: Ms Meg Lee Hsin-Hsuan

Acknowledgements

I would like to take the opportunity to thank everyone who supported me conducting this research. Particularly I would like to thank Mr Joris Demmers for his expertise and for giving me professional direction. I appreciate his guidance and support.

Further, I would like to thank Ms Anna-Katharina Strauch for proofreading the German questionnaire as well as the final thesis. Hereby, I also would like to thank Ms Renée Rikhof for translating the questionnaire into the Dutch language.

Next, I would like to thank all participants who took part in the pilot study and who provided me with extensive and helpful feedback. By saying this, I would also like to express my special thanks to everyone who shared my survey and who took part in my research.

Following, I would like to thank all my friends and other students I could share my (research) experiences with.

I would also like to express my gratitude to my father, who supported me during my whole academic career. Lastly, I would like to thank my mother for her strength and principles.

Cultural Differences in the Privacy-Paradox on Social Networking Sites

By Vanessa Schanzmann

26th of February 2014 / 23rd of March 2014 This thesis is submitted in partial fulfilment of the requirements of the Masters in International Management (Business Studies) at the University of Amsterdam 2013/2014.

Declaration

I declare that this is entirely my work based on my own personal study and research.

_____________________________ Vanessa Schanzmann

Table of Contents

Netherlands………….……… Self-Disclosure……….………..……. Privacy……….………..……. . Privacy Concerns………...……. Privacy-Paradox……….………..…... Privacy Calculus……….………... Concept Development and Hypotheses………..

13 13 15 18 19 21 23 24 26 Methodology……….……….. . Sampling Frame………... Survey Design………... Measurements………. Data collection……… Missing data……… Analysis……….………. . 38 38 39 40 42 43 43 Results……….……… . Reliability... . Normality……… Correlations……… . Hypotheses……….……… 46 46 47 47 48 53

. Additional Analysis…...……….. Discussion……… Contribution……….………... Limitations……….………. Implications……….……… … Future Research……….……….. Conclusion……….………. . 57 64 66 67 69 70 Bibliography Appendices LXXI LXXXI

List of Tables

2. Demographic Comparison NRW and the Netherlands……… 17

3. Sample Demographics……….. 39

4. Sample Demographics of non-Facebook users……… 39

5. Data Collection Offline……… 43

6. Variables Conceptual Model 1 & 2……….. 45

7. Variables Conceptual Model 3, 4 & 5……….. 45

8. Sample Country Scores……… 48

9. Results Conceptual Model 1……… 49

10. Results Conceptual Model 2……….. 50

11. Tests of Between-Subjects Effects INV………. 51

12. Tests of Between-Subjects Effects MAS..………. 52

13. Tests of Between-Subjects Effects UAI....………. 53

14. Model Summary………. 55

15. Self-Disclosure and Privacy Concerns………... 56 16. Correlation Matrix………... CII 17. Exploratory Factor Analysis – Structure Matrix……… CIII 18. Correlation Matrix – Including new factors………..…………. CIV

19. FTC Information Principles Summary...………... CV 20. OECD Guidelines Summary...………... CV 21. EU Directive 95/96 EC Summary……….. CVI

List of Figures

Figure 1 Types of information provided………... 18

Figure 2 Conceptual Model 1 – Power Distance...……….... 29

Figure 3 Conceptual Model 2 – Long-Term-Orientation……….. 30

Figure 4 Conceptual Model 3 – Individualism……….. 32

Figure 5 Conceptual Model 4 – Masculinity...……….. 33

Figure 6 Conceptual Model 5 – Uncertainty Avoidance.……….. 34

Abbreviations

CI Confidence Interval DV Dependent variable

E.g. exemplī grātiā / for example EU European Union

FTC Federal Trade Commission IM Instant messengers

IDV Independent variable INV Individualism / Collectivism

IPPR Internet privacy-protective responses IUIOP Internet Users’ Information Privacy Concern IVR Indulgence / Restraint

LTO Long-term orientation MAS Masculinity / Femininity MON Monumentalism

NRW North-Rhine-Westphalia

OECD Organization for Economic Cooperation and Development PDI Power Distance

SC Social contract

SNS(s) Social Networking Site(s) TRA Theory of reasoned action U&G Uses and Gratification UAI Uncertainty Avoidance UK United Kingdom

US(A) United States of America VSM Value Survey Module

Abstract

Public debates on privacy concerns increased dramatically in times of the Internet (Westin, 2003). The so-called privacy-paradox is about the phenomenon that people who show high privacy concerns are yet willing to disclose personal information. In the present study evidence for the privacy-paradox could be found.

Amongst other, it is argued that privacy and self-disclosure on SNSs are cultural phenomena (De Mooij & Hofstede, 2002; Zhao & Jiang, 2011). Previous, discordant results about culture on privacy concerns and self-disclosure, led to the overall research question: What is the

effect of cultural values on the privacy-paradox of active users of social networking sites?

Through an online and offline questionnaire in Germany and the Netherlands, it was tested whether the five Hofstede dimensions, Power Distance (PDI), Long-Term-Orientation (LTO), Individualism (INV), Masculinity (MAS) and Uncertainty Avoidance (UAI), affect the privacy-paradox. Privacy concerns were hypothesized to mediate the relationship between PDI/LTO and self-disclosure respectively. Neither the effect of the dimensions on privacy concerns nor the effect on self-disclosure could be supported. Further, INV, MAS and UAI were tested to have a moderating effect on the relationship between privacy concerns and

self-disclosure. The moderating effect could not be supported in none of the three models either. Moreover, in this setting, there was no relationship found between privacy concerns and self-disclosure.

Because of unreliable measurements of culture, it was not entirely possible to determine cultural differences in the privacy-paradox on SNSs. However, it could be seen that Dutch Facebook users tend to self-disclose more information than German users, whereas privacy concerns are equal amongst both nationalities. Those differences in self-disclosure can possibly be attributed to other gratifications among German and Dutch Facebook users. Overall, the present study highlights the complexity of online privacy issues and self-disclosure.

Introduction

“Due to the fact that the Internet is a global phenomenon, online privacy concerns also become global.” (Wu, Huang, Yen, & Popova, 2012). Westin (2003) states that public debates on privacy concerns increased dramatically in times of the Internet. IMB (1999) researched privacy concerns among American, British and German Internet users and found 79% of the people to be either moderately or highly concerned.

Recently especially Social Networking Sites (SNSs) caught the attention of researchers and academics (Ellison, 2007). Ellison (2007) argues that millions of users have integrated SNSs into their daily practices. In the same light, individuals assert their privacy claims daily (Westin, 2003). Individuals constantly seek an intrapsychic balance between privacy, disclosure and communication needs (Westin, 2003).

Generally, academic literature about online privacy concerns among SNS users is limited, as most research on these concerns is conducted in the field of e-commerce (Zhang, Wang, & Xu, 2011). Zhang et al. (2011) state that SNS are more complex than e-commerce settings, as social media is about interactions similar to social interactions offline. E-commerce is simply

about transactions between companies and their customers. On SNSs, however, many individuals and companies come together at the same time.

Thelwall (2008) and Ellison (2007) describe SNSs as Web server, where users can register, create personal profiles and communicate with selected others. SNSs offer several facilities such as blogs, photo albums, video uploads, and commentaries (Ellison, 2007; Thelwall, 2008). The purposes of SNSs are mainly: maintaining (long-distance) relationships (Tosun, 2012; Utz & Krämer, 2009), self-presentation (Utz & Krämer, 2009), seeking friends, social support, information, convenience (Kim, Sohn, & Choi, 2010), entertainment (Kim et al., 2010; Tosun, 2012), photo-related activities, organisation of social activities, passive observations, building new friendships, and initiating/terminating romantic relationships (Tosun, 2012). Companies use SNSs to target potential clients and to customize advertisements based on demographic information, such as age, location, gender, education, labour and interests (Ellison, 2007; Facebook, 2013).

The Pew Research Center (2013) identified Facebook as the most used SNS (67%). By February 2010, Facebook was available in 70 languages (Vasalou, Joinson, & Courvoisier, 2010). In 2012, Facebook had 1.06 billion monthly active users, which is an increase of 25% compared to 2011 (Facebook, 2013). Facebook’s mission is to connect people and to make the world more open (Facebook, 2013).

Govani & Pashley (2005) roughly identified two categories of reasons for Facebook usage. The first category involves joining Facebook as a result of friend recommendations and peer pressure. The second category relates to the usefulness of Facebook (e.g. meeting new people, keeping in touch). Joinson (2008:1029) identified what facilities users associate with Facebook in the first place, descending: keeping in touch, watching people virtually, re-acquiring lost contacts, communication, photographs, ease of use, perceptual contact (e.g.

checking statuses on a daily basis), and making new contacts. Joinson (2008) argues that those motives for Facebook usage determine the extent to which users disclose information.

Spiliotopoulos & Oakley (2013) and Veltri & Elgarah (2009) state that motives for Facebook use vary substantially across cultures. E.g. Kim et al. (2010) found that the major motives (seeking friends, social support, entertainment, information and convenience) are similar between Korean and US-American SNS users, however the weights placed on these motives are different (Kim et al., 2010). Vasalou et al. (2010) conducted a survey among 423 Facebook users in five countries. They found that experience with the SNS and culture effect a user’s motivation for using Facebook; e.g. users from Greece, Italy and the United Kingdom (UK) rate groups more than US users. Italian users further rate games and applications more than US users. Users from Greece and France find status updates less important. French users also find photographs less important and visited the SNS less frequently (Vasalou et al., 2010).

Next to motives for joining SNS, the desire for online awareness has a broad cross-cultural appeal (De Mooij & Hofstede, 2002; Lowry, Cao & Everard, 2011; Zhao & Jiang, 2011). E.g. Zhao & Jiang (2011) analysed SNS profiles of American and Chinese users. They found that Chinese users are more likely to customize their profile pictures/images than American users. On the contrary, Veltri & Elgarah (2009) found that individualistic US-Americans tend to express themselves more openly and provide more details on SNSs than e.g. Moroccan users, to be able to stand out of the crowd.

Cultural values also influence people’s desired levels of privacy (Lowry et al., 2011). E.g. Kaya & Weber (2003) found that American and Turkish students differed significantly in their desired privacy levels. Turkish students had lower needs for privacy. This is in line with De Mooij & Hofstede (2002) who argue that privacy is a typical issue in individualistic

cultures; the USA is considered to be individualistic, whereas Turkey is an example of a collectivistic country (Hofstede, n.d).

Lowry et al. (2011) researched whether the initial five Hofstede dimensions of culture influence both, privacy concerns and the desire for online awareness. The authors however could not support the influence of each dimension on privacy concerns and online awareness in their complex model, which can be attributed to the complex nature of culture as a construct (Yoo, Donthu & Lenartowicz, 2011). There are, however, indications that privacy concerns and self-disclosure have a cross-cultural appeal.

Previous, discordant results about cultural values on privacy concerns and self-disclosure, lead to the overall research question: What is the effect of cultural values on the

privacy-paradox of active users of social networking sites? This question will be answered through

surveys in Germany and the Netherlands1 by identifying differences in privacy concerns and self-disclosure behaviours of active Facebook users. De Mooij & Hofstede (2002) argue that there is large variation in the use of the Internet in Europe, despite economical homogeneity. Therefore a comparison between two neighbouring European countries can be of interest. E.g. Germany and the Netherlands are economically homogenous (DNHK, 2014a), but they have different cultural origins; which makes them interesting to investigate (continental vs. sea power) (Hofstede & Hofstede, 2008).

The offline questionnaires will be collected in North-Rhine-Westphalia (NRW) and the Netherlands. NRW is the most populous state in Germany and is seen as being representative for Germany (Staatskanzlei des Landes Nordrhein-Westfalen, 2013b).

The aim of this study is to investigate whether Hofstede’s dimensions of culture explain privacy concerns and self-disclosure behaviour on SNSs.

1 _{Germany is mentioned before the Netherlands, because of the alphabetic order. For consistency purposes Germans are mentioned before}

Overall, there are mixed results revealed in academic literature on e.g. who is sharing information with whom and/or what information is shared under which condition. Given the right circumstances, users often disclose personal information despite having online privacy concerns. The present study will shed light into this so-called privacy paradox.

In the following an overview of Hofstede’s cultural dimensions as well as differences between Germany and the Netherlands are given. Following, literature on self-disclosure, privacy, privacy concerns, the privacy-paradox and the privacy calculus is presented.

Literature Review

Hofstede’s Dimensions of Culture

One of the most famous researchers about cultural differences is Geert Hofstede (Lowery et al., 2011; Yoo et al., 2011). Hofstede collected more than 116,000 questionnaires in 72 countries between 1967 and 1973 (Hofstede, n.d.; Hofstede & Hofstede, 2008). The original data derives from employees working at national subsidiaries of the multinational IBM.

Hofstede (n.d.) says: “Culture is the collective programming of the mind distinguishing the members of one group or category of people from others”. He originally identified four national culture dimensions, which allow country comparisons. The dimensions are: Power Distance (PDI), Individualism/Collectivism (IDV), Masculinity/Femininity (MAS) and Uncertainty Avoidance (UAI). In a later study Hofstede added the fifth dimension: Long-Term versus Short-Long-Term Orientation (LTO). Recently, Hofstede, Hofstede, Minkov, & Vinken (2008) found two further dimensions; namely Indulgence versus Restraint (IVR) and Monumentalism (MON). In the following, the seven dimensions are explained briefly.

PDI. Power Distance is defined as the “extent to which less powerful members of a society

accept the fact that power is distributed unequally” (De Mooij & Hofstede, 2002:63). In societies that score high on PDI there is high respect for old age; moreover, status is important and has to be shown. Status is less important in countries that score low on PDI (De Mooij & Hofstede, 2002; Hofstede, n.d.; Hofstede & Hofstede, 2008).

IND. In individualistic cultures people look after themselves and only their immediate family.

In collectivistic societies people see themselves as a part of a group that looks after themselves in exchange for loyalty (De Mooij & Hofstede, 2002; Hofstede & Hofstede, 2008). The identity of people from collectivistic cultures is based on the social network to which they belong (De Mooij & Hofstede, 2002; Hofstede & Hofstede, 2008).

MAS. Achievement and success are the dominant values in masculine societies. In feminine

cultures caring for others and the quality of life are central. There is less role differentiation in feminine cultures (De Mooij & Hofstede, 2002; Hofstede & Hofstede, 2008).

UAI. Uncertainty Avoidance is “the extent to which people feel threatened by uncertainty and

ambiguity and try to avoid them” (De Mooij & Hofstede, 2002:64). In countries that score high on UAI people need rules and formality. In countries that score low on UAI people tend to be more innovative and have a sense of entrepreneurship (De Mooij & Hofstede, 2002; Hofstede & Hofstede, 2008).

LTO. Long-Term Orientation refers to future-oriented perspectives, acceptance of change,

perseverance and thrift. Short-Term oriented countries are rather conventional historic (De Mooij & Hofstede, 2002).

IVR. Indulgence versus Restraint stands for a society which allows relatively free

gratification of certain desires and feelings; in particular in the field of leisure, merrymaking with friends, spending, consumption and sex (Hofstede et al., 2008).

MON. Monumentalism refers to a society where people are rewarded, proud and

unchangeable (Hofstede et al., 2008).

Overall, those “dimensions relate to country differences in the motives for buying products and services, the degree of dependence on brands, adoption of new technology, and media use” (De Mooij & Hofstede, 2002:63). For example, De Mooij & Hofstede (2002) argue that people from countries that score high on LTO are less receptive to e-commerce. Those people are sparing with resources and prefer to go to stores to pick up merchandise rather than having products delivered at home. De Mooij & Hofstede (2002) also suggest that an early adoption of the Internet is weak in cultures that score low on UAI, as the Internet is an unstructured means of communication.

Cultural Differences between Germany and the Netherlands

“Culture only exists by comparison” (Hofstede, n.d.). A comparison between two European countries in terms of privacy disclosure behaviour on SNSs can be of interest, as large variation in the usage of the Internet in Europe is assumed despite economical homogeneity (De Mooij and Hofstede, 2002). Companies need to mind the (un-)availability of customer’s information when expanding overseas; information that is necessary for market research and targeting e.g. through SNSs.

Germany and the Netherlands are interesting to research as they are geographically and economically close, but display (moderate) cultural distance. Germany and the Netherlands have the second strongest trade relationship worldwide; after Canada and the USA (DNHK, 2014a). Because of political and economic collaborations, Germany is often chosen as an expansion market by Dutch companies and vice versa (DNHK, 2014a). However, the German-Dutch Chamber of Commerce points out that many businesses underestimate the cultural distance between the two neighbouring countries when entering the respective market (DNHK, 2014b). The Chamber refers to differences in communication styles; Germans are rather formal und restrained, whereas Dutch are rather informal and tend to provide (private) information (DNHK, 2014b).

Table 1 (p. 17) presents the Hofstede scores of Germany, the Netherlands and the world average. Kogut & Singh (1988) use the Hofstede dimensions for their cultural distance index; indicating the extent of cultural (dis-)similarity. Using their index, based on the available country scores of 65 countries (Hofstede, 2010), Germany and the Netherlands score 1.61 on cultural distance. Comparatively, despite larger geographical distances, the Netherlands is culturally closer to e.g. Sweden (0.49), Norway (0.93), and Denmark (0.94). This is because of similar cultural origins (sea power) (Hofstede & Hofstede, 2008). Within the EU, Germany is culturally close to Luxembourg (0.31), Italy (0.35) and the Czech Republic (0.38)

(continental powers) (Hofstede & Hofstede, 2008).

According to Hofstede & Hofstede (2008) and Inglehart (2012) masculinity vs. femininity is of particular interest in cross-cultural studies, as this is the only dimension that does not correlate with national income / wealth. Germany and the Netherlands are economically homogenous and different in particular on the MAS dimension.

The World Value Survey by Inglehart (2012) indicates similar cultural differences between Germany and the Netherlands. Germany scores 1.31 on Traditional vs. Secular-rational values and 0.74 on Survival vs. Self-expression values, whereas the Netherlands score 0.71 on Traditional vs. Secular-rational values and 1.39 on Survival vs. Self-expression values (Inglehart, 2012). The dimension Traditional vs. Secular-rational values is mainly about the importance of religion. Societies near the traditional pole value family ties and authorities (comparable to high MAS) (Inglehart, 2012). Germany scores higher on MAS and on traditional values than the Netherlands. Survival vs. Self-expression values is about security and self-realization. Self-expression is about increasing emphasis on subjective well-being and the quality of life (comparable to low MAS) (Inglehart, 2012). The Netherlands score lower on MAS and higher on self-expression than Germany. According to Inglehart (2012), these two dimensions of the World Value Survey explain about 70% of cross-cultural variance.

The index by Roose (20012) also makes cultural differences between Germany and the Netherlands apparent. Roose (2012) proposes a new index to identify cultural similarity in Europe. The index is based on the results of the well-accepted European Social Survey. It shows the relationship between two countries based on answers to value issues in question. Germany and the Netherlands rank 124 out of 325 ‘country couples’ on cultural similarity in Europe (index 0.69). The most similar countries are Great Britain and Ireland (Roose, 2012).

Comparability. In the present study Germany and the Netherlands are in direct comparison;

focusing on the Netherlands as a whole and North Rhine-Westphalia (NRW). NRW is the most populous and fourth biggest state in Germany, and is seen as being representative for Germany from demographic and economical perspectives (Staatskanzlei des Landes Nordrhein-Westfalen, 2013b). The NRW government stresses the similarities with the Netherlands (table 2) in terms of population, area and gross domestic product (GDP) (Staatskanzlei des Landes Nordrhein-Westfalen, 2013a). The NRW government as well as the Dutch Rijksoverheid (n.d.) stress the importance of political, economic, social and cultural relations (Rijksoverheid, n.d.; Staatskanzlei des Landes Nordrhein-Westfalen, 2013b).

Table 1. Hofstede Dimensions

GER NL World Average PDI 35 38 55 LTO 31 44 44 INV 67 80 54 MAS 66 14 53 UAI 65 53 58

Note: From Hofstede n.d.; derived from Veltri & Elgarah (2009)

Table 2. Demographic Comparison NRW and the Netherlands

North Rhine-Westphalia 2011 The Netherlands 2012

GDP 582.1 billion (2012) 599.3 billion Population Men Women 17,538,251 8,521,230 9,017,020 16,730,348 8,282,871 8,371,513 # Of Foreigners Migration Background 1,607,080 4,215,000 N/A 3,494,193

Note: From Centraal Bureau voor de Statistiek, 2013a; Centraal Bureau voor de Statistiek, 2013c; NRW.Invest.Germany, 2013; Zensusdatenbank Zensus 2011 der Statistischen Ämter des Bundes und der Länder, 2011

Self-Disclosure

According to Lowry et al. (2011) self-disclosure occurs during interpersonal communication. Self-disclosure refers to personal information an individual reveals intentionally and voluntary to others about himself/herself. This is in line with an ancient definition by Worthy, Gray & Kahn (1969:59), who say: “Self-disclosure may be defined as that which occurs when A knowingly communicates to B information about A which is not generally known and is not otherwise available to B.” Both definitions point out the awareness of information disclosure.

Govani & Pashley (2005) conducted a research among US students in which they surveyed Facebook users about their provided information and privacy settings. The disclosed information they identified is illustrated in figure 1.

Next to written content, users also share videos and photographs. Facebook (2013) reveals that users upload more than 350 million photos daily. In total over 240 billion photographs have been shared. In August 2012, Facebook acquired Instagram, a photo-sharing service, which can be connected to a users’ Facebook account. There are 100 million registered users on Instagram who share pictures (Facebook, 2013).

Chen (2012) states that to date there are little empirical findings available that show why users expose or withhold information on a public network. Through survey data among undergraduate students in the US, Chen (2012) found that SNS members’ attitudes promote the extent of their self-disclosure. The members’ attitude depends on (1) personality traits (e.g. extroversion), (2) service attributes (e.g. perceived critical mass), and (3) the external environment (e.g. perceived Internet risk) (Chen, 2012).

Tow, Dell, & Venable (2010) also researched why users of SNSs are willing to disclose personal information. They argue that ‘context’ and ‘value’ influence user’s behaviour. ‘Context’ includes users’ objectives for using SNSs, previous experience and outside influences, such as friends, co-workers, and mass media. ‘Values’ refers to values that are relevant to a user’s personal sense of privacy (Tow et al., 2010). Chen (2012) points out that privacy values attenuate the relationship between attitude and self-disclosure. De Mooij & Hofstede (2002) argue that the value-system derives from culture. For example, it is argued that privacy is a typical issue in individualistic cultures (De Mooij & Hofstede, 2002).

Privacy

Pedersen (1999:397) defines privacy as a “boundary control process in which an individual regulates with whom contact will occur, and how much and what type of contact it will be”. Moreover, it is seen as “the claim of an individual to determine what information about himself or herself should be known to others” (Westin; 2003:431). Many customers feel that they have a right for privacy, also electronically, and may see online privacy as a basic human right (Goodwin, 1991).

Zhang et al. (2011) argue that online privacy is a multi-dimensional construct. The authors propose that SNS users are concerned about four aspects, namely: (1) virtual territorial

privacy, (2) factual privacy, (3) interactional privacy, and (4) psychological privacy (Zhang et al., 2011).

(1) Virtual territory privacy points to the fact that unlike in offline social contexts, there are no physical boundaries in the virtual social context. Zhang et al. (2011) state that virtual territory privacy refers to the control a user can have over the virtual private territory.

(2) Factual privacy is about the control a user can have about the access, collection, dissemination and use of personal information. (3) Interactional privacy refers to the control an individual user can have about the other users he/she wants to interact with. (4) Psychological privacy is about the intrusion upon a user’s thoughts, feelings and values. In the field of online transactions Goodwin (1991) stresses a consumer’s ability to control (1) the presence of other people in the environment during the transaction (comparable to virtual private territory privacy and interactional privacy) and (2) the dissemination of information and behaviour to those who were not present (comparable to factual privacy and psychological privacy).

Overall, Zhang et al. (2011) and Acquisti & Grossklags (2005) emphasise the fact that privacy is subjective based on perceptions, not a rational construct. Acquisti & Grossklags (2005) argue that not all individuals protect their privacy all the time, nor do they have the same strategies or motivations. Demographics, behavioural economic activities (risk and discounting attitudes), past behaviours, knowledge of and attitudes towards privacy explain privacy protective behaviour (Acquisti & Grossklags, 2005; Norberg & Horne, 2007). Through protective strategies users guard themselves against social privacy threats (privacy concerns) (Young & Quan-Haase, 2013).

Privacy Concerns

Privacy concerns can be defined as “an unwillingness to disclose personal information” (Culnan & Armstrong, 2000:20). Culnan & Armstrong (2000) see personal information as information that can identify an individual. Alike, Phelps, Nowak, & Ferrell (2000:28) argue that individual specific information comprises: names, addresses, demographic characteristics,

lifestyle interests, shopping preferences and purchase histories of identifiable persons. Li, Sarathy, & Xu (2011) state that not all information cause privacy concerns. Privacy concerns revolve primarily around individual specific data (Phelps et al., 2000).

Chellappa & Sin (2005) make a distinction between (1) anonymous information (e.g. IP

address; domain type; browser version, type, language; local time; operating system), which

is information gathered about a page visitor without any use of invasive technology; (2)

personally unidentifiable information (e.g. age, date of birth, gender, occupation, education,

income, ZIP code, interests and hobbies), which taken alone cannot identify or locate an

individual; and (3) personally identifiable information (email address, name, address,

phone/fax number, credit card details, social security number), which can be used to identify

an individual. In line with Phelps et al. (2000), Li et al. (2011) argue that consumers are less concerned about providing basic demographic information (e.g. gender, age, education, material status). Consumers are moderately protective of information concerning their purchasing behaviour, hobbies, occupation, name, and email address; and strongly protective of financial information and phone numbers (Li et al., 2011).

According to Campbell (1997), information privacy concerns are subjective views about

information privacy fairness (perception). Son & Kim (2008) argue that, amongst other,

perceived justice (distributive, procedural and interactional justice) predict users’ protective

and interactional justices are part of the IUIPC (Internet Users’ Information Privacy Concern)

(Malhotra, Kim, & Agarwal, 2004). (1) Distributive justice includes the collection of

individual specific data, which must be done without angering the customer (Malhotra et al.,

2004; Pitta, Franzak, & Laric, 2003; Smith, Milberg, & Burke, 1996); (2) procedural justice comprises a user’s control and freedom to voice an opinion or to exit (Malhotra et al., 2004);

and (3) interactional justice is about a user’s awareness and understanding of conditions,

actual practices and transparency of procedures (Malhotra et al., 2004). Other factors that

might influence perceived fairness are: (a) unauthorized secondary use, such as a shift in data

usage purposes e.g. by a third external party; (b) improper access of a persons’ data by other

individuals; (c) deliberate and accidental errors in the data set; (d) reduced judgement, which

is an individuals’ feeling to be just a ‘number’ alongside others; and (e) the combination of

databases e.g. within a growing organization (Smith et al., 1996).

As pointed out, perceived justice predicts a users’ (un-)willingness to disclose personal information. In that light, Ackerman, Cranor, & Reagle (1999) and Westin (2003) classified users into three categories based on their (un-)willingness to provide personal data: privacy fundamentalists, unconcerned and pragmatists. (1) The privacy fundamentalists (25% of the public) are extremely concerned about the use of the data and are unwilling to provide information even when privacy protection measurements are in place. (2) The privacy unconcerned (20% of the public) are willing to provide data under almost any condition. (3) Privacy pragmatists (55% of the public) are also concerned about their data, but are less concerned than the fundamentalist. If privacy settings are in place, they are more willing to provide data. The pragmatist tends to examine benefits and privacy risks. As they are willing to provide data under certain conditions, the privacy pragmatists are of special interests for researchers, academics and organizations (Ackerman et al., 1999; Westin, 2003).

Privacy-Paradox

The privacy-paradox is an example of an attitude-behaviour incongruity (Norberg & Horne, 2007). On the one hand, consumers are becoming sensitive to safeguarding their privacy needs (Pitta et al., 2003). On the other hand, SNS users often state that they are concerned about their privacy but yet disclose private information on their profiles (Utz & Krämer, 2009). This discrepancy in privacy concerns and self-disclosure describes the so-called privacy paradox, which Young & Quan-Haase (2013:479) define as a “people’s willingness to disclose personal information on social network sites despite expressing high levels of concern”. Norberg et al. (2007) found evidence for the privacy paradox and argue that despite privacy complaints, consumers freely provide personal data.

Incidents, such as consumer fraud and identity theft, sensitized consumers to protect their privacy and to be more discrete (Pitta et al., 2003). Therefore, privacy concerns are seen as the major determinant of the so-called information privacy-protective responses (IPPR) (Son & Kim, 2008). Son & Kim (2008:505) state that the “main response for protection of information privacy is to refuse to disclose personal information”. However, Larose & Rifon (2007:129) argue that “although online privacy is undeniably a ‘‘concern,’’ one might well ask if it really influences a consumer’s privacy protection behavior”. According to Utz & Krämer (2009) strong privacy concerns result in restrictive profiles on the one side and impression management results in less restrictive privacy settings on the other side. Utz & Krämer (2009) argue that the privacy-paradox might arise, as users have to find a trade-off between those two opposing needs (privacy protection and impression management). SNS users care about their privacy and (constantly) adjust their privacy settings (Utz & Krämer, 2009). Trading-off costs and benefits are the major aspects of the privacy calculus.

Privacy Calculus

The privacy calculus is defined as a “summation of personal beliefs regarding the expected positive and negative outcomes of information disclosure” (Li, 2012:475). The privacy calculus describes the risk-return exchange of self-disclosure (Xu, Michael, & Chen, 2013). Privacy risk refers to the misuse of personal information as a consequence of opportunistic behaviour and self-disclosure (Xu et al., 2013). Users aim to achieve an equilibrium point between perceived costs and benefits of information disclosure; this is part of the utility maximization theory (Acquisti & Grossklags, 2005; Sheehan & Hoy, 2000). The tent of the utility maximization theory is “to maximize the total utility or satisfaction by a person” (Li, 2012:475). In that light, online users often forget about their privacy concerns (Berendt, Günther, & Spiekermann, 2005; Young & Quan-Haase, 2013), but share personal details given the right circumstances e.g. when the online exchange is entertaining and benefits are offered in return (Berendt et al., 2005). White, Zahay, Thorbjørnsen, & Shavitt (2008) researched personalized email solicitations and they found that consumers were more likely to accept those emails when the perceived utility (“benefits”) offset the psychological “costs” e.g. receiving inappropriate personal messages. Consumers favour personalized communication when they perceive a high fit between their personal characteristics and the personalized offer (White et al., 2008). Consequently, they are more willing to disclose information when they perceive the return to be beneficial. If the arrangement is consensual, there is no privacy issue. The situation must be win-win for the consumer and the marketer (Pitta et al., 2003).

Acquisti (2004) and Acquisti & Grossklags (2005) however argue that an individual might not be able to estimate all costs and benefits at all times. This might be because of (1) the incomplete information available to users, (2) the bounded rationality to process all available information as well as (3) the deviation from utility-maximisation. The later reason

emphasises that even if users had all information and could process the information thoroughly, users may still behave against their better judgements. Also De Mooij & Hofstede (2002), who mainly researched consumer purchase decisions, argue that consumers often do not make rational decisions that maximize utility. In that light, considering costs and benefits influence individuals’ intentions to disclose, however those considerations may not be strong enough to influence actual behaviour (derived from Norberg, Horne, & Horne, 2007). According to the theory of reasoned action (TRA), the stronger a person’s intention to perform (e.g. intention to take on protective strategies), the greater the likelihood of actual performance (e.g. using protection measurements) (Armitage & Christian, 2003; Son & Kim, 2008). Intentions, however, are determined by (1) the attitude towards the behaviour, which is the result of a favourable or unfavourable evaluation of the behaviour (e.g. costs and benefits); and (2) perceived social pressure to (not) perform the behaviour (Ajzen & Madden, 1986). E.g. Tow et al. (2010) argue that previous experience and outside influencers, as friends and co-workers, influence disclosure behaviour on SNSs. Govani & Pashley (2005) continue by saying that users who made an informed decision to join Facebook might see the benefits of the information sharing and outweigh the costs of the loss in privacy, but on the contrary, even though they might be informed about the benefits and risks, they could still be influenced to join Facebook because of peer pressure (Govani & Pashley, 2005). This refers to the social exchange theory, which says that there are greater interpersonal interactions if greater rewards are expected (Worthy et al., 1969). This indicates that persons, who get much from others, are under pressure to give much back. This process works out at an equilibrium and balances the social exchange (Homans, 1958). Also Xu et al. (2013) state that people wish to be accepted and to be valued by their social networks. Connecting with others is a form of social capital. To increase and manifest the social capital, users are ‘forced’ to provide authentic personal information (Xu et al., 2013). In that light, Young & Quan-Haase

(2013) argue that the falsification of information is often not chosen as a protective strategy, as friends and peers could question the validity of the disclosed information.

Concept Development and Hypotheses

Privacy Concerns and Self-disclosure. Self-disclosure is defined as disclosing information

intentionally and voluntarily to others (Lowry et al., 2011; Young & Quan-Haase, 2013). Privacy protection measurements facilitate the refusal of personal information disclosure. Users who show privacy concerns tend to take on those strategies to protect their information (Malhotra et al. 2004; Smith et al., 1996; Young & Quan-Haase, 2013). Utz & Krämer (2009) found that the higher the privacy concerns, the stricter the privacy settings; meaning the less self-disclosure. As Wu et al. (2012:890) point it: “Online privacy concern leads to a lack of willingness to provide personal information online, rejection of e-commerce, or even unwillingness to use the Internet”. Therefore,

H1: Privacy concerns are negatively related to self-disclosure

Culture. Milberg, Smith, & Burke (2000) found significant differences in cultural-values in

relation to information privacy concerns. Lowry et al. (2010) go further and state that cross-cultural dimensions are not just predictors of information privacy concerns but also of the desire for online awareness.

Recently, especially IND and UAI caught the attention of researchers that examined cultural differences in privacy disclosure behaviour (e.g. Kaya & Weber, 2003; Kransnova, Veltri & Günther, 2012; Spiliotopoulous & Oakley, 2013; Veltri & Elgarah, 2009). According to Acquisti (2004) and Acquisti & Grossklags (2005) LTO is of particular interest in future research about e-commerce and Internet consumption. People who score low on LTO usually trade-off costs for short-term benefits and vice versa (Hofstede, n.d; Hofstede & Hofstede,

2008). For example, the privacy calculus is about outweighing those costs and benefits.

As far as could be researched, there is no single model available explaining online privacy disclosure behaviour including all five Hofstede dimensions at the same time; this can be attributed to the complex and multi-level nature of culture as a construct (Yoo et al., 2011). Lowry et al. (2011) conducted their research among instant messenger (IM) users from China and the US. They researched the four initial cultural dimensions by Hofstede. They used PDI, IDV, MAS and UAI as independent variables, which in their model were hypothesized to influence information privacy concerns and the desire for online awareness (self-disclosure). They suggested that information privacy concerns and the desire for online awareness influence the attitude towards instant messaging. They argued that the behavioural intentions to use instant messengers mediate the relationship between attitudes towards the IM and the usage of the IM (referring back to the TRA). However, the authors could not support their complex model. They found neither significant influence of MAS on information privacy concerns, nor significant influence of MAS and PDI on the desire for online awareness.

The study by Lowry et al (2011) makes the complexity of Hofstede’s dimensions in relation to online privacy disclosure behaviour apparent. For simplification, in the following there are five separate models proposed.

Overall, the literature suggests that culture might influence privacy concerns, and that those privacy concerns might influence self-disclosure. Therefore, a direct and an indirect effect of culture on self-disclosure are expected in the present study. In conceptual model 1 and 2 privacy concerns are expected to mediate the relationship between self-disclosure and PDI/LTO; as PDI/LTO are expected to influence privacy concerns, and privacy concerns are expected to influence self-disclosure.

between privacy concerns and self-disclosure respectively. Individuals that score high on INV/MAS/UAI trade-off opposing motives, namely protecting online privacy and impression management (INV), obtaining valuable information to stay competitive (MAS), and providing information in response to peer pressure (UAI). The relationship between privacy concerns and self-disclosure is expected to be influenced by the degree of INV/MAS/UAI.

As only limited and discordant results could be found on PDI, MAS and LTO in relation to online privacy disclosure behaviour, the theoretical background for those three dimensions is less profound than for INV and UAI. Moreover, no literature was found on IVR and MON referring to differences in Internet consumption. The scores for Germany and the Netherlands are not available for country comparison either (Hofstede, n.d.). For those reasons, those dimensions are not incorporated in the present study.

Conceptual Model 1. It can be assumed that PDI does not provide evidence for the

privacy-paradox. Lowry et al. (2011) suggest that people that score high on PDI value control and authority. They are accustomed to the fact that authorities might access their personal information (Lowry et al., 2011). Individuals in highly regulated societies have fewer concerns about distributed information (H2b) (Lowry et al., 2011). There is less privacy needed and information is widely shared with a broad social network (Kaya & Weber, 2003; Hofstede & Hofstede, 2008).

Lowry et al. (2011) argue that awareness through self-disclosure technology adds transparency, structure and control to the interpersonal communication process, as without those technologies it is not possible to know what others are doing. Transparency eases information disclosure (Malhotra et al. 2004; Lowry et al., 2011). Moreover, interpersonal information exchange and status are critical in societies that score high on PDI (De Mooij & Hofstede, 2002). To be able to show status and to exchange valuable information, individuals have to disclose information (H2a).

Therefore,

H2a: Power Distance is positively related to self-disclosure H2b: Power Distance is negatively related to privacy concerns

Power Distance is hypothesized to influence privacy concerns (H2b), and privacy concerns are assumed to influence self-disclosure (H1). Following, privacy concerns are expected to mediate the relationship between PDI and self-disclosure (H2c).

H2c: Privacy concerns mediate the relationship between Power Distance and self-disclosure

Figure 2. Conceptual Model 1 – Power Distance

Conceptual Model 2. Chellappa & Sin (2005) argue that privacy is not an absolute concept,

as consumers may give up some privacy for perceived benefits. In general, short-term oriented people seek for quick results and short-term benefits (Hofstede & Hofstede, 2008). This translates into less privacy concerns as they and more self-disclosure. Less privacy concerns (H3b), as short-term oriented people usually ignore possible future consequences, such as e.g. identity theft (Hofstede & Hofstede, 2008); and more self-disclosure (H3a), as short-term oriented people try to achieve short-term goals (Hofstede & Hofstede, 2008), such as e.g. better-targeted advertisements. Overall, costs involved in online transactions are monetary (adoption costs, usage costs of protective technologies or identity theft) or immaterial (learning costs of a protective technology or switching costs between applications)

(Acquisti, 2004). The benefits in the field of e-commerce are monetary (e.g. discounts) or intangible (feeling of protection while sending encrypted emails) (Acquisti, 2004).

Opposing to short-term oriented people, people that score high on LTO are less receptive to e-commerce (De Mooij & Hofstede, 2002). They tend to be more resistant and might not outweigh costs for benefits. This might involve more privacy concerns (H3b) and less self-disclosure (H3a). More privacy concerns, as they think forward and mind consequences; and less self-disclosure, as they do not see short-term benefits.

Thus,

H3a: Long-Term Orientation is negatively related to self-disclosure H3b: Long-Term Orientation is positively related to privacy concerns

Long-Term-Orientation is hypothesized to impact privacy concerns (H3b). Alike in conceptual model 1, privacy concerns are assumed to influence self-disclosure (H1). Following, privacy concerns are expected to mediate the relationship between LTO and self-disclosure (H3c).

H3c: Privacy concerns mediate the relationship between Long-Term-Orientation and self-disclosure

Figure 3. Conceptual Model 2 – Long-Term-Orientation

Conceptual Model 3. Kaya & Weber (2003) researched American and Turkish students on

and Turkish students differed significantly in their desired privacy levels. Turkish students had lower needs for privacy. Kaya & Weber (2003) explain this through the strong inter- and intra-family ties and social dense environments in non-Western countries. This is in line with De Mooij & Hofstede (2002), who argue that privacy is a typical issue in individualistic cultures. People from individualistic countries tend to be in their private circle, whereas people from collectivistic countries (mostly non-Western) tend to spend more time with all of their group members. Supporting, the USA (91/100) score higher on INV than Turkey (37/100) (Hofstede, 2010).

Bellman, Johnson, Kobrin, & Lohse (2004) also argue that individualistic individuals have a lower acceptance of groups intruding on his/her personal life. Milberg, Burke, Smith, & Kallman (1998) continue by saying that people that score high on INV also have less acceptance of organizations to invade one’s private sphere.

Lowry et al. (2011) argue that individualistic individuals might desire to protect their privacy / personal information to achieve own goals. On the contrary however, more self-disclosure might help to assert themselves better; saying that INV will strengthen the desire for information exchange for own benefits (Lowry et al., 2011). Elgarah & Veltri (2009) stress that individualistic users are more likely to express themselves openly by providing details about their personality to stand out of the crowd. Also Krasnova et al. (2012) argue that users from cultures that score higher on INV tend to share more private information than users from collectivistic countries.

Besides, the desire for pleasure is common in societies with high INV (Krasnova et al., 2012). According to Berendt et al. (2005) and Young & Quan-Haase (2013) SNS users often forget their privacy concerns but share information given the right circumstances, e.g. when the online exchange is entertaining. There might be a trade-off between privacy concerns and impression management (showing oneself to others) and entertainment (Utz & Krämer, 2009). The degree of INV is expected to influence the relationship between privacy concerns and

self-disclosure. Usually individualistic individuals are concerned about their privacy, but still share information given the right circumstances. Generally,

H4: Individualism moderates the relationship between privacy concerns and self-disclosure

Figure 4. Conceptual Model 3 – Individualism

Conceptual Model 4. According to Lowry et al. (2011) the literature on the effect of MAS on

information privacy concerns is mixed. Moreover, according to the authors, there are no reasons revealed in the literature about why people that score low or high on MAS might be more or less concerned than the counterpart.

In the present study, it is assumed that people who score high on MAS tend to be more concerned than people from feminine countries. As already referred to, Kaya & Weber (2003) researched American and Turkish students on their desired levels of privacy. They found significant differences and attributed those to differences on the INV dimension. Interestingly, America and Turkey also differ on the MAS dimension. According to Hofstede (2010), America scores relatively high (62/100), and Turkey relatively low (45/100) on MAS; comparing to the world average (50/100) (Hofstede, 2010; Veltri & Elgarah, 2009). Supporting, IBM (1999) researched privacy concerns among American, British and German Internet users and found respondents to be either moderately or highly concerned about their online privacy (79%). According to Hofstede (2010), those three countries score relatively high on MAS (USA 62/100; GB 66/100; GER 66/100). It has to be acknowledged that in

IBM’s (1999) study there are no numbers revealed about countries that score low on MAS.

People that score high on MAS place greater emphasis on material success, achievement, and benefits of using private information than on caring for relationships and the quality of life (Bellman et al., 2004; Lowry et al., 2011; Milberg et al., 2000). In that light, people that score high on MAS might want to protect their privacy to achieve own goals. But, as Lowry et al. (2011:174) also state: “to achieve work goals, highly masculine individuals may understand the need to forego a certain amount of privacy […]”. The authors continue by saying that individuals in high masculine cultures might appreciate interpersonal awareness (self-disclosure) to assert themselves better. In assertive societies, acceptance and social capital are precious (Xu et al., 2013). In less competitive and less assertive cultures, information might be less valuable. Similar to high INV, individuals that score high on MAS are expected to be concerned about their privacy, but provide information if the risk-return exchange is perceived to be beneficial (Sheehan & Hoy, 2000). It can be hypothesized that,

H5: Masculinity moderates the relationship between privacy concerns and self-disclosure

Figure 5. Conceptual Model 4 – Masculinity

Conceptual Model 5. People that score high on UAI require formalities and rules; the Internet

however is an unstructured means of communication (De Mooij & Hofstede, 2002). De Mooij (2000) also argues that an early adoption of the Internet would be weak in UAI cultures.

Grabner-Kräuter & Kaluscha (2003) argue that online-transaction and exchange relationships are characterized by uncertainty, anonymity and lack of control. Krasnova et al. (2012) state that cultures with high UAI are known for averseness for risks, pessimism and concern in terms of actions and consequences. This is in line with Milberg et al. (1998) who argue that a high UAI score can be associated with high levels of anxiety, stress, and concern for security. Following, it can be expected that people that score high on UAI are concerned about their online privacy.

Possibly, users that score high on UAI disclose information in response to peers and other members; by copying the behaviour of others, they can avoid uncertainty (Lowry et al., 2011). Krasnova et al. (2012) argue that users from countries that score high on UAI tend to ignore their privacy concerns. Lowry et al. (2011) also suggest that individuals from countries with high UAI have more privacy concerns while the desire for online awareness is higher, saying that users who score high on UAI disclose information despite privacy concerns.

Following,

H6: Uncertainty Avoidance moderates the relationship between privacy concerns and self-disclosure

Figure 6. Conceptual Model 5 – Uncertainty Avoidance

Control variables. Culture might not be the only factor that influences privacy concerns

Armstrong, 2000; Hann, Hui, Lee & Png, 20002), knowledge (Hann et al., 2002; Liao, Liu, & Chen, 2011) and demographic characteristics (age, education, gender) (Berendt et al., 2005; Gilbert, Karahalios, & Sandvig, 2008; Kaya & Weber, 2003; Nosko, Wood, & Molema, 2010; Thelwall, 2008) on privacy concerns and self-disclosure. Therefore those variables should be controlled for, when examining variance in privacy concern and/or self-disclosure. Further, contrary to most previous researches, the present research is conducted in an online and offline setting; therefore it is controlled whether online and offline respondents differ in their online privacy disclosure behaviour. In the following, literature on the control variables is presented.

Trust. “Trust, in general, is an important factor in many social interactions, involving

uncertainty and dependency. On-line transactions and exchange relationships are not only characterized by uncertainty, but also by anonymity, lack of control and potential opportunism, making risk and trust crucial elements of electronic commerce” (Grabner-Kräuter & Kaluscha, 2003:784). Information about basic functioning and security of the e-commerce system is important (Grabner-Kräuter & Kaluscha, 2003). Also Culnan & Armstrong (2000) argue that customers are more willing to disclose personal information when they believe that fair procedures are in place to protect their privacy. Hoy & Milne (2010) asked their respondents whether they believe that SNSs do a good job in protecting their privacy. The authors also identified to what extent users provide false information. They distinguish between providing false information to set up the profile and false information on the actual profile. For example, Young & Quan-Haase (2013) found greater concern about social privacy than about institutional privacy. Hoy & Milne (2010) label this construct ‘beliefs’. Li et al. (2011) argue that privacy protection beliefs (consumer beliefs that a vendor will protect personal information) act as benefit factors in the privacy calculus.

Knowledge. Liao et al. (2011) found that people are less concerned when Internet literacy

goes up; possibly owing to users’ familiarity with the available control mechanisms. Park, Campbell & Kwak (2012) argue that cognition (knowledge) and emotions are central in decision-making and social behaviour. Emotions guide people to resort to anger, fear, happiness and trust (Park et al., 2012). Cognition helps to modify thinking and rationality, and to control behaviour (Park et al., 2012). The authors examined the interplay between a user’ privacy knowledge (cognition) and a user’s level of concern (affect) in the Internet usage for the control of privacy. Their empirical results indicate that privacy concerns do not directly influence users’ protective behaviours. Knowledge is revealed to be a significant moderator of concern. This might be in line with Tow et al. (2010), as they identified that some users are not aware of privacy issues or they perceive risks (e.g. identity theft) as very low, possibly because of lack of knowledge.

Hoy & Milne (2010) developed the construct ‘awareness’ to test whether respondents think that their online behaviour is monitored and used for targeting purposes; basically showing their familiarity and knowledge with/about the SNS. Awareness is about the understanding of conditions, actual practices and transparency of the procedures (Malhotra et al., 2004). To test knowledge, privacy policy readership is also used as an additional control variable.

Demographics. The amount of personal information decreases when age increases (Nosko et

al., 2010). This suggests that younger SNSs users share more private information. Moreover, Nosko et al. (2010) propose that older people might be less willing to share information, as it might be seen as inappropriate (e.g. women and age in Western societies). Zhang et al. (2011) argue that older social network users are less concerned about information sharing. Moreover, the authors argue that users who have more online friends tend to have stricter privacy settings than people who have fewer friends. Female and younger users have more friends than other members (Thelwall, 2008). This is in line with McAndrew & Joeng (2012), who

argue that young females spend more time on Facebook, have more contacts and are more likely to use profile photographs for impression management.

Nosko et al. (2010) conducted three independent studies in which they found no significant difference between the information disclosure of males and females. Also Pedersen (1999) and Kransnova et al. (2012) found no differences between sexes in information disclosure. According to McAndrew & Joeng (2012) and the Pew Research Center (2013), however, females are more active on Facebook than man. Females are also more likely to maintain private profiles (Thelwall, 2008). Gilbert et al. (2008) agree by saying that in particular women tend to have higher privacy settings. Hoy & Milne (2010) also found that women are more concerned about their online privacy than men. Kaya & Weber (2003) on the contrary argue that, regardless of culture, males report a greater desire for privacy than females (Kaya & Weber, 2003).

It is further controlled for education, nationality and place of living, as many researches to date were conducted among students in the United States (e.g. Acquisti & Grossklags, 2005; Chen, 2012, Thelwall, 2008); limiting the research to younger well-educated US-American SNS users.

Online/Offline. Previous researchers claim that online surveys usually limit the

generalizability of the results (e.g. Pedersen, 1999; Zhao & Jiang, 2011). Some researchers simply had a look at Facebook profiles (e.g. Zhao & Jiang, 2011) and reported that they stayed within the boundaries of their own network, as not all users share their information with people outside their own network. Moreover, it can be argued that researchers would have to rely on the information provided by the users. However, some users might unintentionally share information. Self-disclosure on the contrary is about the intentional und voluntary disclosure of personal information to others.

Methodology

The sample consists of 278 respondents in total, of which 29.1% offline and 70.9% online respondents. 76% of all respondents completed the questionnaire. The completeness of one questionnaire is 91%. The offline survey response rate is 48.2%. 52.9% of the respondents live in Germany and 25.2% live in the Netherlands; 21.9% live outside Germany and the Netherlands. 52.9% of the sample has been German since their birth, 2.5% have become German, and 1.3% of the sample has more than one citizenship under which one is German. 20.6% of the sample has been Dutch since their birth, 0.4% has become Dutch, and 2.1% has more than one citizenship under which one is Dutch. In this study a German or Dutch person is considered as someone who has a German or Dutch citizenship respectively; this includes everyone who received the respective citizenship over time.

20.2% of the respondents are neither German nor Dutch. In total there are 32 nationalities represented in the present study.

Overall, 80.2% of the respondents have a Facebook account. The Facebook sample is presented in table 3 (p.38). If respondents did not have a Facebook account they were asked to skip the SNS-related questions, and to continue with the questions necessary to identify their cultural profiles. The Non-Facebook user sample is illustrated in table 4 (p.38).

The respondents who are active on Facebook indicated moderate privacy concerns (M = 3.32, SD = .67) and moderate self-disclosure (M = 3.17, SD = 1.05).

Table 3. Sample Demographics Facebook Sample Gender Men 48.60% 39.6% Women 51.40% 60.4% Age < 18 3.26% 3.1% 18-29 19.07% 63.9% 30-49 29.53% 24.2% 50-64 26.86% 6.7% 65+ 21.28% 2.1%

Education Less than high school / high school grad 30.49% 15.5%

Some college 29.21% 29.9%

College + 40.30% 54.6%

Note: Adapted from Pew Research Center, 2013:8

Table 4. Sample Demographics of non-Facebook users

Gender Men 48.9% Women 51.1% Age < 18 0% 18-29 13.3% 30-49 37.8% 50-64 46.7% 65+ 2.2%

Education Less than high school / high school grad 13.3%

Some college 42.2%

College + 44.5%

Note: Adapted from Pew Research Center, 2013:8

Survey Design

The questionnaire is based on the constructs by Hoy & Milne (2010) and Hofstede (1994) (Appendix A). All Hoy & Milne (2010) questions are recoded to match the Hofstede scales. Native German and Dutch speakers have translated the questions by Hoy & Milne (2010) into German and Dutch respectively (Appendix A2 and A3). Both translators have an advanced English proficiency. One additional German and one additional Dutch native speaker have proofread the respective translated questionnaire. The Hofstede Value Survey (VSM 1994) questionnaire is available in different languages. The author approved the German, Dutch and English questionnaire (Hofstede, 1994). The authors of the VSM 2008, which is a further development of the VSM 1994, do not approve all translations (Hofstede et al., 2008).

Because of translation discrepancies of the VSM 2008, the VSM 1994 is used in the present study.

Measurements

Dependent variable. Self-disclosure is measured by testing to what extent Facebook users

make use of the available self-disclosure options e.g. full name, address, languages, friends, events etc. This is leaning towards the study by Govani & Pashley (2005), who identified what users share with whom using self-disclosure options to date. Additionally, Hoy & Milne (2010) developed a construct to measure privacy protection behaviour controlled by self. Privacy protection measurements facilitate the refusal of personal information disclosure (Malhotra et al. 2004; Smith et al., 1996; Young & Quan-Haase, 2013). Self-disclosure is about revealing personal information intentionally and voluntary to others. Therefore, disclosure is measured through controlled privacy protection behaviour and self-disclosure options. The self-controlled privacy protection behaviour consists of 13 items, which are measured on a five-point Likert scale. 18 self-disclosure options are tested through a six-point Likert scale ranging from “I don’t share this information” to ‘available for the “public”’ (Appendix A).

In conceptual model 1 and 2 (p.29/30) the mediator is privacy concerns and the independent

variables are PDI and LTO respectively. In conceptual model 3 (p.32), 4 (p.33) and 5 (p.34),

privacy concerns are the independent variable and IND, MAS, and UAI are the respective

moderators.

Dimensions. The cultural dimensions are tested through the Value Survey (VSM 1994) used

by Hofstede. Lowry et al. (2011) argue that the most adopted cross-cultural dimensions come from Hofstede’s studies. The VSM (1994) was developed for comparing culturally

determined values from respondents from two or more countries (Hofstede, 1994). Through the questions it is possible to calculate the index scores of each of the dimensions. The questions are measured on a five-point Likert scale. The index scores are calculated based on the available formulas (Appendix A).

Hofstede (1994) argues that when the Value Survey is used to compare responses at individual level, the answers should be examined question by question. The dimensions cannot be found at individual level but at country level. This is because answers depend on other respondent’s characteristics, such as gender, age, level of education, occupation, or employer.

Steel & Taras (2010) also argue that many scholars recognized substantial differences in cultural values within countries among different demographic groups. In the present study, actual individual continuous scores on the dimensions are not used, because of Hofstede’s proposed limitations in terms of level of analysis. For comparisons at individual level, in the present study, respondents are put into categories (alike MacCallum, Zhang, Preacher, & Rucker, 2002), namely low or high on PDI, LTO, INV, MAS and UAI. Respondents are seen as low on either PDI, LTO, INV, MAS and UAI if they score one-standard-deviation below the respective mean and vice versa (Aiken & West, 1991); if respondents score one-standard-deviation above the respective mean, they are seen as scoring high on PDI, LTO, INV, MAS and UAI. All others are seen as scoring moderately on the dimensions.

Privacy Concerns. Privacy concerns are measured through the construct used by Hoy &

Milne (2010), who examined gender differences regarding online privacy and the personal data usage on Facebook. Respondents indicate on a scale from 1 to 5 (‘very concerned’ to ‘very unconcerned’) to what extent they are concerned about the privacy of the information about themselves that is posted on Facebook. The scale is re-recoded leading to a 1 to 5 scale